DISCUSSION AND CONCLUSION

In high sedimentation rate areas of the continental margins that typically host submarine gas hydrates, burial rates generally exceed diffusion rates, suppressing effective diffusional solute transport (Gieskes, 1975; Hesse, 1990). The hydrochemical and isotopic data obtained during Leg 164 from the Blake Ridge submarine gas hydrate field, however, provide firm evidence for long-range downward chloride diffusion at Site 997. Diffusion is driven by an average chloride gradient (downward decrease) of about 0.1 mM Cl^-/m across the 428-m-thick hydrate zone.

The relative importance of diffusion as a solute-transporting mechanism in the Blake Ridge gas hydrate field, which appears to have been active on a >100-m scale, may be due to the low hydrate concentrations encountered, which leave most of the pore space open for diffusion. This contrasts with gas hydrate fields on active margins with considerably higher hydrate concentrations and hitherto little or no evidence of diffusion (Hesse et al., 1985).

The maximum in the Cl-isotope profile of 0.13 ³⁷Cl at 30 mbsf that is close to, but slightly deeper than the chloride maximum at ~24 mbsf is interesting because it would indicate upward diffusion of chloride towards the seafloor from the roof of the hydrate zone, where salt exclusion occurs during hydrate formation (Hesse and Harrison, 1981). Without this upward diffusion back to the bottom water of the ocean, freshening of the pore waters in hydrate zones would be caused exclusively by advection of low-chlorinity water from below the hydrate zone. If the freshening of this deeper water was due to hydrate melting alone, then no net chlorinity loss from the section would have occurred, in contradiction to the measured chlorinities for Site 997, which on average are below seawater chlorinity. The ³⁷Cl maximum occurs slightly below the Cl^- maximum. However, all Cl-isotope measurements in the upper 30 mbsf of Site 997 are within the error limits of the analytical method with respect to SMOC (Fig. 1). Because of this and the glacial overprint on connate water chlorinity, no conclusive statement is possible with respect to the long-standing problem of Cl^- loss to the ocean from the hydrate zone, although this upward diffusion is the prerequisite for any chlorinity (or salinity) loss from the hydrate field (Hesse and Harrison, 1981).

On the ³⁷Cl vs. Cl^- plot (Fig. 7), sample points associated with low-chlorinity peaks plot on the left half of the diagram ([Cl^-] < 490 mM), indicating that hydrate dissociation does not affect chlorine isotopic ratios, as expected, because the hydrate crystals are supposed to be free of chloride. The tiny offset in the ³⁷Cl curve at the low-chlorinity peak near 250 mbsf (Fig. 1), which is related to samples with higher amounts of hydrate that melted (although within the range of analytical noise), might indicate that real-world methane hydrates do not behave 100 ideally with respect to Cl^- exclusion, as previously speculated by Hesse et al. (1985) and Pavlova and Pashkina (1989).

In conclusion, the chlorine isotope results of this study provide support for the advection-diffusion mechanism proposed by Egeberg and Dickens (1999) to explain the downward pore-water freshening at Site 997, which involves only a small contribution from gas-hydrate dissociation; vice versa, the chlorinity maximum at the top of the section is primarily due to glacial effects and only subordinately caused by hydrate formation. Cl-isotope fractionation due to diffusion is minor. Diffusion is driven by a vertical chlorinity gradient. Where this gradient disappears below the base of the hydrate zone at 452 mbsf, the isotope concentration gradient that is related to a low-chlorinity, ³⁷Cl-depleted water reservoir below the cored section still drives Cl-isotope diffusion, causing the ³⁷Cl to continue to decrease towards the base of the drill site.