INTERPRETATION OF THE RESULTS

Diffusion- and Advection-Dominated Chloride Profile

The downward chlorinity decrease of ~10% (Fig. 1) within the upper part of the hydrate zone (Zone 2) of the Blake Ridge field has been postulated by Egeberg and Dickens (1999) as primarily resulting from diffusive interaction between a low-chlorinity pool of water below the hydrate base (Zone 4), which has been advected upwards into the hydrate zone (Zone 3), and the chlorinity maximum in Zone 1. This conclusion is constrained by in situ water samples that were collected with the water sampling temperature probe (WSTP) during Leg 164 and show a consistently higher chlorinity than the samples squeezed on board for Site 997. Measurements on the in situ samples provide the constraints for modeling the Cl^- profile (Fig. 4) with the combined diffusion/advection model of Egeberg and Dickens (1999). The interpretation of the chlorinity profile by these authors is outlined here, because their model is being tested with the Cl-isotope data. The difference between the chloride measurements on samples squeezed on board and the model trend of Egeberg and Dickens (1999) corresponds to the dilution caused by freshwater release from hydrate melting in the sampling process. The amount of in situ gas hydrate predicted from this difference is small, reflecting the low concentration of hydrate at the site (<4% of the pore space for the hydrate-bearing section between 24 and 452 mbsf). The downward freshening in the shipboard samples that is caused by hydrate decomposition during sampling (sampling artifact) thus is minor except in hydrate-rich layers or nodules. Much of the freshening is due to upward advection and diffusion of water from the low-chlorinity water pool below the hydrate zone, which has formed probably as a result of the long-term existence of the Blake Ridge hydrate field. Isolated low-chlorinity peaks such as those at 255 and 450 mbsf (with Cl^- minima of 454 and 405 mM, respectively) reflect greater freshwater release from zones of higher hydrate concentration (up to 24.5%) as a sampling artifact. Without these hydrate-rich layers, an average hydrate concentration of only 2.3% of the pore space of the hydrate zone at Site 997 is obtained by Egeberg and Dickens (1999). This amount of gas hydrate appears to be low in view of the classical seismic profile of Shipley et al. (1979) along the Blake Ridge, which originally seemed to suggest the presence of solidly frozen hydrate-rich sediment in the hydrate zone. However, the low gas hydrate concentrations are no surprise in view of the equally low chlorinity reductions found during previous deep-sea drilling in the region (Jenden and Gieskes, 1983). Although the increase in hydrate concentration from 1.7% at ~50 mbsf to 3.7% at the bottom of the hydrate zone (at ~450 mbsf) has been ascribed primarily to burial compaction (Egeberg and Dickens, 1999), upward diffusion and advection of methane from below the hydrate base and its recapture and incorporation as solid hydrate within the hydrate zone cannot be excluded and could have contributed to the downward increase in hydrate concentration, particularly in local zones of higher hydrate concentration. If these zones are included in the calculations, the average hydrate concentration is raised from 2.3% to 3.8% of the pore space (Egeberg and Dickens, 1999). In fact, spectacular, but rare, massive hydrate nodules or layers were encountered during core retrieval at sea (Paull, Matsumoto, Wallace, et al., 1996).

Significance of the Chlorine Isotopic Results

The ³⁷Cl data for Site 997 provide support for an advection-dominated chloride profile modified by diffusion. In contrast to the highly negative ³⁷Cl values of the active-margin pore waters of Ransom et al. (1995), the Pliocene-Pleistocene Blake Ridge sediments do not contain volcanic glass or other highly reactive volcanogenic detritus prone to early diagenetic alteration that could lead to preferential incorporation of ³⁷Cl in clay minerals. Detrital clay minerals undergo little early diagenetic change at shallow burial levels with temperatures below 50°C (e.g., Hower et al., 1976; Weaver, 1989, table 7-1), except for some layer-charge development (Ko and Hesse, 1998) that would counteract chloride involvement. Small amounts of authigenic smectites and sepiolite may form in hemipelagic sediments, but their presence has been difficult to demonstrate against an overwhelming detrital background (e.g., Kastner, 1981, pp. 917-918, 953-956; De Lange and Rispens, 1986). The Blake Ridge sediments at first glance are thus unlikely to have undergone mineral or chemical reactions involving fractionation of the chlorine isotopes, at least the shallow-burial sequence penetrated at Site 997. This leaves physical processes such as diffusion and advection as being the cause of the isotope profile, in accordance with the conclusions drawn about the behavior of the in situ pore-water ions. Using chloride diffusion coefficients from the literature (Li and Gregori, 1974), adjusted to the Blake Ridge subbottom temperature field (Egeberg and Dickens, 1999), the best fit for our Cl isotope trend is obtained with an upward pore-water advection rate of 0.18 mm/yr and an assumed ratio of the diffusion coefficients for the light and heavy Cl isotopes of 1.0023 (corresponding to the highest of the ratios employed by Eggenkamp et al., 1995; Fig. 5). This is in good agreement with an estimated advection rate of 0.2 mm/yr by Egeberg and Dickens (1999) based on Br^-/I^- ratios and provides an independent test of their combined advection-diffusion model that was used to generate Figure 5.

Advection has been incorporated in the Blake Ridge hydrochemical model based on the assumption that the vertical trend of Br^-/I^- ratios, which show a dramatic increase in concentration of these halogens compared to seawater (Fig. 6), cannot be obtained by reactions (organic matter decomposition) within the sediment, because the amount of organic carbon present (1.2% on average for Site 997) is insufficient (Egeberg and Dickens, 1999). The high concentrations of Br^-, I^-, acetate, and dissolved organic carbon, and low concentration of Cl^- in pore waters sampled from the base of the drilled sequence have been attributed to influx of highly modified pore water from below. Advection of water enriched in these ions and depleted in Cl^- is therefore required. The source of this water is not known, as are the mechanisms for generating the large ³⁷Cl depletion. Because this water may rise from deeper levels where clay reactions occur, water-mineral reactions and/or interaction with organic matter are indicated.

The Cl-isotope profile therefore probably represents a two end-member mixing line between a paleo-seawater end member with a ³⁷Cl close to 0 and a low-concentration pore-water reservoir that is depleted in the heavy Cl isotope and from which water is presently advected upwards. The difference in chlorinity and Cl-isotopic ratios between the two reservoirs is bridged by diffusion. In the upper part of the hole down to 373 mbsf, the decrease in ³⁷Cl is linearly correlated with Cl^- with a gradient of about 0.50/10 mM Cl^- (Fig. 7). Although the chlorinity gradient in the raw data seems to disappear in Zone 3, the simulated chloride profile of Figure 4A suggests that a gradient exists right to the bottom of the hydrate zone (base of Zone 3 at 452 mbsf). Below this depth, no Cl^- gradient exists, but a separate plot of the concentrations of the two chlorine isotopes (Fig. 6) shows the continuing, though slight, downward change in concentration of the isotopes, which would maintain diffusion as a mixing mechanism between the two reservoirs. Mixing between two end-member pore-water reservoirs with contrasting isotopic signatures similar to the ones proposed for this study has been encountered in North Sea formation waters (Eggenkamp and Coleman, 1998).

Diffusion as an isotope-fractionating mechanism itself is insufficient to explain the observed Cl-isotope profile, as a run of the model without the advection term shows, where the model was allowed to find its own solution to the lower boundary chloride concentration and Cl-isotope signature (at the bottom of Zone 3) without introducing the measured isotopic ratios into the model (Fig. 8). Note that in the absence of advection both the Cl concentration and Cl-isotope profiles are convex down (below the Cl^- maximum at 30 mbsf, see "Discussion and Conclusion" section, this chapter) as opposed to the observed convex-up profiles.

Oxygen and Hydrogen Isotopes

The positive ¹⁸O and D values at the top of Site 997 are within range of isotopic excursions of seawater during the Pleistocene glaciations (up to 0.9 for ¹⁸O) and can be explained by inheritance from buried connate waters, like the chloride maximum found at the same depth (see below). At the depth where these anomalies occur, the sediments are Quaternary in age (Fig. 9). Only part of this glacial seawater signature has been diffused away in the time available (McDuff, 1985). Chlorinity of the pore water is also expected to be raised by hydrate formation though salt exclusion; at the same time, pore-water oxygen and hydrogen isotopes should become lighter due to preferential extraction of the heavy isotopes in the solid hydrate. In view of the low hydrate concentrations, the effects related to hydrate formation are probably minor and overprinted by the glacial signatures.

The largely mirror-symmetrical shape of the middle part of the depth curves with respect to the chlorinity curve (below 200 mbsf, with the positive ¹⁸O peaks at 247-254 and 452 mbsf being matched by low-chlorinity peaks; Fig. 2, Fig. 3) is what is predicted by the hydrate dissociation hypothesis (Hesse and Harrison, 1981). Hydrate melting produces low-chlorinity and isotopically heavy water, as corroborated for Leg 164 results by Matsumoto et al. (Chap. 2, this volume). The negative delta values for the oxygen isotopes in the middle part of the profile are indicative of mineral reactions, but as for the chlorine isotopes, the reaction partners are difficult to pinpoint (see previous section). Carbonate diagenesis has produced authigenic carbonates which would have been in isotopic equilibrium with weakly negative or near 0 pore waters concerning their ¹⁸O, at the time of formation (Pierre et al., Chap. 13, this volume), that are slightly less negative than the observed anomaly of down to -0.6. Negative D values as low as -12 in the lower part of the hole below the hydrate base confirm findings from nearby DSDP Site 533 on Blake Ridge (Jenden and Gieskes, 1983), but still remain unexplained. These depletions are even more pronounced in the deep-basin Site 534, and the results for Sites 997 and 533 could reflect advection of isotopically light waters; however, in view of the advective pore-water flow advocated for Site 997, the spikes in the profiles point to ongoing local, but unidentified, reactions.