DISCUSSION

In this section, discussion is limited to those aspects of the pore-fluid data that are relevant to the fluid flow objectives and that highlight the significant diagenetic results of the Bahamas Transect. The discussion has been divided into four parts that cover the most intriguing hydrogeochemical features of the Transect. These are (1) the shallow pore-fluid gradient shifts, (2) the Na⁺ and Cl^- profiles, (3) the carbonate diagenesis, and (4) the organic matter diagenesis.

Pore-fluid profiles display a dramatic shift in the upper 100 mbsf at all sites. In the upper 20-40 m, profiles of typically conservative elements (Cl^-, Na⁺, and K⁺) as well as many nonconservative elements (SO₄^2-, Ca²⁺, and Mg²⁺) are nearly vertical (Fig. 5). Below a depth of ~40 mbsf, sharp changes occur in the concentration of nearly all pore-fluid constituents and continue down to the base of the hole (Fig. 4). Slight gradients do exist in constituents of the uppermost sediments such as alkalinity, NH₄⁺, and HPO₄^2-, which are most sensitive to small amounts of microbial degradation (Fig. 5). A review of other ODP drilling sites reveals that this type of pore-fluid shift in typically conservative elements is not a common occurrence. Three mechanisms which may contribute to the nature of these shallow profiles are examined here: (1) changing sedimentation rate or a hiatus, (2) active bottom water flushing, and (3) changing reactivity of the sediment.

Large differences in pore-fluid gradients for conservative elements like Cl^- may be explained by a sudden change in sedimentation rates, such as a period of nondeposition followed by a large input of sediment (Goldhaber and Kaplin, 1980; Aller et al., 1986). Sedimentation rates are known to have strong influences on the evolution of pore-fluid profiles by influencing the distance constituents can migrate by diffusion (Berner, 1980). Rates of sedimentation are expected to be substantially higher within the upper 20-40 mbsf if they are responsible for the vertical gradients with this zone. However, sedimentation rates based on biostratigraphic data appear nearly constant throughout the upper 100-180 m of three of the four sites, showing no hiatus or significant increase where the pore-fluid profiles change (Fig. 2). Only at Site 1007 is there a marked hiatus in deposition which roughly coincides (~40 mbsf) with the change in pore-fluid gradients (Fig. 2). Overall, sedimentation rates are higher (10-15 cm/k.y.) in the shallow intervals of the more proximal sites (Sites 1003 and 1005) compared to the more distal sites (Sites 1006 and 1007) (5 cm/k.y.). This difference in sedimentation rate may, in part, explain why the vertical pore-fluid gradients extend to a deeper depth closer to the platform margin. So whereas differences in sedimentation rates may influence the thickness of zones with vertical gradients, they fail to explain why the change in pore-fluid shift occurs where it does.

Active bottom water flushing throughout the upper 20-40 mbsf could explain the lack of significant gradients in the uppermost sediments. Bioirrigation is known to move water through fine sediments and has been documented to produce similar profiles on a much smaller scale (Aller, 1977, 1980). During Leg 166, the term "flush zone" was used to describe this shallow interval. However, no plausible mechanism was developed to explain the advection process. Although sediment dwellers typically influence the upper 1 m of sediment (Wetzel, 1981), it has been hypothesized that they influence profiles down to a depth of 8 m in high-productivity areas (Schulz et al., 1994). However, it is unlikely that bioirrigation mechanisms could produce movement of bottom water down to depths of 2040 m, particularly as both biostratigraphic and lithostratigraphic data indicate a relatively undisturbed sedimentation and the presence of semi-lithified horizons and hardgrounds (Eberli, Swart, Malone, et al., 1997). A more plausible mechanism to advect bottom water through these upper sediments would involve the strong bottom currents that sweep the western slope of the Bahamas Bank and which may have been active since the late Pliocene. Studies of the Florida Current around 27°N indicate bottom currents with velocities in the 10-20 cm/s range (Wang and Mooers, 1997). Whereas bottom currents are not known within the Santaren Channel, the overall flow through the channel is roughly 10% that of the Florida Current. By extension, Santaren bottom currents might be expected to have velocities in the 1-2 cm/s range. Currents in this range could be strong enough to entrain pore-fluids and rework surface sediments. This could have the effect of (1) enhancing exchange of pore-fluids with oxygenated bottom water and (2) increasing the degree of organic matter destruction prior to sediment burial.

A third explanation for the absence of significant chemical gradients in the upper 40 mbsf is that the upper sediments are simply less reactive compared to the sediments below. There is fairly good evidence that the upper sediments down to 20-40 mbsf are oxic to suboxic, following a depth succession of deoxygenation and denitrification. Narrow zones near ~10 mbsf at both Sites 1005 and 1003 show small but significant increases in alkalinity, NH₄⁺, and HPO₄^2-, but no changes in SO₄^2-, indicating small amounts of organic matter oxidation probably by NO₃^- reduction (Fig. 5). Incubation tests done onboard JOIDES Resolution on samples taken from above 30 mbsf at Site 1007 showed no substantial change in pore-fluid constituents after 60 hr, supporting the idea that sediment reactivity is fairly low (P. Kramer and N. Schovsbo, unpubl. data). The lack of reactivity in the uppermost sediments could result from a change in sediment character (labile organic carbon, mineralogy, or grain size) compared to the sediments below. Shipboard lithostratigraphic data do not indicate a major change in the sediment grain size or mineralogy through the interval where the pore-fluid shift occurs, although partly lithified horizons of HMC are more common below 40 mbsf. Total organic carbon (TOC) concentrations do appear to increase below 40 mbsf, particularly at Sites 1003-1005 (Fig. 2). This would suggest that increases in organic content may cause a shift in sediment reactivity, which is reflected in many of the nonconservative pore-fluid constituents (SO₄^2-, alkalinity, NH₄⁺, PO₄^2-, Ca²⁺, Mg²⁺, and Sr²⁺). It may also explain the shift seen in Cl^- and Na⁺ contents if these elements are not behaving conservatively, as is explored in the next section.

Large (up to two-fold) increases in dissolved Cl^- and Na⁺ concentrations from present bottom-water composition occur throughout the Bahamas Transect below 40 mbsf (Fig. 4). The largest gradients occur near the platform margin, whereas the smallest gradients are found at the more distal sites. The possibility that the higher salinity pore fluids were emplaced by fluid-flow advection of brines from deeper sequences is not likely for two reasons. First, the isotopic composition of dissolved strontium (⁸⁷Sr/⁸⁶Sr) in pore fluids at Sites 1003-1007 is in equilibrium with the contemporaneous seawater curve throughout the Pliocene-Miocene sediments (P. Swart and H. Elderfield, unpubl. data). Second, throughout the Pliocene-Miocene sections, pore-fluid constituents are dominated by diffusional gradients and appear to be in equilibrium with the surrounding sediments. For example, within Sites 1003 and 1005, celestite (SrSO₄) is mainly found in core intervals high in dissolved SO₄^2- and saturated with respect to celestite. At Site 1007, pore-fluid Ba²⁺ concentrations are elevated only in the lower intervals where significant amounts of sedimentary acid-soluble barite are present (Fig. 6). This implies that Leg 166 pore fluids are probably in situ and have evolved to their present composition through interaction with the surrounding sediments and diffusion.

Increases observed in pore-fluid Cl^- and Na⁺ concentrations with depth along the Bahamas Transect are a relatively common occurrence along continental and bank margins. During Leg 101, a series of shallow (~<400 mbsf) holes throughout the Bahamas archipelago were drilled and similar increases in salinity at many of the sites were found (Austin, Schlager, Palmer, et al., 1986). During Leg 133, periplatform carbonate sediments drilled off the northeastern coast of Australia revealed a substantial increase in Cl^- at Sites 715 and 823; the increase was interpreted to be caused by diffusion from an underlying evaporite unit (Davies, McKenzie, Palmer-Julson, et al., 1991). Similarly, Leg 150 scientists interpreted the nearly twofold increases in Cl^- and Na⁺ content to result from upward diffusion of deeply buried Jurassic salt along the New Jersey Margin (Miller and Mountain, 1994). Triassic to Lower Jurassic sediments underlying the Bahamas/Florida region are thought to contain evaporites (Sheridan, 1974), and deep salt diapirs are evident on seismic profiles further south. Therefore, it is likely that at Sites 1003-1007 some of the increase in salinity with depth may be caused by upward diffusion of Cl^- and Na⁺. However, an examination of pore-fluid Na⁺/Cl^- ratios shows no significant trend with depth from bottom-water ratios (0.86) except at Site 1006, where there is probably a high degree of clay-mineral interactions involving Na⁺ (Fig. 7). The pore-fluid ¹⁸O data do show an increase by nearly 2.5 at the base of Sites 1003 and 1007, but this increase is interpreted to reflect influences of carbonate recrystallization rather than the influence of an enriched, deep-seated brine or evaporites (Swart, Chap. 8, this volume). In addition, it is not easy to explain how diffusion alone would cause the marked differences in concentration gradients for Na⁺ and Cl^- along the transect (Fig. 4). For example, at Site 1005 Cl^- concentrations increase to 796 mM at a depth of 122.4 mbsf, then decrease to 742 mM at a depth of 178 mbsf, all within Pleistocene sediments. At Site 1006, this range of Cl^- concentration is not reached until a depth of ~453 mbsf within Miocene sediments.

Here, it is postulated that another possible source for the increased Na⁺ and Cl^- concentrations with depth might be salt inclusions contained within defects of the biogenic aragonite, HMC carbonate structure. During diagenetic recrystallization of LMC and dolomite, these salt inclusions may be excluded into the pore water. This process was documented by Malone et al. (1990) at Site 716 in the Maldives archipelago. In this case, sediments originally composed of ~25% aragonite and HMC decreased in sodium content by ~1200 ppm as a result of conversion to LMC over a period of 2.5 Ma. The effect on pore-fluid Na⁺ content cannot be determined because this constituent was not analyzed, but dissolved Cl^- does show some localized increases, although no systematic trend is evident.

A rough calculation can be made to determine the possible influence of sodium expulsion on pore-water concentrations. Sodium concentrations are known to be enriched in biogenic aragonite and HMC compared to LMC (foraminifers and coccoliths) (Busenberg and Plummer, 1985). Sodium concentrations are ~4000 ppm in aragonite, whereas they are much lower in LMC (~500 ppm) (Milliman, 1974). Assuming sediments are originally 100% aragonite containing 4000 ppm sodium, complete recrystallization to 500 ppm LMC could release enough sodium to raise pore-fluid concentrations by nearly 1.5 x seawater values. Sediment porosity values can be used as a proxy indicator for the degree of carbonate recrystallization. An examination of the solid-phase data from Sites 1005 and 1007 shows that sodium concentration and sediment porosity are well correlated, particularly in samples with <30% porosity (Fig. 8). Sodium values are higher than expected, particularly for high-porosity samples, and much of the scatter probably results from incomplete removal of dried salts during sample preparation. However, the well-defined trend in samples with <30% porosity seems to support the idea that, as biogenic carbonates are recrystallized, sodium (and presumably chloride) are expelled from the mineral structure, which should cause small increases in pore-fluid salinity. This process would certainly explain why there are larger (and steeper) gradients in Cl^- and Na⁺ near the platform margin (Site 1005), where carbonate recrystallization is more prevalent compared to the more distal sites. It could also explain why shifts around 20-40 mbsf in Cl^- and Na⁺ contents coincide with shifts seen in other pore-fluid constituents influenced by carbonate remineralization such as Sr²⁺.

Lithostratigraphy and mineralogy results indicate that all sites examined along the Bahamas Transect are heavily influenced by carbonate recrystallization. In general, the more proximal sites show a higher degree of diagenetic carbonate alteration than the more distal sites (Eberli, Swart, Malone, et al., 1997). This higher degree of alteration probably results from the fact that the diagenetic potential is higher along the platform margin as a result of (1) higher input of metastable carbonate and organic matter during highstands along the platform margin, and (2) decreased influence of clay minerals. The chemistry of interstitial waters below 40 mbsf also indicates extensive carbonate alteration, principally aragonite dissolution and calcite and dolomite precipitation.

All sites show a large increase (up to 70-fold) in dissolved Sr²⁺ derived from the recrystallization of metastable aragonite to LMC. Strontium concentrations are much higher in aragonite (12,000 ppm) than LMC (2000 ppm) (Milliman, 1974). The amount of Sr²⁺ able to remain in solution is largely controlled by the solubility of celestite (SrSO₄) and, therefore, can only be used as a measure of the degree of carbonate alteration when pore-fluid SO₄^2- concentrations are very low or absent (Baker and Bloomer, 1988; Swart and Guzakowski, 1988). A plot of Cl^- vs. Sr²⁺ for all sites shows that the two are well correlated at Site 1006, the only site where SO₄^2- is absent from pore fluids below 200 mbsf and where no celestite was detected in the cores (Fig. 9). Again, this seems to support the idea that dissolved chloride and sodium are influenced by carbonate remineralization.

Dolomite formation is also an important diagenetic process occurring within sediments below 40 mbsf along the Bahamas Transect. The largest increases in the Mg²⁺/Ca²⁺ ratios appear to coincide with intermediate-depth intervals (50-200 mbsf), where extensive sulfate reduction is occurring (Fig. 10) Within these intervals, the Mg²⁺/Ca²⁺ ratios increase on the order of 1:3 suggesting dolomite formation by recrystallization of aragonite and HMC (Baker and Kastner, 1981). Significant dolomite (up to 20%) was detected in the upper Pliocene-Pleistocene sediments near sequence boundaries (Fig. 3). In lower sediment regimes (middle Pliocene-Miocene units), there is a progressive loss of Mg²⁺ and increase in Ca²⁺ concentrations leading to a decrease in the Mg²⁺/Ca²⁺ ratios (Fig. 10). Small amounts of dolomite (background dolomite) are probably forming throughout these intervals but are limited by the availability of Mg²⁺ supplied by carbonate recrystallization and diffusion from the overlying seawater.

Pore-fluid chemistry and headspace analyses indicate that the remineralization of organic matter is an important process along the Bahamas Transect. Oxidation of organic matter by sulfate reduction is evident at all sites below 40 mbsf, based on the presence of H₂S gas (Eberli, Swart, Malone et al., 1997). Sulfate reduction is most pronounced in the shallow Pliocene-Recent intervals close to the platform margin, where high rates of sedimentation (15 cm/k.y.) result in the burial of substantial quantities of organic material (Fig. 2). At Site 1005, sulfate content is reduced and below the limits of detection (~1 mM) by a depth of 87 mbsf (Fig. 4). The concomitant increase in alkalinity within this zone of sulfate depletion follows a 2:1 ratio (i.e., a 2-mole increase in total alkalinity per one mole of sulfate lost), which agrees with the predicted model of microbial sulfate reduction (Berner, 1971). What is unusual about Site 1005 (along with Sites 1003 and 1007) is that rather than remain depleted, dissolved SO₄^2- increases in excess of bottom-water concentrations (30 mM) below this zone of sulfate reduction (Fig. 4). This alternation between sulfate-reducing and non-sulfate-reducing zones is very unusual and is examined in more detail below.

Sulfate reduction is normally limited by the availability of labile organic matter and dissolved SO₄^2-. Therefore, we might expect that sulfate-reducing intervals should have a higher abundance of labile organic matter compared to non-sulfate-reducing intervals. An examination of Figure 2 for Sites 1003 and 1005 shows a rough correspondence, but more samples need to be analyzed to test this hypothesis. The availability of dissolved SO₄^2- in deeper sequences (>200 mbsf) is believed to be largely limited by the presence of available dissolved Fe²⁺. Dissolved Fe²⁺ will sequester reduced sulfate (HS^-) to form sedimentary pyrite (Berner, 1966; Canfield, 1989). Significant amounts of pyrite were observed at Sites 1006 and 1007 and below 1070 mbsf at Site 1003. The origin of the dissolved Fe²⁺ is believed to be from Fe-Mn-rich phases associated with siliciclastic clays (smectite, feldspar, quartz, and kaolinite) deposited by channel currents at the more distal sites (Eberli, Swart, Malone, et al., 1997). In contrast, Site 1005 and intervals of Site 1003 probably receive only minimal fluxes of Fe²⁺ delivered by detrital mineral phases and, correspondingly, show only trace levels of iron-sulfide formation. Sulfide produced by SO₄^2- reduction is built up to high concentrations thereby reducing the sediment pH (Ben-Yaakov, 1973). As levels of free hydrogen-sulfide continue to build up in the interstitial waters, reaction pathways are dominated by the oxidation of elemental sulfur back to sulfate, which further reduces the pH (Goldhaber and Kaplan, 1980; Aller, 1982). This results in enhanced carbonate dissolution (Canfield and Raiswell, 1991) and may explain why the platform margin sediments are much more reactive in terms of carbonate recrystallization than the sediments of the more distal sites. Once all labile organic material is oxidized, sulfate reduction is inhibited and all reduced forms of sulfide are gradually transformed back to dissolved SO₄^2-, which can remain at high concentrations.

Therefore, the combined influences of availability of labile organic carbon and dissolved Fe²⁺ are believed to control the alternations between sulfate-rich and sulfate-depleted zones. Diffusion of constituents from above and below these boundaries leads to the formation of many diagenetic sulfur minerals at these interfaces, such as barite, celestite, and elemental sulfur. Figure 11 shows Sr²⁺ and SO₄^2- profiles for Site 1007 illustrating where celestite is forming as waters become saturated with respect to this mineral.

Degradation of organic matter by methane oxidation is also an important process along the Bahamas Transect. The high concentration of methane gas detected within the Pleistocene-Pliocene intervals of Sites 1003-1005 indicates that oxidation of organic matter by methane is occurring immediately below the sulfate-reduction zones. The peak concentrations of NH₄⁺ are commonly below the alkalinity maximum and below the base of the sulfate-reduction zone, as has been observed in other settings where methanogenesis is an important process (Mackin and Aller, 1984). It is also highly probable that partial sulfate reduction may also be occurring by oxidation of upwardly migrating methane (Burns, 1998).

An examination of the pore-fluid ¹³C of the DIC shows behavior similar to other carbonate-dominated sites where sulfate reduction and methanogenesis occur (Swart, 1993). While DIC ¹³C profiles reflect these processes, they are strongly buffered by the host carbonate and consequently only display minor changes in the ¹³C. Surprisingly, one of the regions displaying the most negative ¹³C values is the upper nonreactive zone, supporting the idea that, whereas there may be small amounts of organic matter oxidation using oxygen, there is little carbonate diagenesis taking place. Near the base of the flush zone (20-40 mbsf), all DIC ¹³C values increase slightly to between +1 and +1.5, then show a decrease to a minimum of -1.2 at Site 1005, -0.8 at Site 1004, and +0.81 at Site 1003 corresponding to the loss of SO₄^2-. Although this minimum reflects an increasing contribution from the oxidation of organic material by sulfate, it is masked by the input of carbon from the dissolution of carbonate. The ¹³C values increase with increasing depth, perhaps reflecting equilibration with CO₂ produced during methanogenesis. At Site 1005, the ¹³C values decrease with increasing depth reflecting the reduced influence of methanogenesis below 300 mbsf (Fig. 10).