DISCUSSION

Profiles of minor and trace dissolved constituents are shown in figures with the following order: Site 1005, Site 1007, and Site 1006. This presentation is designed to place the profiles in a geographical context from east to west along the western margin of the GBB (Fig. 1). Site 1005, the easternmost and most proximal Bahamas Transect site on the edge of the platform, penetrates downlapping prograding sediments, whereas Site 1006 is the westernmost site and penetrates a more pelagic sediment sequence (see Fig. 2) in the northern portion of the Santaren Channel (Fig. 1). Site 1007 is intermediate to the other two (Eberli, Swart, Malone, et al., 1997). Discussion of the results will follow this format.

Lithium

Absolute and relative differences between the concentrations of dissolved Li⁺ determined aboard the JOIDES Resolution and in our laboratory are plotted as a function of depth downhole in Figure 8. From a purely analytical standpoint, and in the absence of gross calibration errors, discrepancies in the data sets can arise from either inadequate matrix matching of samples and calibration standards or insufficient background correction during the shipboard measurements. Shipboard standards for Li⁺ determinations were prepared in seawater in an attempt to match the matrix of the IW samples. As described in the methods section, shore-based analyses were performed by AES with calibration by the method of standard addition and with offline emission measurements to correct for background. The latter allows us to correct effectively broad band emission from the matrix, as well as ionization interference effects that are often encountered with alkali metal analysis by flame atomic spectrometry.

We initially suspected that the shipboard analytical problems resulted from increased salinity of pore fluids downhole and the inability of the Varian AAS instrument to deal with the concomitant large increase in background signal. However, no simple relationship is apparent in Figure 8 between depth (i.e., salinity) and -Li, suggesting that increased salinity is not solely the cause of the problem. Rather, other chemical species, likely nonconservative IW constituents, may be responsible for the observed discrepancy. The greatest relative and absolute differences between the two data sets occur in samples from Site 1006 (Fig. 8). At this site the greatest relative difference is observed in IW that displays total SO₄^2- depletion (~200 mbsf). The greatest absolute deviation, however, occurs in samples recovered near 450 mbsf. This depth corresponds to the NH₄⁺ and (second) broad alkalinity maxima. In samples from Site 1005, differences are intermediate to those observed at Sites 1006 and 1007; however, the largest discrepancy also occurs in samples collected near the NH₄⁺ and alkalinity maxima. At all three sites, the greatest absolute difference between the two data sets occurs in samples with the highest dissolved Sr²⁺ concentrations. This is consistent with the known problem of SrOH broad band emission causing an interference on the Li 670.8-nm line (Heinrichs and Herrmann, 1990). The general covariance between the shipboard dissolved Li⁺ and Sr²⁺ profiles (Eberli, Swart, Malone, et al., 1997) also suggests strongly that the latter is responsible for the analytical interference, although a plot of shipboard determinations of Li⁺ vs. Sr²⁺ (not shown), while generally linear, shows significant deviations that suggest that another source of interference also exists in the shipboard analyses. The occurrence of significant alkalinity and NH₄⁺ maxima where large analytical discrepancies in Li⁺ determinations were observed may simply be coincidental and reflect the fact that such conditions lead to enhanced dissolution of aragonite, which is the primary Sr²⁺ carrier in sediments from the GBB. Nonetheless, molecular species (e.g., NH4⁺) are well known to produce broad absorption and/or emission bands.

Although it appears that other species may contribute to the increased background signal, because IW recovered during Leg 166 exhibits the highest Sr²⁺ concentrations observed in ODP to date, Sr²⁺ likely contributed most to the observed analytical problems. Nonetheless, it is clear that care must be taken to account for large variations in any species likely to produce broad band absorption/emission when conducting Li⁺ determinations aboard the JOIDES Resolution.

Iron and Manganese

Some important differences exist in the sample handling and analytical procedures used for determinations of dissolved Fe and Mn²⁺ in our laboratory and onboard the JOIDES Resolution. These differences likely contribute to a large extent to the differences observed between shipboard and shore-based data (Fig. 4). First, splits left over from shipboard alkalinity titrations were sealed in plastic tubes with pliers and subsequently cut open for FAAS analysis using scissors or a carpet knife. Whenever available, samples analyzed in our laboratory by ICP/OES and FIA were taken from Nalgene HDPE bottles that had been filled with sample under the laminar hood aboard ship and preserved with ultra-high purity HNO₃. Second, splits analyzed onboard the JOIDES Resolution were filtered through 0.45-µm acrodiscs, whereas those collected for shore-based studies were filtered through 0.2-µm acrodiscs.

Some of the difference between the two data sets is attributable to contamination of the shipboard samples with Fe during handling (alkalinity titration and subsequent storage and reopening of tubes). Because Fe is also notorious for existing as colloidal particles in natural waters, filtration through 0.2-µm acrodiscs likely removed a substantial portion of the larger colloidal particles. Thus, data from shore-based determinations of Fe in trace element splits are more representative of the dissolved fraction, and some of the difference between shipboard and shore-based results can be attributed to filtration effects. In spite of these important differences, the general trends defined by the shipboard data sets still correlate somewhat with those defined by shore-based determinations, albeit only crudely so.

Insufficient IW was recovered from whole-round core samples collected below 400 mbsf at Site 1005 to acquire trace metal splits; thus, the stored alkalinity splits that were not analyzed aboard the JOIDES Resolution were analyzed on shore for their dissolved Fe²⁺ and Mn²⁺ content. The transition trace metal data from these samples are, therefore, subject to a high bias because of potential contamination (i.e., Fe and to a lesser extent Mn), as described above, and filtration effects. However, shore-based determinations of dissolved Fe yielded significantly lower concentrations than those measured aboard the JOIDES Resolution. It is likely that the shore-based analyses were not as adversely affected by matrix effects. Also, because ICP/OES is more sensitive than AAS for many elements, dilutions of the samples analyzed by ICP/OES were possible, which substantially reduced matrix effects. Furthermore, offline background correction is used in ICP/OES, thereby minimizing matrix interferences and further increasing the reliability of the analyses. Thus, the shore-based analyses of the alkalinity split residues are deemed more reliable than their shipboard counterparts.

Lithium

The profiles shown in Figure 3 reflect a mediation of dissolved Li⁺ concentrations by early diagenesis and by local lithological control at the various Bahamas transect sites. At Site 1005, a series of zones, which correspond well to lithologic units and to the chemical reaction zones identified during Leg 166 (Eberli, Swart, Malone, et al., 1997), display either invariant concentrations or positive or negative Li⁺ gradients. The dissolved Li⁺ profile covaries with several nonconservative dissolved constituents (e.g., alkalinity, NH₄⁺, Sr²⁺) in selected parts of the sediment column and mirrors remarkably the dissolved SO₄²⁺ profile (Fig. 9). Where the Li⁺ profile does not appear to covary with those of other dissolved constituents, breaks in concentration gradients occur at the same depth and reflect stratigraphic horizons (i.e., physical barriers) that affect lateral fluid flow and inhibit the establishment of smooth vertical diffusive profiles. For example, the upper 40 mbsf displays relatively invariant profiles of many IW constituents. This section was interpreted to represent a flushed zone throughout which seawater flows freely and maintains near-seawater concentrations of conservative and most nonconservative IW constituents (Eberli, Swart, Malone, et al., 1997). Dissolved Li⁺ concentrations display, nonetheless, small but significant changes in this interval. Between ~50 and 100 mbsf in lithologic Subunit IA of Site 1005, dissolved Li⁺ and several other constituents increase in concentration as SO₄²⁺ (Fig. 9) becomes strongly depleted during the bacterially mediated oxidation of organic matter (Claypool and Kaplan, 1974).

Conditions of high alkalinity and lowered pH favor the recrystallization of metastable aragonite and high magnesian calcite. These phases typically release trace elements upon dissolution (e.g., Baker, 1986; Baker et al., 1982). Thus, it is tempting to invoke biogenic carbonate as the primary source of Li⁺ to pore fluids. The general shape of the Sr²⁺ and Li⁺ profiles appears similar throughout large portions of the sediment column (Eberli, Swart, Malone, et al., 1997) and supports this inference, although greater deviations exist between the profiles of these constituents at Site 1007.

At Site 1005, dissolved Li⁺ concentrations increase sharply within partially lithified and dolomitized sediments of Subunits IB and lithified sediments of Subunit IC, consistent with expulsion of Li⁺ from solid phases during carbonate recrystallization reactions that formed the lithified intervals. Also, the lower dissolved Li⁺ concentrations below 260 mbsf in the less lithified sediments of Units II and III may reflect a much-decreased rate of diagenesis of biogenic carbonate and concomitant release therefrom. Reduced microbial activity in this interval is evidenced by a return of elevated SO₄^2- concentrations, which approach seawater values below 500 mbsf. Deeper downhole, the dissolved Li⁺ profile largely parallels that of Sr²⁺, and a final change in the Li⁺ gradient below ~525 mbsf also coincides with depletion of SO₄^2-.

Despite the qualitative correlation between dissolved Li⁺ and Sr²⁺ profiles described above, the proposed exclusion of Li⁺ during recrystallization of metastable carbonate minerals in the sediments cannot account for all of the observed increases in concentration downhole. For example, if we were to assume a conservative behavior for Li⁺, just the increase in salinity downhole at Site 1005 could account for approximately half the observed increase in dissolved Li⁺ above the normal seawater concentration. Furthermore, dissolved Li⁺ and Sr²⁺ do not covary that strongly. This is especially true in the uppermost portion of the sediment column. Quite sharp variations in the dissolved Li/Sr value occur as a function of depth in the upper 20 m of sediments at Sites 1005 and 1007 (Fig. 10). Clearly different processes must govern their behavior here. The dissolution of biogenic silica may release Li⁺, at least in the shallowest sediments (e.g., Gieskes, 1983; De Carlo, 1992, and references therein). Indeed, in the range of 15-20 mbsf at Sites 1005 and 1007 and between 122 and 164 mbsf at Site 1005, concentrations of Li⁺ and SiO₂ in pore water increase, whereas that of Sr²⁺ does not.

Reactions involving biogenic silica, however, are unlikely to explain the Li⁺ distribution in deeper sediments. The same can likely be said for diagenetic reactions of aragonite. Rather, a possible mechanism for the release of Li⁺ to solution is by ion exchange with NH₄⁺. At all GBB sites, there is a notable coincidence of high dissolved Li⁺ concentration with the zones displaying high NH₄⁺ concentrations. This is particularly true deeper downhole, where acid insoluble residues, interpreted here to represent largely detrital matter, tend to be more abundant than in the shallower portions of the holes. Examination of Figure 10 reveals features that are consistent with these inferences. For example, the Li/Sr value decreases substantially at Site 1007 below 450 mbsf, probably because of an increase in the abundance of aragonite in the sediments that are a source of Sr²⁺. The interval between ~200 and 500 mbsf, where the Li/Sr value is highest (except for that in the uppermost 50 mbsf), contains abundant detrital minerals (e.g., clays) (Eberli, Swart, Malone, et al., 1997) that represent a potentially more important source of Li⁺ in this hole. The substantial drop in the Li/Sr value below 475 mbsf, although coinciding with a zone still containing significant quantities of detrital minerals, occurs where aragonite is sufficiently abundant that its recrystallization could also supply the pore water with dissolved Sr²⁺. Further support for a terrigenous/detrital source material for Li⁺ is found near the bottom of the hole where the abundance of terrigenous/detrital minerals, as determined on the basis of weak acid-insoluble residues, increases sharply and an increase in the Li/Sr value is observed.

It should also be borne in mind that the low solubility of SrSO₄ exerts a strong control on the abundance of dissolved Sr²⁺ (e.g., Baker and Bloomer, 1988) and complicates the interpretation of the dissolved Li/Sr profiles shown in Figure 10. Solubility control by celestite should have the greatest influence at Site 1005, where pore fluids between 450 and 600 mbsf are near saturation with respect to this mineral (see fig. 24 in Shipboard Scientific Party, 1997b). Thus, the increase in the Li/Sr value below 400 mbsf is not entirely attributable to increases in dissolved Li⁺ derived from terrigenous/detrital minerals. More abundant dissolved SO₄^2- in this interval (Fig. 9), especially near 500 mbsf, likely limits the concentration of dissolved Sr²⁺. At Site 1007, dissolved SO₄^2- is also present near 475 mbsf, but at substantially lower concentrations than at the same depth in Site 1005. Thus, celestite solubility should not control dissolved Sr²⁺ in this depth interval nor should it influence the Li/Sr value.

The mediation of dissolved Sr²⁺ by celestite solubility becomes unimportant below 200 mbsf at Site 1006 because SO₄^2- is fully depleted. Thus, any changes in the Li/Sr value must result from differences in the sources of these two elements. Site 1006 is positioned on a continuous thick section of Neogene-aged drift deposits that interfingered with prograding sediments from the GBB (Eberli, Swart, Malone, et al., 1997). It is also characterized by a greater abundance of insoluble/detrital minerals throughout most of the sedimentary column than either Site 1005 or 1007. Sediments at Site 1007, however, contain a greater abundance of detrital minerals than those at Site 1005, even reaching greater concentrations in some sections than observed at Site 1006. Nonetheless, detrital minerals are most uniformly distributed throughout the sediments of Site 1006. Thus, this material can be invoked as a source of Li to the pore water. The Li/Sr value, however, always remains lower at Site 1006 than at Site 1005 and is lower than above 600 mbsf at Site 1007.

Variations in Li/Sr should be viewed with some caution when used as a tool to evaluate sources of these two elements. Exceptions include relatively anoxic sediments in the uppermost portions of the sediment column, where extreme changes in the Li/Sr value occur over relatively shallow depth intervals. Nonetheless, dissolved Li⁺ can derive primarily from three sources: (1) early diagenesis of biogenic silica, (2) terrigenous material (especially clay minerals), and (3) exclusion during the recrystallization of carbonates. This is consistent with earlier suggestions by Froelich et al. (1991). A possible fourth source, which decreases in importance from Site 1005 to Sites 1007 and 1006, is simply the increased salinity of pore fluids that might contain correspondingly greater dissolved Li⁺ with increasing depth downhole.

Hoefs and Sywall (1997) report that marine carbonates average 2 µg/g Li. De Carlo (1992) showed that sediments from the Exmouth Plateau that are rich in terrigenous matter contain 50-130 µg/g Li, whereas predominantly carbonate-rich sediments from this area often displayed <10 µg/g Li. Kabata-Pendias and Pendias (1992) compiled the Li content of various type soils and reported concentrations ranging from <1 µg/g to 130 µg/g for clay-rich soils from New Zealand.

The weak-acid soluble fraction of sediments from Site 1007 contains between 8 and 38 µg/g, with most samples exhibiting concentrations of <20 µg/g (P. Kramer, unpubl. data). The lower values are representative of predominantly biogenic carbonates of the GBB, whereas values greater than 15 µg/g Li usually occur in sediments from deeper than 600 mbsf. The latter must contain an important Li contribution from detrital minerals. This inference is supported by an increase in the amount of insoluble residues (terrigenous/detrital minerals) downhole, reaching 30-35 wt% in certain samples from below 1000 mbsf. These sediments likely contain interlayer Li bound in clay minerals, which can be released to solution by ion exchange with NH₄⁺_. Thus, even in sediments with substantially more biogenic silica and carbonate than terrigenous material, the Li contribution from terrigenous material could be proportionally greater than that from the biogenic components.

We calculate below the potential of the solid phase to contribute Li to the pore water of sediments from the GBB. Several assumptions are made to simplify this calculation. First, we assume that sediments recovered during Leg 166 contain ~25% terrigenous/detrital material by mass throughout all of Site 1006 and below 600 mbsf at Site 1007. Abundances are considerably lower at Site 1005 where we assume a <5%-10% contribution of terrigenous/detrital matter to the sediment. Second, we assume conservatively that terrigenous/detrital material contains 50 µg/g of potentially available Li and use a biogenic carbonate value of 2 µg/g Li. We further assume no significant contribution to dissolved Li⁺ from biogenic silica, and an equal propensity of each of the two other phases for providing Li⁺ to the pore fluids. Based on these assumptions, we calculate that the terrigenous fraction could contribute nearly 90% of the Li⁺ throughout all of Site 1006 and below 600 mbsf at Site 1007. At Site 1005, the dissolved Li⁺ contribution derived from detrital material could still reach 58%-74% of the total, depending on whether a 5% or 10% terrigenous contribution were considered.

A caveat to the above calculations regards whether terrigenous material has an equal propensity to release its Li to solution as does biogenic carbonate. Although terrigenous material is typically more resistant to solution than biogenic carbonates, because clay minerals are an important carrier of Li, have high surface areas, and readily exchange univalent cations, it is likely that Li bound in clays is readily accessible and, therefore, available for release to solution. This is particularly true where interstitial fluid concentrations of NH₄⁺ are in the millimolar range. Additionally, a significant portion of the Li might have been originally associated with Fe- and Mn-rich phases which are readily solubilized, especially under the sub-oxic and anoxic conditions encountered at these GBB sites. Support for this inference comes from the observation that the Li content of the easily solubilized phases in sediments recovered deep within Site 1007 can reach more than 20 µg/g by 600 mbsf. Thus, we estimate that nonbiogenic phases at or below this depth can easily provide at least 50%, if not more, of the dissolved Li⁺ to pore fluids.

Rubidium

Dissolved Rb⁺ is a minor constituent (1.4 µM) of seawater and displays a conservative behavior in the oceans (Bruland, 1983). High-temperature seawater-rock interactions, however, are known to enrich Rb⁺ in fluids up to several orders of magnitude above seawater concentrations (e.g., Mottl and Holland, 1978). Enrichment of this element in pore fluids of deep-sea sediments has usually been attributed to alteration of volcanic matter (e.g., Gieskes, 1983). For example, Mottl (1992) observed more than five-fold enrichment of Rb on conical seamount (Site 780), a serpentinite in the Mariana forearc. More recently, two- to nearly three-fold enrichments of dissolved Rb⁺ over seawater were observed in pore water from Sites 998 and 999 (Sigurdson, Leckie, Acton, et al., 1997). These were correlated with the occurrence of volcanic ash layers and dispersed volcanic minerals within the sediments, indicating that low temperature alteration of such minerals in sediments also leads to enrichment of Rb⁺ in pore fluids.

The dissolved Rb⁺ concentration at Site 1006 (Fig. 7) reaches nearly twice the seawater value by 700 mbsf, yet pore fluids at all depths, except immediately below the mudline, are depleted on a chloride-normalized basis relative to a fluid with seawater proportions. Dissolved Rb⁺ becomes increasingly more depleted relative to seawater as a function of depth in the upper 250 mbsf of the profile. The cause of this depletion remains elusive. Increases in dissolved Rb⁺ and the Rb/Cl ratio occur only below ~320 mbsf. The very slight enrichment of Rb⁺ immediately below the mudline may be an artifact derived from a temperature of squeezing effect as noted for K⁺ in interstitial waters from the first 50 m of deep-sea sediments (Bischoff et al., 1970). Because volcanic matter is absent from sediments of the GBB, its alteration cannot be invoked to account for the increase in the concentration of Rb⁺ downhole. A potential source of Rb⁺ that might lead to enrichment in interstitial fluids is detrital matter (K. Lackschewitz, pers. comm.). Mineralogical analyses conducted during Leg 166 indicate a substantial increase in detrital matter below 300 mbsf at Site 1006. It is also interesting to note that the onset of dissolved Rb⁺ enrichment coincides with the zone (300-420 mbsf) where dolomite was most abundant (Eberli, Swart, Malone et al., 1997). A submaximum in the dissolved Rb⁺ profile exists at 420 mbsf. Thus, it is tempting to contemplate exclusion of Rb⁺ from dolomite during carbonate recrystallization reactions to account for its enrichment. Such an input of Rb⁺ to pore fluids further downhole, however, is unlikely as the abundance of dolomite decreases substantially below 420 mbsf, whereas the concentration of dissolved Rb⁺ continues to increase. Because the abundance of detrital and clay minerals continues to increase and peaks just below 600 mbsf, it is more likely that this material is the source of the Rb⁺ enrichment observed here. The slightly lower dissolved Rb⁺ concentration observed in the two deepest samples supports this inference, because these samples also coincide with a lower abundance of detrital matter (Eberli, Swart, Malone, et al., 1997). Ostensibly Rb⁺ may be introduced to pore water by a mechanism similar to that invoked for Li⁺, i.e. ion exchange with NH₄⁺ in clay minerals.

Barium

Throughout the sediments of the GBB, dissolved Ba²⁺ concentration increases wherever SO₄^2- becomes depleted (e.g., von Breymann, et al., 1992). This antithetical relationship is expected because the concentration of dissolved Ba²⁺, much like that of Sr²⁺, is mediated largely by the solubility of its sulfate salt. The source of Ba²⁺ to the pore water, however, is likely different than that of Sr²⁺. The enrichment of dissolved Sr²⁺ in pore water from the GBB is caused primarily by its release from aragonite during early diagenesis and exclusion from the lattice during recrystallization of low magnesian calcite (e.g., Baker et al., 1982; Baker, 1986). The source of Ba²⁺ in sediments of the GBB, on the other hand, is more likely to be barite as well as detrital matter for the reasons explained below.

Barium concentrations can reach several thousand µg/g in carbonate-rich sediments (e.g., Boström and Backman, 1990; Schroeder et al., 1997) and sedimentary Ba has been used as a proxy for barite whose formation has been shown to be associated with biological productivity (Dehairs et al., 1980, 1987; Bishop, 1988; Francois et al., 1995; Paytan, 1995; Schroeder et al., 1997). However, Ba is abundant in many phases. Known associations exist between elevated Ba concentrations and hydrothermal mineralization (Bonatti et al., 1972), biogenic sediments/silica (Schmitz, 1987; Bishop, 1988), biogenic barite (Dehairs et al., 1987), and also terrigenous matter (De Carlo, 1992). Mixed sediments, however, can display highly variable concentrations of Ba. For example, De Carlo (1992) reported Ba contents of several hundred to a maximum of 9000 µg/g in hemipelagic sediments of the Exmouth Plateau containing variable amounts of siliciclastic matter. Greater Ba concentrations generally correlated with greater siliciclastic contents, although certain carbonate- and silica-rich intervals also displayed high concentrations of Ba. In sediments from the Japan Sea, von Breymann et al. reported solid-phase Ba concentrations ranging from <300 µg/g to as high as 12.8 mg/g. The latter, along with the 13.2 mg/g reported by Schroeder et al. (1997) in equatorial sediments from Site 850 probably represent the highest observed Ba concentrations in sediments recovered during the Ocean Drilling Program. These studies demonstrate the variable nature of the Ba content of marine sediments; thus, depending on the depositional environment, several sources can be invoked to account for the release of Ba to pore fluids under anoxic conditions.

Three sources of Ba appear likely in sediments from the GBB: (1) the crystal lattice of biogenic carbonates, (2) barite derived from remineralization of biogenic/organic matter, and (3) Ba in detrital/terrigenous matter that generally comprises no more than ~25% of the sediment. A measure of the lattice-bound Ba content of biogenic carbonates can be evaluated from the fraction of the sediments from Site 1007 that is soluble in dilute acetic acid (P. Kramer, unpubl. data). In the upper 400 mbsf of the sediments, this fraction generally contains <20 µg/g Ba. Below this depth, however, readily solubilized Ba concentrations increase substantially to 50 µg/g by 800 mbsf, where detrital minerals represent <10% of the sediment, peak around 140 µg/g near 900 mbsf, where detrital minerals constitute ~10%-15% of the sediment, and then return to <50 µg/g in the interval from 1000 to 1100 mbsf (P. Kramer, unpubl. data). A sharp increase to over 150 µg/g Ba is observed near the bottom of the hole where the abundance of detrital minerals is ~20%-25% of the total. Based on these data, relatively pure biogenic carbonates contain no more than 20 µg/g Ba. Dissolution and recrystallization of this material under anoxic conditions could provide Ba²⁺ to pore fluids, if an exclusion mechanism such as that described for the release of Sr²⁺ from biogenic carbonates, is invoked. However, other sources must be more important.

Because complete dissolutions of sediments from GBB sites were not performed, the total Ba content of the sediments is not known and it is not possible to estimate how much Ba is present in the form of barite derived from remineralization of marine organic matter or organic matter associated with the terrigenous matter hypothesized to originate from Hispaniola and Cuba. Thus, it is not useful to attempt a calculation analogous to that described above for Li⁺. There clearly is a greater propensity of detrital minerals than the lattice of carbonate minerals for providing the dissolved Ba²⁺ observed in pore water from the GBB sites. The observed 150 µg/g of releasable Ba from sediments deep in Site 1007 indicates that the easily solubilized component of these sediments can provide approximately seven times more Ba²⁺ to solution than can be derived from the lattice of carbonates. The covariance between weak-acid soluble Ba and Fe, and Mn to a lesser extent, in sediments from Site 1007 (P. Kramer, unpubl. data) implies that easily solubilized Fe and Mn containing minerals possibly associated with but much more reactive than other detrital minerals do comprise a significant source of Ba. Our observations are consistent with Schroeder et al. (1997), who found a strong correlation between Ba and both Fe and terrigenous accumulation in carbonate sediments of the deep equatorial Pacific Ocean, although the vast majority of Ba was associated with biological productivity. The covariance between dissolved concentrations of Ba and Fe and detrital mineral content observed here is much stronger than that for Li. At Site 1006, where sediments typically contain more detrital matter than at Sites 1005 and 1007, the absence of a correlation between dissolved Ba²⁺ and Fe is explained by the removal of the latter from solution and its sequestration in the sediments as pyrite, other metastable Fe-sulfides, or carbonate phases (Eberli, Swart, Malone, et al., 1997). Nonetheless, barite formed during remineralization of biogenic/organic matter, either in the water column or upon shallow burial diagenesis, likely remains the most important source of Ba to pore fluids of the GBB sites.

Manganese and Iron

Concentrations of dissolved Fe and Mn²⁺ (Table 1; Fig. 4, Fig. 5) are considerably lower than observed in many previously sampled deep-sea sediments (Gieskes, 1981; De Carlo, 1992; Mottl, 1992; Sigurdson, Leckie, Acton, et al., 1997, and references therein). The profiles from the GBB sites do not appear to closely follow the conventional redox sequence observed in shallow sediments associated with the degradation of organic matter. The subsurface enrichments of dissolved Fe and Mn²⁺ that are normally observed in pore fluids as more favorable terminal electron acceptors become depleted do not seem to be manifested strongly here. Only at Site 1006 is the subsurface maximum of Mn significant, yet it only reaches <2 µM. This is considerably lower than observed at many other DSDP and ODP sites, where dissolved Mn²⁺ concentrations of 20-80 µM are not uncommon.

The low concentrations of dissolved Mn²⁺ observed here compared to many other deep-sea sediments likely reflect a combination of the predominance of biogenic carbonate in the sediments of the GBB transect sites and rapid rates of microbial degradation of organic matter. The concentration of Mn²⁺ increases at each site, primarily deeper downhole, where noncarbonate minerals become more abundant. As expected from its redox chemistry, the intervals displaying higher concentrations of Mn²⁺ in these highly variable profiles occur principally where SO₄^2- becomes substantially depleted or in the anoxic portions of the holes (e.g., near 300 mbsf at Site 1105, below 200 mbsf at Site 1006, and below 450 mbsf at Site 1007). Even then, dissolved Mn²⁺ does not exceed 2 µM, except in the deepest sample analyzed from Site 1006.

Concentrations of dissolved Fe are also quite variable; they do not display clear subsurface maxima expected from a classical redox/electron acceptor sequence, yet show a general increase downhole. Throughout most of the sedimentary column, however, concentrations of Fe in the pore fluids remain approximately one order of magnitude lower than observed in many other deep-sea sediments (e.g., Gieskes, 1981; De Carlo, 1992; Sigurdson, Leckie, Acton, et al., 1997, and references therein). Dissolved Fe also increases substantially, as expected and in a manner similar to the behavior displayed by Mn²⁺, wherever SO₄^2- becomes strongly depleted. Notable exceptions include the first SO₄^2- minimum at Site 1005 (near 316 mbsf) and below 250 mbsf throughout most of Site 1006. In these intervals, Fe has been mostly sequestered into the sediments by precipitation as pyrite, other metastable sulfides, or incorporation into carbonate minerals (Eberli, Swart, Malone, et al., 1997).

Clearly, the low concentrations of Mn²⁺ and Fe in interstitial water from GBB transect sites reflect high rates of microbial oxidation of organic matter in a system characterized by a low abundance of easily reduced Fe- and Mn-rich phases that can act as a source of these elements to the pore water. Peaks in the weak-acid extractable Fe and Mn in sediments of Site 1007 (P. Kramer, unpubl. data) coincide reasonably with the intervals showing more elevated concentrations of dissolved Fe and Mn²⁺. The general correlation between the dissolved concentrations of these elements as a function of depth with the abundance of detrital minerals substantiates this inference.

Likely Fe-rich lateritic soils derived from the weathering of volcanics on Cuba and Hispaniola and carried by the northward currents are an important source of Fe and Mn. Their occurrence in alternating layers throughout the sediments of the GBB is consistent with changing sea level, possibly in a periodicity of low-amplitude Milankovitch cycles. Another potential source of detrital minerals, although of a more constant nature, is an aeolian input of African dust carried across the Atlantic (e.g., Duce et al., 1991; Prospero, 1996).

Fluctuations in sea level influence the delivery of biogenic carbonates to the GBB transect sites. Higher productivity on the Bahamas platform during sea-level highstands leads to a dilution of aeolian inputs by abundant biogenic carbonate. Conversely, the aeolian contribution becomes proportionally more important during sea-level lowstands, and the detrital fraction of the sediment from such intervals can provide a source of Fe and Mn to pore fluids during diagenesis. Aeolian dust has been reported to be more soluble than typical detrital matter and may release up to 10% of its Fe content because of an enhanced solubility imparted by chemical alteration during its atmospheric transport (Duce et al., 1991). This solubilization, however, is kinetically rapid and, whatever its extent, it is unlikely to be of great importance in pore waters whose composition has evolved over tens of thousands to millions of years.

In comparison with dissolved Ba²⁺, however, concentrations of Mn²⁺ and Fe in interstitial water remain much lower than might be predicted on the basis of their abundance in the readily solubilized minerals (and likely more so in the more refractory minerals) present in the sedimentary column. For example, readily solubilized Ba and Mn in sediments of Site 1007 vary within a similar range of a few to ~200 µg/g, whereas the Fe range is approximately one order of magnitude greater. The concentrations of dissolved Ba²⁺ fluctuate from 10 µM to nearly 80 µM in the anoxic sediments of Site 1007 (they are substantially higher, 10-228 µM, in anoxic sediments of Site 1006). Concentrations of dissolved Mn²⁺ and Fe, however, rarely exceed 2 and 10 µM, respectively. Although it is clear that the Ba-bearing minerals in the sediments are different (e.g., BaSO₄) and more readily release Ba²⁺ than those hosting Mn and Fe, the order of magnitude lower dissolved concentrations of the latter are more likely attributable to solubility controls by metal sulfides. Evidence in support of this inference is available, particularly at Site 1006, where common pyrite in the sediments exemplifies the many orders of magnitude lower solubility product of this compound compared to that of Mn sulfide. This is reflected by a near absence of dissolved interstitial Fe²⁺ even where abundant source material is present.