DISCUSSION

The ¹⁸O records for both sites along the leeward side of the Great Bahama Bank show that there have been numerous periods during the past 1.4 Ma where changes in sea level have occurred in response to climatic variability. In turn, the way in which the Bahama Platform has responded to these changes has been recorded in the mineralogy and grain-size distribution of the sediment. The ¹⁸O record for the distal hole (Hole 1006A) has shown that there has also been a general climatic warming during the late Pleistocene (after isotope Stage 11). This is indicated by the lightest ¹⁸O values that were interpreted as surface-water warming by Kroon et al. (Chap. 2, this volume). However, at the more proximal hole (Hole 1003A), this warming trend is more difficult to observe because of missing sequences in the stratigraphic record and an isotopically "heavy" record, which indicates diagenetic overprinting below the Holocene (>12 mbsf). This is marked by a 3 shift in the oxygen isotope values. A possible reason for this shift could be diagenetic overprinting caused by early diagenesis. This is supported by evidence found in the mineralogy, discussed later. Further discussion concerning the isotope curve at the distal site (Site 1006) in terms of its paleoceanographic and climatologic significance can be found in the paper by Kroon et al. (Chap. 2, this volume).

It has been noted that the aragonite cycles at Site 1006 display the same sawtooth pattern as the ¹⁸O record, while at Site 1003 only the aragonite curve shows this pattern. However, the reason for this aragonite pattern is still under debate since the Bahama Bank has a flat-topped morphology; thus the aragonite variations should resemble a more "square wave" pattern (Droxler et al., 1983; Haddad and Droxler, 1996). It has been suggested that these unexpected differences between the predicted and actual aragonite patterns could be the result of seafloor aragonite dissolution (Droxler et al., 1983; Droxler, 1985). The idea for dissolution was first put forward by Lynts et al. (1973) as a potentially important factor controlling the periplatform carbonate mineralogy of the Great Bahama Bank. However, the two cores at Holes 1006A and 1003A were recovered at 658 and 481 mbsl respectively, and are, therefore, too shallow to adequately discuss possible changes that could result from seafloor aragonite dissolution. This seems to support the observation by Boardman and Neumann (1984) that stated that carbonate dissolution would not occur at shallow water depths (e.g., 200-2000 mbsl; Providence Channel, Bahamas) and that the aragonite content is solely an input signal. Within the periplatform ooze (Schlager and James, 1978), aragonite was found to be the most abundant mineral on the leeward side of the Great Bahama Bank. As known from literature, its main sources are neritic, bank-top aragonite needles and pelagic-derived pteropods (e.g. Neumann and Land, 1975; Droxler et al., 1983; Boardman and Neumann, 1984, 1986; Boardman et al., 1986; Macintyre and Reid, 1992; Milliman et al., 1993). However, a small decrease in the aragonite content of these periplatform sediments was found to occur with increasing distance and water depth from the platform top (Figs. 6A, 6D). Boardman and Neumann (1984) found a similar pattern for Northwest Providence Channel. These observations were also made for sediments from the Bahama Channels (Droxler et al., 1988; Reijmer et al., 1988; Haddad and Droxler, 1996) and the Bermuda Pedestal (Berner et al., 1976), as well as for the Nicaragua Rise (Droxler et al., 1991). If the decrease observed is solely the result of input variations or if minor diagenesis might have modified the signal remains to be seen (deMol et al., 1998). We found that, although this general off-bank decrease occurred, the general fluctuations in aragonite concentrations, coupled with the ¹⁸O records, indicate that aragonite is in its highest concentrations at both sites during sea-level highstands and in its lowest concentrations during sea-level lowstands (Fig. 6). This difference related to sea level also occurs in a spacial context. It seems that sediment deposited during sea-level highstands does not vary spatially between Holes 1003A and 1006A, but it is the lowstand concentrations that vary (Fig. 6). Therefore, with increasing proximity to the shallow-water platform realm, the difference between the highstand and lowstand aragonite concentrations is less pronounced (Fig. 6). For example, at the distal site (Hole 1006A), there was a 77% increase in aragonite concentration during sea-level highstands, whereas this difference was only 50% at the more proximal site (Hole 1003A). Thus, the decrease in the aragonite content of these periplatform sediments with increasing distance from the platform top seems to be controlled by changes within the aragonite content of the sediment deposited during sea-level lowstands. These results further support the idea that the aragonite signal is an input signal at these shallow depths and not a dissolution modified signal.

In the literature, it is described that HMC input occurs preferentially during interglacials and shows a positive correlation to aragonite (Boardman et al., 1986; Glaser and Droxler, 1993; Emery, 1996; Haddad and Droxler, 1996) or a negative correlation (Droxler, 1985, 1986). At both sites along the platform slope, HMC is in highest concentrations during glacials (e.g., 5% at Hole 1006A and 13% at Hole 1003A; Fig. 6) and, therefore, runs reverse to the aragonite input pattern. Grammer et al. (1993a, 1993b) demonstrated that, where botryoidal aragonite cements form on steep marginal slopes (Tongue of Ocean, Bahamas), discontinuity horizons exist that consist of thin layers of magnesium-calcite micrite. Similar observations were made at the Belize Barrier Reef (Ginsburg and James, 1976). If these then form part of the eroded slope debris that is deposited during glacials (see "Glacials" and "Grain Size" sections), this could account for the higher HMC concentrations in these intervals. In addition, as observed at Site 1005 (deMol et al., 1998), at CLINO and UNDA (Westphal, 1997; Westphal et al., 1999), and at the deep seafloor of Tongue of the Ocean (Schlager and James, 1978), HMC cements are formed during early diagenesis that would also increase its relative concentration, especially at the more proximal hole (Hole 1003A), where early diagenesis has been implied by the isotope curve. This would also account for the proximal site having higher HMC concentrations when compared to the distal site (i.e., a spacial discontinuity), whereby HMC decreases in concentration with increasing distance from the platform (Fig. 6). A final reason for this spacial discontinuity is that the more proximal site is closer to the source area of the magnesium-calcite micrite.

Along the leeward flanks of the Great Bahama Banks, LMC is found to be in highest concentrations during sea-level lowstands (e.g., 29% at Hole 1006A, and 41% at Hole 1003A; Fig. 6). The highstand shedding theory states that during sea-level lowstands there is relatively little neritic carbonate production and low aragonite content (Schlager et al., 1994; Droxler and Schlager, 1985; Haddad and Droxler, 1996). Thus, the sediments are dominated by pelagic input such as pelagic foraminifers and coccoliths that are high in LMC (Johnson, 1961; Scholle et al., 1983; Tucker and Wright, 1990). In addition, the lowstand deposits are dominated by coarse-grained, eroded, slope debris (see "Glacials" and "Grain Size" sections) which could contain lithified rock (i.e., calcite cements). During sea-level highstands, the LMC signal is diluted by the high influx of aragonite-rich sediment. However, this neritic input is drastically reduced during lowstands allowing the LMC signal to be relatively enriched during these periods. The spacial discontinuity in the concentration of LMC in the sediment shows that LMC increases preferentially with increasing distance from the platform, showing a negative correlation to the HMC (Fig. 6). For example, the average LMC concentration at Hole 1006A equates to 26%, compared to only 20.5% seen at Hole 1003A (Fig. 6). This is probably connected to the pelagic input, which increases with increasing distance from the platform. This would also account for the difference between the interglacial and glacial concentrations at the distal site being greater than that seen at the proximal site (Fig. 6).

Dolomite, as a whole, shows very little relationship to the other carbonate minerals. However it does show some trends in terms of glacial/interglacial interplay and spacial variations. At both sites, the first appearance of dolomite is at 17 mbsf, which implies that its presence could be depth dependent; however, that is where the similarity ends. It appears during sea-level lowstands at Hole 1006A, thereby showing a positive correlation to LMC, HMC (Fig. 6), and also to quartz, which appears during these periods. Because quartz represents terrigenous material, this suggests that the dolomite could be of detrital origin. This was also seen in Hole 632A of ODP Leg 101 on the windward side of the Great Bahama Bank (Reijmer et al., 1988). In contrast, at Hole 1003A, dolomite appears in almost every sample from isotope Stage 9 down (with the exception of the upper part of isotope Stage 11 (11.1)), which suggests that its formation is caused by early diagenesis (Mullins et al., 1985). Droxler et al. (1988) and Reijmer et al. (1988) observed a similar pattern. Quartz at the proximal site only appears at two intervals that correspond to sea-level lowstands (isotope Stages 2 and 12).

Sediments from different environments tend to have distinctive grain-size characteristics. The ultimate characteristic of the sediment grain-size distribution is a function of several interacting processes, beginning with growth/production and subsequent erosion in the source area. Thus, the ultimate size distribution of grains in the site of deposition is a function of (1) availability of grains of different sizes at the source, (2) transport and depositional processes, and (3) postdepositional diagenetic changes (Boggs, 1987). To date, the majority of research into grain-size analysis has been based in siliciclastic environments. In carbonate environments, grain size usually only classifies the sediment broadly using terms which combine grain type and depositional texture (e.g., boundstone, grainstone, packstone, etc. [Dunham, 1962]). Therefore, we have attempted to look at the grain-size distribution to provide further insight into its importance as a proxy to aid in understanding the carbonate sedimentary environment. Although this may be relatively premature, we find that it is possible to use grain-size characteristics as "fingerprints" of certain environments. In the carbonate regime, the highest, most variable off-bank sediment transport occurs during interglacials (sea-level highstands), as shown by numerous authors like Mullins (1983), Droxler and Schlager (1985), Reijmer et al. (1988), and Eberli, Swart, Malone, et al. (1997), whereas evidence for low sedimentation rates during glacials was shown by Kendall and Schlager (1981) and Handford and Loucks (1993). Emery and Myers (1996) stated that in both siliciclastic and carbonate environments this variability in sedimentation rate between interglacials and glacials is probably caused by gravity-controlled sedimentation, rather than by a uniform, continuous rain of pelagic sediment. The grain size results will show that this is true for the glacials, but not the interglacials.

Although we find that both sites on the leeward side of the Great Bahama Bank are dominated by the fine fraction (<63 µm), there is some spacial variation (Figs. 9A, 9D). Furthermore, the difference in the percentage of coarse to fine sediment is greater at the distal site than at the proximal site, which indicates that less coarse-grained sediment is deposited at the distal site, thus confirming dominance by the fine fraction. These spacial variations could be related to the site locations in relation to their position on the leeward side of the Great Bahama Bank.

On windward margins of carbonate platforms, wave activity tends to push most of the fine sediment onto the platform, leaving the coarser fraction to accumulate on the windward side (e.g. Emery and Myers, 1996). Hine et al. (1981a, 1981b) and Tucker (1985) found that accumulation of the finer sediment tends to occur on the leeward margins. The large-scale westward progradation of the Great Bahama Bank as shown by Eberli and Ginsburg (1989) and Eberli, Swart, Malone, et al. (1997) demonstrates this sedimentary process. Similar observations were made for the Devonian Iberg Reef (Gischler, 1995), Pedro Bank (northern Nicaragua Rise; Glaser and Droxler, 1993) and the Queensland Plateau (northeastern Australia; Betzler et al., 1995). The fine-fraction dominance at both sites (Figs. 9A, 9D) confirms these observations. The proximal site lies on the slope, which forms a progradational wedge, whereas the distal site is located more basinward in the Santaren Channel. This progradational phase of the platform described by Eberli and Ginsburg (1987), Wilber et al. (1990), and Eberli et al. (1994) was found to contain a high concentration of fine (<63 µm) aragonite needles, which, although originating from the platform top, can nevertheless be deposited well into the basin (Wilber et al., 1990; Droxler et al., 1991; Glaser and Droxler, 1993; Milliman et al., 1993). In other words, the sediment grain sizes supplied to the slope are dependent on the orientation of the slope with respect to the wind regime. Depending on their velocity and their position within the basin, surface and bottom currents might modify this input signal (Mullins et al., 1980).

Work carried out by Kenter (1990) researched the relationship between slope angle and dominant sediment fabric. The resultant model showed that coarse-grained sediment could maintain much higher angles of repose and hence steeper slopes than fine-grained deposits. Thus, at Site 1003 where the slope angle is low (= 3.5°), fine-grained mud would be the representative sediment type. This environment is also characterized by reorganization of the slope deposits that could explain the observed hiatuses seen at this proximal site. The slightly marked increase in the average percentage concentration of fine fraction (Fig. 9A) and the decrease in slope angle at the distal site could be the result of higher pelagic input of fine-grained material compared to the steeper slope and coarser grained deposits seen at the proximal site (Fig. 9D). The high percentage of aragonite of these deposits, however, points to a more platform-related input pattern. The dominance of the fine fraction at the basinal site could result from (1) resedimentation of fine-grained sediment originally deposited at the slope, (2) the fine end-tails of platform-derived turbidites, or (3) result from normal off-bank transport (Wilson and Roberts, 1995).

During the interglacials or sea-level highstands, the proportions of fine (<63 µm) to coarse (>63 µm) sediment seen at both sites (Holes 1006A and 1003A) are very similar (Figs. 9B, 9E). This is probably caused by large export of aragonite-rich muds from the platform top by "density cascading," whereby waters are released from the platform together with entrained sediments that later fall out of suspension as pelagic rain to form "pelagic draping" of the slope and basin (Wilson and Roberts, 1995). In addition, even though the coarse fraction forms the smallest percentage concentration of the sediment (Figs. 9B, 9E), normal sorting occurs within this fraction so that with increases in the grain size there is a decrease in percentage concentration of each consecutive grain-size subdivision (Figs. 12B, 12E). This indicates that all grain sizes are being eroded from the platform top and transported downslope to be deposited across the entire platform-to-basin transect during sea-level highstands and that sorting increases with increasing distance from the source. Therefore, the differences seen in the average coarse/fine grain-size distribution (Figs. 9A, 9D) and within the coarse-fraction subdivisions for the two sites (Figs. 12A, 12D) is solely effected by the volume of the coarse fraction (>63 µm). In addition, Westphal (1997) found that fine-grained, highstand slope sediments from the Great Bahama Bank (the CLINO Core) were protected against large initial fluid-flow alterations because of their low permeability, and thus were subjected to less diagenetic alteration. Therefore, diagenesis within these deposits, as well as being affected on a vertical scale (i.e., through time), is also affected spatially (if to a lesser extent) so that with increasing distance from the platform, and thus decreasing grain size, the diagenetic potential of the sediment is reduced.

When considering the glacial sediments deposited during sea-level lowstands, it was observed that the coarse fraction (>63 µm) constituted 38% of the sediment at the distal site and 46% of the sediment at the proximal site (Figs. 9C, 9F). Thus, the concentration of the coarse fraction is of greater significance at both sites during sea-level lowstands. Mullins et al. (1980) showed that maximum bottom-current strengths in the Straits of Florida correlate with glacials. They propose that the system responds to compressed worldwide climatic belts and increased wind regimes, which intensify ocean circulation during these times. The pattern in the grain-size distributions observed in our cores is in full agreement with these observations. Glaser and Droxler (1993) found a similar pattern for the Walton Basin north of Pedro Bank. However, there is a change in the coarse/fine proportions across the platform during glacials that was not so prominent during the interglacials. This decrease in coarse fraction from the proximal to distal site during the glacials is caused by the types, frequencies, and strength of the mechanisms that transport the sediment away from the platform. Normally, with increasing distance from the source, these gravity-induced transport mechanisms lose energy and are unable to support the larger particles in suspension, thus depositing them nearer to the platform (e.g. Boardman and Neumann, 1984; Boggs, 1987). This would account for the reduced concentration of both the bulk coarse fraction (Fig. 9C) and the larger subfractions (Fig. 12C) at the distal site, which is supported by the reduced input of turbidites at this site (Eberli, Swart, Malone, et al., 1997). Another important point is that carbonate systems do not experience the same type of physical erosion as seen in siliciclastic environments. For example, subaerial exposure of the carbonate platform top will tend to cause chemical erosion, resulting in grain and cement dissolution and reprecipitation of dissolved material (Emery and Myers, 1996). Westphal (1997) found that these coarse glacial deposits, with their higher permeabilities, were subjected to intense diagenetic alteration. They also contained high HMC cements which, together with the eroded slope deposits containing magnesium-calcite micrite (Grammer et al., 1993a, 1993b), would explain the high HMC concentrations during sea-level lowstands. Diagenesis would once again be effected spatially in the same way as during interglacials (i.e., the diagenetic potential of the sediments would be reduced [not to the same degree] with distance from the platform.).

The facies distribution pattern of the Great Bahama Bank of Purdy (1963) clearly showed the grain-size differences across the platform top. Relatively coarse-grained sediments are found at the platform edge, both on the windward and leeward side, whereas mud-size sediment dominates the platform interior facies. Thus, flooding and exposure of the platform will introduce variability in the grain sizes that can be produced and subsequently exported. Dravis (1996) clearly demonstrated that fresh-water calcite cementation will rapidly freeze platform interior deposits, while Grammer et al. (1993b) showed that rapid lithification also occurs at the steep marginal slopes of Great Bahama Bank. Sea-level fluctuations will, therefore, automatically cause variations in the availability of certain grain types for export (as demonstrated for the Bahamas by Haak and Schlager, 1989) and the amount of certain grain sizes within the system (highstand shedding, Schlager et al., 1994; turbidite bundling, Droxler and Schlager, 1985, and Reijmer et al., 1988). Therefore, the way in which the different grain sizes are precisely distributed along the platform-to-basin transect from Site 1005 to Site 1006 during glacials and interglacials, forms the topic of ongoing research.

In addition, the composition of the coarse grains (>63 µm) exported from the platform during glacials and interglacials form the topic of further work in progress. However, first results confirm to a large extent the compositional variations within slope deposits of the Great Bahama Bank presented by Westphal et al. (1999) and within turbidites deposited in the Tongue of the Ocean (Haak and Schlager, 1989). The interglacial sediments are dominated by planktonic and benthic foraminifers and pteropods. The foraminifers present include Globigerinina, Miliolina, and, to a lesser extent, Textulariina, Lagenina, and Rotaliina, while the pteropods represented lie within the suborder Eutheocosomata. Also present are minor amounts of Echinoderm and Bryozoan fragments and some cortoids. In contrast, the glacial sediments are characterized by cortoids, whereby the bioclasts, peloids, and other particles are coated with a relatively thin micrite envelope. The cortoid grains consist typically of peloids, various planktonic foraminifers such as Globigerinina and Miliolina, and subangular quartz grains. These preliminary findings in compositional variations seen in interglacial and glacial sediments form the basis for a more detailed analysis, which is still in progress.

Kroon et al. (Chap. 2, this volume) observed a distinct difference in the isotope stratigraphy for the upper (>isotope Stage 11) and lower (= isotope Stage 11) part of Site 1006 and attributed these changes to increased sea-surface temperatures. The carbonate mineralogy and the grain-size distribution also seem to register this paleoceanographic change. The aragonite values show reduced values for Stages 9 and 11. In the upper-core section, the HMC percentages show a marked increase while the LMC percentages show a decrease. The grain-size distribution shows a sharp increase in the input of coarse (>63 µm) sediment since Stage 11. However, a more detailed analysis is needed to clarify the origin of these changes in greater detail.