OBJECTIVE 2: STRATIGRAPHIC RESPONSE OF CARBONATES TO SEA-LEVEL CHANGES

In the icehouse world of the Neogene, waxing and waning of polar ice sheets have caused numerous high-amplitude sea-level changes. The ¹⁸O content of planktonic and benthic foraminifers has been shown to be an excellent proxy record for sea-level changes that are related to orbitally controlled climate cycles (e.g., Hays et al., 1976; Miller et al., 1987, 1991; Tiedemann et al., 1994). Carbonates can be an equally good recorder of high-frequency sea-level changes. Facies analysis enables water-depth estimates, usually to within 10 m. This helps determine paleobathymetry and, thus, amplitude of relative sea-level changes (Kendall and Schlager, 1981; Schlager, 1981). Shallowing-upward cycles in shallow-water carbonates have often been interpreted as the sedimentological record of successive high-frequency sea-level changes (Fischer, 1964; Grotzinger, 1986; Goldhammer et al., 1987; Strasser, 1988). Likewise, marl/limestone alternations have been considered the record of successive high-frequency sea-level changes on the slopes and in the basin surrounding shallow-water platforms (Fischer, 1991; Fischer et al., 1991). This comparison reveals an interesting discrepancy. Spectral analyses of the oxygen isotope data indicate that Earth's obliquity is the major control on glaciation and, therefore, on sea-level changes (Hays et al., 1976; Ruddiman et al., 1986; Raymo et al., 1990). Similar analyses of the sedimentary cycles, however, suggest that precession is the controlling orbital parameter (e.g., Hinnov and Goldhammer, 1991). This discrepancy is also present in the material from Leg 166 that allows a direct comparison of the ¹⁸O and the sedimentological record.

Sea-level changes of a lower frequency are recorded as unconformity-bounded depositional sequences at the continental margins. The carbonate environment records alterations in climate and relative change in sea level in its own characteristic way, resulting in a system-specific depositional sequence architecture (Sarg, 1988; Eberli and Ginsburg, 1989; Schlager, 1992). Flat-topped carbonate platforms and shelves produce and export more sediment during sea-level highstands. How important the highstand shedding is in controlling sequence architecture has been a matter of debate. In particular, the amount and architecture of carbonate turbidites in such settings has been controversial (e.g., Droxler and Schlager, 1985; Sarg, 1988; Schlager, 1992). The five sites along the Bahamas Transect together with the seismic data have produced a comprehensive data set that gives a better understanding of the sedimentary record of both high- and low-frequency sea-level changes.

Cores along the Bahamas Transect display the record of high-frequency sea-level changes along the entire depositional profile (Fig. 3). With the completion of the transect, it is now possible to document the lithologic expression of these sea-level changes in the different depositional environments. The precisely dated, continuous, and expanded section at the basinal Site 1006 for the first time provides a direct correlation between (1) the sedimentary record and the oxygen isotope record of high-frequency sea-level changes back to the Miocene and (McKenzie et al., 1999; Kroon et al., Chap. 2, and Rendle et al., Chap. 6, both this volume) (2) the seismic stratigraphic record of lower frequency sea-level changes.

On the banktop, in the cores of Unda and Clino, shallow-water carbonate packages are separated by subaerial exposure horizons (Kievman, 1996, 1998). These individual sediment packages generally do not show a shallowing-upward trend in their facies; instead, exposure is observed on a variety of platform facies. Eight exposure horizons are found in the top section back to the Brunhes/Matuyama boundary at 0.8 Ma, which seems to suggest that during each of the last eight high-amplitude sea-level fluctuations, a sedimentary unit was deposited on the platform top (Kievman, 1998). This finding provides a Quaternary analogue for studies that attribute rhythmic bedding of Mesozoic shallow-water carbonates to sea-level fluctuations in the Milankovitch frequency band (Fischer, 1991).

On the slopes and in the basins surrounding the Great Bahama Bank, aragonite cycles and turbidite composition are equally good indicators of high-frequency sea-level fluctuations (Droxler et al., 1983; Reijmer et al., 1988; Haddad et al., 1993). In particular, aragonite content has been shown to closely monitor orbitally induced climate changes. Some scientists interpret the variations as a result of increased dissolution (Droxler et al., 1983), while others consider these variations a result of fluctuations in neritic input to the off-bank areas during interglacial times (Kier and Pilkey, 1971; Boardman and Neumann, 1984; Reijmer et al., 1988; Milliman et al., 1993). Rendle et al. (Chap. 6, this volume), based on cores retrieved during Leg 166, favor the input interpretation based on the facts that (1) their investigated cores at Sites 1003 and 1006 are from water depths deemed too shallow for aragonite dissolution (658 and 481 m) and (2) banktop-derived aragonite needles are most abundant during the interglacials along the entire transect. The mineralogical assemblage during glacial times (identified by oxygen isotope signals) shows a decreased amount of aragonite of 48% on average, whereas high- and low-magnesium calcite, dolomite, quartz, and insolubles account for the rest of the assemblage (Rendle et al., Chap. 6, this volume). The fact that aragonite does not decrease further indicates that production of neritic components never ceased completely during glacial periods.

The comprehensive data set collected during Leg 166, consisting of high-recovery cores, precisely dated sections, and a continuous suite of logs, helps evaluate these sea-level controlled cycles throughout the Neogene. A typical unlithified cycle (Pleistocene-Pliocene) consists of a unit of aragonite-rich, neritic carbonates followed by an interval of aragonite-poor carbonates rich in pelagic foraminifers, nannofossils, and siliciclastics. Miocene cycles consist of decimeter- to meter-scale alternations between light gray, better cemented limestone and dark gray, less cemented, compacted marl/marlstone (Betzler et al., 1999; Frank and Bernet, in press). The platform-derived sediment in the cycles is interpreted as being produced during the highstand of sea level when the platform was flooded, whereas the increased amount of pelagic and siliciclastic sediments are interpreted as being deposited during sea-level lowstands. With increasing distance from the platform margin, these carbonate cycles evolve into typical marl/limestone alternations (Fig. 3). At the basinal sites of Leg 166 (1006 and 1007), pulses of aragonite-rich neritic sediments are separated by dark clay and quartz-rich layers that contain as much as 80% clay, quartz, organic material, and detrital dolomite in aragonite-poor nannofossil ooze. The highly decreased amounts of neritic material in these layers indicate deposition during times when the carbonate production on the bank was reduced, as during sea-level lowstands. Betzler et al. (1999) propose that the marly portion of the cycles could also form during the transgression and represent a condensed interval. In both interpretations, however, each marl/limestone alternation is believed to have formed during one high-frequency cycle of sea-level fall and rise. The amplitudes of these changes might be relatively small. On a flat-topped platform, a sea-level fall of 5-10 m would be sufficient to significantly reduce the production of neritic components.

The cycles can be recognized in the cores and on logs. In particular, resistivity and gamma logs can record these alternations continuously (Eberli, Swart, Malone, et al., 1997; Isern et al., in press; Williams and Pirmez, 1999; Bernet, 2000). The physical properties of the two intervals are different in the uncemented portions of the cores and alter with ongoing diagenesis (Isern et al., in press; Frank and Bernet, in press; Bernet, 2000). Petrographic and isotopic analyses clearly show that the original composition is controlling the lithification processes and burial diagenesis (Frank and Bernet, in press). The disparity in the diagenetic potential of the different layers (i.e., the neritic- vs. the pelagic-dominated intervals) controls subsequent diagenesis. Therefore, burial diagenesis enhances the initial sedimentary differences but is not responsible for creating marl/limestone alternations. The enhancement by diagenesis also increases the response in geophysical logs. For example, resistivity values are high in the carbonate-rich portion of the cycle and low in the siliciclastic portion, and gamma-ray signals record the marly intervals (Williams and Pirmez, 1999; Bernet, 2000).

The recognition of these alternations in logs and cores throughout the Neogene has major implications for two reasons. First, it shows that such marl/limestone alternations often found in the ancient rock record may be the result of changes in input of both the carbonates and the siliciclastics that are controlled by climate and sea-level changes (Bernet et al., 1998; Frank and Bernet, in press; Betzler et al., 1999). In addition, the uninterrupted recovery of these alternations from an unlithified to a lithified stage reveal for the first time their diagenetic behavior from early to burial diagenesis with pressure solution (Frank and Bernet, in press). Limestone beds are usually early cemented and little compacted. Their isotopic and petrographic characteristics are interpreted to reflect cement precipitation from cold seawater during the first ~100-200 m of burial. In the adjacent marlstones, diagenesis is inhibited because of higher proportions of insoluble materials, in spite of significant compaction and pressure solution during burial (Frank and Bernet, in press). Aragonite, for example, is still present in Miocene marlstones, whereas the limestones are completely altered to low-Mg calcite (Eberli, Swart, Malone, et al., 1997). This indicates that the diagenesis of the limestones and marlstones is not coeval and that limestones are formed before the marls undergo their alternations. Consequently, it is not likely that the formation of marl/limestone alternations is the result of a process in which carbonate dissolution in the marls would lead to precipitation in the limestones. With these results in hand, there seems to be less room to explain these alternations as produced solely by diagenesis (Ricken, 1986; Bathurst, 1987). Our results instead corroborated the findings of Diester-Haass (1991) that Neogene marl/limestone alternations result from diagenetically altered input variations produced by glacial-interglacial fluctuations in sea level, climate, and ocean circulation.

Second, the continuous record of the logs allows for a frequency analysis of the alternations and a comparison with the isotope record. Spectral analyses of resistivity and gamma-ray data of marl/limestone alternations from the Santaren Channel (Leg 166) show strong power spectra at 23 and 19 k.y. throughout most of the Miocene and Pliocene (Fig. 4; Bernet et al., 1998; Kroon et al., Chap. 15, this volume). This frequency is in concert with orbital precession but is in contrast to the obliquity frequency of glacial cycles recorded by stable isotope data (see below).

The recognition that fluctuating sea level divides the strata in genetically related units not only makes sequence stratigraphy a method to date sequence boundaries and global sea-level changes (see above) but also adds a predictive capability to sequence stratigraphy that is lacking in the other stratigraphic methods. By combining depositional models with time, sequence stratigraphy is capable of documenting the dynamics in a depositional system and, therefore, the distribution and architecture of facies belts through time (Vail et al., 1977; Sarg, 1988; Handford and Loucks, 1993). The sequence architecture in each depositional system is controlled by the sea-level change that acts with other variables such as climate, subsidence, and sediment supply. These parameters largely control the depositional character of sedimentary rocks (sediment and facies type, geometry, continuity, etc.) and thus the resulting sequence architecture. Carbonate environments, for example, record changes in climate and relative sea level in a characteristic way, resulting in a system-specific depositional sequence architecture (Sarg, 1988; Eberli and Ginsburg, 1989; Schlager, 1992; Handford and Loucks, 1993). The difference between the carbonate and siliciclastic systems arises because the flat-topped carbonate platforms and shelves can produce and thus export more sediment during sea-level highstands when they are flooded. Export of sediment during sea-level highstands (known as highstand shedding) places the carbonate environment 180° out of phase with the siliciclastic environment where most of the sediment is exported into deeper water during sea-level lowstands (known as lowstand shedding) (Schlager and Chermak, 1979; Mullins, 1983; Droxler and Schlager, 1985).

The degree to which these differences influence the facies distribution in carbonate systems has been a matter of debate (Mullins, 1983; Schlager, 1991). There is general agreement that carbonate depositional sequences, like their siliciclastic counterparts, are unconformity-bounded depositional packages. Indeed, every sequence boundary along the Bahamas Transect was identified by a seismic unconformity at or near the platform margin (Eberli et al., 1997; in press). In the cores of Leg 166, each one of these boundaries coincided with a change in facies in one or more holes (Eberli et al., 1997). The internal sequence architecture is more controversial. Many scientists argue that as a result of highstand shedding, carbonate sequences develop relatively thin sediment packages during sea-level lowstands (lowstand system tracts) with small or nonexisting turbidite fan systems (Schlager, 1991). Others believe that platforms are also eroded and these erosional products would form lowstand turbidite fans (Sarg, 1988; Jacquin et al., 1991).

The Bahamas Transect was designed to document the facies variations related to sea-level oscillations from shallow water to a water depth of ~600 m. This range allows a full assessment of the sedimentary response of carbonates to sea-level changes and, in particular, an assessment of the amount of highstand vs. lowstand shedding. Core data and seismic analyses now document a much more complicated picture than previously expected (Betzler et al., 1999; Bernet et al., Chap. 5, this volume). The distribution of turbidites within sedimentary sequences varies strongly. Generally, turbidites are clustered at the upper and/or lower portions of the sequences, indicating deposition of carbonate turbidites during both the highstand and lowstand of sea level and a suppression of turbidite sedimentation during transgressions. To complicate matters, highstand and lowstand turbidites seem to be deposited at different locations on the slope. Fifty percent more highstand turbidites (309 vs. 209) were deposited at the lower slope (Site 1003) than at the toe-of-slope (Site 1007), but twice as many lowstand turbidites were deposited at the toe-of-slope than at the lower slope (Bernet et al., Chap. 5, this volume). In addition, depocenters change from a lower slope position in the early and middle Miocene to a toe-of-slope position in the late Miocene and early Pliocene, probably as a result of the change in slope geometry from concave to more convex (Betzler et al., 1999).

Highstand and lowstand turbidites can be distinguished by slight differences in composition. In general, highstand turbidites contain abundant shallow-water allochems such as green algae, red algae, shallow-water benthic foraminifers (miliolids), and intraclasts. In lowstand turbidites, shallow-water allochems are often negligible but consist mostly of planktonic foraminifers and micrite. These compositional differences are similar to the well-documented Quaternary turbidites along the steep-sided Great Bahama Bank. Throughout most of the Neogene, however, the Great Bahama Bank had more of a ramplike geometry. This morphology resulted in reduced production of shallow-water sediment but not a complete shut-off. It also explains the abundance of lowstand turbidites in the older sequences (Fig. 5). Betzler et al. (1999) propose the following model for calciturbidite deposition during a third-order sea-level fluctuation. The lowering of sea level produces redeposition along the distally steepened ramp of the Great Bahama Bank. With progressive flooding, turbidite shedding is suppressed or reduced. During sea-level highstand, decreased accommodation space on the bank results in increased off-bank transport and renewed turbidite deposition.

Figure 5 shows the distribution of highstand sediments along the Bahamas transect. The lowstand and transgressive portion of the seismic sequences are grouped together into one unit (Bernet, 2000). The separation of highstand into lowstand/transgressive systems tracts is based on mineralogical criteria and gamma-ray peaks. The onset of increased aragonite and/or high-Mg calcite is taken as an indicator of the onset of platform top production (i.e., as a sea-level highstand). A gamma peak often coincides with this mineralogical transition, and this criterion is used in the older strata to separate the highstand systems tract from the lowstand and transgressive systems tracts, which are not separable in most sequences (Bernet, 2000). Using these criteria, the highstand portion of the sequences is approximately half of the volume of the sequences except in the youngest sequences, where it is significantly thicker. The thickening of the highstand systems tract is probably related to the transformation of the bank morphology from a steepened ramp into a steep-sided platform.

Calibration of cores along the complete transect makes it possible to define the facies from the proximal to the distal portion of the seismic sequences along the Bahamas Transect. The internal facies architecture of these carbonate depositional sequences displays five major elements (Fig. 6). In the undathem, or the platform top, the sediments are arranged in shallow-water packages separated by exposure horizons (Kievman, 1996; 1998). Very few of these packages display a shallowing-upward succession, although most of them are capped by an exposure horizon (Kievman, 1996). The uppermost slope at borehole Clino is characterized by thick sections of fine-grained, platform-derived material. Hardgrounds and firmgrounds associated with thin intervals of coarser grained skeletal packstone to grainstone indicate changes of sedimentation rates and sequence boundaries in this environment (Eberli et al., 1997; Kenter et al., in press). At Site 1005, the most proximal site drilled on the upper slope during Leg 166, a similar sedimentary succession consists of unlithified to partially lithified wackestones and slightly coarser grained intervals composed of packstones and grainstones. Compositional variations document an alternating pattern of bank flooding, concomitant shedding to the slope with periods of exposed banks, a shutdown of shallow-water carbonate production, and largely pelagic sedimentation (Eberli, Swart, Malone, et al., 1997). These pulses of sedimentation produce the prograding clinoforms seen on seismic data (Fig. 1). Few mass gravity flows are deposited at the proximal sites. Most of the platform-derived turbidites bypassed the upper slope. In this upper slope setting, incisions are common. They form a series of canyons perpendicular to the platform edge (Fig. 7A; Anselmetti et al., in press). Major incisions mark several seismic sequence boundaries; for example, at the Miocene/Pliocene and the early/late Pliocene boundaries (Fig. 7A).

On the middle to lower slope (Site 1003), sedimentation is more variable. Sedimentary units consist of either fine-grained, more neritic sediments; pelagic deposits; or turbidite successions. Channels and incisions are more abundant but of smaller dimension. The variability in facies and small-scale channeling is reflected in the seismic facies that shows a characteristic, discontinuous seismic reflection pattern (Fig. 7B; Anselmetti et al., in press). Seismic sequence boundaries correlate with breaks in the pulses of progradation.

The toe-of-slope (Site 1007) is the main depositional location for redeposited carbonates that accumulate during both sea-level highstands and lowstands (Bernet et al., Chap. 5, this volume; Betzler et al., 1999). The turbidite lenses are discontinuous within the marl/limestone alternations that form the background sedimentation. Seismic data that were calibrated with the core data reveal that the turbidite lenses are arranged in mounded lobes that partly coalesce (Betzler et al., 1999; Anselmetti et al., in press). Internal geometry of some of the lobes can be interpreted as a feeder channel. Canyon incisions on the upper and middle slope could act as a point source for some of these channels (Fig. 7A).

The distal portion of the sequences is dominated by cyclic marl/limestone alternations with few turbidites. In addition, drift deposits dominate the basinal sediments from the late middle Miocene onward (12.4 Ma) at the Bahamas Transect.