In this study, we combined seismic-sequence stratigraphic information with the analysis of the calcareous turbidites to examine the relationship between turbidite frequency and sea-level fluctuations. This methodology differs from other studies in that the separation of sea-level highstand and lowstand is based strongly on the geometry seen on the seismic section and not on the indirect evidence such as grain composition.
Seismic-sequence boundaries were used to separate the cores into highstand and lowstand systems tracts. The depth of the seismic-sequence boundary was calculated through the time/depth conversion from the vertical seismic profile shot at each site. This conversion has a resolution of 5-10 m maximum. The exact position of the sequence boundary in the core is determined by sedimentologic criteria within the interval given by the seismic resolution (Eberli, Swart, Malone, et al., 1997). In this study, the transgressive systems tract is included in the lowstand portion of the sequence. This can be justified by the fact that generally the platform was only flooded during highstands. To place the maximum flooding, or the turn-around from lowstand to highstand system tracts within each sequence, we relied on changes in texture together with mineralogy (low-Mg calcite [LMC], high-Mg calcite [HMC], aragonite, and quartz) and standard gamma-ray logs. This separation was based on the following evidence:
Supko (1963) and Boardman (1978) found that aragonite with a high strontium content originates on the bank top (from green algae, ooids, and inorganic precipitation), whereas pteropods show a low-strontium variety and represent planktonic aragonite. Droxler et al. (1983) document that 90% of the Bahamian aragonite in fine fraction is derived from a shallow-water source. Similarly, several authors assume a shallow-water source for HMC in the Bahamian troughs (Kier and Pilkey, 1971; Boardman, 1978; Droxler et al., 1983). Therefore, we assume that sections with high aragonite and HMC content, which are currently produced on the platform top, are indicative of deposits of high sea level when the platform is flooded.
In contrast, sections with increased LMC content and siliciclastic material probably represent lowstand system tracts for the following reasons: (1) Several authors showed that coccoliths and globigerinids, which predominantly occur in the open ocean, can account for nearly all the LMC (Kier and Pilkey, 1971; Boardman 1978; Droxler et al., 1983); (2) further, high contents of quartz, which appear whenever aragonite contents are low (Droxler et al., 1983), are likely to represent sea-level lowstand when the platform production is shut down and no aragonite or HMC are produced on the bank and exported into the basin. We acknowledge that diagenetic alteration of the mineralogic composition of the sediments may contribute to the difficulty of applying the mineralogic criteria but found no inconsistent behavior in our cores (Table 1, Table 2).
A second criterion for separating lowstand from highstand systems tracts is taken from the gamma-ray log. Standard gamma-ray logs record the presence of potassium, thorium, and uranium present in potassium feldspars, micas, glauconite, and phosphate (Asquith and Gibson, 1982). Glauconite and phosphate are diagenetic minerals that precipitate from sea water in deeper parts of open platforms or nonrimmed margins (Scoffin, 1987). Uranium and potassium are present in green algae, certain corals (zoantharias and calcitic octocorals), pelecypods, gastropods, and arthropods found on hardground surfaces (Milliman et al., 1974). Hence, the natural gamma radiation increases at hardgrounds. Kenter et al. (in press) showed that in the proximal part of the transect at Unda and Clino, each sea-level cycle has an associated change in gamma radiation, and that the gamma-ray excursions also represent maximum flooding surfaces. The standard gamma-ray log can be correlated to the texture and lithology of the sediment and the mineralogy. Sequence k at Site 1003 (Fig. 2), for example, documents that the texture of the carbonate sediments changes up-section from bioclastic wackestone to packstone, to peloidal packstone to grainstone with bioclasts, and is capped by a peloidal grainstone to rudstone with bioclasts. At the first increase in texture, the aragonite values decrease to zero and the amount of quartz increases dramatically. At the same depth, the standard gamma-ray log shows the highest peak underlain by two smaller but also dominant peaks. Based on a sudden appearance of peloids following the quartz peak and the low value of aragonite (Betzler et al., in press), we assume that the high excursion in gamma radiation that often coincides with clay layers represents the maximum flooding surface within the sequence. In such cases, the gamma-ray peaks are used to separate lowstand from highstand systems tracts. The separation of sea-level highstand vs. lowstand packages based on the above criteria enables us to assess the distribution of the turbidite composition and frequency within the individual sequences.
The core recovery at Sites 1003 and 1007 was 55% and 73%, respectively. To reveal the total amount of turbidites and their thickness, we calibrated the sedimentary description to the log data (Fig. 3). Carbonate turbidites generally consist of coarser material than the sedimentary composition of the background sediments and, therefore, show differences in resistivity values. The Formation MicroScanner (FMS) records resistivity where highly resistive intervals are displayed in bright colors, and layers with low resistivity values are shown in dark colors. Thus, FMS images have been powerful in distinguishing turbidite deposits from background sediments (Williams and Pirmez, in press).
In high-recovery cores, we characterized the resistivity pattern of the FMS log and compared it to the turbidite deposits and surrounding sediment. Changes in texture in normally or inversely graded turbidite deposits were recognized as subtle color changes from bright to dark colors or from dark to bright colors. The strongest color contrast is represented by the sudden change from fine-grained background sediment to a coarse-grained turbidite layer. The resolution of the resistivity tool is, however, not high enough to record other sedimentary structures. For example, in the cores, the sediments showed parallel laminations that were rather difficult to recognize in the FMS images. This may also be related to Mullins' (1983) and Eberli's (1991) observation that complete Bouma sequences may or may not be present and that there appears to be a lateral segregation of top-cut-out turbidites in more proximal positions (Section 166-1003B-56X-1, 12-32 cm) and base-cut-out turbidites in more distal regions (Section 166-1007C-43R-3, 85-110 cm). In any rate, the FMS could be used to identify turbidite deposits that were not recorded in low-recovery sections. Nevertheless, we estimate that a certain number of turbidites was not detected by the FMS logs. One of the factors of "miss-measuring" the thickness may be related to the difficulties of recognizing the finer grained turbidites in the FMS images, because it is rather difficult to separate the top of a fining-upward turbidite sequence from the fine background sediment. Knowing these difficulties, we have to accept that the thickness and the actual number of turbidites in the sedimentary section exceeds the counted number and measured thickness of tabulated turbidites (confirmed by C. Betzler, pers. comm., 1998). We are confident, though, that the relative proportions of both the thickness and number of turbidites at the two sites are true, since the same methods were used for both sites.
We counted all the carbonate turbidites at Sites 1003 and 1007 and measured their thicknesses using the core description and/or the FMS log. The frequency of turbidite shedding was calculated in events per meter section and events per sedimentary sequence.
A total number of 241 samples were taken throughout the cores from Sites 1003 and 1007. The samples represent different events of turbidite shedding and background sediments in the Miocene and early Pliocene. We concentrated our compositional analyses on the Miocene. In each thin section, 300 points were counted to determine the compositional differences of turbidites and background sediments and to describe the compositional differences of highstand and lowstand deposits (Table 3, Table 4). The term "background sediment" includes three out of four facies that characterize the sediment along the Bahamas Transect and represent initial sediment that has not been redeposited, such as periplatform ooze, peloidal wackestone to packstone, and siliciclastic sediments. Grains were counted once or more, according to their size (volumetrical counting). The quantities of the characteristic carbonate grains have been analyzed and were assigned to point-count groups that characterize the different depositional settings of the platform margin, (i.e., lagoon, reef complex, and basin). Nonskeletal grains and embedding sediment were counted in separate groups (Table 5).