DISCUSSION

Variations in carbonate content on the Bermuda Rise can be correlated with fluctuations found in the ¹⁸O of ice in cores from Greenland (Adkins et al., 1997). Variations in grayscale at Site 1063 can be correlated similarly (Keigwin, Rio, Acton, et al., 1998). The ¹⁸O features in the Greenland ice cores have been dated (Dansgaard et al., 1993). Therefore, a preliminary time scale was assigned to the sediment interval between 25.0 and 29.5 mbsf by aligning variations in grayscale at Hole 1063D with changes in the ¹⁸O record at the Greenland Ice-core Project (GRIP) core (Fig. F11). Actual ages of the sediment were calculated by spline interpolation between control points using the software package Analyseries (Paillard et al., 1996). No time lag between the carbonate record at Site 1063 and the ¹⁸O record of the GRIP ice core was incorporated into the age model. According to this model, the carbonate abundance peaks at 25.8, 26.8, and 28 mbsf correspond to interstadials 19, 20, and 21, respectively (Dansgaard et al., 1993). Thus, as suggested by Keigwin, Rio, Acton et al., (1998), changes in sediment physical properties, especially bulk density, occur during D-O cycles.

The sediment record of D-O cycles at Site 1063 does not reflect simple fluctuations in the amount of carbonate and aluminosilicates, however. The densities of calcite and all common aluminosilicate minerals are between 2.6 and 2.8 g/cm³. Thus, even though calcite and aluminosilicate content are changing across D-O cycles, the observed density variations exceeding 0.3 g/cm³ indicate additional complexity. This complexity arises because late Quaternary sediment on the Bermuda Rise is a three-component system comprised of calcite, aluminosilicate minerals, and biogenic silica, and the relative abundance of all three phases varies between stadials and interstadials. Moreover, the relative abundance of various aluminosilicate minerals is also changing across D-O cycles.

The calcite fraction of Bermuda Rise sediment is mostly comprised of tests of foraminifers and coccolithophorids. Theoretically, the fluctuations in calcite could arise from increased terrigenous dilution or enhanced calcite dissolution. Keigwin and Jones (1994) suggested that terrigenous dilution is the major control on the formation of D-O cycles in Bermuda Rise sediment, although they acknowledged some dissolution during isotope Stage 4. The work presented here does not address this issue directly.

The aluminosilicate mineralogy at Site 1063 is comprised of large quantities of illite with subordinate amounts of kaolinite, chlorite, and muscovite. This result is consistent with findings of previous workers in the western North Atlantic (Ericson et al., 1961; Laine and Hollister, 1981; Keigwin and Jones, 1989). Most of the aluminosilicate minerals probably originate in eastern Canada (e.g., Laine and Hollister, 1981). However, the elemental ratios determined here suggest subtle changes in the aluminosilicate composition over D-O cycles, independent of dilution by carbonate and biogenic silica. In particular, there is an increase in the K₂O/Al₂O₃ ratio during stadials (Fig. F5F). This increase is possibly caused by an increase in the muscovite/illite ratio.

Silva et al. (1976) and Baker (1986) have suggested that low bulk density in glacial-age sediments reflects abundant sand-sized biogenic silica. The reason is that biogenic silica lowers grain density and increases sediment porosity (and thus water content). Such an interpretation is consistent with results presented here for Site 1063.

High concentrations of sand-sized biogenic silica might also be expected to result in a high P-wave velocity (e.g., Hamilton and Bachman, 1982). The reason is that coarse, angular grains in contact form a rigid framework conducive to the transmission of sound waves. The correspondence of biogenic silica and P-wave velocity is consistent with this interpretation. Thus, D-O cycles on the Bermuda Rise are partly the result of fluctuations in biogenic silica. The combination of low density and high compressional-wave velocity has been attributed to the presence of abundant biogenic silica by other workers (e.g., Weber et al., 1997).