MATERIAL AND METHODS

Core Material and Sampling

During ODP Leg 178, a nearly complete upper Miocene to Quaternary sequence was recovered from Holes 1095A and 1095B on the distal flank of Drift 7, whereas a complete lower Pliocene to Quaternary sequence was recovered from Holes 1096A, 1096B, and 1096C near the crest of this drift (Fig. F1; Table T1). A complete upper Pliocene to Quaternary sequence was recovered from Hole 1101A at the center of the smaller Drift 4 in the northeast (Fig. F1; Table T1). At all three sites, the recovered sediments generally consist of thin-bedded, bioturbated to massive diatom-bearing muds with common foraminifers, alternating with thick-bedded, lithogenic, mainly fine-grained laminites (Barker, Camerlenghi, Acton, et al., 1999). The intervals with higher biogenic contents were interpreted as hemipelagites deposited during interglacials, whereas the lithogenic laminites were described as distal turbidites and contourites deposited during glacials. Throughout the Neogene and Quaternary, all sediments were deposited in a glaciomarine environment as indicated by the presence of ice-rafted debris (IRD). At distal Site 1095, turbidites are more abundant than at the drift-crest Sites 1096 and 1101. According to smear slide analyses, the siliceous microfossil assemblages at all three sites predominantly consist of diatoms with abundant sponge spicules, rare silicoflagellates, and rare radiolarians (Barker, Camerlenghi, Acton, et al., 1999).

Opal contents were analyzed on 863 samples from sediment cores recovered at ODP Leg 178 Sites 1095, 1096, and 1101 (Table T1). The sample spacings correspond to time intervals of 31 k.y. for Site 1095, 11 k.y. for Site 1096, and 22 k.y. for Site 1101. At Site 1095, the samples were taken from Holes 1095A and 1095B, and at Site 1096, the samples were taken from Holes 1096A, 1096B, and 1096C. Thus, for these sites, depths in meters below seafloor (mbsf) were recovery corrected and converted to meters composite depths (mcd) using the corrections given in Barker, Camerlenghi, Acton, et al. (1999). At Site 1101, samples were only taken from one hole (Hole 1101A), and depths are therefore given as recovery-corrected meters below seafloor (rmbsf). The samples used in this study represent splits from 10-cm3 samples taken on board for determinations of CaCO3 and total organic carbon (TOC) contents (Wolf-Welling et al., Chap. 15, this volume).

Contents and accumulation rates of biogenic opal, TOC, CaCO3, and biogenic barium (Babio) from sediments deposited at gravity core site PS1565 are shown to compare various palaeoproductivity proxies. Site PS1565 is located at the seaward termination of Drift 3 (Fig. F1). The data presented for this sediment core were previously published in Hillenbrand (2000). All raw data presented in this study are available from the PANGAEA data bank of the Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, Germany (http://www.pangaea.de).

Analytical Procedure

We analyzed the opal-A content of 20-mg homogenized dry bulk sediment samples using an automated leaching technique after Müller and Schneider (1993), which was shown to be reliable in an interlaboratory comparison (Conley, 1998). Silicon was extracted by wet-chemical means with 1-M NaOH, and concentrations of dissolved silicon were measured by molybdate-blue spectrophotometry. The proportion of biogenic silicon was determined by graphical analysis of the absorbance vs. time plot (DeMaster, 1981). Resulting silicon values were converted into weight percent biogenic SiO2 and are given as weight percent opal by assuming a uniform bound water content of 10 wt% within the opaline substance. According to Müller and Schneider (1993), the relative accuracy of the method is better than 2% for samples rich in biogenic silica and 4%-10% for samples with <10 wt% biogenic silica, respectively. To estimate the precision of our laboratory array we analyzed three sediment samples repeatedly, yielding a relative accuracy between 2% and 5% with no systematic variation (Table T2). For the method used, Bonn et al. (1998) assumed an artificial background value of 1-2 wt% "pseudo-opal," caused by a partial leaching of clay minerals. In contrast, we measured opal contents down to 0.4 wt% on discrete samples, suggesting a negligible background opal signal. Therefore, no correction for possible background values was made.

Stratigraphy

The age models for Sites 1095, 1096, and 1101 were computed by linear interpolation between magnetostratigraphic datums given in Barker, Camerlenghi, Acton, et al. (1999), using the timescale of Berggren et al. (1995). Our estimated linear sedimentation rates differ slightly from the sedimentation rates given by Barker, Camerlenghi, Acton, et al. (1999) because we referred the paleomagnetic ages to recovery-corrected composite depths and recovery-corrected depths below seafloor, respectively. The paleomagnetic age model for Site 1096 was refined by including the Reunion Event (Chron 2r.1n), which was identified in Core 178-1096C-2H (G. Acton, pers. comm., 1999). At Site 1095, we additionally considered a possible hiatus within the upper Pliocene sedimentary sequence (Barker, Camerlenghi, Acton, et al., 1999). The paleomagnetic data indicate that the hiatus spans the earliest Matuyama Chron (C2r) and the onset of the Olduvai Event (C2n). Assuming nondeposition or erosion for the corresponding time between 2.581 and 1.770 ka, we calculated a linear sedimentation rate of 5.0 cm/k.y. for the late Gauss Chron (C2An.1n). This value agrees well with the sedimentation rate of 4.8 cm/k.y. calculated for the sediments deposited during the late Matuyama Chron. There is biostratigraphic evidence for another hiatus at Site 1095, spanning the time interval between 6.140 and 5.040 ka (M. Iwai, pers. comm., 2001). We did not consider this possible hiatus in the calculation of the age model for Site 1095, but we considered nondeposition or erosion during that time interval in the discussion section. For a better accentuation of long-term trends, we additionally resampled the data sets of opal at Sites 1095, 1096, and 1101 by linear integration at equal time increments of 200 k.y., applying the AnalySeries software (Paillard et al., 1996).

Calculation of Opal Accumulation Rates

Opal accumulation rates (MARopal, in grams per square meter per year) were calculated by multiplying linear sedimentation rates (LSRs) with the proportions of biogenic silica and dry bulk densities measured on board ship (Barker, Camerlenghi, Acton, et al., 1999). For this purpose, the data sets of opal and dry density were resampled by linear integration at equal time increments, applying the AnalySeries software (Paillard et al., 1996). For Site 1095, time increments of 20 k.y. were used for the interval from 9.580 to 2.580 ka and increments of 50 k.y. were chosen for the interval from 1.750 to 0 ka. Equal time increments of 10 and 20 k.y. were used for Site 1096 and Site 1101, respectively. The chosen time increments correspond to the average temporal resolution of the opal data sets.

NEXT