METHODS

Sediment samples were analyzed at a frequency of ~1.5 m from Sites 1008 and 1009 and ~10 m from Site 1006. At all three sites, selected lithified horizons were also sampled and analyzed. Before analyses, each sample was examined and classified for the relative degree of lithification. Because the sediments in the present study are not deep-sea oozes, and to be consistent with shipboard descriptions (Eberli, Swart, Malone, et al., 1997), the nongenetic descriptors—unlithified, partially lithified, and lithified—are used rather than ooze, chalk, and limestone. All 317 samples were subjected to the same cleaning and analytical procedures. The outer edge of the sample was scraped away to avoid any contamination obtained during sampling, and then approximately 1 g of bulk sediment was rinsed twice in deionized water, centrifuged and decanted, and dried overnight at 60°C. Lithified samples were crushed prior to rinsing.

A portion of each sample was analyzed by powder X-ray diffraction (XRD) using CuK radiation on a Rigaku D-Max 111V-B X-ray diffractometer equipped with a graphite monochromator. Samples were ground in acetone, then smear-mounted onto glass plates and step-scanned from 20°-80° 2, collecting data every 0.03° 2 at 2 s/step. Quantitative proportions of aragonite, high-Mg calcite (HMC), low-Mg calcite (LMC), and dolomite (normalized to 100% carbonate) were determined by Rietveld refinement of XRD patterns (Rietveld, 1969; Post and Bish, 1989; Bish and Post, 1993). Reported accuracy of the method for carbonate minerals is better than ±3% (Bish and Post, 1993; Reid et al., 1992). Replicate analyses indicate that the precision is better than 1% when the phase is present in quantities >~40 wt%. Precision subsequently decreases with decreasing weight percent. In addition to the chemical analyses described below, Mg content of HMC and dolomite was determined from the d_{10.4} shift using the idealized curve of Goldsmith and Graf (1958) after correcting for specimen displacement by Rietveld refinement. Precision determined from replicates is ± 0.3 mol% Mg.

Each sample was analyzed for stable oxygen and carbon isotopic ratios. Approximately 120 mg of powdered sample was reacted in "100%" phosphoric acid at 70°C in an online, automated Kiel device coupled to a Finnigan MAT 251 stable isotope-ratio mass spectrometer. The carbonate standard NBS-19 (¹³C = 1.95, ¹⁸O = -2.20) was used to calibrate to the Peedee belemnite (PDB) standard. Repeated analyses of NBS-19 yielded reproducibility of better than 0.1 for ¹⁸O and ¹³C (N = 34).

For major and minor elemental compositions, ~50 mg of each sample was leached for 30 min in 25 ml of 1M acetic acid buffered with 1M ammonium acetate (pH of ~5). The buffered acetic acid was chosen to minimize contamination from noncarbonate phases. The leachate was centrifuged, decanted, and stored in HDPE bottles for analyses. After appropriate dilution, the solutions were analyzed for Ca, Mg, Sr, and Na by flame atomic absorption spectroscopy using a Perkin-Elmer Model 603 spectrophotometer. Standardization was achieved with SPEX plasma grade standards, coupled with the following internal check standards: reagent grade calcium carbonate, NBS-1C, and two previously well-characterized periplatform carbonate sediment samples from the Maldives (Malone et al., 1990; Malone, unpubl. data). Replicate analyses of samples yielded the following mean relative error for the entire procedure: <1% for mol% CaCO₃, 3% for mol% MgCO₃, 3% for Na, and 2% for Sr.