ANALYTICAL METHODS AND RESULTS

The coulometrics titration technique measures all of the CO₂ that is liberated by acidifying and heating sediment samples in a closed system and back-titrating to a coulometric end point. To determine carbonate carbon concentrations, ~10 mg of sample was reacted with a 50% phosphoric acid solution in a heated reaction vessel. The percentage of carbonate in each sample reflects the amount of inorganic carbon (IC) liberated as CO₂ with the assumption that all inorganic carbon is present as calcium carbonate, such that CaCO₃ = IC × 8.332. No corrections were made for the presence of either siderite or dolomite. Total carbon (TC) concentrations reflect the amount of carbon released as CO₂ during combustion of a ~5-mg sample in oxygen at 1000°C. The amount of organic carbon (C_org) was calculated as the difference between TC and IC: C_org (wt%) = TC (wt%) - IC (wt%). Reported silt/clay concentrations were calculated as the amount of insoluble residue during acidification that cannot be attributed to organic carbon: Silt/clay (wt%) = 100 - [CaCO₃ (wt%) + C_org (wt%)]. The precision of these acidification and combustion methods is better than 0.5% and was monitored by multiple analyses of a pure carbonate laboratory standard. The precision of the C_org determinations is defined by the combined precision of the TC and IC methods and is generally not better than 1.5%-2%.

With minor exceptions, carbonate contents at Site 1006 fall within a narrow range between 80 and 90 wt% (Fig. 1). Within the upper and middle Miocene sections at Site 1007, carbonate contents generally range between 80 and 100 wt% (Fig. 1). Contents within the lower Miocene section at this site are more variable and range between ~60 and 100 wt% (Fig. 1). Within the section at Site 1007 are occasional 2- to 5-cm-thick intervals characterized by exceptionally low carbonate contents (as low as 18 wt%; Eberli, Swart, Malone, et al., 1997) and, in some cases, concentrations of organic carbon that exceed 1.0 wt%. Variations in carbonate, clay/silt, and organic carbon concentrations correspond to decimeter- to meter-scale cyclic color changes that characterize the Miocene sections recovered during Leg 166 (Eberli, Swart, Malone, et al., 1997). In general, lighter intervals have higher carbonate and lower siliciclastic and organic carbon contents, whereas darker intervals are characterized by lower carbonate and higher clay/silt and organic carbon concentrations. Such compositional contrasts between light and dark intervals are more pronounced at Site 1007 than at Site 1006.

Quantitative carbonate mineralogy was determined by X-ray diffraction using the methods outlined in Eberli, Swart, Malone, et al. (1997). Analytical precision is within 5% of the actual weight percent, with a standard deviation of ~3%. The carbonate mineralogy at Sites 1006 and 1007 is dominated by LMC with lesser amounts of aragonite and dolomite (Fig. 2). Aragonite contents at Site 1006 are relatively uniform and range from 15 to 35 wt% between 450 and 550 mbsf and to <15 wt% at deeper depths. At Site 1007, aragonite contents in the Miocene section are generally <10 wt%, with spikes as high as 40 wt%. No aragonite was detected below ~1000 mbsf at Site 1007. Elevated aragonite contents at Site 1007 correspond to lower carbonate contents and increased concentrations of acid-insoluble residues (e.g., clays and silts). Aragonite concentrations at Site 1006 are relatively uniform and show little correlation with clay, silt, and organic matter contents. Dolomite contents at Site 1006 are consistently below ~5 wt%. At Site 1007, dolomite contents range to slightly higher concentrations (average 10 wt%) with occasional spikes as high as 40 wt%.

Concentrations of Ca, Ba, Fe, Mg, Mn, and Sr in the carbonate fraction of selected samples over a 100-m interval at Site 1006 and throughout the core at Site 1007 were determined using inductively coupled plasma-atomic emission spectrometry. In preparation for analysis, ~25 mg of sample was reacted in a buffered (pH = 4.4), ~6% acetic acid solution. The insoluble, noncarbonate fraction was subsequently removed by filtration and dried overnight in a 50°C oven. The amount of carbonate in each sample was calculated as the difference between the dry weights of the original sample and the filtered, insoluble fraction. To avoid any uncertainties associated with weighing small samples, trace metal (TM) concentrations described below and listed in Tables 1 and 2 are reported relative to the major element, Ca: (TM/Ca)´1000. Analytical precision, determined using gravimetric standards, was ± 2% for Ca, ± 3% for Mg, Sr and Ba, ±4% for Mn and ±5% for Fe.

Sr concentrations at Site 1006 average 14.3 ± 1.7 (Table 1). Sr contents are generally lower at Site 1007 (average = 6.7 ± 3.8), but exhibit an increase to concentrations as high as 15.2 in the interval spanning ~500-800 mbsf (Fig. 3). Mg concentrations at Sites 1006 and 1007 are virtually identical and, on average, range between 10 and 11 (Table 1; Fig. 3). Ba concentrations at Site 1006 are uniform and average 0.7 ± 0.2 (Table 1). Concentrations of Ba at Site 1007 generally range between 1 and 1.5, but exhibit two downcore increases (Fig. 3). Between 500 and 600 mbsf, contents increase to 4, and between ~1050 and 1200 mbsf, concentrations increase to as high as 5.4 (Fig. 3). Fe and Mn concentrations at Site 1006 are quite variable and, on average, slightly higher than those at Site 1007 (Table 1).

Carbon and oxygen isotope analyses were performed in stable isotope facilities at the University of Michigan (UM) and Pennsylvania State University (PSU). At UM, samples were reacted with anhydrous phosphoric acid at 73°C in individual reaction vessels of an online, automated Kiel-type device coupled to a Finnigan-MAT 251 mass spectrometer; whereas at PSU, samples were reacted at 90°C in an automated carbonate device (common acid bath) coupled to a Finnigan-MAT 252 mass spectrometer. Duplicate sample analyses reveal no laboratory-dependent trends in measured compositions. Carbon and oxygen isotope ratios are reported in parts per thousand () relative to the Vienna Peedee belemnite standard (Table 1, Table 2). Precision is better than 0.05 for ¹³C and ⁸O values and was monitored through multiple analyses of National Bureau of Standards 19 and other powdered calcite standards. A comparison of isotopic data from upper Miocene sections at Sites 1006 and 1007 reveals that whereas ¹³C values from both sites are virtually indistinguishable, ¹⁸O values of whole-rock samples from Site 1006 are generally more negative than values of whole-rock samples of equivalent age from Site 1007 (Fig. 4).