METHODS

Coring and Stratigraphy

Details of the coring procedures, the use of planktonic foraminifers and calcareous nannofossil datums and magnetostratigraphic polarity zone boundaries to obtain sedimentation rates, and calculations of the bulk accumulation rates may be found in Sigurdsson, Leckie, Acton, et al. (1997).

The age-to-depth and mass accumulation rates for each sample were calculated using interpolated ages for each core interval. Although our records cover approximately the last 80 m.y., in some cases they are not complete. At Site 1001, middle Miocene nannofossil ooze overlies middle Eocene chalk, and the middle Eocene chalk overlies early Eocene chalk, corresponding to hiatuses of 30 and 8 m.y., respectively (Sigurdsson, Leckie, Acton, et al., 1997).

Normative Calculations of Sediment Composition

The absolute concentration of the terrigenous component was obtained from normative calculations based on the concentration of Cr in a given sample and the concentration of Cr in average shale (Table 1; Cr 110 ppm; Taylor and McLennan, 1985) according to

(%Terrigenous)sample = 100 x (Crsample) / (Craverage shale).

This calculation compares the composition of a given sample to that of an "average shale." In Sigurdsson, Leckie, Acton, et al. (1997), the terrigenous component at Site 998 was calculated using a model-based Ti/Zr value. Although there are no significant differences in results from these two methods, the terrigenous component was recalculated here using the Cr-based normative approach for consistency. We use the value of Cr for average shale based on the Post Archean Average Shale (PAAS) of Taylor and McLennan (1985). Although this is a somewhat arbitrary choice, it is important to note that other average shales (e.g., North American Shale Composite with Cr 125 ppm, Gromet et al., 1984; Proterozoic and Phanerozoic cratonic shales with Cr 110 ppm, Condie, 1993; Continental Crust with Cr = 119 ppm, Rudnick and Fountain, 1995) have similar concentrations of Cr. In further support of our selection of this particular Cr concentration, we note that Dobson et al. (1997) determined the chemical composition of operationally derived (i.e., using a sequential extraction procedure) terrigenous matter at the Ceara Rise, thought to represent sediment transported down the Amazon River, and the average of their determinations yields Cr 100 ppm. Thus, the results described below are not overly dependent on the choice of any particular average shale approximation.

Chromium was selected as the reference element because a large majority of the discrete ash layers contain <3 ppm Cr (Sigurdsson, Leckie, Acton, et al., 1997), compared to 110 ppm in average shale (Taylor and McLennan, 1985). Thus, given these end-member values, even in a sediment sample with a large amount of dispersed ash and a small terrigenous load, the use of Cr to quantitatively determine the terrigenous abundance is appropriate. Furthermore, as a refractory element, Cr is unlikely to be affected by diagenetic remobilization (Taylor and McLennan, 1985). Our use of Cr assumes that the original chemistry of the dispersed ash was similar to that of the discrete ash layers, at least with respect to Cr. The use of major elements such as Al or Ti to calculate the terrigenous load leads to erroneous results because these elements are found in large concentrations in both average shale (Taylor and McLennan, 1985) and Caribbean ash (Sigurdsson, Leckie, Acton, et al., 1997). Similarly, the use of other trace elements found in relatively low abundance in ash is precluded. For example, Ni was not chosen because the range of Ni concentrations in the ash layers is 30-90 ppm even in relatively unaltered ash (those with low MgO concentrations; Sigurdsson, Leckie, Acton, et al., 1997) and the concentration in average shale is 54 ppm (Taylor and McLennan, 1985). Vanadium was not chosen for the same reason; the V concentration in the ash (1-50 ppm) is not significantly lower than that of average shale (100 ppm).

Dispersed ash was calculated by difference according to

(%Ash)sample = 100 - %CaCO3 - %Terrigenous.

This yields a maximum estimate for the dispersed ash because the calculation could potentially include oxide and biogenic opal components in the remaining portion, as well as sea salt. However, Fe and Mn oxides and biogenic opal components only account for a maximum of 7% of the bulk sediment deposited here (Sigurdsson, Leckie, Acton, et al., 1997). Because such additional components may have stratigraphically local importance, care is taken to not overly rely on single data points.

Our normative calculations are in close agreement with petrographic analysis of the sediment. For example, at certain stratigraphic horizons where petrographic analysis indicates only the presence of terrigenous matter and CaCO3, our normative calculations yield numerical closure. Our calculations also agree with the qualitative smear-slide descriptions of the lithology (Sigurdsson, Leckie, Acton, et al., 1997). We estimate the uncertainty in the normative calculation of dispersed ash to be ~5%-10% of the measured weight percent value. We arrive at this estimate because of the ~3 ppm Cr in ash compared to the 110 ppm Cr in PAAS (3/110 3%), as well as analytical precision of the Cr analysis (<2% of the measured value), the potential for variation in the detrital component, and when taking into consideration that other components are not necessarily included in the "by difference" calculation. Given the arithmetic, the realistic "detection limit" of the normative approach is also ~5 wt% dispersed ash. All data are given in Table 1.

Chemical Analyses

Shipboard analysis of selected major and trace elements was performed by X-ray fluorescence (XRF). For Site 998, analysis was done on two samples per core for the first 150 meters below seafloor (mbsf) and then one per core below. At Site 999, samples were analyzed at a frequency of one per core, and at Site 1001, one sample every three cores. Samples were selected from intervals that appeared representative (so called "background" sediment) and had no obvious marker beds. Details of the shipboard analytical procedures may be found in Sigurdsson, Leckie, Acton, et al. (1997). Eighty-seven samples (~30 per site) were selected for reanalysis by inductively coupled plasma source emission spectrometry (ICP-ES) at Boston University to compare the shipboard XRF-generated data with ICP-ES. The ICP-ES analytical procedures were similar to those used by Schroeder et al. (1997), Murray and Leinen (1993, 1996), and Murray et al. (1995). Samples targeted for replicate analysis were chosen to bracket a range of Cr, Ti, and CaCO3 concentrations. The shore-based ICP-ES analyses confirmed the shipboard XRF data determinations. Although shipboard XRF quality control was maintained at the highest possible level, including blind replicate analyses, the replicate analyses by the different technique gives us increased confidence in the analytical results.

NEXT