We used 5-cm3 samples for phosphorus determinations and 10-cm3 samples for barite extractions at continuous 2-cm intervals from 153.40 to 154.8 mbsf, which corresponds to the PETM (Section 199-1221C-11H-3) recovered at Leg 199 Site 1221. The 10-cm3 samples were washed through a 63-µm sieve, and both size fractions were retained. The >63-µm fraction was oven-dried and analyzed (by other researchers) for foraminiferal isotopes. We used the <63-µm fraction for barite separations. Individual fine fraction sample weights ranged from 5 to 10 g. Because sample weights were much smaller than those typically used for barite separations (e.g., 20–30 g) (Paytan et al., 1996; Eagle et al., 2003), a subset of initial samples were combined to represent 4-cm intervals (50–54, 54–58, 58–62, 62–66, and 66–70 cm) to ensure that enough barite would be extracted for analysis. However, barite yields were so high that combining samples was deemed unnecessary and was not done for the remainder of the samples.
We separated barite from sediment samples using a sequential leaching procedure that included reaction with hydrochloric acid (6 N), warm (50°C) sodium hypochlorite (5 wt%), warm (80°C) hydroxylamine (0.02 N) in acetic acid (0.05 M), and 1:2, 1:1, and 2:1 hydrofluoric acid (40 wt%):nitric acid (1 N) mixtures (Paytan et al., 1993; Eagle et al., 2003). This procedure dissolves carbonates, oxidizes organic matter, removes transition metal oxyhydroxides, dissolves siliceous material, and removes fluorides in order to isolate barite. After weighing the sample separation residues, we used a back-scattered electron imaging detector mounted on a scanning electron microscope (SEM) and the EDAX Image/Mapping program to determine the percent barite in the residues and to examine barite crystal morphology. Most sample residues in this study were found to be ~100% barite upon examination. This procedure is detailed in Eagle et al. (2003).
We used a four-step, operationally defined sequential P extraction (Anderson and Delaney, 2000) modified from a five-step P extraction procedure (SEDEX) (Ruttenberg, 1992) to determine P concentrations in four sedimentary components: oxide-associated P (includes P sorbed to and incorporated in oxyhydroxides), authigenic P (authigenic carbonate fluoroapatite), organic P (acid insoluble P), and detrital P (terrestrial silicates and detrital apatite). An additional first step of water-soluble P was extracted for randomly chosen samples (Table T1) to determine whether there is a contribution of water-soluble P to total P in deep-sea sedimentary samples. Water-soluble P is a significant source of P in sediment trap samples (Faul et al., in press). However, since most water-soluble P is typically remobilized before reaching the sediments, this fraction is insignificant for deep-sea sedimentary samples and will not be discussed further.
We determined P concentrations on splits of the 5-cm3 bulk samples. After freeze-drying the samples, we crushed the samples and passed them through a 150-µm sieve to ensure uniformity in particle size. After extracting the four components of P into known volumes of extractant from replicate samples (~0.1 g), we used a Lachat Quick Chem 8000 automated spectrophotometric flow injection analysis system to measure P concentrations. Results are reported as the means ±1s (sample standard deviation) of duplicate runs. Reactive P is calculated as the sum of water-soluble P (when appropriate), oxide-associated P, authigenic P, and organic P, with errors propagated from errors on individual components. Total P is the sum of reactive P and detrital P.
The long-term analytical reproducibility was assessed by calculation of the mean P concentrations ±1s for two sedimentary consistency standards (Table T2). One of the two was processed and measured with each run as a sample. Consistency standard relative errors are similar to those from other sedimentary P studies (e.g., Filippelli and Delaney, 1995, 1996; Anderson and Delaney, 2000). The relative errors of the long-term means of the consistency standards were high for components that were close to detection limit and constituted small fractions of total P for the standards. Relative fractions in these consistency standards are very similar to those of the samples analyzed here. Typical sample concentrations well exceeded the detection limits by a minimum factor of 10 for oxide-associated P and up to a factor of 350 for authigenic P (Table T2).
CaCO3-free concentrations of reactive P, detrital P, barite, and shipboard Mn were calculated using CaCO3 values generated by Murphy et al. (this volume). Organic C to reactive P values were calculated using CaCO3-free organic C (Murphy et al., this volume) and CaCO3-free reactive P.