Samples (~0.5 g dry weight) were freeze dried and extracted in a 3:1 (v/v) methanol:hexane solution using an automated pressurized fluid extraction device (ASE 200, Dionex, Inc., Salt Lake City). This instrument exposes samples to small volumes of solvent at elevated pressure and temperature; it extracts organic compounds rapidly and delivers the extracts to annealed glass vials with polytetra-fluoroethylene (PTFE) septa. Extractions were carried out for 15 min at 150°C. Organic extracts were dried under a nitrogen stream before dilution for gas chromatography (GC). Although we generally avoid silica gel column treatment of extracts in our procedure, the complex makeup of Cariaco Basin lipid extracts forced us to use a short-column procedure. The column consisted of a Pasteur pipette packed with glass wool at the base. A silica gel column of ~3.75 cm was rinsed with methanol, methylene chloride, and hexane before loading organic extracts. Next, 50 µL of concentrated sample was transferred onto the column and eluted with 5.5 mL of methylene chloride.
Alkenone analyses employed gas chromatography (HP 5890) with flame-ionization detection. Samples were prepared using toluene as a solvent, and autosampler vials were sealed with PTFE septa to prevent the appearance of ghost peaks observed with rubber or silicone septa. The chromatographic column used was a J&W DB-1 (60-m length, 0.32-mm diameter, and 0.11-µm film thickness), with a 10-m deactivated silica guard column on the injection side of the GC. Temperature programming was 12.5°C/min to 250°C followed by 1°C/min to 290°C, with an isothermal holding step at 320°C for 15 min. Calculation of the Uk´37 index followed graphic determination of C37:3 and C37:2 peak areas using Hewlett-Packard Chemstation software. Quantification of abundance was achieved by adding C36 and C37 n-alkanes as external standards.
Paleotemperature estimates come from the Prahl et al. (1988) calibration of unsaturation index to growth temperature. We routinely achieved a reproducibility of better than ± 0.01 Uk´37 units, as determined by multiple extractions of sediment standards from Santa Barbara Basin and the central California Margin. This corresponds to a nominal temperature uncertainty of <0.3°C using the Prahl et al. (1988) calibration, barring other complications in interpreting the unsaturation index. For several cases of very low alkenone concentrations (see Table 1), we re-extracted larger (2 g) quantities of sediment to ensure that the low alkenone concentration did not bias our gas chromatographic results. Reanalyses of these leaner samples yielded unsaturation indices identical to the initial values determined.