Bulk sediments from Sites 1170-1172 were analyzed for their carbonate content and stable isotope composition in an Analytical Precision AP2003 mass spectrometer. Benthic isotope measurements were performed on ~10 specimens of Cibicidoides mundulus (in very rare cases on a mix of Cibicidoides bradyi and Cibicidoides kullenbergi) from the >212-µm size range. This genus is generally accepted as living within the sediment surface (0-1 cm), although C. bradyi might prefer a slightly more infaunal habitat. Analyses were performed using standard procedures for Prism mass spectrometers. Precision is better than ±0.06
for 
13C, ±0.10 for 18O, and ±2% for carbonate. All stable isotope results were calibrated to the international standard Vienna Pedee belemnite through standard laboratory carbonates including correction for vital effects (Shackleton and Hall, 1997). A number of outlying measurements were subjected to repeat analysis. All measurements were performed at the Godwin Laboratory, Cambridge University.
Fine-fraction carbon content (Ctotal) was measured by CHN analysis in a Carlo Erba EA1106 at the Godwin Laboratory, Cambridge University, UK. Additional measurements of total carbon and inorganic carbon content (TIC) were performed in an Elementar VarioMAX at the Max Planck Institute for Biogeochemistry, Jena, Germany. Samples for inorganic carbon content were combusted at ~400°C for 16 hr, as combustion of only 4 hr did not remove all nitrogen. In order to use both measurements of Ctotal we calculated a linear regression based on 32 parallel measurements from various intervals at Site 1168, arriving at a regression coefficient of R2 = 0.925. All CHN measurements were subsequently converted to VarioMAX-equivalent values of Ctotal (y = 0.9229x + 0.2674), where y = VarioMAX-equivalent Ctotal and x = CHN value of total C. Total organic carbon (TOC) was approximated by
Four to eight C. mundulus specimens from the >150-µm size fraction were analyzed in VG Optima or Prism gas-source mass spectrometers at the University of California, Santa Cruz. All values are reported in permil notation relative to Vienna Peedee belemnite standard, with 18O measurements including a correction for vital effects (+0.5) (Shackleton and Hall, 1997). Analytical precision for both 13C and 18O averaged better than 0.1. No interlaboratory corrections were applied, as these are usually constant and should not affect our interpretations; however, we do concede that these offsets exist.
Faunal and lithogenic reconnaissance is based on random order inspection of the >63-µm fraction at 75- to 300-cm resolution. We distinguished between
the same scheme as applied in shipboard benthic foraminiferal studies at 9-m resolution (Exon, Kennett, Malone, et al., 2001).
