METHODS

All samples were freeze dried, weighed, shaken 2 hr in a 4% Calgon solution, and then washed with deionized water through a 64-µm sieve; the coarse fraction was also dried at 50°C and weighed. Average weight percentages of the >64-µm fraction were generally low, averaging 6 wt% at Site 1014 and 1.5 wt% at Site 1020. Benthic foraminifers were picked for isotopic analysis from the >150-µm fraction. At Site 1014, specimens of Cibicidoides wuellerstorfi were very rare, so other Cibicidoides species were picked, including C. pachyderma and C. mckannai as well as Planulina ariminensis. Separate species of benthic foraminifers were analyzed at Site 1014 at 18 different levels to estimate possible interspecific ¹⁸O and ¹³C offsets. Interspecific ¹⁸O differences were negligible, and no systematic ¹³C offsets were observed between C. wuellerstorfi, C. pachyderma, C. mckannai, and P. ariminensis. It should be noted, however, that both C. mckannai and C. pachyderma ¹³C values were typically ~0.1-0.2 lighter than C. wuellerstorfi and P. ariminensis, but the data density was insufficient to reliably quantify systematic interspecific ¹³C offsets. Specimens of C. wuellerstorfi were common at Site 1020, and these were exclusively picked for isotopic analysis at this site. An average of 12 specimens was picked for analysis for each sample, although some samples yielded as few as 3-5 specimens, which were also analyzed.

Samples from Site 1014 were analyzed on the Woods Hole Oceanographic Institution Finnegan/MAT 252 mass spectrometer with the Kiel automated carbonate preparation device. Samples were not roasted before analysis. The external precision of the isotopic analyses was ±0.03 for ¹³C and ±0.07 for ¹⁸O, based on more than 1200 measurements of NBS-19 standard. Samples from Site 1020 were analyzed on the new Lamont-Doherty Earth Observatory (LDEO) VG Optima mass spectrometer equipped with the automated multiprep carbonate preparation device that, like the Kiel device, also acidifies samples individually. The external precision for this instrument was ±0.05 for ¹³C and ±0.06 for ¹⁸O, based on numerous analyses of NBS-19.