METHODS

To achieve a target sample resolution of ~2 k.y., the Site 1146 splice was sampled every 50 cm from 0 to ~57 mcd, followed by every 20 cm from ~58 to ~83 mcd, every 15 cm from ~84 to ~109 mcd, and every 10 cm from ~110 to ~185 mcd. These 10-cm³ samples were freeze dried, disaggregated, washed in tap water, sieved at 150 µm, dried, and stored in glass vials.

Two separate size fractions of G. ruber (white) were selected (212 to 355 µm and >355 µm) in order to have sufficient material for isotopic analysis with replicates where necessary. Isotopic analyses were completed on five to twelve individuals (typically eight) in the 212- to 355-µm size range and three to eight individuals (typically four) in the >355-µm size range, depending on availability. For each sample, foraminifers were placed into the bottom of a reaction vial, broken using a glass pestle, and 30 mL of 5% H₂O₂ was added and allowed to react for 1 hr. Then 60 mL of ethanol was added, the sample was sonified for 30 s, and the solution was drawn off using a micropipette. The sample was then dried overnight at 35°C.

A total of 1936 analyses (including duplicate and triplicate analyses) were conducted on 1170 samples; ~30% of the samples have been replicated at least once. Of the 1936 analyses, 601 were on the 212- to 355-µm size fraction and 1335 on the >355-µm size fraction. A total of 58 samples were measured for both the 212- to 355-µm and >355-µm size fractions. Based on the median difference within these 58 paired analyses, we convert data in the 212- to 355-µm size fraction to values consistent with the >355-µm size fraction by adding 0.44 to the ¹³C values (making them heavier) and subtracting 0.16 from the ¹⁸O values (making them lighter). All data reported are relative to the >355-µm size fraction (Table T1; Fig. F1).

Two species of benthic foraminifers were picked in the >150-µm size fraction (C. wuellerstorfi and U. peregrina) in order to have sufficient material for isotopic analysis with replicates where necessary. Isotopic analyses were completed on one to three individuals depending on availability. For each sample, foraminifers were placed into the bottom of a reaction vial, broken using a glass pestle, and 30 µL 5% H₂O₂ was added and allowed to react for 1 hr. Then 60 µL ethanol was added, the sample was sonified for 30 s, and the solution was drawn off using a micropipette. The sample was then dried overnight at 35°C.

A total of 1224 analyses (including duplicate and triplicate analyses) were conducted on 1045 samples; ~15% of the samples have been replicated at least once. Of the 1224 analyses, 307 were on C. wuellerstorfi and 917 on U. peregrina. A total of 135 samples were measured for both species. Based on the median difference between these 135 paired analyses, we convert C. wuellerstorfi to values consistent with U. peregrina by subtracting 0.74 from the ¹³C values (making them lighter) and adding 0.64 to the ¹⁸O values (making them heavier). All data reported are relative to U. peregrina (Table T2; Fig. F1).

Samples were run in batches of ~40 on a Finnigan MAT 252 equipped with a carbonate (Kiel) III autosampler that reacts samples in individual reaction vessels at 70°C using H₃PO₄. External reproducibility based on repeated analysis of National Institute of Standards and Technology isotopic reference material NBS-19 (N = 10), Carrara marble (N = 242), and Brown Yule marble (BYM; N = 129) is ±0.02 and ±0.06 for ¹³C and ¹⁸O, respectively (1 ). External reproducibility based on replicate analysis of planktonic foraminifer samples is ±0.12 and ±0.10 for ¹³C and ¹⁸O, respectively (average half-range, N = 32). External reproducibility based on replicate analysis of benthic foraminifer samples is ±0.08 and ±0.05 for ¹³C and ¹⁸O, respectively (average half-range, N = 26). The Carrara and BYM in-house standards have been calibrated to NBS-19 for conversion to the Vienna Peedee belemnite (VPDB) scale. All data reported here are relative to VPDB.