The samples were wet sieved using a 63-µm sieve. The dried sample residue >63 µm was sieved into five fractions (125-200, 200-250, 250-315, 315-400, and >400 µm) for micropaleontological studies. These five fractions were then quantitatively subsampled with a microsplitter until ~100 whole planktonic foraminiferal specimens were left for microscopic investigation. All planktonic foraminifers in these subsamples were identified and counted, following the taxonomic concepts of Bé (1977), Bé and Tolderlund (1971), Kennett and Srinivasan (1983), and Hemleben et al., (1989) (see "Taxonomic List" in the "Appendix"). Using this method proposed by Pflaumann et al. (1996), a minimum of 500 planktonic foraminifers were counted per sample. Percent abundances of planktonic foraminifer species found within each sample are listed in Table T1.
Because of selective carbonate dissolution, the composition of planktonic foraminiferal assemblages observed in the sediment core may differ significantly from the in situ living foraminiferal fauna. As a result, the SSST calculated based on such biased faunal composition would result in values that do not mirror the adequate sea-surface conditions (Le and Shackleton, 1992, 1994; Le and Thunell, 1996). To monitor carbonate dissolution we applied the fragmentation ratio proposed by Le and Shackleton (1992) (Table T2):
Fragmentation was estimated by counting the number of planktonic foraminiferal fragments in a split that contained 500 or more whole planktonic foraminifers.
Lithic fragments in the >125-µm fraction, interpreted as IRD, were also counted. The number of lithic grains per gram of dry sediment is listed in Table T2.
To evaluate the temperature estimates, we chose the MAT (Hutson, 1980; Prentice, 1980; Overpeck et al., 1985; Le, 1992; Pflaumann et al., 1996). The SSST estimate, the dissimilarity coefficient (which measures the difference between the assemblage of a downcore sample and the assemblage of the analog), and the standard deviation are listed in Table T2.
Isotopic analysis for Site 1090 was performed on benthic foraminifers (Uvigerina peregrina, Fontbotia wuellerstorfi, and species of the related genus, Cibicidoides) (Table T3; data are not corrected for specific fractionation). The isotope measurements were performed with a Finnigan MAT 251 mass spectrometer coupled to an automatic carbonate preparation device at the Alfred Wegener Institute. Data are related to the Peedee belemnite (PDB) standard through repeated analyses of National Bureau of Standards isotopic standard reference material SRM-19 (Hut, 1987). The standard deviation of replicate sample analyses run on different days for Site 1090 is 0.07
for 
18O. The benthic foraminiferal 18O record has been combined from both F. wuellerstorfi (and Cibicidoides spp.) and U. peregrina 18O data after correction for specific fractionation (+0.64 for F. wuellerstorfi and Cibicidoides spp., according to Shackleton, 1974) to produce a continuous record.
