METHODS

The oxygen isotope record is based on Globigerinoides ruber, a near-surface-dwelling planktonic foraminifer (Bé and Tolderlund, 1971) that records changes in surface water ¹⁸O and ¹³C (Linsley and von Breymann, 1991). Stable oxygen and carbon isotope analyses at Site 1144 were made on 1667 samples at uneven sampling intervals (Table T1).

Approximately 15–20 cm³ of wet bulk sediment was freeze-dried, weighed, and wet-sieved to remove the <63-µm fraction. The sand fraction (>63 µm) was oven dried at 40°C for 24 hr and then weighed to determine the percent sand fraction.

Because foraminifer tests are rare in the common size class 315–400 µm, we chose to use the 250- to 315-µm fraction (Linsley and Dunbar, 1994; Lee et al., 1999). For each sample, ~15–20 tests of the planktonic foraminifer G. ruber (white) variety, were selected following the morphotype classification of Wang (2000).

In a few samples, not enough specimens complying with the sensu stricto morphotype requirements were found. These samples were supplemented with as many sensu latu tests as necessary to reach the amount of carbonate (150–200 µg) required for reliable measurements. The number of specimens (15–20) and the rather narrow size fraction were chosen to minimize the influence of vital effects (Berger et al., 1978) and, moreover, to average out the signal noise linked to differential bioturbational mixing.

Only tests that were intact and without visible dissolution were selected. A substantial effort was made to avoid G. ruber (pink), which is generally isotopically lighter (¹⁸O = 0.17‰) (Thompson et al., 1979) than G. ruber (white). At Site 1144, G. ruber (pink) is either light pink throughout or, more often, the pigmentation is very weak and only the very first chambers are pink colored, which complicates identification.

Before isotope analysis, the foraminifer tests of each sample were immersed in ultrapure ethanol, carefully cracked to expose the interior of the chambers, and ultrasonicated for 20 s. Subsequently, the ethanol, with possible contaminants still in suspension, was siphoned off with a syringe and the remaining foraminifer fragments were dried at 40°C. The cleaned samples for isotopic measurements weighed 150–200 µg. For isotope analysis, standard techniques were used: the carbonate was reacted with 100% orthophosphoric acid at 70°C, and the isotopic ratios were determined using a Finnigan MAT 251 micromass spectrometer with the Carbo Kiel device (Kiel I type) at the Leibniz Laboratory at Kiel University (Germany). Samples from the interval 172.16–182.17 meters composite depth (mcd) were measured on a Finnigan Delta Plus XL mass spectrometer combined with a Kiel GasBench II continuous flow interface. Precision was regularly checked by running (internal) Solnhofen limestone standards. Conversion to the Peedee belemnite (PDB) scale was performed using the National Bureau of Standards NBS-20 carbonate standard. For the MAT 251, the external standard errors over 1 yr (2000–2001) were <0.08‰ for ¹⁸O and <0.05‰ for ¹³C (both 1- values); on the Delta Plus XL mass spectrometer, the error reached <0.096‰ for ¹⁸O (NBS-19) and <0.06‰ for ¹³C. The true uncertainty for G. ruber ¹⁸O data ranges from ±0.04‰ to ±0.08‰ on the basis of quasi-replicate analyses of centimeter-spaced samples from the Holocene section in neighboring SONNE-95 core 17940 (Wang et al., 1999a).

All stable isotope data for Site 1144 are given in the "Appendix" and presented in Figure F2. Data are also available as an electronic file at www.pangaea.de/PangaVista.

Instead of the meters below seafloor (mbsf) depth scale, we used the continuous shipboard (Shipboard Scientific Party, 2000) composite depth scale (mcd), which links sediment profiles of cores from Holes 1144A, 1144B, and 1144C to 235.41 mcd. Splicing of cores below this interval was precluded by incomplete core recovery and incidental alignment of core gaps (Shipboard Scientific Party, 2000). Although the cores below 235 mcd cannot be tied directly to the composite depth scale, the cores can be correlated with each other so that correlative features of multisensor track (MST) records and sediment color are matched in depth. This relative, or "floating," composite depth scale is not tied to the overlying composite depth scale (Shipboard Scientific Party, 2000). The majority of samples below 235 mcd are from Hole 1144A; some gaps resulting from the floating composite depth scale were bridged with samples from Hole 1144B.

Seventeen samples were radiocarbon dated to better constrain the age control of the Holocene to last glacial sediment section between 1.270 and 47.200 ka (see Table T2; Fig. F3). Five samples were dated on mixed bulk planktonic foraminifers (Chen and Shyu, unpubl. data), and twelve were dated on mixed planktonic G. ruber and Globigerinoides sacculifer (this study). The nine ¹⁴C datings constrain ages between 1.27 and 40.95 ka. Note that most dates of MIS3 come close to the tuned ages of 20–50 ka, which were assigned to the core depths by simple correlation of the ¹⁸O record to the GISP2 record right below the top of Dansgaard-Oeschger (DO) Interstadial 2 (see "Appendix"; Figs. F2A, F3).

During the Last Glacial Maximum (LGM), the dates show an age reversal from 18.12 ka (¹⁴C) at 14.97 mcd to 16.42 ka (¹⁴C) at 17.22 mcd (see Table T2). In part, this reversal may stem from differences in the species analyzed. The lower sample consisted of specimens from G. ruber and G. sacculifer, which dwell in the surface layer with a modern ¹⁴C reservoir age of ~465 a (Southon et al., 2002). In contrast, the upper sample consisted of mixed bulk planktonic foraminifers that possibly included specimens of species that dwell near and below the thermocline, where the ¹⁴C reservoir age exceeds that of surface water by several hundred years (P.M. Grootes et al., unpubl. data). Moreover, local ¹⁴C reservoir ages and their vertical gradient may have experienced significant short-term increases toward the end of the LGM, whereas primary ¹⁴C production did not change significantly during this time (Hughen et al., 2004). Since these factors cannot yet be quantified, the precise origin of the age reversal remains unknown.