RESULTS AND DISCUSSION

The 518-m-long sediment section reaches back to at least 1.03 Ma. Figure F3 shows the uppermost 120 m, spanning the last 140 k.y.

In general, the sawtooth-shaped ¹⁸O curve for Site 1144 records the 100-k.y. climatic cycles that are well established for the last 900 k.y. (Prell et al., 1986) (Figs. F2, F4). High-amplitude ¹⁸O variations occur throughout the entire Site 1144 record, with slightly decreasing amplitudes in the lowermost 100 m. The age model for Site 1144 was constructed by combining evidence from various reference records for age correlation. The uppermost 106 m of the Site 1144 ¹⁸O record was correlated to the ¹⁸O record of GISP2 (Grootes and Stuiver, 1997) and dated according to the GISP2 timescale (Alley et al., 1997; Meese et al., 1997) (Fig. F3). This correlation is largely supported by radiocarbon dates, although they show a slight age reversal near 15 mcd (see the "Appendix"; Figs. F2, F3).

In the interval 110–504 mcd, MIS 5.52–29 (128–1027 ka) (Emiliani, 1955; Shackleton and Opdyke, 1973) were identified by visual correlation to orbitally tuned ¹⁸O records of ODP Site 658 (Sarnthein and Tiedemann, 1990), ODP Site 677 (Shackleton et al., 1990, 2002), and Marion Dufresne core MD90-0963 (Bassinot et al., 1994) (Figs. F2, F4). Ages for the whole record were defined through linear interpolation between the tie points listed in Table T2.

Below 480 mcd, age control was based on two biostratigraphic datums. MIS 31 (Figs. F2, F4) was used as a tie point to constrain the age of the upper end of the small Gephyrocapsa acme Zone (Gartner, 1988) at 500.66 mcd to 1100 ka (Tables T3, T4) (compared to 1020 ka, according to Shipboard Scientific Party, 2000). A major hiatus between 503.76 and 506.82 mcd is inferred from the last occurrence (LO) of Discoaster brouweri/triradiatus at 506.82 mcd (1960 ka, according to Shipboard Scientific Party, 2000). The isotope stage below the hiatus is possibly MIS 73 but cannot be identified with certainty.

Holocene

The Site 1144 Holocene ¹⁸O record displays a high-amplitude internal variability reaching 0.6‰–0.8‰ (Fig. F3). The lightest values of –2.85‰ were measured at 5.8 mcd (near 10 ka). They may stem from a freshwater plume near the Pearl River mouth that resulted from high precipitation and runoff from China (Wang et al., 1999a, 1999b, 1999c). In general, the Holocene oxygen isotope curve is similar to the ¹⁸O record of neighboring core 17940 reported in Wang et al. (1999a). An early suspicion that the uppermost 2 m of sediment may be lost at Site 1144 (Shipboard Scientific Party, 2000) seems disproved by the ¹⁴C age of 1.7 ka (1.272 ± 0.060 k.y.) at 0.74 mcd. This is the approximate age expected for this core depth when taking into account the average Holocene sedimentation rate of ~0.5 m/k.y. at Site 1144. However, a small portion of the topmost sediment, we estimate ~7 cm, is still missing, since the oxidized reddish sediment surface was not recovered.

The planktonic Holocene to LGM ¹⁸O shift amounts to approximately 1.8‰, with maximum differences of 2.4‰ between short-term extremes in the Holocene and LGM. The ice volume effect on the ¹⁸O record for the most recent glacial–interglacial transition (MIS 1–2) is estimated to be 1.2‰, assuming that 10 m of sea level rise is equivalent to 0.11‰ change in ¹⁸O (Labeyrie et al., 1987; Fairbanks, 1989). Thus, the overall glacial–interglacial ¹⁸O shift at Site 1144 primarily records the global ice volume signal; in addition, fluctuations in local surface water temperature and/or salinity account for >0.6‰.

The Younger Dryas (YD) event (6.5–8.1 mcd) is recorded as a prominent isotopic shift of 0.7‰ back to more positive ¹⁸O values (Figs. F2A, F3) and shows the characteristic double-peak structure also known from the Greenland ice core records (Alley et al., 1993). Prior to the YD, a distinct ¹⁸O minimum near –2‰ marks the Bølling/Allerød warm phase (8.1–10.2 mcd). The top of MIS 2 is assigned to the major ¹⁸O shift of 0.8‰ toward the Bølling/Allerød level near 10.4 mcd (Termination IA).

Marine Isotope Stages 2–5

MIS 2–5 in the Site 1144 ¹⁸O record (Figs. F2A, F3) are characterized by frequent narrow-standing negative oscillations that reach 0.7‰–0.85‰ in MIS 2 and 3, up to 0.9‰ in MIS 4, and 1.1‰ in late MIS 5. These short-term ¹⁸O excursions are not analytical noise but instead form a reliable record of true climate change because

1. They exceed the analytical error by almost an order of magnitude;
2. Most spikes are not one-point peaks but are replicated by various data points at their peaks and/or along the flanks;
3. The given sampling space of 20–30 cm, equal to 200–300 yr, provides sufficient resolution for storing short-term differential climate signals in the sediment record; and
4. The tuned age correlation of the ¹⁸O oscillations between 20 and 41 ka was clearly confirmed by several calibrated ¹⁴C ages within the error range of calibration (see the "Appendix"; Fig. F3).

In Figure F3 we correlated the most prominent positive ¹⁸O excursions in the uppermost 110 mcd of Site 1144 to North Atlantic Heinrich events 1–6 (H1–H6). On the other hand, the negative ¹⁸O excursions were correlated with DO interstadials (IS) in the ¹⁸O record of core GISP2 (sensu Schulz et al., 1998; Wang et al., 1999a). Accordingly, the analogs of IS events 3–7, 9–16, and 20–23 are well represented in the Site 1144 ¹⁸O record as prominent negative excursions. Their internal structure appears slightly more rugged than the IS in the GISP2 record.

Starting from the top of MIS 2 (10.45 mcd), the first major negative ¹⁸O excursion occurs between 16.9 and 19.0 mcd and was termed "20k IS" because it corresponds to a highly similar (yet unnamed) ¹⁸O excursion near 20 ka in the GISP2 record. The broad negative ¹⁸O interval between 23.9 and 26.4 mcd was correlated to IS 2 of core GISP2 near 23 ka, although it lacks the high negative spikes that mark the ice core record. However, bracketing ¹⁴C datings (equal to 22.17 and 25.8 ka) (Fig. F3; see the "Appendix") and the preceding strongly positive ¹⁸O excursions near 26.89 mcd, which correspond to H2, strongly support this interpretation. Similar to core GISP2, the analog to IS 8 at Site 1144 covers a broad bipartite interval but lacks the typical sawtooth pattern, perhaps because of still insufficient sampling resolution (267 yr). DO IS 10 is directly supported by a calibrated ¹⁴C date of 40.05 ka (see the "Appendix"). The shape of the DO IS 16–17 equivalents at Site 1144 slightly differs from GISP2, but the ¹⁸O drastic increase below 66 mcd is a clear counterpart to the top of event H6 and marks the bottom of IS 17 at the top of MIS 4 (Fig. F3). Here, the structure between IS 18 and 19 is richer than in core GISP2. The bipartite IS 19 in core GISP2 was correlated to the broad, deeply structured interval from 73.5 to 78.8 mcd. The rugged equivalents of IS 21, 22, and 23 at Site 1144 represent marine isotope Substages 5.1, 5.3.1, and 5.3.3, respectively.

The H3 event is weak in comparison to its equivalent in core GISP2; it is, however, well confirmed by a calibrated ¹⁴C age of 30.72 ka (Sarnthein et al., 2000) at 33.07 mcd. Farther downcore, the prominent positive ¹⁸O excursion at 44.5 mcd corresponds to event H4 (see the "Appendix"; Fig. F3). The H5 equivalent at 53.6 mcd has a proper calibrated ¹⁴C age of –47 ka, but the broad sawtooth structure of ¹⁸O minima farther below, including IS 14, clearly matches the pertinent ¹⁸O record of IS 14 in core GISP2. A pronounced increase in ¹⁸O close to 61 mcd is tentatively assigned to Heinrich subevent 5.2 (van Kreveld et al., 2000; Sarnthein et al, 2000).

Only the very top of MIS 5.5 and the MIS 5.4–5.5 transition are preserved downcore from MIS 5.4 (centered around 107 mcd; ~108 ka) subsequent to ~116 ka (Figs. F2A, F4). Most of MIS 5.5 is missing according to the stratigraphic definition of Shackleton et al. (2002). This major hiatus, spanning ~12 k.y., is deduced from an obvious lack of sediment thickness and, in particular, from the fact that the ¹⁸O minimum of approximately –2.5‰ misses by far the negative extreme of –2.9‰/–3.2‰ characteristic of MIS 5.51 in other cores from this region (Linsley, 1996: ODP Hole 769A; Wang et al., 1999a: SONNE-95 core 17954; Lee et al., 1999: Marion Dufresne core MD972151) and at neighboring Site 1145 (Oppo and McIntyre, unpubl. data; McIntyre and Oppo, 2001). The hiatus is also documented by the LO of G. ruber (pink). It lies at the top of a pronounced positive ¹⁸O excursion at 110–111 mcd, whereas it has been generally established for the top of an extreme ¹⁸O minimum indicating MIS 5.51 (116 ka). Finally, the hiatus at 110 mcd corresponds to an abrupt ¹³C shift of more than 0.5‰ (Figs. F2B, F4B). Farther downcore, the ¹³C record shows a broad extreme minimum similar to that observed for late MIS 6 at Atlantic ODP Site 658 (Sarnthein and Tiedemann, 1990).

Below the hiatus, the short-term positive ¹⁸O excursion near 110–111 mcd most likely represents a YD-style climatic oscillation (event 5.52 or H11 at >128 ka) (Sarnthein and Tiedemann, 1990; Shackleton et al., 2002). The preceding negative peak is assigned to warm event 5.53 near 130 ka. The transition from MIS 6 to MIS 5.53 (Termination II) near 113 mcd reveals an almost instantaneous ¹⁸O shift of 1.25‰ (Figs. F2A, F4A).

The millennial- to submillennial-scale variations in the planktonic stable isotope record of MIS 2–5 at Site 1144 match the high-resolution sea-surface salinity record of neighboring core 17940 (back to DO IS 9) (Wang et al., 1999a) and the ¹⁸O record of SONNE-95 core 17924 (150 km farther east; M. Sarnthein, unpubl. data). Thus the short-term ¹⁸O changes at Site 1144 can be interpreted as large-scale changes in sea-surface salinity in addition to minor changes in sea-surface temperature and global ice volume (Lambeck and Chappell, 2001). Because of the close vicinity to the Pearl River mouth, the short-lasting freshwater pulses at Site 1144 especially record short increases in fluvial runoff induced by enhanced precipitation and summer monsoon intensity on the Chinese mainland. On the other hand, any decrease in summer monsoon intensity has resulted in decreased precipitation and/or in sea-surface temperature reduction at Site 1144. Thus the short-term maxima in planktonic ¹⁸O provide a record of high salinity off South China during major and minor stadials.

The question is not fully answered yet (An, 2000) as to whether sudden ¹⁸O variations in the SCS were precisely coeval with the DO interstadial and stadial events in the North Atlantic and the Greenland Ice Sheet (Dansgaard et al., 1993; Adkins et al., 1997; Bond et al., 1997; van Kreveld et al., 2000). However, we regard it as most likely that the short-term changes in Asian Monsoon intensity were intimately tied to high-latitude climate forcing via atmospheric signal transfer (Wang et al., 1999b; Kudrass et al., 2001). This concept is corroborated by recent U/Th datings of a speleothem record in China (Wang et al., 2001) and a model study of Mikolajewicz et al. (1997). They showed that atmospheric transfer of Greenland temperature variations would have resulted in quasisynchronous signals across the entire Northern Hemisphere.

In contrast to a scenario in which the SCS was primarily responding to climate changes in northern high latitudes, an opposite scenario may be considered, in which tropical climate and the monsoon system were the driving forces for Northern Hemisphere millennial-scale climate variations. Recent models of Clement and Cane (1999) and Clement et al. (1999) indeed suggest that long-term oscillations of the El Niño–Southern Oscillation may have been an important forcing factor for millennial-scale changes in the climate system of tropical monsoon and, furthermore, may have exerted an important influence on global climate change.

Marine Isotope Stages 6–11

Below the hiatus at MIS 5.5, the ¹⁸O record of marine isotope Stage 6 shows well-developed Substages 6.2–6.6 (Figs. F2, F4) as in various other records from the South China and Sulu Seas (Linsley and Dunbar, 1994; Wang et al., 1999a). Moreover, there are ongoing millennial-scale oscillations with internal amplitude variations reaching 1.0‰, similar to (possibly coeval?) ¹⁸O oscillations at neighboring Site 1145 (Oppo and McIntyre, unpubl. data) and the short-term variations in MIS 3–4.

The precise stratigraphic assignment of MIS 6.5 and the MIS 6/7 boundary (~190 ka) were subject to extensive discussion among the Leg 184 shipboard scientific party because of planktonic ¹⁸O values for MIS 6.5, which are as negative as those in MIS 7.1. Final correlations were established by means of (1) a benthic ¹⁸O record for the interval 153–185 mcd, which shows a massive ¹⁸O increase by more than 1.0‰ near 175 mcd to be used as stage boundary (Clemens et al., unpubl. data); and (2) a precise correlation of the Site 1144 planktonic ¹⁸O record with that of neighboring Site 1145 (Oppo and McIntyre, unpubl. data).

Planktonic ¹⁸O values of MIS 7.1 and 7.3 do not quite reach the interglacial level of the Holocene (–2.6‰), only the values of MIS 7.5 near 234 ka, a trend similar to that of other ¹⁸O records from this sea region (Linsley and von Breymann, 1991; Wang et al., 1999a). Sedimentation rates in warm MIS 7 are clearly reduced compared to MIS 6 (Table T2). The amplitude of cold Substage 7.4 (221 ka) (Figs. F2A, F4) is similar to that at Site 1145 (Oppo and McIntyre, unpubl. data) but appears less "glacial" than in SPECMAP (Martinson et al., 1987) and Site 769 isotope curves.

The detailed and accurate fit of the planktonic ¹⁸O curves from Sites 1144 and 1145 (Oppo and McIntyre, unpubl. data) also led to the definition of a major hiatus reaching from the top of MIS 7.5 down to the lower top of MIS 8.5 at Site 1144. Like the stratigraphic gap at peak MIS 5.51, the erosional event near MIS 7.5 appears linked to peak interglacial conditions. Unlike MIS 5.5, erosion had ended shortly before the peak interglacial during late glacial Termination III and deleted the complete antecedent glacial sediment section.

Unlike various stratigraphic reference curves (Bassinot et al., 1994) MIS 9.1 is broad and bipartite in the ¹⁸O records of Sites 1144 and 1145 (307–313 ka) (Figs. F2, F4) (Oppo and McIntyre, unpubl. data). Two warm Substages 9.11 and 9.13 are clearly detached by a ¹⁸O increase of 1.3‰, identified as cold Substage 9.12. Below, ¹⁸O values reach the Holocene ¹⁸O level in warm MIS 9.3, which is separated from MIS 9.1 by a positive 1‰ excursion in MIS 9.2. A major ¹⁸O shift of 1.5‰ constitutes the MIS 9/10 boundary (Termination IV) at 208 mcd (334 ka) (Figs. F2A, F4). Oxygen isotope values for antecedent cold MIS 10.2 (up to –0.5‰) are slightly higher than for MIS 2. High-frequency ¹⁸O variations during MIS 10.2–11.1 reach 1‰, similar to the ¹⁸O amplitudes of DO cycles in MIS 2–4.

Despite sufficient sampling resolution, most of the characteristically broad interglacial MIS 11.3 is missing (Figs. F2, F4), similar to the stratigraphic gaps outlined for peak interglacials MIS 5.5 and 7.5. This gap is also suggested by minimum isotope values of –2.5‰ that are far more positive than the values of –3.1‰ recorded at neighboring Site 1145 (Oppo and McIntyre, unpubl. data). Based on detailed comparison with other planktonic high-resolution records of MIS 11.3 (Berger et al., 1993a, 1993b; McManus et al., 1999), we place the major hiatus somewhere between 246.33 and 247.13 mcd (~413–397 ka) (Figs. F2A, F4), comprising ~15 k.y. Similar to the base of the hiatus in Stage 5.5, the base of this hiatus lies within a YD-style cold episode right after the short warm event 11.33 (sensu Sarnthein and Tiedemann, 1990), which is comparable to the Bølling/Allerød warm phase at the end of the last glacial.

Marine Isotope Stages 12–25

MIS 12 (250–289 mcd; 423–471 ka) (Fig. F2A, F4) is the most prominent glacial of the last 800 k.y., with average ¹⁸O values as much as 0.4‰ heavier than during MIS 2. The ¹⁸O record separates three major cold events (MIS 12.22, 12.24, and 12.4) and numerous short-term climatic oscillations similar to those found in MIS 2–4.

MIS 13 and 15 do not display the characteristic trident shape expected on the basis of the stratigraphic reference record of Bassinot et al. (1994). The ¹⁸O record of MIS 13 and 14 may be complete. On the other hand, we strongly surmise that MIS 15 has lost peak interglacial Substage 15.5 near 610 ka, similar to the hiatuses found at peak interglacial Substages 11.3, 7.5, and 5.51. In Figure F2 we placed the hiatus to the ¹⁸O jump at 325 mcd. This postulation receives independent support from an abrupt major drop in sedimentation rates by a factor of 4 for lower MIS 15, if no hiatus is assumed (Fig. F5).

Peak glacial MIS 16 (623–650 ka) is broadly developed, consisting of Substages 16.2 and 16.4 with average ¹⁸O values as high as those during MIS 2. MIS 17 is clearly tripartite and MIS 18 is bipartite with a broad Substage 18.3, as in the reference record of Bassinot et al. (1994). This general coherency of records also applies to MIS 19–21. A layer of Australasian microtektites (Zhao, unpubl. data) provides an important age marker within the lower part of the MIS 19–20 transition at 386.2 ± 0.03 mcd, right below the Brunhes/Matuyama boundary.

Narrow-standing isotope Stages 22–33 were identified through comparison with ¹⁸O records of ODP Sites 607, 677, and 659 (Tiedemann et al., 1994), where MIS 27 turns out to be clearly bipartite.

Millennial-scale climate variability over MIS 13–19 shows ¹⁸O amplitudes of 0.6‰–0.9‰, which are as high as those in MIS 2–4 and other Pleistocene stages (Fig. F2A).

Marine Isotope Stages 26–(?)34 and Features near the Base of the Core Profile

Below 422 mcd (~970 ka), sampling intervals still range between 1.5 and 2 m (Figs. F2, F4). Thus, MIS 26–33 are assigned only tentatively but bear the potential for a high-resolution climatic record because of high sedimentation rates (Fig. F5).

A major hiatus, deduced from biostratigraphy (Shipboard Scientific Party, 2000), occurs below MIS 33 (Figs. F2A, F4), which, accordingly, may be incomplete. Based on two biostratigraphic ages (Tables T3, T4), the hiatus spans almost 0.85 m.y. Farther downcore a definitive designation of isotope stages for the interval 509.3–517.9 mcd was not possible. The LO of D. brouweri/triradiatus (1.96 Ma) possibly represents MIS 73 (Sarnthein and Tiedemann, 1989) at 506.82 mcd. Assuming a constant average sedimentation rate of 0.50 m/k.y., the lowermost depth at Site 1144 (517.56 mcd) may reach an age of 1.982 Ma (Tables T3, T4).

Site 1144 is located on a thick sediment drift (Sarnthein et al., 1994; Shipboard Scientific Party, 2000) where changes in sedimentation rate mainly reflect variations in downslope terrestrial input and/or lateral reworking and transport by bottom current activity. Accordingly, Site 1144 is characterized by generally high sedimentation rates (Table T2; Fig. F5). From 507 to 100 mcd (MIS 33 to MIS 6 top) linearly interpolated sedimentation rates vary between 0.2 and 0.9 m/k.y. Higher rates mark glacial MIS 20, 18.4, 16, and, especially, MIS 10 and 6.6; lower rates mark interglacial sediment sections.

This trend culminates in hiatuses during peak interglacial MIS 15.3, 11.3, 9.1, lower MIS 7.5, and MIS 5.5, where the ¹⁸O minima neither reach a level analogous to the Holocene nor cover sediment sections that are as thick as those in the Holocene section (Figs. F2, F4). Thus, major portions of the interglacial sediment sections appear to be lost in major and minor stratigraphic gaps as also revealed by direct comparison with ¹⁸O records of neighboring Sites 1144 and 1145 (S. Clemens and M. Sarnthein, pers. corr., 2002). A similar sediment loss may also apply to MIS 9.3, where the linearly interpolated sedimentation rate decreases to a rare low of 0.15 m/k.y. and may contain short stratigraphic gaps. Most hiatuses correspond to minor or major unconformities in the seismic record at Site 1144 (Shipboard Scientific Party, 2000).

We tend to ascribe the striking stratigraphic gaps to shifts in the intensity and/or vertical shifts in the axis of a deep boundary current, the incursion of Pacific Intermediate Water. The current shifts were probably linked to changes in global thermohaline circulation characteristic of early to peak interglacial stages. Indeed, Sarnthein et al. (1994) and Lehmann (1996) found a pertinent PARASOUND echo character, erosional microfurrows that today are widespread and form a record of modern strong bottom current action at water depths below Site 1144.

From 100 mcd to near the sediment surface, sedimentation rates gradually increase from 0.5 to 1.8 m/k.y. (Fig. F5). This increase may be apparent, partly as a result of still modest sediment compaction and higher water contents (Shipboard Scientific Party, 2000), and partly because maximum sedimentation rates are tied to peak glacial MIS 4 and 2, similar to the maxima of glacial stages farther downcore.

Note that the high maximum sedimentation rates in MIS 2 contrast with the rates found at neighboring site 17940 farther upslope (Wang et al., 1999a). At site 17940, maximum rates of ~0.7 m/k.y. mark the Preboreal and early Holocene section, whereas glacial sedimentation rates drop to 0.2 m/k.y. during MIS 2–3. This difference in the trend of sedimentation rates (1) may be an artifact near the core top of the two sites, resulting from different coring techniques, or (2) may stem from the slightly greater water depth at Site 1144 (Fig. F1), which was subject to enhanced deposition when the glacial sea level reached a minimum. Lüdmann et al. (2001) suggested that differences in sedimentation rates may be linked to complex interactions between sea level stand and the location of maximum sediment discharge from the Pearl River mouth. They surmise that the synglacial exposure of the Dongsha Island area, northwest of Site 1144, led to a deviation of the sediment plume from east to west of the island group during the LGM and back to east when the shelf again was flooded during the early Holocene. In this way, the deeper Site 1144 would possibly receive much more hemipelagic sediment during times of low sea level, whereas site 17940, ~300 m farther upslope, would receive more sediment during sea level high stands.

Glacial–interglacial variations in planktonic ¹³C at Site 1144 reach 0.6‰–1.1‰ over the past 1.1 m.y. These variations form a record of changes in the composition of dissolved inorganic carbon (DIC), which in turn depends on the nutrient concentration and the isotopic composition of dissolved atmospheric CO₂ and CH₄ near the sea surface (~30 m water depth). Long-term ¹³C changes roughly parallel the general climate cyclicity of the last 1 m.y. and are dominated by the ~100-k.y. eccentricity cycle.

High ¹³C levels were more common during (late) warm ¹⁸O stages and interstadials. However, very high ¹³C values also marked early glacial MIS 2, semiglacial MIS 3, and large parts of MIS 12–13, where ¹³C values reached or approached the late Holocene level of 1.2‰–1.4‰. Over longer times, maximum ¹³C values culminated at ~30 (MIS 2/3 boundary), ~480 (MIS 13.1), and ~1035 ka (top MIS 31) (see the "Appendix"; Figs. F2B, F4B) and thus formed a sort of overlying 450-k.y. cycle possibly controlled by orbital eccentricity.

On the other hand, minimum ¹³C values of 0.0‰–0.25‰ were most common near the end of each peak glacial stage and the subsequent glacial Terminations I, II, IV, and V. Similar ¹³C minima also marked the glacial terminations leading from MIS 7.4 to 7.3, MIS 18.4 to 18.3, and, especially, MIS 22 to 21. Further potential ¹³C minima at Terminations III and VI were lost in the hiatuses outlined above. In addition, there was a ¹³C minimum at early MIS 16. Immediately after the ¹³C minimum at each glacial termination, we find the largest shifts in ¹³C (1.1‰) (Figs. F2B, F4B). Some ¹³C gradients are spurious since they were steepened by hiatuses, in particular after glacial Stages 2, 6, 10, and 12.

The younger portion of the planktonic ¹³C record at Site 1144 (Figs. F2B, F4B) is consistent with records from neighboring core 17940 and various other SONNE-95 cores from the SCS (Wang et al., 1999a). Moreover, the major ¹³C minima (Figs. F2B, F4B) are consistent with deglacial minima found in many other low-latitude planktonic records (Linsley and Dunbar, 1994; Oppo and Fairbanks, 1989), in the subarctic North Pacific (M. Sarnthein, unpubl. data), and many Atlantic ¹³C records (Tiedemann, 1994). Thus, the long-term overall ¹³C reduction along Pleistocene glacial terminations may in part represent a global feature.

In part, the minima in ¹³C toward the end of the glacial maxima and early interglacial times may also result from local factors such as short-term enhanced riverine nutrient input (Wang et al., 1999a), triggering high productivity and enriching remineralized organic carbon and ¹²C in nearshore surface water. During Termination I, the switch back to high ¹³C values only occurred after the Preboreal, when a prominent maximum in local freshwater and nutrient input had ended and/or shifted landward on the inundated shelf as a result of ongoing sea level rise. This freshwater plume is clearly recorded as massive sea-surface salinity reduction in neighboring core 17940 during the Preboreal (Wang et al., 1999b) and documents a maximum in summer monsoon precipitation over China. This conclusion is corroborated by a short-term culmination in clay accumulation rates, probably the result of prominent sediment discharge from the Pearl River (Wang et al., 1999a).

To refine the age assignment of Pleistocene biostratigraphic datums (Table T3) and the Australasian microtektite layer in the SCS, we employed the age model of Table T2, which in turn is tuned to the astronomical timescale developed for the stacked ¹⁸O records of ODP Site 677 and core MD90-0963 by Bassinot et al. (1994). This approach was partially successful; it was hampered by frequent stratigraphic gaps in the Site 1144 sediment section.

The latter problem applies to the LO of G. ruber (pink) at Site 1144 at 110.30 mcd. Since this depth forms the base of a hiatus across Interglacial Stage 5.5, the actual LO position lies within the lost sediment section, somewhere between 128 and 116 ka (Table T2; Fig. F2A, F4A).

The first occurrence (FO) of Emiliania huxleyi was found at 192.82 mcd (Shipboard Scientific Party, 2000), which corresponds to the middle of MIS 7.4 with an assigned age of ~227 ka. Because of the 50-k.y. hiatus at Site 1144 below MIS 7.5, this FO age is much younger than the ages generally assigned to the FO of E. huxleyi within MIS 8.4–8.5 at 268 to 285 ± 3.5 ka (Thierstein et al., 1977; Vergnaud Grazzini et al., 1990; Ahagon et al., 1993).

The number of G. ruber (pink) used for defining the G. ruber (pink) acme Zone increases abruptly just above MIS 11.3 in the Site 1144 ¹⁸O record at 248.39 mcd (417 ka). By contrast, the Shipboard Scientific Party (2000) described this prominent increase in G. ruber (pink) farther upcore, at 261 mcd (400 ka) (Li, 1997). However, specimens of G. ruber (pink) also occur farther downcore to 273 mcd (early MIS 12; 470 ka) and increase in several deeper samples, although the coloration of these specimens is very faint and mostly restricted to juvenile chambers. A similar gradual increase was first reported by Thompson et al. (1979) from the western equatorial Pacific. The clearly recognizable increase in frequency and coloration of G. ruber at 248.39 mcd may also reflect an improved preservation of the pigment (Thompson et al., 1979). In summary, we propose a new age of 417 ka for the onset of the G. ruber (pink) acme Zone in the SCS, an age that is slightly older than that generally reported for sediment records of the low-latitude West Pacific.

The LO of Pseudoemiliania lacunosa lies close to the MIS 12/13 boundary at 288.54 mcd (Shipboard Scientific Party, 2000), which corresponds to an age of 493 ka (Table T2). This date clearly precedes the dates defined through ¹⁸O records of cores V28-238 and 239 (Shackleton and Opdyke, 1973), where P. lacunosa only becomes extinct in MIS 12.4 (Thierstein et al., 1977).

The Brunhes/Matuyama boundary was not identified at Site 1144 by the Leg 184 Shipboard Party. Subsequent shore-based paleomagnetic measurements (Solheid et al., this volume) have neither confirmed the shipboard results (Shipboard Scientific Party, 2000) nor furnished new evidence of the most important paleomagnetic events. Instead, we used a layer of Australasian microtektites at 386.18–386.23 mcd, centered at ~386.21 mcd (Zhao, unpubl. data) as an independent age marker to establish solid age control for the MIS stratigraphy of Site 1144 near the Brunhes/Matuyama boundary. Published ⁴⁰Ar/³⁹Ar ages of the Australasian microtektites range from 783 ± 21 ka (Izett and Obradovich, 1992) and 784 ± 12 ka (Kunz et al., 1995) to 761 ± 17 ka and 816 ± 7 ka (Yamei et al., 2000). The microtektite layer was also found in SONNE-95 sediment core 17957 from the southern SCS (Jian et al., 2000; Zhao et al., 1999) where the abundance of microtektites culminates within the MIS 19–20 transition, 10 cm (~11.6 k.y.) below the Brunhes/Matuyama reversal (Wang et al., 2000). At Site 1144 the microtektite layer also lies within the lower MIS 19–20 transition. On the basis of our age model (Table T2), the layer has an age of 787 ka, similar to the age of 793 ka reported by Lee and Wei (2000).

The LO of Reticulofenestra asanoi occurs in the Site 1144 sediment profile at 417.23 mcd within MIS 24, which is slightly deeper than the definition of Wei (1993) (MIS 22–23) and leads to an assigned age of 912 ka.