High-resolution Carbonate Stratigraphy of IMAGES Core MD972151 from South China Sea

High-resolution carbonate stratigraphy of the deep-sea core MD972151 from the southwestern South China Sea shows millennhil-scale variability similar to oxygen isotopic fluctuations recorded in ice cores from Greenland. In a long term of glacial-interglacial scale, carbonate contents in the inter glacial time (up to 25 o/c by weight) were higher than that in the glacial periods (5-15 o/c by weight). Even for the last glacial, carbonate contents are relatively high in the interstadial events and low in the stadial horizons. This demonstrates clearly that carbonate content in the continental slope above the lysocline in the southwestern South China Sea is primarily con trolled by dilution of terrigenous inputs, which in turn is due to sea-Ievel fluctuations in response to changes of ice volume in high latitude regions.


INTRODUCTION
The South China Sea (SCS; 5°_22°N, 109°-121°B) is the largest marginal sea off the Asian continent ( Figure 1). The SCS. spread in NE-SW and B-W directions in late Oligocene -middle Miocene (Taylor and Hayes, 1983). This has resulted in the rhombic shape of the SCS. Wide passive continental shelves exist off Chinese coast (65 x 10 4 km 2 ) in the north and on the Sunda Shelf (115x 10 4 km 2 ) off the Indochina Peninsula in the southwest. The Central Basin (80 x lQ4 km 2 , > 4000 rn) is below the present carbonate compensation depth (CCD; Figure 1).
Ta the east, the SCS oceanic crust is subducting beneath the Philippine Sea plate a10ng the Manila trench. In this geological context, the SCS marginal sea is a semi-c1osed marginal sea bounded by the landmasses of SB China to the north, the Indachina Penin sula ta the west, the micro-continent of the Kalimantan-Palawan islands ta the south, and the Luzon-Taiwan islands ta the east (Figure 1).
The SCS exchanges its surface water with ather water masses through several water gates. For example, the lndian Ocean surface water flows in the SCS through the Sund a Shelf in the summer, while the Chinese Coastal Water flows in the SCS through the Taiwan Straït in the winter. A branch of the Kuroshio Currents flows in the SCS through the southem part of the Luzon Strait off northem Luzon and then retums to join the main Kuroshio through the northem part of the Luzon Straït off southern Taiwan (Metzger and Hurlburt, 1996). In comparison, SCS surface water flows out ta the Sulu sea through Mindoro and Balabac Straits (Figure 1).
sm depths of these water gates, however, are all shallow (less than 100 meters) except for the Luzon Straït. Therefore, during the last glacial time, when the global sea-Ievel fell by more than 100 meters (Fairbanks, 1989), configuration of the ses and water circulation would have been very different from the present one ( Figure 2; Wang et al., 1995;Jian et al., 1998). However, the sill depth of the Luzon Strait is 2500 m, above which bath surface and intennediate/ deep waters of the ses could exchange with the Pacifie Ocean throughout the entire Quater-nary.

.
Presently, climate and surface water circulation in the ses marginal sea are controlled by seasonal reversaI monsoan systems ( Figure 3). During the summer monsoon season (May to September), when the warm Indian Ocean water invades the SCS through the Sunda SheIf, the ses surface water flows from the southwest to northeast and sea surface temperature (SST) is homageneously high (28-29°C). Local upwelling occurs off the coast of Vietnam (Figure 3). In contrast, during the winter monsoon season (October to April), the cold Chinese Coastai Water flows southward and the ses surface water flows from the northeast to the southwest.
The cold water, along with cold wind moving south ward from the Siberian High Pressure, results in low SST and a steep temperature gradient in the ses (21-24°C in the north and 25-27°C in the south). Previous studies have shown that on the Asian continent, weathering and precipitation have changed significantly in respanse to fluctuations of paleomonsoon systems in the last 150 kyrs (An et al., 1990;Huang et al., 1997a;Liewet al., 1998). These changes will have further given rise to changes in the sedimentation rate of terrigenous materials along the SCS continental slope. Sedimentation rate on the continental slope of the SCS is relatively high (10-60 cmJky; Wang et al., 1995), which pennits a high-resolution study of the IMAGES (International Marine Past Global Change Studies) goals. During Leg III of the IMAGES III (MD 106)-IPHIS cruise in June 1997, seven deep-seacores were raised from the ses (Figure 1). The purposes of this cruise across the ses was to obtain high quality deep-sea cores for a high-resolution  (after NOAA, 1994;Chen et al., 1985).
paleoceanography/paleomonsoon study in low latitude oceans and marginal seas. The results are then used to link with middle-high latitude paleoclimate records in the terrestrialloess sequences in central China and the Greenland ice cores as suggested in PAGES-PEP II Huang and Chen, 1998). Here we present a high-resolution carbonate stratigraphy of the core MD972151 off Vietnam to construct a basic stratigraphy for a further detailed paleoc1imatic correlation with Chinese loess and Greenland ice cores in the near future ).

CORE SITE LOCATION AND GENERAL OCEANOGRAPHY
The core MD972151 was taken from a water depth 1,598 mat 8°43.73'N, 109°52.17'E which is north ofWan-An-Tan, southeast off Vietnam ( Figure 1). The site is on the continental slope to the east of the Sunda Shelf, where thick terrigenous sediments derived from the Indochina Peninsula and northem Kalimantan have accumulated ( Figure 1). The core length is 26.72 m (down to Stage 6) with continuous and homogenous olive to dark clay. Sorne thin sandy layers occur in the lower part (>23 m, in Stage 6), which do not obviously disturb any . significantly paIeoclimatic signaIs registered in the core.

4.RESULTS
The carbonate stratigraphy (Table 1) of the core MD972151 plotted against depth and age are shown in Figure 4a and 4b. Carbonate content varies between 25 % and 3 % by weight over the whole length of the core as in the other cores from the SCS continental slope (Wang et a1., 1995;Wang, 1998). In a long-term interglacial-glacial cycle, the carbonate curve is almost parallel to the oxygen isotope curve for planktic G. sacculifer  and   SPECMAP, Carbonate contents are high (15-25%) in the interglacial (Stages 1 and 5) but low (3-15%) in the glacial intervals (Stages 2-3-4 and 6). The three intervals ofhigh carbonate content in the lower part of the core (16-24 m) corresponding to Stage 5a (17%), 5c (21 %) and Se (25%), are intervened by two low intervals of Stage 5b (10%) and 5d (5%). However the carbonate content of these two lows in the last interglacial interval is still higher than that in the glacial intervals of Stage 4 (3-5%) above and Stage 6 (3%) below. Carbonate content in Stage 3 (8-15 %), especially in its early part, is higher than in Stage 2. In the upper part of the core, carbonate content rapidly increases from 5% in the LGM to 25% in theearly Holocene, and then decreases to 15 % in the core top (938 years). On a short-term millennial scale, carbonate content in MD972151 fluctuates at a high frequency (Figure 4b). Sorne events correlate with warm interstadial events found in the Greenland ice cores. For examples, the two peaks at 1520 and 1576 cm aboveand below the interval with the Toba eruption event of 1556-1558 cm correlate with IS 19 and 20 of GISP II, respectively (Schulz et al., 1998). The Toba eruption event is characterized by very low «1 %) carbonate content. Carbonate peaks at 1252, 1172, 1060 and 906 cm are conspicuously higher than the interval below and above these peaks. These high carbonate peaks may correspond ta IS 16,14,12 and 8, respectively, where alkenone sea surface temperature (U\7SST) are also reIatively high . The low carbonate content in 322-362 cm corresponds to Younger Dryas where oxygen isotope values are relatively heavier . In comparison, high carbonate content at 362-370 cm correlates with Aller~d warm event (14 Ka) of the GISP II ice core (Grootes, et al., 1993;Grootes and Stuiver, 1997

DISCUSSION
Carbonate content in marine environments is controlled by three factors: dissolution by corrosive bottom water, biological productivity in surface water, and dilution by terrigenous inputs.
Significant carbonate dissolution occurs below the carbonate lysocline, while carbonate is near zero or very low « 3%) near the CCD (Berger, 1975;Bramlette, 1961). In the SCS, the depths of the lysocline and the CCD are about 3000 m and 3500 m, respectively (Rottmann, 1979). At present, the aragonite compensation depth (ACD) in the SCS is about 1000-1200 m. The core MD972151 is located at a water depth of 1598 m, which is below the ACD but far above the carbonate lysocline depth. In the last glacial, the ACD and CCD deepened to 2500 m and 4000 m, respectively (Thunell et al., 1992;Miao et al., 1994). Although there is no infor~ mation about the carbonate lysocline in the glacial SCS, the depth of the carbonate lysocline in the glacial Pacific Ocean deepened as the ACD and CCD did (Farrell and PreU, 1989). An increase in the carbonate lysocline and CCD will have caused the biogenic carbonate fauna to be much better preserved Chen et a1., 1997) in the last glacial. Thus if dissolution is an important factor in controlling carbonate content of the core MD972151, the carbonate content in the last glacial interval should be greater than in the Holocene. However, this is not apparent in Figure 4. Therefore, dissolution is unlikely as a major factor controlling carbonate content in MD972151.
Previous studies have suggested that surface productivity increased when the winter monsoon was enhaneed in the last glacial (Thunell et al., 1992;Huang et al., 1997a, b). Moreover in the last glacial, a fall in sea-levelled to the exposure of a wide continental shelffrom which the ses derived many nutrients. This would also have favored an increase in biological productivity, and thus would have increased the biogenic carbonate in the core. However in comparison to the interglacial intervals, carbonate contents in the last glacial were relatively low ( Figure 4). Therefore, surface productivity is neither a major factor for controlling carbonate content in the study core.
Wide continental shelves (1,82 x10 4 km 2 ) occupy 52 % of the ses's total area (350 x10 4 km 2 ). They exist off the ehinese coast (67 xl Q4 km 2 ) in the north and off the Indochina Peninsula (Sunda Shelf,15 xl 0 4 km 2 ) in the southwest. At present, terrigenous sediments (200 xl0 9 Tons/year) derived from southern China are deposited on the northern shelf and slope via the Red River and Pearl River systems. Meanwhile, 160x10 9 Tons of sediments are annually transported via the Mekong River system from the Tibetan Plateau, and then deposited on the Sunda Shelf and its slope (Milliman and Meade, 1983). These present data indicate that sedimentation in the ses is greatly influenced by dilution of terrigenous inputs -the larger terrigenous inputs, the lower the carbonate content in the shelf-slope sediments. Furthermore, sealevel fluctuates in response to changes of ice volume. During an interglacial period, such as the present one, sea-Ievel rises due to contraction of the iee volume in the high latitude regions. In contrast, during the glacial periods, expansion of the ice in high latitude regions causes a fall in sea-Ievel and exposes the wide continental shelf. Thus much more terrigenous materials derived from both the continent and the exposed shelves, will be finally deposited on continental slope. This will strongly dilute biogenic carbonate components in the core MD972151 to as low as 3-5 % by weight in Stages 2-3-4 and 6. In comparison, during interglacial periods (sedimentation rate: Stage 1: 29.0 cm/kyrs and Stage 5: 10.6 cm/kyrs; Table 2), the Sunda Shelf was not exposed and even became a depositional site for the terrigenous sediments. Therefore, in the Stages 1 and 5 of the core MD9721511ess continent-derived sediments were deposited on the continental slope and the carbonate content is relatively high. Similarly, the ice volume factor also plays a major role in controlling millennial-scale variations of carbonate content in interstadial-stadial cycles in the last glacial. Exposure of the continental she1f remains in balance with fluctuation of both sea-Ievel and ice volume during the last glaciation. Much more of the shelf was exposed in the cool stadial event than in the warm interstadial event, and this led ta a much lower carbonate content in the cool stadial events.

CONCLUSION
Using multiple age constraints, carbonate stratigraphy of the deep-sea core MD972151 shows millennial-scale fluctuations which cOITelate with D-O cycles found in the Greenland ice cores. Carbonate contents in the core MD972151 are primarily contra lIed by fluctuations of ice volume. During the glacial and cool stadial events of the last glacial, extensive glaciation in the high latitude regions resulted in a low sea-level which exposed the wide Sunda she1f, and thus provided much more of terrigenous sediments to dilute biologieal carbonate in the continental slope of the SCS. In contrast, during the interglacial periods and wartn interstadial events of the last glacial, less glaciation in the ArcHc, Greenland, and Antarctic regions resulted in a high sea-Ievel, and therefore less teITestrial"sediments were transported to the continental slope and thus carbonate contents were high.