Major element contents of the sediment (Fig. F7) at Site 1144 are very similar to those of a granite. This suggests that the detrital material of the northern part of the SCS could be derived from a siliceous igneous source with no major contribution of basic igneous rocks, such as the basaltic volcanoes of the Luzon arc (Fig. F1).
The Pearl River is the second largest river supplying the SCS in water discharge (320 km3/yr) with a modern sediment discharge of ~95 x 106 t/yr. Its catchment area (~442,600 km2) is mainly composed of Precambrian and Phanerozoic granitic rocks at the outcrop. The Pearl River thus appears to be the main contributor of detrital material to the northern part of the SCS. However, detrital material inputs from Taiwan ranges and/or from the East China Sea through the Taiwan Strait (Fig. F1) may also have contributed to Site 1144 sediments. Eolian sediment is an another potential source of detrital material to the SCS (Wang et al., 1999) and will be discussed below.
Numerous paleoclimatic studies have shown that clay mineral contents change through time as a result of climate modifications (Bouquillon et al., 1990; Fagel et al., 1994; Colin et al., 1999, 2001). To understand the meaning of changes in clay mineral composition, it is necessary to document the origins and potential source areas of minerals in the SCS sediment. Two groups of minerals can be distinguished:
Consequently, the smectite/(illite + chlorite) ratio can be used as a proxy for the intensity of chemical weathering and/or physical erosion on the continent. For Site 1144, the smectite/(illite + chlorite) ratio values exhibit a restricted range from 0 to 0.6 with no clear relation to glacial-interglacial changes except for a slight increase during the warm isotopic Substages 1, 5.1, 6.3, 6.5, and the end of MIS 7.
In order to assess the degree of chemical weathering of terrigenous detritus experienced prior to marine deposition, the chemical index of alteration (CIA)
introduced by Nesbitt and Young (1982), was also calculated from major element results (Fig. F8). This last parameter quantifies more precisely the effect of chemical weathering on the rocks by loss of the labile elements Na, Ca, and K. CIA for all feldspars = 50%, and the mafic minerals biotite, hornblende, and pyroxenes have CIA values = 50%-55%, 10%-30%, and 0%-10%, respectively. The secondary clay minerals and chlorite CIA values = 100%, and illite and smectite CIA values = 70%-85%. Consequently, the CIA reflects the proportions of primary and secondary minerals in bulk samples. It has been used as a climatic indicator in the Andaman Sea sediment where it is well correlated with the mineralogical record (Colin et al., 1998). CIA values obtained from Site 1144 exhibit a restricted range between 73% and 79% (Fig. F8), typical of altered rocks, suggesting low variations of the proportion of primary and secondary minerals in bulk sediment. This is in agreement with the small changes observed in the smectite/(illite + chlorite) ratio. A slight increase in the CIA values can be observed only in the earliest Holocene, during the warm MIS 5.1 and 5.3, and during the MIS 6-MIS 5 transition. At Site 1144, the sediment does not seem to have recorded any important changes in the intensity of the chemical weathering affecting the continent.
From the quartz concentration records through the last 180 k.y., displayed on Figure F9, it appears that minima coincide with interglacial periods, whereas during isotopic Stages 2, 3, and 6, quartz values are higher. There is no clear correlation between quartz and specific clay mineral composition (Fig. F6). However, long-term fluctuations of quartz are well correlated to SiO2 variations (Fig. F9). An increase in quartz is characterized by an increase in SiO2, suggesting that variations of SiO2 mainly depend on quartz input. The SiO2/Al2O3 ratio can also be controlled by hydrolysis during chemical weathering. As CIA values and clay distribution exhibit minor changes, we suggest that this ratio mainly reflects changes in the proportion of silt quartz and clay mineral content proportions. An increase in the SiO2/Al2O3 ratio is attributed to an increase of the quartz proportion and a decrease of the aluminosilicate fraction. As K2O is mainly linked to the aluminosilicate fraction, variations of the SiO2/K2O ratio could also signify the same pattern.
Site 1144 grain size variations are similar to those obtained from cores 17940 and 17939 (Fig. F1) collected during the Sonne-95 cruise in the same area (Wang et al., 1999). In core 17940, the clay (<6 µm fraction) and silt (>6 µm fraction) contents present the same variations as those at Site 1144 for the last 40 k.y. The Holocene period is characterized by higher proportions of clay (70%-75%) than during the last glacial maximum (LGM) (50%-55%). Based on a simple empirical relationship (Koopmann, 1981) between the percentage of siliciclastic fine fraction (<6 µm) and the primary modal grain size of siliciclastic silt (>6 µm), Wang et al. (1999) distinguish two different sources for the clay and silt fractions. Clay and silt fractions have been mainly attributed to fluvial sediment supply and eolian input, respectively. The intensity of wet summer and dry winter monsoons thus was reconstructed from the history of continental aridity in South China, which in turn controls the fluvial and/or eolian sediment supply to the continental margin of the Pearl River mouth. On a global scale, an increase in summer monsoon rainfall implies an increase in clay supply by the Pearl River, whereas drier conditions associated with an intensification of winter monsoon transport carry more eolian loess to the northern part of the SCS.
During the last 180 k.y., grain size changes are correlated to quartz and SiO2/Al2O3 variations (Fig. F9), suggesting that the silt fraction (20- to 40-µm grain size class) is mainly dependent on quartz input. Quartz is a common mineral in loess deposits of Central China (Porter and An, 1995) and has been widely studied on land or in marine sediments as an indicator of continental aridity and winter monsoon strength (e.g., Rea and Leinen, 1988; Xiao et al., 1995; Wang et al., 1999). Glacial stages are characterized by an enhanced winter monsoon that drives eastward eolian particles from the Central China deserts to the Pacific Ocean (Pettke et al., 2000; Jones et al., 1994). These may have an eolian origin at Site 1144, already supposed by Wang et al. (1999). Glacial stages are associated with a decrease in the summer monsoon rainfall and/or an increase in the winter monsoon transport. The long-term grain size increase observed between 1000 and 600 ka would be associated with global cooling, inducing drier conditions and/or an increase in the winter monsoon transport.
However, this increased supply of quartz and silt size class appears also to be correlated with a time of lowered sea level, which implies that sea level changes could also have an effect on detrital material transport from the continent or from the shelf to the deep ocean. The northern SCS shelf is wide at the Pearl River mouth and is mainly composed of sandy sediments (Wang et al., 1992). The distance from the shore to the -120-m bathymetry corresponds to ~250 km (Fig. F1), and a large part of the continental shelf emerged during the glacial low sea levels (-120 m for the LGM) (Wang et al., 1995). During high sea level stands, sediment could have been impounded on the shelf and coarser clastic deposition cut off from the margin. In contrast, during low sea level stands, the shelf was exposed to erosion and remobilized sediments were then redeposited in the deep sea. This is in agreement with higher sedimentation rates during glacial periods (Fig. F2) as well as with a lower reflectance (L*, lightness) than that during interglacial stages (Fig. F9). This last parameter often increases with increasing carbonate calcium content (Fig. F9) and, therefore, could be a proxy of either the CaCO3 productivity by both benthic and planktonic organisms or the detrital input. Consequently, glacial times are expected to show an increase of terrigenous supply to the SCS. In addition, during low sea level stands, the Pearl River mouth is located closer to Site 1144 than during high sea level, suggesting intensified bypass of suspended matter from shelf/slope into the basin. This would transport higher proportions of silt to the continental margin in front of the Pearl River mouth.