For the present study, late Pliocene sediment cores from ODP Site 1143 were sampled at 10-cm resolution. At a sedimentation rate of ~4.5 cm/k.y. (Shipboard Scientific Party, 2000), this equals a temporal resolution of ~2 k.y. Site 1143 is located in a water depth of 2772 m in the Nansha Islands or Dangerous Grounds area (Fig. F1). This area is riddled with reefs, shoals, and small islands. Some of the islands are within 20 to 30 mi of the site (Shipboard Scientific Party, 2000). Site 1143 is also located within the region of relatively stable and warm sea-surface temperatures, the West Pacific Warm Pool. The most important sediment source for the southern South China Sea is the Mekong River (Fig. F1), with a draining area of 810,000 km2, delivering 160 million tons of sediment per year (Milliman and Meade, 1983). Smaller rivers of Borneo may also contribute material to the Site 1143 area. During sea level lowstand, rivers from the emerging shelf have also contributed material to the sediments of the southern South China Sea (Molengraaff, 1921). Assuming a mineral aerosol flux equal to or less than the estimated modern value (0.5 g/cm2 per k.y.) (Duce et al., 1991), eolian sediment contributions, at maximum, may account for 12% of the total sediment accumulation (~4.1 g/cm2 per k.y.) (Shipboard Scientific Party, 2000).
A primary productivity of 300 to 500 g C/m2 per day has been observed for periods of upwelling (spring/summer) in the Nansha Islands area. The annual average is 100 to 200 g C/m2 per day (Guo, 1994).
Prior to chemical analysis, the samples were freeze-dried, ground, and homogenized in agate ball mills. For X-ray fluorescence (XRF) analysis, 600 mg of the sample powder was mixed with 3600 mg dilithium tetraborate (Li2B4O7, Spectromelt by Merck), preoxidized at 500°C with NH4NO3, and fused to glass beads in Pt crucibles. For the analysis of sulfur in three selected samples, a mixture of 50% lithium tetraborate and 50% lithium metaborate (Spectroflux by Alpha) was used as a flux. All the beads were analyzed using a Philips PW 2400 X-ray spectrometer. Calibration was done with a set of up to 50 carefully chosen international reference samples. Analytical precision, as checked by parallel analysis of one international (GSR-6) and several in-house standards, was <1% for major and <4% for minor elements (except for As and Co; 10%).
Total carbon (Ctotal) contents of selected samples were determined by coulometric titration following combustion with a Stöhlein-Instrument. Analytical precision of this method was <3%. Carbonate carbon (Ccarb) was determined by coulometric titration following release of CO2 with 2-N HClO4 at 70°C with a UI-carbon analyzer. The analytical precision of this method was <3%. Total organic carbon values were calculated as the difference between Ctotal and Ccarb (Prakash Babu et al., 1999).
Calcium carbonate contents were estimated for all samples based on the assumption that all Ca is present as calcite (CaCO3 = CaO x 1.7848). This approach is supported by the good correlation between CaO contents obtained by XRF and Ccarb values obtained by coulometry (Fig. F2).
Elemental contents were calculated on a carbonate-free basis for each sample according to the following equation:
where
For geochemical comparison and as an age model, we use the oxygen isotope record of
Cheng et al. (this volume) and the stratigraphy of
Tian et al. (this volume). The data set is based on measurements of the benthic foraminifer species Cibicidoides wuellerstorfi (one to three specimens of 0.3 to 0.9 mm in diameter) in a core sampling interval of 10 cm (Cheng et al., this volume). The chronological framework is based on the comparison of the benthic 
18O record with the 6-Ma composite oxygen isotope curve provided by Shackleton (Tian et al., this volume). Essentially, the part of the stratigraphy between 1.811 and 6 Ma is based on the comparison with ODP Site 846 data (Shackleton et al., 1995a, 1995b).
