RESULTS/DISCUSSION

Alkenones were not detected in samples older than 31 Ma; only diunsaturated ketones were detected in samples between 8 and 31 Ma, and both di- and triunsaturated ketones were present for the last 8 Ma (Table T1). A statistical analysis of the abundances of C_37:2 and C_37:3 as related to age is shown in Figure F4 and clearly suggests three significantly different time periods. Separate categories were used for samples containing both C_37:2 and C_37:3 alkenones, samples containing only C_37:2 alkenones, and samples containing no detectable alkenones.

Samples containing no detectable alkenones had a median age of ~30 Ma with a small range and variance (Fig. F4). It is possible that these samples represent the pre-alkenone conditions in the northern SCS where either coccolithophores had not yet started to produce the compounds (either due to temperatures that were too warm or due to evolutionary processes) and/or migration of the alkenone-producing species into the SCS had not yet occurred (S. Brassell, pers. comm., 2001). It is most likely that the SCS was just too warm during this early time for coccolithophores to produce either of the alkenones. Climate records show a general trend of cooling and increased variability since ~65 Ma, with long periods of steady conditions (Zachos et al., 2001). The Oligocene and early Miocene conditions were comparably "steady," whereas late Miocene records (from ~15 Ma) indicate a globalwide cooling that occurred through the Pliocene (i.e., global-scale Neogene cooling) (Zachos et al., 2001). Further, an alkenone study of the past 4.5 m.y. off the coast of southwest Africa showed steadily high SSTs (>25°C) from ~4.5 to 3.2 Ma and a decreasing trend from ~3.2 to 1 Ma (partial plot in Fig. F5B) (Marlow et al., 2000). Similarly, an alkenone study from 2.2 to 6.5 Ma off the northwest coast of Africa showed a cooling trend beginning around 3.2 Ma (Herbert and Schuffert, 1998).

The possibility that the lack of alkenone production was due to evolutionary processes cannot be ruled out. Recent studies have shown that alkenones appear in the sedimentary records first in high latitudes during the early Eocene and later in mid to low latitudes (S. Brassell, pers. comm., 2001). Further, the lack of production may just be due to the lack of alkenone-producing species, as the SCS is believed to have opened at ~32 Ma (Tamaki and Honza, 1991; Briais et al., 1993); thus, this may have been when alkenone-producing coccolithophores were introduced into the SCS. Finally, it is entirely possible that in this environment, these samples are just too old and alkenone concentrations are too low for detection.

Only C_37:2 was detected in samples from 8 to 31 Ma (Fig. F4). With the onset of marine transgression at ~32 Ma, alkenone-producing coccolithophores would have been established in the SCS by the time C_37:2 occurs in the sediments. There are several possible explanations for the presence of only C_37:2. If we assume the modern U^k´₃₇ SST calibration is valid, the first explanation would simply be that warmer SSTs (>28°C) existed and coccolithophores only produced diunsaturated ketones at these points in history. The occurrence of only C_37:2 is intermittent between ~18 and 23 Ma, which could further indicate an environment that was too warm for the production of alkenones most of the time but, at times, dropped below the required 28°C and allowed for the production of C_37:2. This could indicate that the SCS was completely free by this time to exchange warm water with the Pacific ocean and/or the summer monsoon system had developed, but a winter monsoon was weak, absent all together, or even fully developed but not detected through alkenone stratigraphy due to average seasonal temperatures >28°C. The second possibility is that the trend shows the evolution of alkenone compounds in coccolithophores. Because it is possible that SST controls on alkenone unsaturation developed as a response to Eocene cooling (S. Brassell, pers. comm., 2001), the large time span with only C_37:2 present in detectable amounts might be explained as an evolutionary adjustment. Additionally, C_37:3 compounds may exist in these samples but are undetectable because of very low concentrations and small preferential adsorption of C_37:3 onto the GC column (Villanueva and Grimalt, 1996). Even if C_37:3 is present in these samples and is adsorbed onto the GC column, the fact still remains that a U^k´₃₇ value cannot be determined because the sample consists of virtually only C_37:2.

C_37:2 and C_37:3 were detected in samples younger than 8 Ma (Fig. F4). The appearance of C_37:3 can be used to indicate either the further cooling of the northern SCS or the full evolvement of the temperature control on alkenone production by this time. The SST range for 0–2.5 Ma is between 19° and 26°C. U^k´₃₇ values and SSTs for these samples are plotted in Figure F3. The time resolution for this record is quite low (~50–100 k.y. per sample), so only limited interpretations are attempted here. Samples with ages older than 1.2 Ma give SST values of ~24°C with less variation. This is during the early stage of the winter monsoon intensification (An et al., 2001) where cold wind and waters from the north may not yet have had a significant effect on SST. For the last ~1 m.y., the U^k´₃₇ record shows large variations with a temperature range of 19° to 26°C (Figs. F3, F5). Although the lower resolution of the record does not allow us to clearly identify the glacial–interglacial stages within this time span, these temperature ranges are consistent with glacial–interglacial temperature ranges recorded in the northern SCS (Wang and Wang, 1990; Huang et al., 1997a, 1997b; Chen and Huang, 1998; Pelejero et al., 1999a, 1999b). Fluctuations in the SST record also show relatively good agreement with benthic foraminiferal (Cibicides wuellerstorfi and Uvigerina peregrina) ¹⁸O values for the past 2 m.y. from Site 1146 during the same ODP leg (Fig. F5A) (Clemens and Prell, this volume). Comparison of our record with an alkenone SST record from ODP Site 1084 off the southwest coast of Africa (Fig. F5B) (Marlow et al., 2000) shows similar SST trends, even though the absolute SSTs are different. Both show an increase in SST oscillations for the last 1 m.y. These marine records are also in agreement with terrestrial indicators of climate change (Cerling et al., 1997; Rea et al., 1998; Ma et al., 1998; An et al., 2001) and support the data that indicate an enhanced winter monsoon since the late stage of Himalayan uplift. Therefore, the most likely explanation for the detection of both types of alkenones is the cooling of the northern SCS because of intensification of the winter monsoon or even broader-scale ocean cooling due to an increase in Northern Hemisphere ice sheets (Shackleton et al., 1984; Raymo, 1994). In addition, sea level drops and the emergence of the islands surrounding the SCS (~6.5 Ma), which would have restricted the exchange of water between the SCS and western Pacific Ocean, could also have enhanced the cooling effect of the winter monsoon (Huang et al., 1997c).