DISCUSSION

The ¹⁸O stratigraphy and biostratigraphic datums reported here indicate a late Quaternary sequence at Site 1002 that is considerably older than what was initially thought when the "Site 1002"chapter of the Initial Reports volume was prepared and published (Shipboard Scientific Party, 1997b). Correlation of the ¹⁸O record to the SPECMAP standard now suggests a basal age for Site 1002 of ~580 ka, terminating in sediments deposited at about the MIS 15/14 boundary. Time series of tropical ocean and climate variability presently being generated from this site will provide important comparisons to other high-resolution records and should yield new insights into tropical climate processes and extra-tropical teleconnections.

As previously noted, Leg 15 scientists postulated that the rhythmic pattern of facies variations recorded in DSDP Site 147 sediments is the result of large-scale changes in climate and sea level associated with recent glacial-interglacial cycles. The correlations observed between the ¹⁸O record in Hole 1002C and the relative proportions of carbonate, TOC, and terrigenous material (Fig. 9), plus the regular appearance of diatom-rich intervals (Fig. 4), clearly support this general view. Given the relatively shallow topography of the shelf and banks surrounding the Cariaco Basin, it is logical to focus on glacioeustatic sea level change as a prime candidate for driving first-order changes in biogenic productivity and terrigenous sedimentation.

During glacial periods, the Cariaco Basin would have been increasingly isolated from the open Caribbean by lowered sea level. At the maximum lowstand of the LGM (-121 ± 5 m; Fairbanks, 1989), the principal connection between the basin and the open Caribbean would have been near the western end by Cabo Codera (Fig. 1) at a depth of <30 m, while the Tortuga Bank to the north and topography to the east between Margarita Island and the Araya Peninsula must have been at least partially emergent. Upon first consideration, the clear evidence for oxic LGM conditions in the Cariaco Basin at a time when the basin was so physically isolated is odd. Peterson et al. (1991), citing foraminiferal census data, attributed the oxic conditions of the last glacial to lower surface productivity and reduced oxygen demand resulting from a shift in the locus of upwelling seaward of the then exposed banks. Recently, Lin et al. (1997) and Haug et al. (1998) have extended the productivity argument by focusing on the more direct effect of reduced glacial sill depths on property distributions in the Cariaco Basin. At present, nutrients are advected from depths of 100 m or more to the surface, where they stimulate high productivity along the coast and over the basin. During the LGM lowstand, however, waters spilling over the sill and filling the basin could only have been derived from the upper 30 m or so of the open Caribbean water column, presumably from the very nutrient-depleted waters that characterize the surface. Thus, even if upwelling was physically active within the more restricted confines of the LGM Cariaco Basin, a likely scenario given evidence for stronger and perhaps even more zonal trade winds (e.g., Mix et al., 1986), the upwelled waters should have been nutrient limited to begin with and have sustained only low biogenic production. Such a situation is likely to have existed during earlier glacials as well, as appears to be reflected in the lower TOC and carbonate contents of the corresponding sediments.

As sea level rose during glacial terminations and the thermocline connection to the open Caribbean became re-established, nutrient supply would be expected to increase and surface production be stimulated. This is clearly illustrated for the most recent deglaciation by the sharp lithologic transition between Subunits IB and IA which occurred at ~12.6 ka. According to the reconstruction of Fairbanks (1989), sea level rose little in the early phase of deglaciation between 17.1 and ~12.5 ka. This interval was subsequently terminated by a rapid sea-level rise of 24 m in <1000 yr, a rise that correlates in time with peak discharge rates of meltwater (Meltwater Pulse 1-A) into the Gulf of Mexico as recorded by ¹⁸O records (e.g., Emiliani et al., 1975; Kennett and Shackleton, 1975; Broecker et al., 1989). The onset of high productivity in the Cariaco Basin, indicated by an abrupt increase in abundance of the upwelling-sensitive Globigerina bulloides (Peterson et al., 1991) and by high sediment TOC content and deposition of biogenic opal (Fig. 4, Fig. 9), reflects the development of more normal marine conditions as rising sea level enhanced connections with the adjacent Caribbean. Anoxia quickly set in as high productivity and an increased supply of organic detritus overwhelmed the capacity of circulation to ventilate the deep basin. The strong association of well-laminated, diatom-rich sediments and periods of maximum interglacial conditions following abrupt deglaciations (Fig. 4) is consistent with a model of sea level rise causing increased availability of nutrients.

Anoxia occurs whenever the rate of oxygen consumption exceeds the rate of oxygen supply. The presence or absence of anoxic conditions in the marine environment involves complex interactions and balances between biological, chemical, and physical-transport processes. Although efforts to date to explain the most recent transition from oxic to anoxic conditions (i.e., 12.6 ka) in the Cariaco Basin have largely focused on oxygen consumption through productivity-related mechanisms, the higher LGM salinities implied by the combination of ¹⁸O and alkenone data discussed earlier could also have helped to drive local downwelling that led to greater ventilation of the deep basin. Hólmen and Rooth (1990), based on the analysis of tritium distributions, have implicated the periodic input of warm, salty shelf water as an important ventilation source for the modern Cariaco Basin. Such a process is likely to have been enhanced during the LGM and earlier glacials, given the more restricted nature of the basin and the more arid regional climate (e.g., Van der Hammen, 1974; Schubert, 1988; Clapperton, 1993) that developed in response to a southward shift in the ITCZ and its accompanying rain belt.

In considering the potential role that salinity variations may play in ventilating silled basins, it is of interest to compare the Cariaco Basin setting with that of the much larger Japan Sea (present sill depths of <130 m), which was also relatively isolated during sea-level lowstands of the LGM and earlier glacials. In contrast to the Cariaco Basin, which is anoxic today but was oxygenated during the LGM, the presently well-ventilated Japan Sea experienced anoxic bottom conditions during glacial maxima as indicated by the deposition of dark, organic-rich laminated sediment layers (Oba et al., 1991; Tada et al., 1992). Very light ¹⁸O values recorded in planktonic foraminifers during the last glacial led Oba et al. (1991) to suggest that these layers were deposited when the upper water column became stratified by the influx of low-salinity waters, presumably derived from rivers like the Huang Ho whose input was diverted directly into the then more restricted basin.

Despite evidence for increased glacial aridity in northern South America, the relative increase in the terrigenous content of sediments in the Cariaco Basin during sea-level lowstands (Fig. 9) should probably come as no surprise. Today, sources of terrigenous material entering the basin include local rivers that drain directly onto the broad inner shelf region, such as the Manzanares, Tuy, Unare, and Neveri Rivers, as well as the major downstream rivers, the Orinoco and Amazon. During sea-level lowstands, contributions of fine-grained sediments sourced from the more distant Orinoco and Amazon Rivers are expected to have been reduced as the shelf narrowed and their sediments were discharged directly onto the Atlantic continental rise. Observations of Milliman et al. (1982) support this view; carbonate reefs buried beneath upper Holocene muds along the northern Venezuelan coast east of the Cariaco Basin yielded radiocarbon dates between 9.3 and 9.9 ka, leading these authors to speculate that demise of the reefs was caused by some combination of drowning and/or smothering by mud as rising sea level opened connections near Trinidad that permitted Holocene sediment transport from the Orinoco/Amazon systems to begin. Clay mineral studies of Site 1002 sediments indicate that the relative contribution from local rivers to the Cariaco Basin greatly increased during all glacials as lowered sea level reduced the width of the inner shelf from ~50 km to only a few kilometers (Clayton et al., in press). Given the proximity of the local river mouths to the edge of the basin at these times, a significant increase in the overall volume of terrigenous sediment input during glacials would probably have been expected.

The long-term ventilation history of the Cariaco Basin can be inferred from the variable distribution at Site 1002 of sediment intervals that are distinctly laminated and intervals that are massive or visibly bioturbated. This is illustrated in Figure 4 where the downhole occurrence of laminated sequences is shown relative to the ¹⁸O record of Hole 1002C. Counter to what might be predicted based on earlier discussion of the LGM, the long-term record of anoxia is clearly not a simple one related to glacial-interglacial extremes of sea level, productivity, and/or salinity; much of the time period spanning glacial MIS 3-4 is characterized by intervals of laminated sediments, whereas much of the sediment in interglacial MIS 5, 7, and 9 preserves a clear record of bioturbation. Laminated sediments make up much of the Site 1002 sequence deposited before MIS 9, but appear to be less commonly distributed in MIS 6-9. The reasons for this apparently complicated oxygenation history are not yet clear, but are the focus of active investigation. As data from our arsenal of biotic, lithologic, and geochemical proxies continue to grow in shore-based studies, we hope to better understand the relationships and connections between climate, sea level, and depositional history in this small, climatically sensitive basin.