RESULTS: CLIMATE- AND PRODUCTIVITY-RELATED PROXIES

The record of oxygen isotope ratios that emerges after age assignment as described may be taken as depicting climate state (Fig. F3). Well-expressed cycles are seen in Core 175-1085A-7H. In the older sediments, cyclicity is less well expressed and overall the amplitudes tend to be lower as well. The record of Core 175-1085A-8H has substantial gaps owing to the scarcity of C. wuellerstorfi in certain intervals. (In what follows, this is remedied by merging with adjusted Uvigerina. We have not assigned isotope stages to these data, as the precise stratigraphic position of the records within the cores is in some doubt and is immaterial to the argument of this paper.)

The record of the carbon isotope ratios reflects both deep-ocean ventilation and the regional addition of carbon dioxide from the combustion of organic matter on the way to and on the seafloor. The carbon isotope series is seen to be shifted by somewhat different amounts, with respect to oxygen isotopes, in the three cores for which data are available (Fig. F4). The shift consists in a slight lag of ¹³C, from which one might tentatively conclude that the carbon cycle (and deep ventilation) follows rather than leads climate change. However, the relationship of the carbon isotope record at any one location to the global change in carbon cycling (and atmospheric carbon dioxide) is not readily ascertained.

As described in "Materials, Methods, and Data," the cycle closest to a 40-k.y. period was used for age assignment. Among the four cores studied, Core 175-1085A-7H has the most prominent 41-k.y. cycle, as readily seen in the Fourier spectra (Fig. F5). The record of Core 175-1085A-7H also has the only cycle that might qualify as a ~100-k.y. eccentricity cycle, as well as a possible 23-k.y. cycle. No claim is made (or seems advisable) regarding the presence or prominence of cycles in the other cores, given this limited analysis. Neither is this necessary given the purpose of the study.

The index BF/g is a stand-in for the accumulation rate of benthic foraminifers, which in turn is closely related to the supply of organic matter to the seafloor, and therefore, to export production (Altenbach and Sarnthein, 1989; Herguera and Berger, 1991; Loubere, 1994; Fariduddin and Loubere, 1997). We do not use benthic foraminifer accumulation rate (BFAR) because we have not determined changes of sedimentation rate within the 41-k.y. cycles. We assume that BF/g reflects accumulation rate sufficiently well to show the glacial-interglacial contrast in productivity, if any. In principle, the higher the surface productivity, the more carbon reaches the seafloor, and the more benthic foraminifers can thrive (Herguera and Berger, 1991).

In Core 175-1085A-7H, the abundance of benthic foraminifers expressed on a logarithmic scale is perfectly in phase with the oxygen isotope curve, such that warm intervals show low abundance and cold intervals high abundance (Fig. F6A). The regression coefficient is near 0.4 and is highly significant. In the other cores, the same overall relationship is observed but the details are more complicated (Fig. F6B, F6C, F6D). In particular, there is a tendency for peak highs and lows to occur not in line with climate extremes (as seen in the oxygen isotopes) but offset by between 5 and 10 k.y.

EDA, being closely related to diatom abundance, should provide an excellent index for productivity. Estimated diatom abundance on the present-day seafloor is clearly tied to upwelling regions (e.g., Barron and Baldauf, 1989; Baldauf and Barron, 1990; Charles et al., 1991; Berger and Herguera, 1992). We first check the relationship of EDA to the glacial-interglacial cycle as seen in Core 175-1085A-7H (which shows the strongest cycles). To simplify the task of finding phase relationships, we only consider cycles in the 41-k.y. band, which we generate by inverting the portion in the Fourier matrix that surrounds the 41-k.y. cycle (Fig. F7A). After such filtering, different indices are readily compared. Results of the Fourier analysis show a negative correlation between ¹⁸O and EDA, at least for the upper half of Core 175-1085A-7H in the central part of the Pleistocene (see Fig. F7B). This result is reminiscent of the situation reported for the Walvis Ridge (Deep Sea Drilling Project Site 532) (Hay et al., 1984; Diester-Haass 1985). The reasons for the negative correlation are not clear. Taking the data at face value, either productivity is lower during glacial times than during interglacials (contrary to expectations and to the results from the abundance of benthic foraminifers) or the index EDA does not reflect overall productivity (see "Discussion: Walvis Paradox and Phase Relationships").

The ¹³C record of benthic foraminifers is related to oxygen consumption and nutrient content in deep waters (Berger and Vincent, 1986; Zahn et al., 1986; McCorkle et al., 1990), export production (Mackensen and Bickert, 1999), and exchange processes between ocean and atmosphere (Charles et al., 1993). To make some simple comparisons regarding glacial-interglacial contrast, we selected 11 samples from Cores 175-1085A-7H and 10H for detailed analysis; six of these have high ¹⁸O values for C. wuellerstorfi, and five have low values (10 percentile). The samples were analyzed for the corresponding isotope values for Uvigerina peregrina, and these values were use to define the environmental space.

Typical interglacial values for ¹³C of Uvigerina (low ¹⁸O in C. wuellerstorfi) scatter around -0.9, and glacial values around -1.15, for a difference of 0.25. This is slightly more than one-half the glacial-interglacial contrast in deep water for the late Quaternary in this setting (Mackensen and Bickert, 1999) but much less than the overall range expected (Raymo et al., 1997). The change in deep water composition is a general background signal presumably related to glacial input of terrigenous organic carbon (Shackleton, 1977) and to changes in deep ocean oxygen utilization and ventilation (Boyle, 1988; Raymo et al., 1997). From the diminished glacial-interglacial contrast observed, it can be surmised that the oxygen utilization within the water bathing the seafloor at Site 1085 was decreased during maximum glaciation in the early Quaternary.

To test whether EDA is more closely related to Uvigerina ¹³C than to Uvigerina ¹⁸O, with which it is negatively correlated (Fig. F7), we chose 12 samples, 6 with high and 6 with low values of EDA, and which also have isotopic values for Uvigerina. High EDA tends to occur with low ¹⁸O of Uvigerina and high ¹³C of Uvigerina, indicating that diatom production is high during interglacial conditions.

Twelve samples, six with high and six with low values (10 percentile), were selected for investigation of SOA patterns. Only samples containing Uvigerina were chosen, and the Uvigerina specimens picked were analyzed for stable isotopes. High SOA values (mainly pyrite) go with interglacial conditions and vice versa. Surprisingly, there is more coarse pyrite in the interglacial.

Why should there be more SOA associated with the interglacial periods? Several reasons are possible. There may actually be more pyrite in the glacial sections, but it may be more finely disseminated and not emerge in the coarse fraction. Or it may be diluted by increased terrigenous input. Delivery of iron may change greatly on a glacial-interglacial timescale, with chemical weathering (and supply of SOA) less important during glacial periods. Finally, conditions of diagenesis may change markedly from glacial to interglacial conditions. When productivity is high, iron (hydr)oxides are quickly reduced and made mobile, and reduction of sulfate provides sulfide to trap iron and related SOAs. With so many factors potentially playing a role, SOA cannot be expected to be a good productivity indicator. In any case, according to these results, coarse SOA parallels EDA, that is, it is anticorrelated with oxygen isotopes, being high in interglacial periods and low in glacial intervals.