It is well known that enrichments of Re, U, Mo, and As imply the depletion of O2 in bottom water and/or shallow pore waters. However, subtle geochemical contrasts among the group allow the relative intensity of oxygen depletion to be assessed. For example, while Re and U are chemically reduced and sequestered under suboxic conditions (where O2 is absent and nitrate reduction prevails), Mo is chemically reduced only at lower redox potentials (Calvert and Pedersen, 1993; Crusius et al., 1996), and As is concentrated in horizons of diagenetic pyritization (Thomson et al., 1998) where sulfate reduction is occurring at least locally. Thus, the resemblances between the depth profiles of Re and U and their enrichment throughout the sampled section (Fig. 3A) imply that the near-surface sediments at Site 1017 have been at least suboxic throughout the last 25 k.y. The more episodic Mo and As enrichments (Fig. 3B) reflect intervals when sulfate was being reduced in either the bottom waters or the near-surface pore waters. Further evidence for the primarily diagenetic origin of these four elements can be seen in the absence of dilution in the sandy layers, implying postdepositional addition and by the relative enrichments at 220 cmbsf where there is a Ccarb concentration peak (Fig. 3A, B, D).
A low Mo content compared to the Re enrichment in the sediments between 220 and 240 cmbsf implies that bottom waters were suboxic during deposition. Higher Mo concentrations in the deposits at 150 and 300 cmbsf indicate stronger O2 depletion at that time (Crusius et al., 1996). These Mo-enriched horizons are accompanied by high Corg, which suggests that the oxygen concentration in near-interface pore waters and bottom waters was depleted by an enhanced Corg settling flux. However, the lack of As and pyrite maxima at 150 cmbsf implies different processes of fixation for Mo and As. Molybdenum, for example, could be scavenged by labile organic matter still present above a "burn-down" level (Dean et al., 1997; Dean and Gardner, 1998), or it could have coprecipitated with FeS (Calvert and Pedersen, 1993) under conditions where FeS was not converted to pyrite. This could explain the lack of an arsenic concentration peak at this horizon if As enrichment specifically requires pyrite formation (Thomson et al., 1998). In MIS 2, the Mo and As distributions imply episodically low O2 conditions that are not associated with significant Corg maxima. This may suggest that short-lived intervals of oxygen depletion during MIS 2 were not induced only by an enhanced Corg settling flux but also by remote forcing of the O2 content of the intermediate or bottom waters at 1 km depth.
The close interrelationship of the Sr and carbonate carbon profiles (Fig. 3D) suggests that the variation of Sr is mainly controlled by changes in the biogenic carbonate content. A 1% difference in carbonate carbon corresponds to an 80 ppm difference in Sr, which is consistent with usual Sr concentration in biogenic carbonate (Graham et al., 1982). Approximately 200 ppm of Sr might be contributed by some other (probably detrital) phase or phases.
The distributions of elements such as Zr, Nb, Ba, Y, and Cr, which are highly enriched within sand layers and sandy patches (Fig. 4), clearly seem to be controlled by the proportion of coarse-grained terrigenous detritus in the deposits. Significant Cr enrichments relative to "average crust" imply contributions from mafic and/or ultramafic source materials (Cullers, 1994). In contrast, elements depleted within sand layers and patches such as V, Ni, Cu, Zn, Mn, Rb, and Ti seem to be contributed primarily by the fine-grained terrigenous fraction (Fig. 5A, B). Dean et al. (1997) pointed out the possibility that excess (over detrital) concentrations of Cu, Ni, and Zn were associated with an organic fraction based on interrelationships among element concentrations in Core F2-92-P29 collected from off southern California. However, the detrital fraction at Site 1017 is approximately twice as large as that at Core F2-92-P29, whereas the biogenic fraction at Site 1017 is half of that at Core F2-92-P29. Furthermore, concentrations of V, Ni, Cu, and Zn in Site 1017 sediments are much higher than expected based on the Corg content and consideration of the minor element contents in marine plankton (Piper, 1994). Thus, the concentrations and amplitudes of variations of these elements do not seem to be explained by contributions from the organic fraction. The enrichments of Cu, Ni, and Zn as well as V relative to "average crust" at Site 1017 could instead be explained by the contributions from mafic and/or ultramafic terrigenous materials (Cullers, 1994; Matsuo, 1989). Smaller amplitudes of fluctuation of Mn, Rb, and Ti (Fig. 5C, D) compared to V, Ni, Cu, and Zn (Fig. 5A, B) suggest that the former occur in both the fine and coarse fractions. Differences in the average concentrations of Cu, Zn, and Mn between MIS 1 and 2 suggest that the chemical composition of the fine fraction changed in the transition from the glacial to the interglacial. Finally, the common depletion of all 12 detrital elements at 220 cmbsf, where carbonate carbon is at a maximum, confirms their assignment to the detrital category.