Several Leg 175 sites show a distinct opal maximum within the upper Pliocene and lower Quaternary, spanning the lower half of the Matuyama reversed polarity Chron (Sites 1081, 1082, 1084, and 1085; see Fig. 14). Detailed onboard studies reveal that the maximum abundance of diatoms is reached within the early Matuyama Chron, centered on ~2.2 Ma (Fig. 15). At Site 1084, near the most active upwelling cell of the Benguela Current system, the diatom maximum owes its existence to a vigorous proliferation of Thalassiothrix and other pelagic species, in addition to the Chaetoceros spores typical for coastal upwelling. A mixture of warm-water and cold-water pelagic forms and of upwelling species suggests frontal developments and intense mixing in this region during the late Pliocene. At Site 1084, the rich supply of diatoms resulted in the development of diatom mats, reminiscent of those reported from the eastern equatorial Pacific (Kemp and Baldauf, 1993).
Patterns are similar for the Walvis sites, as mentioned earlier, except that the cold-water component is somewhat weaker and diatom mats did not develop (or were not noted). A maximum abundance of diatoms centered near the Pliocene/Pleistocene boundary was earlier reported for DSDP Site 532 by Leg 75 scientists (Fig. 16). This site is located on the Walvis Ridge at 1131 m water depth and was cored by the APC (Dean et al., 1984; Gardner et al., 1984). The patterns recorded at this site for carbonate, organic carbon, and diatoms yield valuable clues for the interpretation of the MOM (Fig. 16). From inspection, it appears that organic carbon and diatom abundances do not show similar trends (dilution effects may play a role in decreasing any correlation). Both the lowest and highest values for organic carbon are measured in the upper Pleistocene sediment, when diatom abundances are well off the maximum. On the whole, carbonate increases within the upper Pleistocene sediment, with the onset of large-amplitude ice-age cycles (i.e., after the mid-Pleistocene climate shift).
The late Pleistocene increase in carbonate values is here interpreted as a decrease in productivity by the arguments given above. A strong negative correlation between carbonate and organic carbon (which sets in just after the diatom maximum) supports this interpretation. As productivity decreases, the silicate supply diminishes more rapidly than the phosphate supply, a pattern that is common in most regions of the ocean (Herzfeld and Berger, 1993; Berger and Lange, in press). Significantly, opal deposition is reduced during glacials at this site (Diester-Haass, 1985), even though upwelling activity has most likely increased (Oberhänsli, 1991). What this means is that the intense late Pleistocene glacial conditions take the system beyond the optimum for opal deposition, which occurs at an intermediate stage of cooling.
Fundamentally, then, the Benguela Current system responded to cooling in the Pliocene with increased upwelling (and increased diatom deposition). However, when cooling passed a certain threshold, upwelling becomes less efficient in pumping nutrients into the photic layer. This drop in efficiency is likely tied to the nutrient content of subsurface waters (that is, to thermocline fertility), as suggested by Hay and Brock (1992). The passing of the system through and beyond a silica optimum would also explain why the correlation between temperature and opal abundance changes sign sometime within the Pliocene, as reported by Diester-Haass et al. (1992). On the warm side of the optimum, cooling produces additional upwelling and hence increases diatom supply to the seafloor. On the cold side of the optimum, cooling removes silicate from the thermocline, canceling any effects of increasing upwelling.
Why should "excess" cooling lead to a lowered silicate content in the Benguela Current system? The answer must lie within the processes supplying silicate (and phosphate) to the Benguela Current system. At present, maximum silicate and phosphate values in subsurface waters are found not within the Benguela Current system, but off Angola, well north of the Walvis Ridge. From there, nutrient-rich waters are brought southward along the upper slope with a poleward undercurrent, at least to the latitude of Lüderitz (Fig. 4). This mechanism of nutrient supply was identified by Hart and Currie (1960), who found evidence of a subsurface current with extremely low oxygen content flowing poleward along the edge of the continental shelf ("compensation current"). The current is centered at ~200–300 m depth and appears to be the replacement source for upwelled water (Shannon, 1985). That a strong oxygen minimum with high silica content does not build up within the Benguela Current system (we assume) is because ventilation by intermediate water currents prevents this: the poleward flow is restricted to a narrow band along the shelf and upper slope (it may be severely curtailed once sea level drops below the shelf edge). If such ventilation runs parallel to the strength of the Benguela Current (as seems reasonable), then intensification of the Benguela Current (and of the underlying northward-flowing intermediate water) would simultaneously result in decreasing the strength of the oxygen minimum, first off South Africa, and then off Namibia. In the extreme, only the region off Angola would continue to collect high-nutrient subsurface waters. Indeed, the correlation between opal deposition and organic matter deposition remains strongly positive off the Congo through glacial/interglacial cycles (Schneider et al., 1997).
The effects of increased intermediate-water currents on upper slope sediments may be seen at Site 1086. Here, the upper Quaternary record is missing entirely, and the upper part of the section (upper Pliocene to Pleistocene) is foraminifer rich; that is, it appears to be winnowed.
In summary, there is evidence for increased Pleistocene intermediate-water flow along the upper slope below the Benguela Current. Relaxing these currents (through general warming) should allow the Angolan oxygen minimum to expand southward. This could provide for increased opal productivity despite a decrease in coastal upwelling. This type of trade-off, we suggest, explains why there is a MOM in the early phase of the ice-age fluctuations.