Upwelling along the western coast of Africa south of the equator may be partitioned into three major areas, each having its own dynamics: (1) the eastern equatorial region comprising the Congo Fan and the area of Mid-Angola; (2) the Namibia upwelling system, extending from the Walvis Ridge to Lüderitz; and (3) the Cape Province region, where upwelling is subdued. The eastern equatorial region has very nutrient-rich thermocline waters, fed by subsurface eastward flow near the equator and by the Angola dome. An extensive oxygen minimum is present throughout the coastal ocean off Angola, which does not, however, have the highest productivity of the region. The thermocline waters of the Namibia upwelling system have but a modestly high nutrient content compared with waters off Angola, and concentrations decrease rapidly going southward from the Walvis Ridge. In essence, the pattern suggests diffusion of nutrients southward, from the Angola maximum, and erosion of poleward-flowing nutrient-rich waters by the Benguela Current as it sweeps past Namibia. The mobility of the phosphate seems somewhat greater than that of silicate in the southward migration (Fig. F4). Thus, the export production at the southern margin of the Namibian system is expected to contain a lower Si-to-C ratio than that in the north near Walvis Bay. The Cape Province upwelling region has relatively low phosphate and silicate values in subsurface waters and is correspondingly less productive.
The highest productivity is reached off Namibia between 20° and 25°S. Here, there is an optimum situation regarding the product of rate of upwelling and nutrient content of upwelled waters. Trade winds are strong, offshore transport is vigorous, and the cold water from below is high in both phosphate and silicate.
It is the product of upwelling and nutrient content that determines productivity, not just the rate of upwelling. Changes in any one of the upwelling regions, therefore, can be produced either by changes in wind strength (or direction) or by changes in the nutrient content of the thermocline "thermocline fertility"), or both.
In the deep-sea deposits of the eastern equatorial region, carbonate content is high for interglacial periods, while opal content is high for glacial intervals. The two components are precisely 180° out of phase, with opal expanding into the noncarbonate space during glacials. Regarding opal, the pattern in the sediments of the Congo Fan region is quite similar to that of the deep sea. (The carbonate pattern is not.) In addition, opal tends to follow organic carbon quite closely, with the proviso that organic carbon has a strong precessional component, while opal does not, instead reflecting general climate change (that is, a general oxygen isotope curve). Despite relatively low opal content, sediments off Angola show the same patterns as those off the Congo; thus, they are part of the same regime. The organic matter content is quite readily modeled as a combination of a global oxygen isotope curve and insolation forcing at low latitudes (15°N, almost all precessional forcing). The same model does not work for opal. Apparently, orbital forcing from high and low latitudes interacts nonlinearly for the silicate (high-latitude processes perhaps being responsible for thermocline fertility, low latitude for wind forcing). The result is a record that is difficult to model and whose spectrum shows lines well off the orbital ones.
On Walvis Ridge, as in the Congo-Angola region, the organic matter record can be modeled using a global oxygen isotope curve and combining it with precessional forcing. Glacial periods and periods of low summer insolation in the Northern Hemisphere are favorable for high export production. In contrast, interglacial periods are favorable for opal deposition. Thus, in moving from Angola to Walvis Ridge, the sign of the opal record reverses with respect to organic matter. The patterns suggest that silicate within the thermocline decreases off Congo and Angola during glacial times, but this decrease is insufficient to result in a reversal of patterns relative to organic matter. On Walvis Ridge, however, the decrease in silicate is sufficient to produce the reversal. The reversed phase (opal abundant during interglacials) persists throughout the entire Pleistocene, not just on Walvis Ridge but all the way to the Oranje River and off the Cape Province (Site 1085) (Anderson et al., Chap. 21, this volume).
It appears that glacial conditions led to a reduction of silicate in the thermocline all along the eastern margin of the South Atlantic, with the reduction weak in the eastern equatorial region but strong from the Walvis Ridge on southward. The situation is reminiscent of the North Pacific (Berger et al., 1997) and the west equatorial Pacific (Berger and Lange, 1998) but with the east equatorial Pacific being exempt, presumably because of silicate imported by the equatorial undercurrent.
The patterns suggest that the upper thermocline of the entire ocean was depleted in silicate relative to now during glacial periods. Presumably, stronger wind-driven mixing exposed silicate more readily to sunlight, which led to precipitation and removal. Increased supply of eolian dust may have helped the process. The lack of shelves further prevented trapping of opal at shallow depths, from where silicate can reenter the thermocline upon dissolution. In this scenario, the Walvis Paradox is not a local phenomenon but one with global implications.
The hypothesis of global thermocline depletion with silicate during glacial intervals does not preclude a regional increase in places where this is appropriate, as, for example, in the North Atlantic, which normally loses silicate through vigorous NADW production. The Walvis Hypothesis simply states that silicate is diminished in the global thermocline on average whenever winds become strong enough to shorten the residence time of silicate in upper waters substantially.
A central discovery of Leg 175 was the documentation of a late Pliocene opal maximum for the entire Namibia upwelling system. The maximum is centered on the period between the end of the Gauss Chron and the beginning of the Olduvai Chron. A rather sharp increase in both organic matter deposition and opal deposition occurs near 3 Ma in the middle of the Gauss Chron, in association with a series of major cooling steps. This association is also reflected in high P/Al and Ba/Al ratios at Site 1085 off the Oranje River (Murray et al., in press). As concerns organic matter, high production persists at least to 1 Ma, when there are large changes in variability, heralding subsequent pulsed production periods. From 3 to 2 Ma, organic matter and opal deposition run more or less parallel, but after 2 Ma opal goes out of phase with organic matter. Apparently, this is the point when silicate becomes strongly limiting to opal production. In contrast, carbon cannot become limiting, and phosphate (which is equally important to opal and organic matter production) remains sufficiently available to maintain high production, aided by increased mixing.
Maximum (carbon) productivity off Namibia is reached after the opal acme within the early Pleistocene. After that time, the quality of the upwelled water may have deteriorated further, for example, by increased erosion (by the Benguela Current) of poleward-moving waters bringing nutrients from the region off Angola and from the northern Namibia upwelling area. However, increased mixing of ever deeper waters during the high-amplitude glacials of the late Pleistocene tended to counteract the drop in thermocline fertility. The fact that (carbon) productivity stayed high from 3 to 1 Ma is also reflected in increased dissolution of carbonate in this interval off Namibia and in a number of productivity-related indices. NADW production is thought to be important in the delivery of silicate to the Antarctic circumpolar water body, which ultimately provides a source for silicate for intermediate waters. It was previously suggested that an Antarctic opal maximum is tied to optimum conditions for NADW production (Berger and Wefer, 1991). The hypothesis is based on the notion of silicate enrichment of Antarctic waters by import of NADW (called "firehose concept" by Boyle and Rosenthal, 1996).
The early Matuyama Diatom Maximum, or Namibia opal acme (NOA), owes it existence and position to sharply increased mixing and upwelling beginning around 3 Ma, owing to global cooling. The thermocline waters delivering upwelled waters were fed by an Antarctic system exceedingly rich in silicate, a condition inherited from a warm ocean and sustained by increased NADW production. As the ocean cooled, silicate content in the entire ocean dropped because of increased mixing and precipitation (in part from increased dust supply to the sea surface) and because opal dissolves more slowly in cold waters than in warm (for a number of reasons). The large-scale frontal activity that characterized the NOA retreated poleward with the spin-up of the Circumpolar Current. The result of diminished silicate supply and decreased frontal activity was a precipitous drop in opal supply to the seafloor ~2 m.y. ago off Namibia.
The above-stated hypothesis concerning the origin of the Namibia opal acme is not fundamentally different from the Walvis Hypothesis, stating that glacial conditions result in removal of silicate from the thermocline. By making use of the Monterey Hypothesis (Vincent and Berger, 1985), which states that increased upwelling produces a drop in atmospheric carbon dioxide, we can produce another corollary: whenever silicate becomes severely limiting for diatom production in most of the ocean, and especially in places of upwelling, the carbon dioxide content of the atmosphere has already been lowered. From this corollary, we deduce that there was a reduced level in pCO2 at ~2 Ma, with a sharp drop taking place between 3 and 2 Ma. (Since cooling produces upwelling and upwelling reduces pCO2, this is a positive feedback on climate change.).
Moore (1969) argued some time ago that the thinning of radiolarian shells in the Cenozoic has to do with the fact that diatoms remove silicate from the ocean. He saw the process in terms of evolution. The view here taken, in accord with the Walvis Hypothesis, is that a cooling ocean automatically results in a decrease of silicate, provided that diatoms are present (Berger, 1991). The great cooling steps of the Cenozoic are intimately related to the production of diatomaceous deposits (Barron and Baldauf, 1989). Increased mixing from increased wind-driven stirring will favor the growth of diatoms as long as silicate is available to make frustules. Localization of upwelling areas will concentrate the deposits such that remobilization is hindered. Cold water will slow reactions leading to the recycling of silicate (Kamatani, 1982), both in upper waters (slowing removal of organic protective layers on the shells) (Bidle and Azam, 1999) and in deep waters (slowing kinetics of dissolution). The overall preservation ratio (opal accumulation vs. production in surface waters) is thought to average near 3% (Tréguer et al., 1995). A small temporary increase in this ratio (say, from cooling) could quickly bring down silicate values, because of the relatively small reservoir of this nutrient. (The many intricacies and difficulties in modeling the Si cycle have recently been reviewed by Nelson et al., 1995; Tréguer et al., 1995; and Ragueneau et al., 2000.)
The geographic distribution of opal and organic matter in the present ocean (Fig. F38) shows the success of processes segregating opal from organic matter. Why should these patterns look different at all?
In a warm ocean, fractionation processes are difficult to visualize. In a cold ocean where production is highly localized in upwelling areas (the largest of which is the Antarctic upwelling ring), fractionation is straightforward. The greater difficulty in recycling silicate, as compared with organic matter and its phosphate, allows opal to become concentrated in regions where organic matter cannot stay because of reoxidation and reworking. Upwelling generates complex interactions between physical and biological processes. High productivity encourages "luxury feeding," which generates rapidly sinking fecal strings and pellets. Pulsed productivity generates senescence and consequent mass settling of diatoms. Frontal submergence brings diatoms into the dark, forcing their settling. In all these scenarios, the result is rapid removal of silicate from the upper water layer. (Thus, silicate is a crucial component of the biological pump, as emphasized by Dugdale and Wilkerson, 1998.)
In general, a cold ocean is much better at fractionating different types of biogenic sediments from each other than is a warm ocean. The contrast between high- and low-productivity systems is greater in a cold ocean, as is the range in density of water masses. Contrast in productivity and the effects of deep circulation, driven by density contrast, result in contrast in facies. In this very general sense, the Namibia opal acme, and other opal maxima in the latest Neogene in other regions of the ocean, mark the interval when a cooling ocean (tied to a strengthened wind field) selectively removes the excess silicate inherited from a warm ocean. When the excess silicate is removed, the process ceases.
The question about the starting time of upwelling has been discussed by a number of authors (e.g., Siesser, 1980; see list in Diester-Haass et al., Chap. 1, this volume). The deposits recovered during Leg 175 suggest shifts to higher production at the Tortonian-Messinian transition (in the late late Miocene, just before 6 Ma) and at 3 Ma, as mentioned. A shift toward overall lower production is indicated for the last million years. The record of the fraction of reactive phosphate at Sites 1082, 1084, and 1085 (Anderson et al., 2001) supports a statement that high-productivity effects begin to show at 6 Ma or later. This does not preclude a gradual ramping up from the middle Miocene into the late Miocene. A choice of a threshold within that interval (say, 10 or 8 Ma) might be possible but would be in need of additional justification beyond the available data.