CAPE BASIN GROUP(SITES 1085, 1086, AND 1087)

General Aspects

The Cape Basin group comprises Sites 1085, 1086, and 1087. They were drilled off South Africa, south of the Oranje River, just north of the Cape of Good Hope. They are situated south of the "Lüderitz boundary," in the so-called "Southern Benguela Region" (Giraudeau, 1993; Dingle 1995). The sites are some 2000 km south of the port of Benguela in Angola and well south of the Namibia upwelling system; they are part of a system including incursions from the South Atlantic Current and eddies from the Agulhas Retroflection. Upwelling is much less prominent than off Namibia. This environment might be reasonably referred to as the "Cape Region," since it is part of the Cape Basin and off the Cape Province of South Africa (Fig. F33).

The prevailing facies in the three cores, despite their proximity to the continent, is nannofossil ooze, a pelagic deposit. Diatoms, radiolarians, and other opaline fossils generally are rare or absent, further demonstrating the lack of a strong upwelling signal. (However, diatoms were indeed present in the early Matuyama at all Cape sites, indicating that the early Matuyama Diatom Maximum extends to these sites as well.) At Site 1086, which is quite shallow (780 m), there is a strong showing of foraminifers, especially toward the top portion of the section, which seems strongly winnowed (much of the late Quaternary is missing altogether). The overall rate of deposition is near 3 cm/k.y., but sedimentation rates tend to decrease in the upper Pliocene and in the Pleistocene. At Site 1085, the calcareous ooze is clay rich and olive in color, as appropriate for a strong hemipelagic influence. The site is on the southern rim of the Oranje River Fan; its sedimentation rate is near 5 cm/k.y., intermediate between hemipelagic and pelagic deposits. Site 1087, at slightly shallower depth than Site 1085 (1380 vs. 1710 m), has similar facies, similar sedimentation rates, and similar olive hues, at least in the late Neogene section.

Carbonate and Organic Carbon Record

An important question is whether the Cape group shows roughly the same patterns, as far as carbonate deposition and organic matter deposition, as the group of cores off Namibia or whether there are striking differences. If similar, we may take this as a hint that wind forcing (productivity) and nutrient supply (thermocline fertility) act similarly all along the coast of southwestern Africa. If different, we might discover that local mechanisms of control on production of sediment are much more important than the concept "Benguela Current System" would imply.

The carbonate records of the three Cape sites are sufficiently similar to allow the recognition of general patterns (Fig. F34). Carbonate is plotted as the log of the ratio of carbonate to noncarbonate to avoid the deceptive nonlinearity of values above 80% as they approach the limit. A value of unity corresponds to Cb/(1 - Cb) = 10; that is, Cb = 10/11 = ~91%. After a start with very low carbonate values near the late Miocene/middle Miocene boundary and in the earliest late Miocene (Carbonate Event CX, well known as a CCD excursion upward in the South Atlantic), there follows a period of high variability within the late Miocene. At ~8 Ma, there is a distinct and quite abrupt rise to a new regime of high carbonate values (Carbonate Event CVIII; marked "shift" in Fig. F34). Maximum carbonate values are attained near 7 Ma, within the late Miocene (CVII), followed by a shift downward near 6.25 Ma (CVI). A highly variable regime follows, apparently with maximum variability around 5 Ma (CV). The productivity shift seen off Namibia is here marked by maximum variability in the carbonate content (CIV). Low values prevail in the early Pleistocene, especially in the earliest part of the epoch (CIII). A shift to higher values occurs almost precisely at 1 Ma (CI). The sequence of events is not fundamentally different from that seen in the Walvis group (Fig. F28) back to 6 Ma.

The organic carbon record (Fig. F34B) poses no surprises in that it fluctuates counter to the carbonate values: carbonate maxima correspond to organic minima and vice versa. Also, the position of shifts is fairly congruent: a drop near 8 Ma (at Site 1085), a rise at 6.25 Ma (at Sites 1086 and 1087), and a drop near 1 Ma—all superposed on a general rise in organic matter content after 10 Ma (in good agreement with Siesser, 1980). However, the resolution here achieved through stacking of the cores (albeit still highly unsatisfactory) allows the statement that productivity increased in steps and that the step just before 6 Ma (CVI) and the one at 3 Ma (CIV) were the most important, with both steps followed by increased variability. Regarding the time after 3 Ma, there is a "pulling down" of values by Site 1086, presumably as a result of increased winnowing. On the whole, the Cape group record can be reconciled with that of the Walvis group, or more cautiously, differences cannot be demonstrated given the spotty data, while the overall trends seem similar.

The paleoproductivity of the late Miocene is explored by Diester-Haass et al. (Chap. 1, this volume) using materials from Site 1085. A general increase in carbonate deposition at this site is well documented, beginning near 7 Ma. This trend and associated long-term cycles in carbonate preservation likely affect the various productivity proxies based on calcareous fossil abundance.

Productivity Record of Site 1086: Preliminary Results

The question regarding the similarity of productivity patterns recorded at the Cape group sites and at the Walvis group sites is unresolved. It is unlikely that it can be answered convincingly on the basis of the available data, for which sample spacing invites severe errors from aliasing (that is, sampling cyclic sediments in ways that prevent random representation of conditions). To escape from this conundrum, high-resolution sampling is needed. Such sampling was done for Site 1086 for the section corresponding to the last 4 m.y. As mentioned above, the late Quaternary record is missing from Site 1086, presumably owing to winnowing, as suggested by the high content of sand in the uppermost cores (Fig. F35).

To obtain diatom-independent information on productivity (for the purpose of assessing the circumstances of the Namibia opal acme), we turn to the faunal composition of benthic foraminifers. Foraminifers are the main component of the sand fraction. Whatever biological processes affect foraminifers also affect the sand content, and whatever physical processes affect the sand fraction also affect foraminiferal abundance and composition. The foraminifer data reported on in this section are part of a larger set of counts by Susan Burke, at SIO. A more comprehensive account of these results is in preparation for publication in the SIO Reference Series.

The sand content at Site 1086 is usually below 9% in the late Miocene and early Pliocene (index of <-1 for the sand% index) (Fig. F35). It rises above background at ~3.5 Ma, reaches a maximum between 3.3 and 3.4 Ma, drops slightly toward 2.5 Ma, and rises after that to rather high values near 50% (for which the log index is 0). Assuming that the BF/PF ratio (and therefore its logarithm) is a proxy for productivity (shaded line in Fig. F35A), we can ascribe the decrease in sand content between 3.2 and 2.5 Ma—against a background of generally elevated values and a rising trend—to depression of sand content by dissolution of foraminifers. If planktonic foraminifers are (on the whole) more easily dissolved than benthic ones (as seems likely) (Parker and Berger, 1971), the BF/PF index would be enhanced by this effect through dissolution of carbonate by the carbonic acid evolved from the combustion of organic matter. High productivity, then, seems indicated for the intervals between 3.2 and 2.5 Ma and very low productivity for the interval between 1.8 and 1.2 Ma. This appears to conflict with the patterns for the Walvis group, reported above. A period of high productivity between 3 and 1 Ma was suggested for that group, with values peaking in the early Pleistocene.

We suggest this discrepancy in the apparent productivity patterns may be owing to the winnowing effect: winnowing removes organic carbon and moves it downslope. Clearly, sediments in the late Pleistocene at this site have no carbon at all, simply because there are no sediments left behind. Because of apparent strong winnowing in the entire Quaternary, useful data regarding the export production of organic carbon (or of diatoms) are unlikely to emerge, even from detailed studies. This is true because winnowing affects the very surface of the sediment and thereby also the living conditions of benthic foraminifers, which will simply record that there is little organic matter around in this region of strong currents.

Additional clues to the presence of organic matter at the interface, at least sporadically, can be found in the abundance of Hoeglundina elegans, a benthic foraminifer with an aragonitic shell (Fig. F35B). Interestingly, it is actually quite abundant (in proportion to its own record, that is) within the interval between 3.2 and 2.5 Ma, which the BF/PF ratio indicated as a time of high production. In contrast, it goes to zero between 2.4 and 2 Ma during the very time of the MDM off Namibia. Throughout the rest of the section (all the Quaternary present), Hoeglundina comes and goes, but it has pulsed maxima around 1.5 Ma. It seems to us that these particular data are compatible with maximum productivity during the Namibia opal maximum off the Cape Province and rather low productivity before and after. We then have two possibilities from reading these high-resolution records: (1) high productivity between 3.2 and 2.5 Ma (early in the Namibia opal maximum) or (2) high productivity between 2.4 and 2 Ma (late in the Namibia opal maximum and congruent with its center).

A useful index of productivity is the accumulation rate of benthic foraminifers (Herguera and Berger, 1991), which reduces to BF/g for sections without variations in sedimentation rates. The use of mass accumulation rates (rather than content) has one drawback: after multiplication with a series of instantaneous sedimentation rates, all indices in a sequence tend to look similar whether the rates are correct or not. Thus, for qualitative characterization, the use of indices based on ratios is much safer than the conversion to mass accumulation rates. Comparing the index for benthic foraminifer abundance (BF/g) with the BF/PF ratio (Fig. F36), we find that the two track quite nicely over much of the record, with the exception of the early part of the Pleistocene. Here, planktonic foraminifers are unusually well preserved and depress the BF/PF ratio even though benthic foraminifers are quite abundant. Of course, since they are components of the sand, they increase with sand content and hence with winnowing. Correcting by this effect, we would see that productivity is in fact low within the Pleistocene and presumably for the reason given: intense winnowing. That winnowing is a source of confusion is also suggested by the fact that benthic foraminifers seem more affected by changes in the Pleistocene than planktonic forms (Fig. F36B). Strong winnowing removes their food. But for this fact, they might just follow sand content as do the planktonic species.

The abundance of ostracodes follows rather closely that of benthic foraminifers (Fig. F36C), presumably as a result of the fact that ostracodes likewise are a component of the sand and are subject to similar effects from removal of nourishment through winnowing.

Different species of benthic foraminifers prefer different environments (Lutze, 1980; Hermelin, 1992; Burke et al., 1993; Loubere, 1994; Fariduddin and Loubere, 1997). Such preferences can be used to construct productivity indices. A good assumption is that benthic foraminifers that are abundant on upper-slope environments in areas of high productivity but rare in the deep sea make excellent markers of high organic carbon supply to the seafloor. Thus, the more abundant species at Lüderitz Site 1084 that are also less abundant in the pelagic environment (e.g., Site 1087) would qualify as productivity indicators.

One of the more abundant species indicating elevated productivity in a pelagic setting is the genus Uvigerina. (It has also been used to indicate reduced oxygen supply) (Burke et al., 1993.) Uvigerina follows the pattern of BF/PF rather closely, except at two intervals in the record (Fig. F37A). The two exceptions are in the early Pleistocene and between 3.2 and 2.9 Ma. In the early Pleistocene, the abundance of Uvigerina suggests that this interval was not one of unusually low productivity, which should have reduced abundance. A minimum does persist near 1.5 Ma, however. The discrepancy near 3 Ma is unexplained. Low Uvigerina content is supported by high values of Cibicides, an open-ocean form, indicating reduced supply of organic matter. However, both indices are contradicted by the rather high BF/PF values and the BF/g maximum centered near 3 Ma (Fig. F36). One possibility is that winnowing during this period (~3 Ma) is responsible for the high values for BF/PF and BF/g and that they say nothing about productivity in this particular setting. Additional checks will be necessary to confirm that this anomaly is real.

Two genera that are very much at home in high-production regions are Bolivina and Bulimina, while species of the genus Cibicides (or Cibicidoides or Planulina) are generally taken to be adapted for the pelagic realm (where they record bottom-water properties in their shells). A comparison of the percentages of these forms at Site 1086 is instructive (Fig. 37B). On the whole, Bolivina and Bulimina are low where Cibicides has high abundance, and vice versa. The most impressive shift in abundance is near 3 Ma, where Cibicides suddenly drops to very low values, within 200 k.y. or so, and Bolivina and Bulimina begin their increase to an overall greater presence within the section. Starting at the Pliocene/Pleistocene boundary, the two groups show positive correlation in places. We interpret this as an expansion of the range of productivity on scales too small to see in these data. When times of high productivity alternate with times of low production, both groups can thrive apparently simultaneously (since each sample is a mixture comprising many centuries). A low-productivity interval centered on the early Pleistocene seems supported by these indices.

If Bolivina and Bulimina respond to high production and Cibicides spp. to low, a ratio between the two should be especially sensitive to productivity variation (Fig. F37B). A plot of this index (log scale) confirms that the biggest step in the last 4 m.y. is the one from low productivity before 3 Ma to high productivity beginning at ~2.7 Ma and persisting since. An impressive peak is seen centered on 2.6 Ma, and a precipitous drop to modest values occurs at 2.5 Ma. The reduced ratio persists to ~2.1 Ma. In essence, the reduced productivity (within the generally elevated background) is congruent with the Namibia opal maximum to the north. The discrepancy is intriguing. Either the mixed layer thickened in this region (for example, by pulling in Agulhas waters around the Cape, with increased winds) or the nature of the food changed unfavorably for the Bolivina and Bulimina group. We think the first alternative is more likely. On the whole, the index offers no evidence for reduced production within the Pleistocene, either around 1.5 Ma or as a trend. Taken together, the abundance patterns of Uvigerina and of Bolivina and Bulimina vs. Cibicides suggest that there was one big change in productivity near 3 Ma and that high production of organic matter persisted ever since, possibly with an increased range of fluctuation. It is noted that the BoBu/Cib index backs up the reality of the strangely low values of Uvigerina around 3 Ma, to a degree.

In summary, the detailed data gathered for Site 1086 suggest that there are two regimes of productivity in the Cape Province region south of the Oranje River: a time of low production in the early Pliocene ending near 3 Ma and a time of high production since, with a peak centered on 2.6 Ma and a minimum lasting from 2.5 to 2.1 Ma. It appears therefore, that the Namibia opal maximum was inaugurated by upwelling both off Namibia and off the Cape Province but that the southern upwelling shut down again while the Namibian upwelling system kept ramping up. We propose that the increased frontal activity off Namibia during the early Matuyama was accompanied by increased import of warm water into the Cape Province region, presumably Agulhas eddy water and water from a strengthened South Atlantic Current. Such input must have been pulsed and alternating with Southern Ocean input to account for the occurrence of Antarctic diatom species in the sediment. The development of fronts (convergences) between nutrient-rich cold and nutrient-poor warm water masses is known to result in intense pulses of diatom sedimentation in the eastern equatorial Pacific in the Neogene (Kemp and Baldauf, 1993).

Notes on the Agulhas Eddy Effect

The contribution of warm water from the Indian Ocean to the Benguela Current in the form of numerous large eddies rounding the Cape (Fig. F33) has been well appreciated for some time. Estimates of the flow of water vary greatly; in recent years values between 6 and 8 sverdrup have been suggested (see reviews by Lutjeharms, 1996; Shannon and Nelson, 1996). If this is so, the Agulhas source contributes substantially to the water volume of the Benguela Current and to the heat transported northward by this current.

The history of the contribution of warm water from Agulhas eddies to the great current off Namibia is of central importance in understanding the history of productivity of the coastal ocean off southwestern Africa. It may be surmised from first principles that this contribution goes through an optimum when conditions are "just right," and it may also be assumed that conditions are not precisely at optimum right now, by happenstance. The strength of the Agulhas Current, a western boundary current like the Gulf Stream and the Brazil Current, depends on the wind field and hence on the temperature gradient in the Southern Hemisphere. This gradient is increased within the Neogene cooling trend. Thus, overall cooling (in high latitudes) and increased winds should drive more warm water southward along the shores of eastern Africa. In contrast, a strengthening of westerly winds driving the South Atlantic Current, and especially a northward motion of the polar and subpolar frontal systems, should make it more difficult for the Agulhas waters to penetrate into the Cape Basin against the prevailing current. The optimum delivery of warm water around the Cape, presumably, will occur when strong seasonal contrast allows appropriate phase shifts in driving and opposing wind forcing. Strong seasonal contrast also stimulates diatom production (as shown by the analysis of cycles contained in the diatom record of Site 1084) (Berger and Wefer, 2001), so that one would expect, under this hypothesis of an Agulhas optimum, that maximum diatom production coincides with maximum delivery of warm eddies.

How can this hypothesis be tested? First, we should find that strong winds deliver warm water to high latitudes on the eastern side of Africa, even during glacial periods. This is indeed the case, judging from the arguments presented by Winter and Martin (1990), based on the study of cores taken off the eastern and southern coasts of South Africa. Winter and Martin suggest from isotopic evidence and nannofossil content that the Agulhas Current did not shift much laterally and that "it did not cool off appreciably during the LGM at least between latitudes 27° and 34°S" (Winter and Martin, 1990, p. 484). They conclude "if the Agulhas Current continued to retroflect in glacial times, then the Holocene resurgence of the Agulhas Current could be obviated and this mechanism would not be necessarily responsible for reseeding the planktonic foraminifer Globorotalia menardii into the Atlantic, nor for pulses in North Atlantic Deep Water production (Berger and Vincent, 1986)" (Winter and Martin, 1990, p. 485). We suggest in response that eddy delivery to the South Atlantic does not readily follow from maintaining the strength of the Agulhas Current through glacial-interglacial cycles. Presumably, the West Wind Drift south of the Cape is greatly strengthened during glacials as the polar front moves northward following sea ice expansion. Thus, access of Agulhas waters is likely to be more difficult under these conditions and perhaps denied entirely. The long absence of G. menardii within the Caribbean and central Atlantic, at times within the late Pleistocene, in cases through times when conditions would seem favorable (see Imbrie et al., 1973), suggests that reseeding from the Indian Ocean (which is the obvious source) is not a common and automatic process.

The complexity of hydrographic conditions along the shores of the Cape Province—presumably influenced by processes associated with eddy delivery through the Cape valve—is well illustrated by the results of the study by Pierre et al. (Chap. 12, this volume). They generated a stable isotope record for the last 500,000 yr at Site 1087. They find (abstract of their contribution) that "oxygen and carbon isotopic gradients between surface and deep waters display long-term changes superimposed to rapid and high-frequency fluctuations which do not follow the regular (glacial-interglacial) pattern." In the oxygen isotope record (Cibicides wuellerstorfi), unusually large warm peaks are associated with Stage 11 and Substage 5e, and also with the present, indicating strong warming of thermocline waters (Site 1087 is at 1370 m water depth) during these periods. In a companion paper on the same section, Giraudeau et al. (Chap. 7, this volume) suggest that delivery of warm waters from the Indian Ocean continued throughout the last 460 k.y. based on the presence of the warm-water foraminifer G. menardii at Site 1087. They propose that interocean exchange was most effective at glacial terminations.