QUATERNARY RECORD OF THE CONGO AND ANGOLA SITES

The northernmost sites of Leg 175 are situated near 5°S on the Congo Fan. They contain the record of sediment supply by the Congo, intercalated with the oceanic record. Pollen, freshwater diatoms, phytoliths, and clay minerals provide clues to climatic change in the inner Congo drainage basin in central Africa. Fluctuations in the accumulation of pelagic diatoms and marine organic matter track the changes in productivity in this periestuarine environment. Much of the sediment here consists of diatomaceous clay with varying admixtures of nannofossil carbonate. Sedimentation rates are high; typically near 10 to 15 cm/k.y., on the high side of the depositional rates mapped for the late Quaternary (Fig. F13). The deepest site (1075; near 3000 m water depth) penetrated the entire Quaternary, reaching roughly 2 Ma at 200 mbsf.

Shipboard studies showed pronounced cyclic sedimentation, evidenced in the stratigraphy of physical properties, notably magnetic susceptibility and reflectance of the sediment surface. For Site 1075, reflectance clearly demonstrates response of the Congo upwelling region to precessional forcing (Berger et al., 1998b). The record of magnetic susceptibility, in turn, is closely allied to a standard oxygen isotope curve, suggesting a strong link to sea level variation. The nature of the links of Congo sedimentation to Quaternary climate has been discussed in previous studies (Jansen et al., 1984; Schneider et al., 1995).

For the deepest site (Site 1075; ~3000 m water depth), the possibility must be considered that a general deep Atlantic signal is present within the record for carbonate, organic carbon, and opal. To compare the actual record found at this site, we should first get an idea of what the general deep Atlantic record looks like. Such a deepwater record is reflected in the results from Site 663 in the eastern equatorial Atlantic (deMenocal et al., 1993). The late Pleistocene record shows an excellent correspondence between carbonate content (plotted as negative log of noncarbonate percent) and the deepwater oxygen isotope curve (Fig. F14). Carbonate is high during interglacials and low during glacial periods. In fact, the record is quite similar to the ¹³C stratigraphy of Cibicidoides wuellerstorfi in the deep western equatorial Atlantic (Curry, 1996) (Fig. F10). The ¹³C record in that region was interpreted in terms of fluctuations of the lower NADW boundary, such that the boundary rises during glacial periods in response to decreased NADW production (Curry, 1996). However, studying the sand content in calcareous Quaternary sediments from the Guinea Basin in the deep west-equatorial Atlantic, Bickert and Wefer (1996) note the strong precessional component and ascribe the changing preservation of foraminifers (that is, sand) to dissolution effects from varying productivity. Increased glacial-time productivity, thus their hypothesis, increases dissolution and destroys sand. The third possible factor, fluctuations in the dilution from the supply of noncarbonate matter, is emphasized by deMenocal et al. (1993) in their explanation of the carbonate cycles.

We may conclude from these contrasting views that while we should expect low carbonate content for glacial periods in deepwater sediments, we might not be able to attribute a mechanism to the fluctuations without further detailed study. The ¹³C stratigraphy of the Guinea Basin cores studied by Bickert and Wefer (1996) shows similar fluctuations to those from Ceara Rise (Curry, 1996), with amplitudes exceeding 1 (more than twice the global-ocean background fluctuation). The most parsimonious explanation is that both sets of cores reflect changes in NADW production, although an additional effect from productivity variation is likely in the eastern set (making up for a reduced influence from NADW).

The opal record at Site 663 (Fig. F14B) also is in synchrony with the oxygen isotope stratigraphy, such that high values of opal (plotted in reverse as log opal percent, standardized) coincide with glacial periods. The agreement between carbonate and opal is virtually perfect, provided one of the two is plotted in reverse and the log transforms are used before standardization (Fig. F14C). This remarkable match in amplitude and phase supports the proposition of Bickert and Wefer (1996) that dissolution effects are caused by varying productivity. Note that this is not simply a matter of most of the noncarbonate consisting of opal. This is not the case. The opal expands into the noncarbonate space during glacial periods. A plot of log(opal/noncarbonate) is virtually identical to the opal record shown, signifying that variation of carbonate is unimportant as far as variation of opal. (The high correlation shown only appears after some smoothing.)

In contrast to the deepwater environment well west of the Congo Fan, the carbonate variations off the Congo are much less regular and therefore more difficult to describe and interpret (Fig. F15). High carbonate values encountered in core GeoB1008 (6°44.9´S, 10°19.1´E; 4124 m water depth), which was taken during preparatory work for Leg 175, are concentrated in glacial intervals (position of peaks is marked by black rectangles on the oxygen isotope curve). However, the relationship is not a simple one: peaks tend to move to the positions of warmings and coolings and apparently avoid maximum glacial conditions. The water depth of the location of core GeoB1008 is distinctly deeper than that of Site 1075 (4124 vs. 2995 m). Nevertheless, any influence from abyssal lysocline fluctuations (good carbonate preservation during interglacials) apparently is eclipsed through effects prevailing above the lysocline. These include, apparently, increased preservation during glacials well above the lysocline (even though that level is now shallower) and changing productivity (which provides for increased dissolution at peak glaciations).

The record of productivity is well reflected in the marine organic carbon within the core, with abundance of organic matter following the oxygen isotopes rather precisely, at least for the last 140 k.y. or so (Fig. F15B). A tendency for large amplitudes in the precessional band in the organic carbon decreases the match with the oxygen isotopes, although the long-period response (glacial periods have high productivity) persists. The opal content follows the organic matter content nicely back to ~140 ka but moves out of phase before that (shaded line in Fig. F15C). Evidently, precessional forcing is less important for the opal deposition than for organic matter deposition. To some degree, the two processes are decoupled. (Uliana et al., Chap. 11, this volume, propose adjustments to the timescale before 150 ka, but this does not affect the match between organic carbon and opal, or lack thereof.)

A strongly dominant eccentricity signal was found at Site 1077 in the opal abundance series by Uliana et al. (Chap. 11, this volume) (power near 98 k.y.) for the late Quaternary. Interestingly, the 400-k.y. cycle is clearly reflected (i.e., Stage 11 and adjacent glacial periods are not clearly expressed). This means that the opal cycle here feeds off precessional amplitudes. Surprisingly, however, precessional periods are not clearly expressed. Neither is the obliquity cycle. Indications are, therefore, that the energy from these forcings ends up in periods produced by interference (that is, 70 k.y. and its harmonic, 35 k.y.). Such interference might be expected at the sites of production of intermediate waters at latitudes where interaction of precessional (eccentricity) forcing and obliquity forcing would be optimal. If so, the silicate content of subsurface waters (and the opal deposition off the Congo) would reflect this interaction.

The precessional effect is commonly attributed to the competing effects from monsoon winds and trade winds (Schneider et al., 1996, and references therein). In the present case, the competition reaches across the equator, with North African monsoon interfering with southwest trade winds. Enhanced heating of the North African land masses by a close-by sun in northern summer (June, July, or August perihel) weakens both the trade winds off northwest Africa and the South Atlantic ones, thereby reducing forcing for upwelling (Kutzbach and Liu, 1997). Relevant land areas being much smaller in the south, the reverse effect (monsoon over southern Africa) is much less important. Thus, the precessional tone of the forcing is preserved in the Congo sediments, as productivity variation, in phase with northern summer monsoon maxima and minima.

Schneider et al. (1997) suggest that much of the opal results from the delivery of dissolved silicate by the river itself rather than from oceanic upwelling (which is shown to dominate the organic production).

As an alternative to the hypothesis of Schneider and co-workers (riverine silicate supply), we propose that the ratio between silicate and phosphate within thermocline waters changes in such a fashion as to counteract the precessional upwelling effect as far as opal production. That is, the silicate content would have to be decreased during glacial periods relative to phosphate. Such a change implies a drop of all nutrients during the glacial period (Berger and Lange, 1998). We envisage increased dust supply during glacials to the areas of convergence, where intermediate waters are formed, with corresponding precipitation of opal from increased availability of iron and of dust particles providing for fecal ballast (Berger and Wefer, 1991, and references therein; see also Young, 1991; Martin, 1994; Harrison, 2000, and references therein).

The question of river input is addressed by Uliana et al. (Chap. 11, this volume) in a comprehensive study of siliceous components of continental and marine origin within the sediments of Site 1077 for the last 460 k.y. Uliana and colleagues find that marine siliceous microfossils entirely dominate opal deposition, especially during glacial periods. Uliana et al. also find that there is considerable pulsed input of freshwater to the surface waters above Site 1077 during certain periods, with a distinct increase in brackish water diatom species. Abrupt changes are observed during Termination II, near 130 k.y. ago, where the assemblage composition varies from predominantly marine to marine/brackish. In addition, evidence obtained from freshwater diatoms and chrysophycean cysts points to changes in the drainage basin from more arid conditions to more humid conditions at that time (a change since reversed). Presumably, two effects have to be considered: increased monsoonal rains in the northern parts of the drainage area, during times when perihel occurs during the monsoon season (precession effect), as well as changes in current direction offshore, which may redirect the freshwater outflow off the Congo River toward the area of Site 1077.

There is a major shift in global climate close to 0.9 Ma following a cooling initiated near 0.95 Ma. This event is referred to as the mid-Pleistocene climate shift. It marks the initiation of high-amplitude glacial-interglacial cycles, which start as 80-k.y. cycles and then move into 100-k.y. cycles upon entering Stage 16, ~700 k.y. ago (Berger and Wefer, 1992; Berger and Jansen, 1994). It is of great interest to find the response of the various upwelling systems to this event. In the case of the Congo group, we would like to know what kind of climate shift may have occurred in central Africa and how the large-scale pattern of sedimentation responded to this event.

The shipboard data allow some statements of a general nature (Fig. F16). Regarding carbonate deposition, it seems that variability of carbonate content increased drastically at the time of the mid-Pleistocene shift, indicating much larger fluctuations in productivity after the shift. The range expands both toward lower values and toward higher values, compared with preshift time. There is a slight relaxation for the period containing isotope Stages 12-15 and a similar reduction of range for the last 100 k.y. However, overall, the range of variation is distinctly greater after the shift than before it. The same kind of range expansion is also seen in the nannofossil abundance, as recorded in smear slides (the index values from the Leg 175 report are log-transformed). Highest carbonate values, according to these visual estimates, are reached between Stage 11 and Stage 8.

Organic matter content at the Congo sites also shows a change at the time of the mid-Pleistocene climate shift but much less prominent than for carbonate. Again, the range expands right after the shift, at 0.9 Ma (Fig. F17A). However, on entering the Milankovitch Chron (at 0.65 Ma), the range tends to contract again. A loss of low values and a gain in higher ones (seen between 0.1 and 0.4 Ma) reflects a general increase in productivity (or better preservation of organic matter within the less-aged sediments). The range of values even before the shift is quite remarkable (a factor of four; note the logarithmic scale). Diatom abundance also shows a distinct expansion of range at the time of the shift. Before the shift, diatom abundances are high (in fact, they tend toward a maximum because of the way the visual index is used when describing smear slides). After the shift, they cluster around an intermediate value, with occasional low values. In the 400 k.y after the mid-Pleistocene climate shift (0.9-0.6 Ma) the range varies between low and high values, with the lowest sustained values (marked by a dashed-line box) centered on the maximum glacial period of the late Quaternary, isotope Stage 16 (Fig. F17B).

From the increase in variation in productivity-related indices, we conclude that the sensitivity of the Congo upwelling system to precessional forcing increased at the time of the mid-Pleistocene climate shift. If so, this would indicate that developments in high latitudes (ice buildup) and global cooling (desert development) provided for changes in response of tropical systems to (comparatively invariant) astronomical forcing. In turn, the greater sensitivity of the tropical systems would have resulted in increased feedback to the high-latitude ice dynamics, with implications for runaway amplification and accelerated melting (terminations).

Lydie Dupont et al. (Bremen) has made a detailed study of changes in the terrestrial record of Sites 1075 and 1077, especially with a view to pollen and dinoflagellate cysts (pers. comm., 2001). She finds that the variation of river discharge increased after 0.94 Ma, suggesting larger contrasts in rainfall between glacial and interglacial periods within the African interior. Before 1.05 Ma, she finds, there is no strong glacial-interglacial rhythm within the pollen record. A strong rise of Podocarpus pollen occurs at 1.05 Ma, suggesting a cooling, in synchrony with distinct changes in the dinoflagellate record. While the dinoflagellate cysts indicate a reduction in river discharge, the contrast between pollen spectra of glacial and interglacial periods increases after 1.05 Ma. Thus, the expansion in the range of fluctuation of productivity off the Congo goes parallel (more or less) with an increase in the range of variation in the vegetation cover in the interior of the continent. The connection, presumably, is the tie-in of both marine and terrestrial photosynthesis to the monsoon/trade wind contrast, at least for the latest Pleistocene. A complication arises for the earlier glacial cycles after the climate shift. Those glacials, according to Lydie Dupont, may have been cool and wet, rather than cool and dry.

The Angola or Lobito sites are situated off the central part of the Angolan coast, near 12°S. They might also be called, with more justification than all others, the "Benguela sites," since they are located not so distant from the port of Benguela. However, the "Benguela Current" does not reach this far north and they are not within the region off Namibia and South Africa usually referred to as the "Benguela upwelling system." The deposits of the Angola group document high productivity, albeit these are low when compared with the upwelling areas immediately to the north and south. While silicate-rich waters are potentially available below the mixed layer (thanks to the nearby Angola Dome), upwelling is seasonal and comparatively weak and opal accumulation is very modest for this coastal environment. Organic matter averages 2.5 wt% at Site 1078 and 4 wt% at Site 1079 (where there is less dilution by terrigenous matter). Sediments at Site 1078 (448 m) consist of gray silty clay with varying amounts of nannofossils and foraminifers. In parts of the section, very high sedimentation rates (up to 60 cm/k.y.) were found; much of the material deposited this rapidly may have been delivered by coastal erosion. Dolomite concretions and laminated intervals are present in places, presumably indicating sporadic expansion of anoxic conditions. Site 1079, with sedimentation rates ~60% lower than those of Site 1078, has uniform olive-gray silty clays with varying amounts of nannofossils and foraminifers.

Site 1080 was drilled off the Kunene River (southern Angola). It is situated near the northernmost coastal upwelling cell of the Namibia upwelling system. Sediments consist of diatom-bearing and diatom-rich silty clays that are accumulating at a rate near 10 cm/k.y. The late Quaternary section is greatly attenuated. Drilling was terminated after a dolomite layer was encountered, slowing the advance of the bit to unacceptable rates. No further reference will be made to this site.

The two Lobito sites terminate in the late Quaternary. From the study of core GeoB1016 (11°46.2´S, 11°40.9´E; 4411 m water depth) and comparison with core GeoB1008 (discussed above), it appears that productivity fluctuations off Mid-Angola are similar to those off the Congo (Schneider et al., 1996). In contrast, the rate of opal deposition is some five times lower (Schneider et al., 1997) (Fig. F8). Not only is the overall deposition of organic carbon similar in the two areas, but the patterns match in some detail when correlated using oxygen isotope stratigraphy for guidance (Fig. F18). Thus, it is likely that forcing in the two regions is the same; that is, in both areas high productivity results from strong trade winds (Schneider et al., 1997). A remarkable feature of both patterns (and of many productivity records in general) is the spiky nature of the record. Fluctuations are by no means sinusoidal, but resemble the shape of a chain hung from a number of nails in series—a sequence of catenaries. This morphology suggests a response to narrow optimum conditions, as will be discussed below.

As in core GeoB1008 off the Congo, the opal record does not follow the organic matter record (TOC) very closely in core GeoB1016 (Fig. F18B). While the overall trends coincide, there is actually a marked anticorrelation on the timescale of precession and shorter. This supports our hypothesis that the silicate content of thermocline waters was lower in this region during glacial periods. Since the ratio of silicate to phosphate is positively correlated with the abundance of nutrients in the modern ocean, a high opal to TOC ratio suggests high thermocline fertility and vice versa. A high TOC content accompanied by low opal content (e.g., near 80 k.y. ago and near 250 k.y. ago) suggests strong upwelling from a silicate-impoverished thermocline (Berger and Lange, 1998).

The organic carbon content of core GeoB1016 is readily modeled by combining a standard oxygen isotope curve describing sea level change (in this case OJsox96) with the amount of irradiation in July at 15°N (from Berger and Loutre, 1991). Upwelling tends to be high when glacial conditions prevail and when summer irradiation at 15°N is at a minimum (Fig. F18C). Actually, during maximum glacial conditions, the response to irradiation seems somewhat diminished. The same model does not work at all for the opal abundance (Fig. F18D). In fact, the opal record seems to chart a compromise between strengthened upwelling during glacial periods combined with a greater availability of silicate during periods of intensified monsoon (low TOC times on the precessional timescale).

The patterns of carbonate content of the Lobito sites presumably are correlated with productivity, in that a high supply of organic matter will lead to dissolution and a low supply to preservation of carbonate, as discussed above. High values at the sites (Fig. F19) occur late during glacial periods near and within terminations (three boxes: Stage 2, 5-6 transition, and 8-7 transition). High preservation of carbonate during termination intervals is well known from off northwest Africa and elsewhere ("deglacial preservation spike"). It may be a global signal indicating reduced productivity from increased stratification and density contrast between freshened wind-mixed upper water masses and underlying cold and saline masses inherited from ice-age conditions (a scenario envisaged by Worthington, 1968). Low carbonate values are situated within interglacials (near 0.1, 0.2, and 0.4 Ma), indicating a tie-in to 100-k.y. cycles. More detailed work will be necessary to assess the usefulness of the carbonate stratigraphy in this area where dilution from terrigenous materials should play a strong role.

Organic matter content at the Lobito sites is confusing, given the expectations from core GeoB1016 (farther offshore) that TOC is high during glacial periods and low during interglacials. Again, dilution effects may be very important at these sites, given the apparent high variation in sedimentation rate and the proximity to an actively eroding coast (Berger, Wefer, Richter, et al., 1998).

Pérez et al., Chap. 19 (this volume) have studied benthic foraminifers at Site 1079 to document the course of changing productivity independently from dilution effects. From biostratigraphy, it appears that there is a sudden change in sedimentation rate between 200 and 400 k.y. ago; thus, there are less than two full glacial-interglacial cycles for reliable documentation. The oxygen isotope stratigraphy is based on Globobulimina spp. and reaches back to ~240 ka. The abundance of benthic foraminifers as well as the benthic foraminifer accumulation rate (BFAR) index (Herguera and Berger, 1991) suggest high productivity during glacials and somewhat lower values for Stages 1, 4 and 5, especially for Substage 5e. The fact that planktonic foraminifers show a similar pattern suggests that benthic and planktonic species may be influenced by the same factors, that is, dilution and dissolution. In fact, the fragmentation is at a minimum during glacial periods, supporting this suggestion. A strong carbonate maximum, as well as a sand percent maximum, is associated with late Stage 6 and the 6-5 transition, supporting the pattern read from Fig. F19 above.

Comparing core GeoB1016 and Site 1079, Pérez et al., Chap. 19 (this volume) find an excellent correlation between BFAR at Site 1079 and percent organic matter in core GeoB1016, supporting the notion that BFAR represents productivity—even though the planktonic foraminifers show similar patterns. Further confirmation comes from the foraminiferal epifauna, which indicates low productivity for high values of abundance. The maxima are found at the major transitions, 6-5 and 2-1, suggesting low productivity during deglaciation. Other high values are seen in Stages 7 and 5 (Pérez et al., Chap. 19, this volume). Organic matter stratigraphy in core GeoB1016 and the sequence of Bolivina pseudopunctata at Site 1079 run parallel, indicating that B. pseudopunctata is a productivity proxy, as the most abundant species. Bolivina dilatata, the second most abundant species, tends to show an inverse pattern, presumably because of the forced negative correlation within percentage space.

The balance of the evidence gathered by Pérez et al., Chap. 19 (this volume) supports the standard productivity pattern: high values for glacial periods and low values for interglacials, with BFAR and benthic foraminifers (BF)/g being the most reliable productivity indices. The spectrum of benthic foraminifer abundance at Site 1079 (Fig. F20), which presumably reflects the spectrum of productivity, shows strong peaks near 100 and 35 k.y. and also power at the precessional lines (24 and 19 k.y.), but very little in the vicinity of obliquity variation (41 k.y.).

How was the power from obliquity moved into power at 35 and 53 k.y.? Power can be moved in various ways, including interference processes and transfer of energy between harmonics. The whole-number harmonics of 100, downward, are 50, 33, 25, and 20, which correspond closely to the power distribution seen. The difference tones between obliquity (41 k.y.) and the main cycle (100 k.y.) is 69.5 k.y. One-half of this period is close to the one seen at 35 k.y. Thus, it is entirely reasonable to expect that energy from both the 100-k.y. cycle and the obliquity-driven cycle are contained in the 35-k.y. cycle. Of course, such reasoning does not address the issue of mechanism; it only asserts that the concept of astronomical forcing need not be abandoned just because one of the important lines (here, 41 k.y.) is missing.