The Walvis group consists of the three Leg 175 sites on Walvis Ridge and in Walvis Bay, which join two sites drilled earlier in this area, Deep Sea Drilling Project (DSDP) Sites 362 and 532 (Legs 50 and 75). The DSDP sites (with water depths near 1300 m) are seaward of the coastal upwelling regions but contain an upwelling record brought by the Namibia Current in eddies and filaments. Glacial-interglacial cycles are represented as cycles of carbonate dissolution, productivity, and terrigenous sediment supply. Site 1081 (at 760 m water depth) has gray clays with varying amounts of diatoms, nannofossils, foraminifers, and radiolarians. Authigenic minerals such as glauconite, pyrite, and dolomite are present. Sedimentation rates vary between 70 and 150 m/m.y. Site 1082 (at 1290 m water depth) has well-developed cyclic sedimentation. Sediments are composed of green clays with varying amounts of diatoms, nannofossils, foraminifers, and radiolarians. Site 1083 (at 2190 m water depth) has hemipelagic sediments consisting of clayey nannofossil ooze. Sedimentation rates vary between 60 and 150 m/m.y. Productivity changes are reflected in dark-light color cycles throughout the drilled sequence.
The Walvis Paradox arises from the observation, first made on Walvis Ridge, that glacial-age sediments contain less opal than interglacial ones (Diester-Haass, 1985a, 1985b). This is precisely opposite to the expected pattern if it is assumed that productivity is at a maximum during glacial times, as we have seen is the case north of Walvis Ridge. To complicate matters, there is actually evidence that productivity was in reality higher during glacial periods in the vicinity of Walvis Ridge. The evidence consists of cold-water foraminifers (Neogloboquadrina pachyderma [sin]) being more abundant during glacial periods within the last 500 k.y., at Site 532 (Oberhänsli, 1991). It seems then that at the Walvis Ridge, the intensity of upwelling is completely decoupled from the deposition of opal—or even more astounding, the least amount of opal is being deposited during the very times when upwelling is most intense. As suggested by Hay and Brock (1992), this conundrum calls for a situation where subsurface waters are impoverished in silicate during glacials in this region.
Increased upwelling results in the deposition of fewer diatoms—this is the Walvis Paradox.
Can we be reasonably sure that productivity was indeed high during times of reduced diatom deposition? If so, the Walvis Paradox will point to fundamental changes in the nutrient chemistry of the eastern Atlantic between glacial and interglacial times.
It is readily established that the organic matter deposition on Walvis Ridge follows the general pattern of increased organic sedimentation during glacial periods (Fig. F21). Core GeoB1028 (20°06.2´S, 09°11.1´E; 2215 m water depth) is situated on Walvis Ridge close to DSDP Site 531, west of Site 532. The stratigraphy of organic matter can be modeled in the fashion exemplified above for core GeoB1016 off Angola (Fig. F18C). Again, a globally valid oxygen isotope standard (806sox, parent to OJsox96, used above) is combined with insolation in July at 15°N in an appropriate fashion to estimate the TOC. In the present case, the formula used is
and all variables are standardized to mean = 1 and deviation = 0.25. When the overall trend is removed, the estimate yields r2 = 0.65 (see Fig. F21A).
Direct evidence for an excellent fit with glacial-interglacial fluctuations is given by the match of the TOC with the oxygen isotope stratigraphy within the core itself (Fig. F21B). The two records are precisely in phase at the terminations. At the warm peaks (low TOC) and the cold peaks (high TOC) there is some indication of a lead of the TOC; that is, warming and cooling are as important as the state itself in determining productivity. This phase relationship suggests that the same process producing warming and cooling (the orbitally driven changes in insolation) strongly affect productivity. Insolation drives seasonal winds. The interplay between monsoon and trade winds (see above) is one available mechanism; a shift in mid- and high-latitude seasonal fronts is another.
It has been suggested (e.g., Diester-Haass et al., 1992) that redeposition of organic matter from shelves during glacial periods when sea level was lowered could have simulated increased productivity within the downslope sediments now enriched in organic carbon. If this were so, one would not expect that the benthic foraminifer content should be similarly enriched within glacial sections. In fact, the benthic foraminifer abundance does follow the oxygen isotope record rather closely (Fig. F22B). Thus, it appears that the TOC represents mostly in situ supply of organic matter and, hence, the productivity of overlying waters.
Weinheimer (Chap. 3, this volume) has studied the uppermost section of Site 1082 in some detail with regard to radiolarian content. From the sedimentation rate for this site (see Fig. F10), her "warm-water" radiolarian curve can be interpreted to reach the last glacial maximum at 3 mbsf, the Stage 5-4 transition at 7 mbsf, and Substage 5e near 12 mbsf at the end of her series. The observation relevant to the discussion of the Walvis Paradox is that "intermediate-water" radiolarians, which presumably indicate upwelling, are abundant throughout the section comprising Stages 4 to 2. Moreover, the intermediate forms become abundant even before the transition into Stage 4, implying that upwelling starts at the end of the warm period rather than within the transition. This is wholly in agreement with the notion that insolation is just as important as the overall glacial-interglacial climate state in driving the upwelling system.
Do these relationships between climate state and productivity apply to pre-Quaternary sediments as well?
Apparently so. Regarding opal, this was already indicated by work on DSDP Site 532 (Diester-Haass, 1985a, 1985b; Diester-Haass et al., 1992). Intriguingly, Diester-Haass et al. (1990) found that the phase between climate state and productivity in the Miocene was the reverse of that found for the Quaternary and Pliocene. In the older sediments, high productivity paralleled opal deposition, assuming that low carbonate content, low sand content, and high benthic/planktonic foraminifer ratios indicate high productivity. This seems reasonable, especially since the organic matter content is high in the several cases where this was checked against the carbonate, sand, and benthic/planktonic (B/P) ratio indices (Diester-Haass et al., 1990).
Sancetta et al. (1992) made a detailed study of the section between 2.7 and 2.35 Ma at Site 532 in an interval containing major early glacial excursions (Stages 96, 98, and 100) and displaying a major cooling step toward Northern Hemisphere glaciation (near 2.53 Ma) (see Fig. F23). This step is identical to the "intensification of Northern Hemisphere glaciation" event or "iNH" of some authors. In what follows, we refer to this abrupt increase in excursion to heavy values as the "post-Gauss cooling step," to avoid direct implications for "Northern Hemisphere" or "glaciation," for which Leg 175 has no evidence.
The cold-warm cycles in the oxygen isotopes (Uvigerina) in the entire interval are wholly dominated by obliquity-related cycles (Fig. F23A, F23B). Although precessional information plays a role (Fig. F23A), it is unimportant as far as spectral power (Fig. F23B) (as previously noted by Sancetta et al., 1992). Extracting the obliquity-related cycles from the Fourier matrix (Fig. F24), it is seen that carbonate content (Fig. F24A) and opal content (Fig. F24B) are close to being in phase with the oxygen isotopes, such that peaks in either coincide with warm periods. However, in the high-amplitude portion of the record after the post-Gauss shift, carbonate tends to lead the oxygen isotopes and opal tends to lag. We interpret this to mean that low productivity (high carbonate) is shifted toward warming, whereas high opal productivity is shifted toward cooling, a phase relationship familiar from the Quaternary (see above).
The spectrum of the opal record shows power at periods other than 41 k.y. (Fig. F24B). The relevant peaks are at 167, 97, and 71 k.y. The peak at 71 k.y. derives its power from interaction of precession and obliquity, as the difference tone between 96 k.y. (precession modulation by eccentricity) and 41 k.y. (obliquity) is 71 k.y. The sum tone between 71 and 125 k.y. (the second eccentricity period modulating precession) yields 164 k.y. The 51-k.y. period is but little above the noise level; it crops up when beating obliquity (41 k.y.) with the main precessional peak (23 k.y.).
The particular mixture of the interference tones in the opal record is of considerable interest. It is very similar to the mixture seen for the deepwater carbon isotope record of the interval in question in the Atlantic Ocean on Ceara Rise (Bickert et al., 1997) (fig. 6: 2.1-2.6 Ma). The peaks of the spectrum in this instance (Sites 925 and 927) are 166, 69, 41, 33, and 23 k.y. Neither the oxygen isotopes of any of the intervals studied in that paper nor the carbon isotopes of any other interval but the one between 2.6 and 2.1 Ma show this mixture of periods. We take this as evidence that the opal record on Walvis Ridge is somehow related to the deep circulation of the Atlantic Ocean, in addition to or in preference to the climatic agents forcing upwelling off Namibia. The message from this evidence, we think, is that the nutrient content of upwelled waters changes through time in this region (supporting suggestions by Hay and Brock, 1992; Berger et al., 1998a; Lange et al., 1999; and Ettwein et al., Chap. 18, this volume).
The change must comprise large areas off southwestern Africa if it is tied to deep-sea circulation. In particular, we propose that the silicate content of glacial-age subsurface waters decreases much more in proportion than phosphate and nitrate and that this discrepancy is at the root of the Walvis Paradox (cf. Berger et al., 1998a; Lange et al., 1999).
A detailed study of glacial-interglacial cycles in the late Pliocene is presented by Ettwein et al. (Chap. 18, this volume) based on material from Site 1083. Their data (Ettwein et al., Chap. 18, this volume, fig. F2) suggest a change in the response of the upwelling system (as seen in diatom deposition) at the time of the post-Gauss cooling step (2.54 Ma). Regarding organic matter, the maximum before the shift is broadly straddling a cool period (Stage 104), with the subsequent three stages too warm for high TOC values. The highest values appear at the post-Gauss cooling step itself, centered between 2.53 and 2.54 Ma. Also, the subsequent maxima are centered on cooling events, while both warm and cold stages have smaller TOC values. Thus, the system is most productive for an optimum condition that is intermediate between warm and cold (but not on the warming side) whenever the warming goes through a substantial range. Since there is no substantial warming before the post-Gauss shift in this sequence, the rule only applies after the shift.
Regarding the diatom abundance—plotted by Ettwein et al. (Chap. 18, this volume) as endemic Antarctic species and Chaetoceros spores—it is noted that the endemic Antarctic species become less abundant after the shift, avoiding Stage 100, but are otherwise without a clear tie-in to warm and cold stages, while before the shift there is a certain preference for warm stages. The Chaetoceros spores, like TOC, tend to be abundant at the cooling events but also favor the mild cold stages before the shift. They avoid the maximum glacial states in Stages 96 and 100. From inspection, there seems to be a negative correlation between endemic Antarctic species and Chaetoceros before the shift and a positive one after the shift. This suggests that after the shift both become strongly dependent on the same factor, that is, the availability of silicate in subsurface waters. Ettwein et al. reach a similar conclusion regarding the importance of the supply of silicate in attempting to explain the patterns of diatom deposition.
The various lines of evidence inherent in all the studies summarized here allow the conclusion that productivity is a function of wind-driven upwelling on Walvis Ridge and that the opal content of the sediments reflects silicate content in subsurface waters more than it reflects intensity of upwelling, at least after the post-Gauss cooling step. The same is apparently true for the early Quaternary off southern Namibia (Anderson et al., Chap. 21, this volume). This conclusion will be important in interpreting the subject of the late Pliocene Diatom Maximum off Namibia, also referred to as the "early Matuyama Diatom Maximum" (Lange et al., 1999). To this subject we turn next.
The Namibia Opal Paradox is analogous to the Walvis Paradox, which derives from the contrasting views on glacial-interglacial productivity patterns held by geologists working on the record of Site 532. Diester-Haass (1985a, 1985b) found that opal deposition was higher during interglacial periods than during glacial ones and argued for lowered glacial productivity. Oberhänsli (1991) found that upwelling increased during glacial periods within the Quaternary. At first glance, the two findings seem incompatible. However, they are resolved if it assumed that the fertility of the thermocline decreases during glacial periods, especially with regard to silicate.
On a scale of several million years, according to the data gathered by Leg 175 scientists, we again see a puzzling pattern. With overall cooling in the late Pliocene, diatom deposition increases off Namibia (unsurprisingly), but with additional cooling after 2 Ma, the abundance of diatoms decreases again, despite indications for strong upwelling. It is as though the cold water supplied by upwelling is impoverished in the silicate necessary to make diatom shells. On the whole then, we are tentatively equating glacial (silica poor) conditions with the late Quaternary and the late Pliocene with interglacial (silica rich) conditions.
The Namibia Opal Paradox was discovered not just by mapping maximum diatom deposition during the late Pliocene (which had been done before for the Walvis Ridge by Leg 75 scientists; see Hay, Sibuet, et al., 1984). Instead, several additional ingredients define this important discovery by Leg 175 scientists:
It is the study of the species composition of the diatom assemblages at several sites along Namibia that lets the paradox emerge. We spell this out to emphasize the importance of diatom paleoecology in the context of upwelling dynamics. Studies on upwelling without reference to the types of diatoms in the sediments are at a disadvantage, as far as the reconstruction of upwelling dynamics.
The question is, can we assume that productivity was higher in the Quaternary than during the Gauss-to-Olduvai diatom maximum because of the signs for strong coastal upwelling? Or was the productivity much the same since the end of the Gauss Chron, and only the silicate content of subsurface water changed? Or else was the productivity, in fact, lower in the Quaternary but insufficiently so to explain the drop in diatom abundance?
Answers to these central questions are not readily available. The data at hand may fall short of providing the necessary information at this time. In principle, the timescale may be sufficiently stable for the last 3 m.y. to allow the use of mass accumulation rates in helping to unravel the pattern. Of course in practice, the age models for any one site depend on the quality of the stratigraphic record available for dating. Fortunately, for Site 1084 the age model seems to be excellent (see below), and the sedimentation rates apparently do not change much in going from the Namibia opal maximum to the Quaternary (Fig. F11). Thus, plotting mass accumulation rates (Lange et al., 1999; Giraudeau et al., 2001) brings refinement but does not change the argument in fundamental ways. In what follows, we discuss the question of changing productivity in terms of facies and content rather than converting to mass accumulation rates.
As one result of this exercise, we shall reaffirm the notion that the various records within the Namibia upwelling system are closely related and subject to similar dynamics. This does not preclude local variations, but it puts them in context as variations on the same theme. In fact, the present-day geographic setting for the Walvis group of Leg 175 (Fig. F1) suggests that we should expect their records to reflect changes in a system, albeit in the manner appropriate to local conditions.
The long-term trend in carbonate content of the record at Site 532 (Gardner et al., 1984) suggests a general increase in productivity on Walvis Ridge throughout the Pliocene, culminating in the early Quaternary. After this, increasing carbonate values point to a general decrease in productivity in that region (Fig. F26). Pulses of high productivity (thought to coincide with minimum carbonate content) occur between 3 and 1.2 Ma, but rarely before or after that interval. Within this pulsed production interval, a time of high carbonate values just after the Gauss/Matuyama reversal stands out (Fig. F26A). The record of organic carbon deposition shows a large amount of scatter (Fig. F26B; note log transform). In general, however, it supports the tentative conclusions based on the carbonate record: a general increase of productivity into the early Quaternary and a drop after 1.2 Ma or so. This apparent drop, however, seems less pronounced in the organic carbon than suggested by the increase in carbonate.
On the scale of 400 k.y. or so, we find substantial excursions in both carbonate and organic carbon, in addition to the overall trend described (Fig. 26C). Three events stand out (two of these are marked on the graph). They are (1) a low-productivity period lasting roughly 250 k.y. just after the Gauss/Matuyama reversal ("early Matuyama productivity minimum"), (2) a high-productivity period centered around 1.4 Ma ("early Pleistocene productivity maximum"), and (3) a distinct productivity minimum in the interval between 1 Ma and Stage 16, that is, roughly between the mid-Pleistocene climate shift and the onset of 100-k.y. cycles. Another important feature is the rather sudden onset of elevated productivity near 3 Ma.
The overall productivity pattern reflected in the record of Site 532 (Fig. F26) must be seen against the background of major changes in climate during the Pliocene-Pleistocene, as reflected, for example, in the oxygen isotope pattern of benthic foraminifers at Sites 925 and 926 on Ceara Rise (Bickert et al., 1997, and unpublished data kindly provided by Torsten Bickert for ages >2.6 Ma) (Fig. F27). The pattern observed is a result of global changes of water chemistry from ice-mass buildup and decay and from local changes in temperature, tied to vertical movements in the boundary between North Atlantic Deep Water and Antarctic Bottom Water at the depth of the sites (3.3 to 4.3 km). Maximum long-period excursions are seen to occur at intermediate stages of overall cooling in the 1-m.y. intervals centered on 3 and 0.5 Ma. The generally low-productivity situation before 3 Ma is well represented by the absence of "cold" excursions, that is, by the narrow range of values fluctuating on the "warm" side of the 18O record.
After the first strong cooling in the late Pliocene (near 3.2 Ma), the deepwater system responds to further cooling and orbital forcing by creating long-wave oscillations related to eccentricity (Fig. F27A). The same kind of thing (although with more regular cycles) is seen again after 1 Ma. Within the period from the end of the Gauss to 1 Ma, however, such fluctuations are subdued, presumably because the 41-k.y. cycle becomes dominant. In a broad brush scenario one might state the hypothesis thus: in the warm part of the Pliocene, productivity is low and cyclicity is precession forced. In the transition stage (3.3 to 2.3 Ma), productivity is increased and large, long-wave fluctuations feed off the energy delivered by the interference of obliquity and precession forcing. Within the first half of the Quaternary (2 to 1 Ma), precessional effects are subdued, which removes energy from long waves (since they depend on eccentricity variation, which expresses itself through precession). In the second half of the Quaternary, interference patterns reemerge, as precessional effects reassert themselves.
Additional ideas for how the system changed can be gleaned by comparing the deepwater stratigraphy of oxygen and carbon isotopes (Fig. 27B). The data are from the eastern equatorial Pacific (Site 849) (Mix et al., 1995). All three series represent comparisons of the two isotope stratigraphies, one is the ratio of variability over a 100-k.y. window, another shows the correlation (r2), and the third shows the product of variability. We see that there is a fundamental transition from a world dominated by variability in carbon isotopes to one dominated by that of oxygen isotopes, with the most striking change happening just before 3 Ma. We might refer to this as the transition from a "carbon-dominated system" to an "ice-dominated system" in the present context (within the Pliocene-Pleistocene sequence). We also note that the correlation between the two responses is quite poor at times before 2.8 Ma but improves overall after that. Within the ice world, correlation is worst during the first half of the Pleistocene, when obliquity dominates fluctuations. Presumably, carbon is more tied to precession and eccentricity, since we find 100-k.y. fluctuations well before these occur in the oxygen isotopes (Schmidt et al., 1993). As far as the sensitivity of the system to outside forcing, it increases all the time, on average, as shown by the product in the variability of oxygen and carbon isotopes (dark gray line in Fig. F27). Comparing the C/O variation ratio with r2, one notes that there is an inverse correlation before 2.2 Ma and a positive correlation after that.
From the foregoing discussion of the data in Figures F26 and F27, we conclude that there are three major regime shifts: the first near 3.2 Ma (increase of carbon-independent oxygen isotope variations; increase in production), near 2.8 Ma (increased coherence between carbon and oxygen isotopes whenever carbon variability is relatively low; decrease of production), and near 2.2 Ma (increased coherence between carbon and oxygen isotopes whenever carbon variability is relatively high; increase in production). What we see as an overall trend is an increasing entrainment of the carbon system (including NADW production, which is reflected in carbon isotopes in the deep Pacific) by the ice-dominated system. The overall result in terms of Namibian production is an increase, albeit with interesting (and puzzling) minima, in the early Matuyama and the late Matuyama.
The productivity-related record of Sites 1081, 1082, and 1083 presents the same elements as those seen at Site 532 in the data of Gardner et al. (1984). Again, carbonate is at a maximum near the Miocene/Pliocene boundary (Figs. F26B, F28A). There is then a more or less gradual drop to low values between 3 and 2.5 Ma, a period of major cooling steps (see Maslin et al., 1998, for review). The following period, between the end of the Gauss and Olduvai Chrons and crucial for the Namibia opal maximum, has both high and low carbonate values, presumably reflecting large-scale climate excursions. The absolute minimum of carbonate content is then reached between 2 and 1.5 Ma (centered on 1.5 Ma at Site 532), followed by a rise toward generally higher values, especially after 1 Ma. A minimum within that interval of the second half of the Quaternary appears near Stages 16 and 15, as it does at Site 532.
Thus, we are satisfied that Sites 532, 1081, 1082, and 1083 show the same general carbonate pattern and may be considered as telling more or less the same story as concerns large-scale patterns of productivity. The distribution of organic matter (Fig. F28B) confirms this impression from comparison of the Leg 175 Walvis group results with those from Site 532. Inspection also confirms the strong tendency for negative correlation between carbonate and organic carbon already noted (see also Gardner et al., 1984). High organic matter and low carbonate, that is high productivity, dominate between 3 and 1 Ma, with the peak of productivity offset toward the younger part of the interval, to ~1.5 Ma. This is not the same as the pattern for the opal abundance.