LÜDERITZ SITE (SITE 1084)

In many ways, Site 1084 may be considered the "flagship" site of Leg 175. It has the highest content of organic matter and the greatest number of diatom mat deposits. Its interstitial water chemistry provides evidence for intense diagenetic activity driven by redox reactions. The site's sediments showed intense sulfate reduction in the upper few meters and had extremely high ammonia values. Also, an unusual amount of gas was produced by the sediment on being depressurized and an offensive odor emanated from much of the section. There was no doubt about the record having a high-productivity signal.

Site 1084 was drilled not far from Lüderitz Bay, a major upwelling center. The sediment sequence consists of clay-rich nannofossil diatom ooze, diatomaceous nannofossil ooze, and clay-rich nannofossil ooze. Conspicuous decimeter-thick intervals of dark organic-rich clay layers are present between 120 and 510 mbsf and are characterized by reduced carbonate content. Diatoms are abundant in these layers, with Chaetoceros resting spores dominant. These spores may be taken as upwelling indicators (Lange et al., 1999). The close proximity of Site 1084 to the Lüderitz upwelling cell results in well-expressed cycles of organic carbon content and diatom and coccolith abundance, as a consequence of cyclic changes in productivity of the type described for the late Quaternary record in this area (Little et al., 1997).

The record of Site 1084, as concerns the important issue of the course of opal deposition, is not fundamentally different from the record of the Walvis group sites. In fact, the diatom data are readily merged when considering general patterns as derived from smear slide data (Fig. F29). Peak values for diatom abundance agree well between Sites 1081 and 1084; they occur between ~ 2.7 and 2.1 Ma; the most persistent peak is offset to the younger part of this interval, between 2.2 and 2.3 Ma. The rise from low values in the early Pliocene to high values in the late Pliocene is extremely rapid; it takes place at 3.2 Ma. For the first part of the high opal condition, oceanic diatom types still dominate. Then, increasingly, cold-water diatoms and spores become important. Finally, Chaetoceros spores and other upwelling indicators dominate from ~2.1 Ma.

The shift to high-productivity conditions near 3 Ma is notable in the carbonate data (Fig. F29B), and another shift to lower productivity occurs just after 1 Ma (following an interval with two unusually low values). Extremely low values of carbonate occur between 2.2 and 1.9 Ma, just after maximum diatom supply. This suggests that maximum export of organic matter followed maximum export of diatoms. The relatively low values of organic matter between 3 and 2.5 Ma are somewhat surprising. The Walvis group as a whole shows high values for this interval. The small number of points, which invites the possibility of aliased sampling overemphasizing one or the other deviation from the mean, makes interpretation difficult.

On the whole, the data from Site 1084 agree with those of the other sites, with maximum opal deposition near 2.2 to 2.3 Ma embedded in an interval of generally increased background productivity between 3 and 1 Ma. A secondary opal maximum near 1 Ma seems associated with a productivity peak (as seen in the TOC). We infer that on timescales of long periods (0.2 m.y. or more), opal deposition and organic productivity are positively correlated, despite the fact that they are anticorrelated on the scale of periods of 0.1 m.y. or less.

When using carbonate as a productivity indicator—by assuming that dissolution intensity is a function of the supply of organic matter—one runs the risk of confusing dilution with dissolution. Although this is not the only risk (the other one refers to the changing undersaturation of deep water), it may be important. It is also readily diminished by noting the preservation of calcareous fossils. No systematic studies of preservation stratigraphy are available as yet, but we can make use of the number of benthic foraminifers counted by the specialist on board, Otto Hermelin. Hermelin desired to obtain a representative estimate of the proportion of the more important species present. For this purpose, one needs to count ~300 specimens. Thus, when foraminifers were plentiful, Hermelin counted this number, or even more. But when his samples did not contain sufficient tests, he counted whatever was available. The list of numbers counted (which he provided) is therefore a measure of the preservation of benthic foraminifers. In particular, periods of low numbers counted are periods of low availability and represent dissolution events.

When plotting Hermelin's counts, it becomes apparent that dissolution events are greatly concentrated between 3 and 1 Ma at Site 1084 (and also at Site 1081) (Fig. 30A). This is precisely the interval previously identified as a high-production period. The diversity of benthic foraminifers is another potential measure of productivity. As in most situations in different types of ecosystems, a high supply of food suppresses diversity and encourages rapidly growing opportunists. As a matter of observation, the number of species in the open ocean environment tends to be much larger than the number of species living under high-productivity regions at the margins of the ocean (Berger et al., 1998a). Diversity values for Hermelin's samples (Shipboard Scientific Party, 1998g, 1998j) were calculated as the logarithm of the inverse of the maximum percentage in a given sample. (The percentage was augmented by one point.) Results show low diversity values between 3 and 1 Ma, with a minimum near 2.2 Ma, rather close to the peak of the diatom deposition (Fig. F30B). Diversities at Site 1084 are especially low, as here productivity is highest. Poor preservation can interfere with this index to some extent. If there are only a few specimens, the most abundant species is bound to have a relatively high number.

Species belonging to the genera or groups Bolivina, Bulimina, Globobulimina, and Praeglobobulimina may be combined to yield yet another index for productivity (Fig. F30B). Maximum values throughout belong to Site 1084, which is situated in the most productive region of the sites drilled off Namibia. Unusually high values are concentrated between 2.5 and 2 Ma (in the Namibia opal maximum) and around and just after 1 Ma. The values after 1 Ma are not in agreement with the scenario of high productivity from 3 to 1 Ma, with lower values following. Interestingly, some of the lowest values of the Bolivina-Bulimina ("BoBu") index also are situated within the Namibia opal maximum itself. This means that the opal maximum is a time of maximum range of fluctuation between high and low productivity. Overall, ignoring the spikes, there is actually little change in the typical value from 2.5 Ma to present. This suggests that the opal maximum represents a period of transition between the pre-late Pliocene low-production time and a subsequent (Quaternary) high-production time, with the transition expressed as a flipping back and forth between the old and the new situation.

The origin of the early Matuyama Diatom Maximum, between the end of the Gauss and the beginning of the Olduvai Chrons, is complex. It involves seasonal mixing and background supply of silicate within the thermocline in an upwelling environment, as well as frontal zone activity of varying degrees of relative importance and with the participation of various water masses. In preparation for numerical modeling of such a system, it is useful to take account of the patterns emerging from available data. The basis for discussion is the "diatom abundance index" (DAI) used in shipboard work and in subsequent more detailed work on the diatom content of Site 1084 and neighboring sites (Lange et al., 1999; Pérez et al., Chap. 4, this volume). It has a pattern similar to the logarithm of percent opal, except that it has an upper limit of value 6 ("extremely abundant"). The stratigraphy of the DAI (edited and smoothed) shows the familiar ramping up, with superposed long-wave cycles, to the Namibia opal maximum between 2.6 and 2.1 Ma (Fig. F31). Opal content (Pérez et al., Chap. 4, this volume) shows that, in fact, the maximum is not at the center of the plateau, but shifted to the younger side, near 2.2 Ma.

The optimum temperature of surface waters at Site 1084, which prevailed during maximum opal deposition between 2.3 and 2.2 Ma, can be read from the alkenone-based temperature curve of Marlow et al. (2000). It is 22.5°C, considerably warmer than typical present temperatures, which fluctuate around a mean of 18°C or so, using the maximum for late Pleistocene values recorded in the alkenones (cf. Tchernia, 1980). The value of 22.5°C is beyond three standard deviations from the mean for the last 0.5 m.y., according to the data of Marlow et al. (2000); that is, a spike of opal deposition as high as during the Namibia optimum (if tied to temperature) might be expected once or twice in the last million years, given the sampling density of the DAI determinations (Fig. F31). (The sea-surface temperature (SST) series shown in Fig. F31, marked 1084SST, is based on data kindly supplied by J. Marlow. The points are generalized from the original data by arbitrarily reassigning positions of sampling points by up to 15 k.y., for more even spacing, and by eliminating points with neighbors at 10 k.y. or less.)

To make the relationship to temperature more precise, we have used the ¹⁸O record of benthic foraminifers at Site 849 in the eastern equatorial Pacific (Mix et al., 1995), reversed the sign, and set the series to the mean and standard deviation of the alkenone data. We then resampled the smoothed curve (40-k.y. boxcar) for the (adjusted) sampling positions of the diatom index. (The adjustment results from smoothing both age and DAI values with the same three-point boxcar.) The resulting curve, labeled "849ox" describes the alkenone data very well over much of the record. Exceptions occur between 2.6 and 2 Ma where the alkenone data seem to scatter in the upper part and above the range defined by the ¹⁸O series. This period of anomalously warm surface water (relative to general trends) is also the period of the Namibia diatom maximum.

It is evident from the history documented in the series plotted (Fig. F31) that the warm waters of the early Pliocene are unfavorable for diatom growth. On entering the "high-productivity interval" between 3 and 1 Ma, identified above, the diatom supply suddenly increases; that is, diatom supply simply follows productivity at this point. The 3-Ma productivity increase is marked by a cooling step, as seen in the deepwater data. The next two global cooling steps are near 2.75 and 2.55 Ma; they result in further increases of diatom supply (and presumably overall productivity). (A strong admixture of allochthonous material makes interpretation difficult.) Near 2.1 Ma, with another marked cooling step, there is a regime shift. No longer does the Namibia upwelling system respond with increased diatom supply to an increase in cooling and production. At the cooling step between 2.1 and 2.0 Ma, there is a distinct reduction of opal accumulation. The reduction persists over the rest of the record, throughout the Quaternary. Thus, the 2-Ma cooling step produces an overall regime shift in the system: the response to forcing changes in fundamental ways (as pointed out above when discussing Fig. F27). We can only guess at this point what this change of response might be, but it resulted in the separation of diatom productivity from general productivity. Thick organic-rich layers occur well before and after the shift, from the post-Gauss cooling step (2.55 Ma) to the Brunhes/Matuyama boundary (0.78 Ma), with maxima at 1.1 to 1.2, near 1.45, 1.7, 1.9, and 2.4 to 2.5 Ma. Organic matter content does not seem to decrease after 2 Ma (Fig. F29) and neither does the abundance of benthic foraminifer indicators of high productivity such as Bolivina and Bulimina (Fig. F30).

We interpret these data to mean that cooling increases upwelling and productivity but at the same time decreases nutrient content in the upper thermocline, especially silicate. It is precisely the same dynamics as postulated in attempting to resolve the Walvis Paradox (Berger et al., 1998a; Lange et al., 1999). We also note that the discrepancy between expected and observed surface temperatures mentioned above (during the diatom maximum in Fig. F31) suggests frontal development, with increased temperature gradients.

This simple conceptual model—cooling increases mixing but decreases thermocline fertility—can produce a pattern of productivity history that is appealingly similar to the one observed (Fig. F32). To fit the DAI index (here represented with a mean of unity and a standard deviation of 0.14), we used two factors. One is the system state (labeled "x") as proxied by the ¹⁸O values of benthic foraminifers at Site 849, eastern equatorial Pacific. We assume that upwelling increases with x (downward), as marked on the graph. At the same time, the sea-surface temperature decreases (as shown on the y-axis) in the fashion prescribed by regression on the alkenone results (Marlow et al., 2000). The other factor is the "distance from optimum conditions," defined as

abs[1/(fzo - x)],

where fzo (for "frontal zone optimum") has the values corresponding to x within the frontal zone (plus 0.5 to avoid dividing by zero), x is the ¹⁸O value as before, and the difference is the distance to the nearest frontal zone value. (Thus, within the frontal zone optimum, the "distance from optimum" factor = 1/0.5). The series labeled "model" results from adding the two factors:

DAI est = f{x, 1/(fzo - x)} = a x ¹⁸O +
b x 1/(0.5 + abs[distance from optimum]) + c,           (4)

where a, b, and c are set for best fit. Distance from optimum turns out to be the more powerful of the two variables, with twice the weighting.

The nature of the residual DAI, after subtracting the smoothed version shown in Fig. F32, is of some interest. It has considerable variation at the scale of 0.4 m.y., the period of modulation of eccentricity amplitude, which in turn modulates precession. The smoothed residual fits quite well with the seasonal contrast potential from orbital parameters (as given by Berger and Loutre, 1991); that is, diatom productivity is high during times when climatic fluctuations are large (Berger and Wefer, in press). The effect is dominated by precession. After 1 Ma the fit deteriorates, presumably as a result of increased decoupling between organic productivity and diatom productivity. A similar residual, based on opal determinations (see Pérez et al., Chap. 4, this volume), shows striking parallelism with the benthic carbon isotope record from Site 849 (in the eastern equatorial Pacific) (Mix et al., 1995). This may be interpreted as another clue to the link of silicate supply in thermocline waters of the eastern South Atlantic to deep circulation.

The same processes enriching Pacific deep waters with ¹²C enrich the upper intermediate waters of the South Atlantic with silicate (Berger et al., in press). We take this as confirmation of our hypothesis that the delivery of silicate from NADW is crucial in explaining optimum diatom production in the Southern Ocean (Berger and Wefer, 1991). Delivery of thermocline water from the Southern Ocean, then, would seem to provide a link to this global system and a starting point for the exploration of mechanisms producing the MDM. A strong 0.4-m.y. cycle, in addition, suggests the possible influence of chemical weathering on land as a factor modulating silica supply. In this respect, the studies of Lin and Chen (in press) on the Ge/Si ratio in diatomaceous shells are of special interest.