Records generated from the Congo Basin and the Walvis Ridge provide an insight into the dominant climatic and marine factors that influenced these areas. These factors include the influence of both gradual changes, the MPR, and short-term glacial-interglacial variability. A synthesis of the main events is provided in Figures F5 and F6.
A comparison of several proxy indicators of terrestrially derived supplies of material to the Congo Basin and the Walvis Ridge supports the use of the magnetic susceptibility record as a reliable indicator of terrigenous input (Durham, 2000). Magnetic susceptibility measures the concentration of magnetizable material in the sediment and the ease with which it can be magnetized (Thompson and Oldfield, 1986). In deep-sea sediments, this material is predominantly terrigenous in origin, as biogenic components are diamagnetic and therefore effectively nonmagnetic. Magnetic susceptibility varies according to the size and source of the lithogenic supply to the ocean and the biogenic/lithogenic ratio of the deep-sea sediments, both of which are essentially climatically controlled (Kent, 1982).
Further evidence for terrigenous input comes from low-resolution shipboard X-ray diffraction (XRD) data (Shipboard Scientific Party, 1998b). The identification and partial quantification of minerals by XRD enables an estimation of the relative abundances of materials such as quartz and clay minerals.
In particular, quartz is continentally derived and, therefore, can be used as an indicator of terrigenous input (Tiedemann et al., 1989). In addition, it is often attributed to eolian input and has therefore been used in some studies as an indicator of dust flux to the deep sea from the continents (Tiedemann et al., 1989), enabling a consideration of aridity and wind strengths. Clay minerals are also major constituents of deep-sea sediments and have a predominantly continental origin. Their distribution in the oceans is related to the sources and transport paths of material from the continents, influenced by climatic regimes and weathering processes (Singer, 1984). In particular, the clay mineral kaolinite has been shown to be associated with aridity-humidity variations (Singer, 1984). Because kaolinite is a known product of chemical weathering of igneous rocks in the tropical rainforest, enhanced supplies of kaolinite are generally assumed to represent increases in humidity. However, kaolinite volumes might also increase if overall supply is increased, and therefore, the relative amount of kaolinite compared to the supply of smectite is more favorably employed as an indicator of aridity and humidity by using kaolinite/(kaolinite + smectite) values (Singer, 1984).
In addition, comparison of the volumes of chlorins vs. porphyrins (the 410/665 porphyrin/chlorin ratio) in marine sediments can be used to consider sources of material and has been shown to be an indicator of terrigenous input (Rosell-Melé and Koç, 1997). A typical marine 410/665 signal in recent material would be expected to show a gradual increase in values downcore as chlorins gradually degrade to porphyrins with time and push the ratio upward. However, if the record shows great variability or high porphyrin/chlorin ratios, this suggests that there has been an input of porphyrins from another source. Often, this has been attributed to the input of degraded continental organic matter (Rosell-Melé and Koç, 1997).
Terrigenous input to the Congo Basin is predominantly driven by input from the Congo River (Jansen et al., 1984; Durham, 2000). Support for the use of the magnetic susceptibility record as an indicator of terrigenous input and, in particular, fluvial flow from the Congo River comes from studies that show that the majority of material transported by the river is fine-grained paramagnetic clay (Frederichs et al., 1999). Concentrations of paramagnetic materials have a large influence on magnetic susceptibility values, particularly in areas where supplies of other magnetic materials are low (Thompson and Oldfield, 1986).
Sedimentation rates in Holes 1076A and 1077A are high compared to those usually observed in deep-sea sediments (Shipboard Scientific Party, 1998a). Sedimentation rates are often assumed to reflect variability in terrigenous supplies (Schneider et al., 1997), especially in areas where terrigenous input is high and variable, yet these rates can also be influenced by supplies of organic matter in areas where productivity is high (Stein et al., 1989). The material from Holes 1076A and 1077A, which contains high volumes of both terrestrially derived clays and organic matter, suggests that enhanced sedimentation rates in this area are the result of both large volumes of terrigenous material from the Congo River and high productivity (Durham, 2000).
Hole 1076A sedimentation rates vary between 20 and 450 m/m.y. with an average of 215 m/m.y., whereas Hole 1077A rates vary between 40 and 280 m/m.y. with a lower average of 116 m/m.y. (Durham, 2000). The higher rates for Hole 1076A may be a result of its proximity to the mouth of the Congo River, causing it to receive enhanced supplies of fluvially derived terrigenous material compared to its more distal neighbor. In addition, Hole 1077A sedimentation rates suggest a change in source or supply between 0.8 and 0.9 Ma. Prior to this, values average ~100 m/m.y. and there is very little variability.
Shipboard and shore-based magnetic susceptibility records from both Holes 1076A and 1077A compare favorably with one another and provide evidence for glacial-interglacial variability, with increased values coinciding with interglacial periods (Figs. F7, F8), indicating that terrigenous input was enhanced during interglacial periods. In conjunction with this, previous studies on the Congo Fan demonstrated that the flow of the Congo River increased during periods of enhanced humidity, coincident with warm interglacial periods (Jansen and van Iperen, 1991).
In addition, there is evidence for intermittent increases in magnetic susceptibility values during glacial periods (Durham, 2000; L.M. Dupont et al., unpubl. data). It has been suggested that these increases during glacial periods provide evidence for lower sea levels and enhanced erosion of the continental shelf (L.M. Dupont et al., unpubl. data). These increases are more evident in the records of Hole 1077A, suggesting that they may be a result of its more distal nature from the river mouth, causing it to receive greater influxes of terrestrial material from other sources.
In addition, the 410/665 ratio record for Hole 1077A compares favorably with the magnetic susceptibility record (Fig. F9) and therefore supports the use of both these parameters as indicators of terrigenous input.
Spectral analysis on the magnetic susceptibility records provides evidence for cycles with periods, indicative of Milankovitch forcing factors (Durham, 2000). In particular, a cycle with a periodicity close to 100 k.y. has been identified in Hole 1077A in both shipboard and shore-based magnetic susceptibility records. In addition, both sites display evidence for strong cycles in the precessional band. The existence of strong cycles in this band supports the suggestion that the production of sediment in the Congo Basin, which controls the sediment load of the Congo River, is driven by low-latitude precessional forcing (Berger et al., 1998b), which may affect aridity and humidity. The presence of a 100-k.y. cycle in Hole 1077A provides evidence for high-latitude forcing and may represent the supply of terrestrial material to this site due to the redistribution of sediment from the continental shelf in association with 100-k.y. variability in sea level changes.
In support of the hypothesis that glacial-interglacial variability in the magnetic susceptibility records is driven by fluctuations in fluvial flow, several previous studies demonstrated that during glacial periods the African continent suffered increased aridity, causing decreased flow of the Congo River (Jansen and van Iperen, 1991; Pastouret et al., 1978). Vegetation reconstructions have also demonstrated that during the last glacial maximum, the drainage basin of the Congo River consisted of savanna and grassland and the coastal zone was a desert, indicative of an arid environment, whereas in the present day, the same area is covered by woodland and tropical rainforests (Jansen and van Iperen, 1991).
Long-term trends in the magnetic susceptibility records provide evidence for a gradual decrease in terrestrial input from 0.9 to 0.4 Ma in Hole 1077A and from the base of the record (0.75 Ma) to 0.4 Ma in Hole 1076A. In the upper 0.4 Ma of the record, values begin to increase (Figs. F5, F8, F9). In addition, values for Hole 1076A are very low from 0.6 to 0.4 Ma, suggesting that fluvial flow reached a minimum at 0.6 Ma. In support of this, records of quartz concentrations show similar variability (Fig. F10). A decreasing trend in quartz values is particularly evident in Hole 1076A from the base of the core to 0.4 Ma, although, in fact, a further small decrease in quartz is also evident until 0.3 Ma. A decreasing trend also occurs from 0.9 to 0.4 Ma in the quartz records for Hole 1077A.
These records suggest that terrestrial input to the Congo Basin gradually decreased from at least 0.9 to 0.4 Ma. This can be attributed to a response to overall increased aridity, which has been shown to have occurred during the Quaternary period (Dupont et al., 1989; Tiedemann et al., 1989). This increase in aridity may have reduced the flow of the Congo River and, therefore, its sediment load. This signal was interrupted, however, at 0.4 Ma, when terrestrial input began to increase. In addition, a change in sediment supply to the Congo Fan was observed at this time (Jansen et al., 1984). Jansen et al. (1984) suggest that this may represent a response to a localized increase in humidity in the drainage basin of the river, but insufficient evidence from land records is available to justify this hypothesis.
In conjunction with this transition in trends, glacial-interglacial cyclicity demonstrates that sediment supply continued to increase during interglacial periods, suggesting that supply was still driven by the river. This increased supply, therefore, could represent a local increase in humidity, as suggested by Jansen et al. (1984), although it may also represent a response to the continuing increase in aridity, which may cause reduced vegetation cover, enabling more erosion and a greater source of material. Therefore, although the flow of the river may have decreased, its sediment load may have increased in response to a greater availability of material (Durham, 2000).
A similar event occurred at 0.9 Ma. Prior to 0.9 Ma, records from Hole 1077A demonstrate that fluctuations in the amplitude of quartz concentrations and quartz values are low. In addition, magnetic susceptibility values are also low and sedimentation rates manifest little variability, suggesting that supply from the Congo River may have been less significant at this time. At 0.9 Ma, increases in magnetic susceptibility and quartz values occur prior to the gradual decreasing trend. In addition, a large drop in the kaolinite/(kaolinite + smectite) record is evident. The change in these records coincides with the timing of the MPR. In many locations, this event has been associated with an increase in aridity, which, in a similar fashion to the event at 0.4 Ma, may have caused the sediment load of the Congo River to increase, although its flow may have been reduced. In support of this, pollen reconstructions from Hole 1077A (L.M. Dupont et al., unpubl. data) suggest that a change in vegetation occurred between 1.0 and 0.9 Ma.
The terrigenous input signal inferred from Hole 1081A shipboard and shore-based magnetic susceptibility records over the past 0.8 Ma on the Walvis Ridge is quite typical of the signal that would be expected at such a location far from fluvial influences (Fig. F11). Input increased during glacial periods in association with global cooling and enhanced aridity, leading to an increase in the supply of eolian material. Eolian transport from the Namib desert by northeasterly to easterly winds is corroborated by wind erosion forms (Diester-Haass et al., 1988). In addition, the lowering of sea levels in response to global cooling and increased ice volume during glacial periods left greater areas of the continental shelf and slope exposed and vulnerable to erosion and, therefore, also able to provide supplies of terrestrial material.
Diester-Haass and Rothe (1987) further argued from clay mineralogy evidence at DSDP Site 532 that a supply of terrigenous material from the Orange River, located to the south at a latitude of 28ºS, may have reached the Walvis Ridge during glacial periods. Over the past 0.7 m.y., therefore, the supply of terrigenous material to the ridge was increased during glacial periods, not only by eolian input and reworking of material from the continental shelf, but also by a supply from the Orange River from the south, transported to the ridge by the intensified strength and flow of the Benguela Current (Meeuwis and Lutjeharms, 1990).
The opposite terrigenous signal, however, is evident in the last 1-1.5 m.y., and a change in mineralogy suggests that a change in conditions occurred between 1.1 and 0.8 Ma. In contrast to increased terrigenous input during glacial periods, from 1.5 to 1.0 Ma, terrigenous input increased during interglacial periods. This suggests that the supply of terrigenous material via eolian input may only have been significant at the Walvis Ridge in the last 0.8 m.y., suggesting that prior to this period a different source of terrigenous material played a major role.
Evidence from mineralogy provides support for an additional or different source of terrestrial material influencing the ridge from 0.8 to 1.5 Ma (Fig. F12). During this period, quartz concentrations decreased and kaolinite values increased. Yet, subsequent to this, the values appear to parallel one another and both manifest an overall decreasing trend from 1.0 to 0.8 Ma. Diester-Haass and Rothe (1987) suggested that kaolinite was a major constituent of the Kunene River, located north of the Walvis Ridge at 17ºS, but was not evident in the material from the Orange River. In addition, it has been demonstrated that the ABF, which controls the latitude at which the Benguela Current turns west, also controls the southerly extent of the southward-flowing Angola Current (Jansen et al., 1996). It has been shown that this front shifted further north during glacial periods (Jansen et al., 1996). Therefore, during interglacial periods in the 1.0- to 1.5-Ma period, the ABF may have been far enough south to allow the supply of terrigenous matter from the Kunene River to be transported via the Angola Current to the Walvis Ridge.
Kaolinite concentrations, however, dropped significantly between 1.1 and 0.8 Ma. Although kaolinite is a measure of humidity, supplies are also strongly dependent on source (Frederichs et al., 1999; Diekmann et al., 1999), and this evidence suggests that between 1.1 and 0.8 Ma the supply of material from the Kunene River via the Angola Current from the north was somehow reduced and was therefore no longer observed at the ridge after 0.8 Ma. This change in supply may be attributed to one of two possible factors, or even both (Durham, 2000)—either that the Angola Current no longer reached the ridge subsequent to this time or that the flow and terrestrial supply from the river was greatly reduced because of increased aridity, meaning that although the Angola Current may have continued to reach the ridge, it no longer brought with it a supply of terrigenous material.
In support of this second factor, Oberhänsli (1991) suggested that intermittent incursions of the Angola Current are evident at the Walvis Ridge over the past 0.5 m.y. This suggests that it is the supply of the river, rather than the influx of the current, that caused this change in source (Durham, 2000). The timing of this change in supply and source of terrigenous input to the Walvis Ridge coincided with the timing of the MPR. In addition, it has been shown that aridity on the African continent significantly increased between 1.1 and 0.9 Ma (L.M. Dupont et al., unpubl. data), and this may have caused a decrease in the flow of the river.
Subsequent to this event, magnetic susceptibility values gradually increased until 0.5 Ma, after which values decreased. The increasing trend in values in conjunction with increased supplies during glacial periods indicates a response to increased aridity, causing an increase in eolian input due to enhanced exposure of the continent and the continental shelf and increased erosion, both on long and short timescales. The termination of this trend at 0.5 Ma, however, indicates that the supply of material began to decrease, although evidence from mineralogy does not indicate a change in source. This may, therefore, perhaps represent a response to several possible factors, including a gradual decrease in the supply of material from the exposed continental shelf in association with a gradual increase in sea levels, a decrease in wind strengths, or even a decrease in the volume of material eroded due to a decrease in humidity-related chemical weathering.
Comparison of several productivity proxy records demonstrates that the most reliable indicator appears to be the chlorin MAR and TOC MAR records. Although these are dependent on sedimentation rate, the close correlation between chlorin MAR and chlorin concentration evident from scatter plots indicates that, in fact, sedimentation rate appears to be highly driven by organic matter and, therefore, that the influence of terrigenous supply on these mass accumulation rates can be ignored (Fig. F13). In addition, evidence from the shipboard evidence for the fossil assemblages may also help us consider the dominant primary producers and dominant water masses and sources (Shipboard Scientific Party, 1998c, 1998d, 1998e).
Upwelling in the Congo Basin is not directly related to the Benguela Current, which does not travel as far north as the basin (Jansen and van Iperen, 1991). Instead, the shifting north of the ABF means that incursions and eddies of upwelled water, associated with the Angola Dome, formed by the ABF, may be seen in this area (Jansen et al., 1996). In addition, over the past 0.2 m.y., southerly excursions of the ABF led to the influence of equatorial doming from the north in the waters overlying the Congo Fan (Peterson and Stramma, 1991; Jansen et al., 1996). The location of the fan and Holes 1076A and 1077A, situated north of the fan, means that they are, therefore, sensitive to the influence of two thermal domes, the equatorial dome to the north and the Angola Dome to the south. Incursions from these domes are associated with the northerly and southerly shifting of the ABF (Peterson and Stramma, 1991; Jansen et al., 1996).
Previous studies also demonstrated that, in general, the nutrient supply from the Congo River to the ocean is greater during humid (interglacial) rather than arid (glacial) periods, yet periods with more variable river fluxes and high coastal upwelling coincide more often with arid periods (Jansen and van Iperen, 1991). High biological productivity off the Congo River is characterized by very high diatom productivity, evident in a plume that reaches out into the open ocean for 800 km (van Bennekom and Berger, 1984; Schneider et al., 1997). In association with this plume, river-induced upwelling has also been shown to occur (Jansen and van Iperen, 1991).
Records from Holes 1076A and 1077A indicate that no clear glacial-interglacial variability in productivity is evident in the Congo Basin (Figs. F14, F15). The first evident peak in the productivity record from Hole 1077A occurs at 1.1 Ma during interglacial Stage 37, and a subsequent peak occurs at 0.98 Ma during Stage 31. Throughout this interval until 0.8 Ma, upwelling diatoms are high in abundance and evidence from foraminifers suggests that waters from the north were apparent above Hole 1077A from 1.2 to 0.6 Ma (Shipboard Scientific Party, 1998d; Durham, 2000). In addition, influxes of colder waters from the south occurred between 1.2 and 1.05 Ma and 0.88 and 0.8 Ma. The incursion between 0.88 and 0.8 Ma may be the cause of the low SSTs evident at this time; SST values in the 1.2- to 0.5-Ma interval are also slightly reduced (Fig. F15).
The productivity peaks at 1.1 and 0.98 Ma, therefore, occur when warm waters from the north are evident in Hole 1077A, although the peak at 1.1 Ma also coincides with an influx of waters from the south. Both coincide with increased terrigenous input, suggesting a fluvial influence (Durham, 2000). The influx of water from the south at 1.1 Ma does not influence this river-induced signal, yet a small increase in productivity in the glacial period prior to the event at 1.1 Ma (isotope Stage 32) may represent enhanced marine upwelling due to this incursion.
At 0.8 Ma, records from Hole 1077A show a decrease in the abundance of upwelling diatoms in conjunction with an increase in productivity, which is enhanced intermittently in both glacial and interglacial periods (Fig. F15). From 0.8 to 0.6 Ma, the area remained bathed in waters from the north, with no evidence of waters from the south. The increases during interglacial periods suggest the influence of the river. Yet, the lack of evidence for waters from the south indicates that the glacially induced productivity during this interval was not driven by southern waters and was, instead, driven by other marine forces such as enhanced coastal upwelling. A brief decrease in productivity coincides with an influx of the warm-water foraminifer G. ruber between 0.75 and 0.65 Ma, suggesting an enhanced incursion of warm waters from the north, may have temporarily impeded these influences.
From 0.6 to 0.4 Ma, productivity in Holes 1076A and 1077A was low, except for a peak at both sites at 0.56 Ma toward the end of interglacial Stage 15 and a further peak at 0.42 Ma in Hole 1076A. Both appear to have a fluvial driving force (Durham, 2000). During this interval, incursions of water from the south are again evident in Hole 1077A. In addition, there is no evidence for incursions from the north, suggesting either that northern waters were cut off by the enhanced strength of the southerly waters or that the southerly waters reached the site because of the reduced flow of the river or even that the strong current from the south may have impeded the flow of the river (Durham, 2000).
At 0.4 Ma, upwelling diatoms increase in abundance in Hole 1077A, foraminifers become abundant in Hole 1076A, and terrigenous input at both sites begins to increase, suggesting sediment supply from the river was enhanced. Foraminifer assemblages from Hole 1076A indicate that an incursion of water from the north occurred at this time and the tropical species G. ruber is also briefly evident, yet productivity is low, and in Hole 1077A, waters from the south remained dominant. At 0.3 Ma, it has been proposed that the position of the ABF stabilized (Shipboard Scientific Party, 1998d, 1998e). After this time, the influence of southern waters appears to have ceased in Hole 1077A, and from 0.24 Ma, there is once again evidence for warmer waters from the north.
In Hole 1076A, incursions of waters from the south began at 0.2 Ma and the influence of northern waters ceased between 0.25 and 0.05 Ma. Therefore, despite its more southerly location, the influence of waters from the south is not evident in Hole 1077A, suggesting that although the ABF shifted to the north and stabilized, the transport of waters from the south to Hole 1077A was somehow impeded (Durham, 2000). In the upper 0.2 m.y. of Hole 1076A, the most evident productivity event occurs at 0.18 Ma, and there is also evidence for such an event in studies from the Congo Fan (Jansen and van Iperen, 1991; Jansen et al., 1996). Yet, this event is not evident in Hole 1077A, suggesting that different water masses affected the two sites at this time. The event has been attributed to a temporary shifting of the ABF to the south and the influence of equatorial doming (Jansen et al., 1996).
The evolution of the productivity records from Holes 1076A and 1077A, therefore, demonstrates that a complex interaction of factors drives productivity records in this region. In Hole 1077A, enhanced productivity is not as extensive in either duration or magnitude as in Hole 1076A, yet there is evidence for more frequent periods of increased productivity, particularly during interglacial periods. In addition, interglacial productivity appears to coincide with enhanced supplies of terrigenous input, suggesting that it was driven in part by the Congo River. Yet, glacial increases in productivity cannot be so simply attributed to one source, as at times they coincide with incursions of waters from the south, whereas at other times, there is no clear evidence for other water masses and a wind-driven coastal upwelling cause is hypothesized (Durham, 2000).
The record for Hole 1076A shows less consistent variability and is dominated by four major productivity events. In addition, the close proximity of the site to the Congo River might suggest that it would be more highly driven by fluvially driven nutrient input, but although it received an enhanced sediment load from the river, it does appear to be located at the very edge of the nutrient-enriched plume and, therefore, was less affected by fluvially induced productivity than Hole 1077A, which, although at a more distal location from the river mouth, was more greatly influenced by the plume. In addition, the proximity of Hole 1076A to the continent means that it was also less affected by marine-induced productivity than Hole 1077A and, therefore, that its productivity record represents responses to significant transitions or events in both the marine and terrestrial environment, rather than a consistent response to gradual or cyclic changes.
Previous evidence from the Walvis Ridge has proved to be contentious over the issue of glacial-interglacial cyclicity because it has been argued by several authors (Oberhänsli, 1991; Summerhayes et al., 1995; Little et al., 1997) that upwelling at the latitude of the Walvis Ridge generally increased during glacial periods, whereas other studies suggest that upwelling was more intense during interglacial periods (Diester-Haass, 1985).
It has since been suggested, however, that the discrepancy between these records may not necessarily reflect a decrease in upwelling and productivity during glacial periods but could instead be an artifact of the location of the site (Diester-Haass et al., 1992) and, therefore, that upwelling may have been enhanced during glacial periods but that the signal may have been transported elsewhere. Alternatively, it has been hypothesized that upwelling persisted but that the upwelled water was depleted in nutrients because of changes in deep-sea circulation (Hay and Brock, 1992). In a similar fashion, Oberhänsli (1991) suggests that the site did experience intensified productivity from increased upwelling during glacial periods but this was not reflected in some records because of poor preservation.
Records from Hole 1081A generally provide evidence for increased productivity during glacial periods (see Fig. F16). A peak occurs in all proxy records of productivity at 1.1 Ma that coincides with a maximum in diatom abundance. Subsequent to this, diatom abundance dramatically decreases, the abundance of foraminifers begins to increase, and values in other productivity proxies remain elevated (Durham, 2000). This change in the dominant microfossil species might suggest that a temperature-associated threshold (Hay and Brock, 1992) was reached at this time, causing diatom productivity to begin to deteriorate, whereas overall productivity remained high until 0.8 Ma when values significantly drop in association with a more rapid decrease in diatom abundance. This diatom maximum has, in fact, been observed in many records from the South Atlantic and is followed by a decrease in abundance that begins between 1.5 and 1.0 Ma (Dean and Parduhn, 1984; Berger and Wefer, 1996), suggesting it is driven by more than just localized variability.
The timing of these changes in productivity suggests that they may be related to the MPR. In association with this event, hypothesized to have caused increased cooling and aridity in this area (L.M. Dupont et al., unpubl. data), it is possible that the water masses or nutrient supplies began to gradually change, causing diatom productivity to begin to decrease following a period of elevated and increasing values, while productivity continued to be elevated. At 0.8 Ma, a significant and rapid drop in both overall productivity and diatom abundance is evident, suggesting that the gradual change was complete.
From 0.8 to 0.5 Ma, productivity begins to increase once again and diatom abundance decreases more rapidly, culminating in a disappearance in diatoms at 0.5 Ma in association with a peak in productivity. Productivity then appears to rapidly drop before gradually recovering in the latest 0.4 m.y. in conjunction with a gradual increase in diatom abundance. Studies suggested that an abrupt decrease in sea level may have occurred at this time (Vail et al., 1977; Haq et al., 1987; van Donk, 1976), and this may have been the cause of this rapid increase and subsequent decrease in productivity at 0.5 Ma. In response to a sea level drop, productivity may be enhanced, as lowered sea level means that nutrient-enriched bottom waters are closer to the surface and also that there is less volume of water in which the nutrients are distributed (Hay and Brock, 1992). The peak is, however, only short-lived, suggesting that the enhanced productivity removed the nutrients before they could be replaced, and productivity, therefore, subsequently rapidly decreased. The gradual subsequent increase in productivity values following this decrease indicates that conditions begin to return to normal as sea level began to rise and nutrient supplies were replaced.
The influence of the MPR is evident in the records from both the Congo Basin and the Walvis Ridge, although its impact on these areas was different. In particular, the effect of the MPR at the Walvis Ridge was more gradual, especially in proxy productivity records (Durham, 2000).
Proxy productivity records from the Congo Basin show a very rapid change at 0.8 Ma, suggesting that the basin responded only when conditions had fully changed. A brief incursion of waters from the south is, however, apparent from 0.9 to 0.8 Ma (Fig. F5), which might suggest that the strength of the Benguela Current intensified at this time, but lack of further evidence for southern waters until much later in the record indicates that this incursion was only temporary. In addition, however, there is evidence for enhanced glacial productivity subsequent to 0.8 Ma, indicating that either coastal or marine upwelling was enhanced in conjunction with increased productivity associated with the flow of the river.
Evidence from terrigenous input to the Congo Basin suggests that the MPR occurred between 1.0 and 0.9 Ma and may have caused a decrease in the flow of the Congo River (L.M. Dupont et al., unpubl. data). Yet, in conjunction with this, the sediment load of the river appears to increase (Durham, 2000). In addition, subsequent to this event, productivity was enhanced, and this, too, may be attributed to an increase in the nutrient load of the river after this time.
At the Walvis Ridge, the MPR appears to have occurred more gradually between 1.1 and 0.8 Ma and caused a change in the dominant source of terrigenous material to the ridge in conjunction with a change in productivity. It is hypothesized that these events are connected (Durham, 2000) and that prior to the MPR, a supply of terrigenous material from the Kunene River, situated north of the ridge, may have been transported to the ridge by the Angola Current and partially driven productivity. The MPR appears to result in a decrease in the flow and sediment load of this river, however, as subsequent to 0.8 Ma, this fluvial source of material was less evident. In conjunction with this, productivity decreased and the main primary producer changed.