INTRODUCTION

The carbon reservoir within the oceans is 60 times greater than that of the atmosphere (e.g., Berger et al., 1989). A small change in the oceanic reservoir may strongly affect the atmospheric reservoir with which it exchanges. Perhaps the most important way in which the ocean can alter its ability to hold CO₂ is through changes in surface water productivity (Broecker, 1982). Such changes set the partial pressure of CO₂ (pCO₂) through both "biological pumping" (in which carbon is removed from surface waters to deep waters), and "biological dumping" (in which organic carbon is removed from the system through sedimentary deposition and burial) (Shipboard Scientific Party, 1998a). Coastal upwelling regions play an important role in this system because nutrient-rich waters, when vertically advected to the surface through the combined actions of wind forcing and Ekman pumping, induce intense surface water productivity. This acts as a huge sink for atmospheric CO₂.

The defining feature of the Quaternary era was the rapid expansion of the Northern Hemisphere ice sheets. This marked a unique period in climate history, when the planet entered into a state of bipolar glaciation. Numerous Ocean Drilling Program (ODP) records taken from the Northern Hemisphere suggest the intensification of Northern Hemisphere glaciation (INHG) took place in three stages at ~2.74, ~2.70, and ~2.54 Ma (reviewed in Maslin et al., 1998). However, the latter step associated with the expansion of the Laurentide Ice Sheet (i.e., the glaciation of northeast America) had the greatest effect on the global system (e.g., Maslin et al., 1998; Shackleton et al., 1995, 1984; Tiedemann et al., 1994). It has been suggested that the INHG occurred as a consequence of tectonically induced long-term cooling and was initiated by changes in orbital forcing (Haug and Tiedemann, 1998; Maslin et al., 1998), however the feedback mechanisms that translated this forcing into global climate change have still to be understood.

Northern Hemisphere glaciation would have altered atmospheric circulation patterns, and therefore wind-driven upwelling regimes, such as the Benguela Current (BC) upwelling system, which is among the five or six major upwelling regions of the world (Berger et al., 1998). Intermediate in intensity between the systems off Peru and California (Lange et al., 1999; Berger et al., 1998), it is located in the subtropical eastern South Atlantic, off the coast of Namibia, southeast Africa, extending from Cape Point (34°S) in the south to Cabo Frio (18°S) in the north (Lange et al., 1999), as shown in Figure F1.

The Benguela system is important because it incorporates the BC, a major surface water vector of heat and salinity toward the Brazilian Coastal Current (BCC) that crosses the equator. Consequently, it has a major influence on the cross-meridional flow of heat energy that influences global climate. Moreover, the BC system is also a major sink for atmospheric CO₂. Both of these processes may be important climate feedbacks that contributed to the INHG.

The modern oceanography of the region consists of a number of surface and deep-water currents and is shown in both aerial and schematic view in Figure F1. The BC is a shallow (<80 m), cool, surface water current flowing toward the Equator. It forms part of the eastern limb of the South Atlantic Gyre (Dowsett and Willard, 1996). The water of the BC is thought to originate from three sources (Garzoli and Gordon, 1996, and references therein): the South Atlantic Current (the southern limb of the South Atlantic subtropical gyre); the Agulhas Current (the warm south Indian Ocean western boundary current), and subantarctic water from the Antarctic Circumpolar Current (Antarctic Intermediate Water, AAIW).

Between 23° and 28°S (varying with the season), the BC splits into two components (Lange et al., 1999; Berger et al., 1998): the BCC, which transports cold, nutrient-rich water from the wind-dominated coastal area northward along the shelf (Schneider et al., 1997), and the Benguela Oceanic Current (BOC), which takes warmer water northwest by geostrophic flow (Holmes et al., 1996). The BCC is the weaker branch of the divergence (Hay and Brock, 1992), whereas the BOC is the stronger vector and transfers heat toward the Brazilian Coastal Current, and thus to the Northern Hemisphere (Berger et al., 1998). This, therefore, has important implications for global climate.

At ~16°S (Cabo Frio), the cold, northward-flowing BCC meets with the warm tropical/equatorial southward-flowing waters of the Angola Current (AC). This marks the position of the Angola Benguela Front (ABF), across which there is a sharp temperature gradient of 7.5°C (Holmes et al., 1996). At this point, the BCC subducts beneath the AC and continues northward into the Angola Basin as a shallow subsurface current (Holmes et al., 1997). The location of the ABF shifts between ~15°S during austral spring and ~17°S in austral autumn (Holmes et al., 1996).

Beneath the BCC is a poorly oxygenated poleward-flowing undercurrent (PUC), which is restricted to the edge of the shelf and upper slope, and centered at ~200-300 m depth (Berger et al., 1998, and references therein). It forms the eastern side of a cyclonic subsurface gyre that upwells the more oxygenated, nutrient-rich AAIW immediately beneath the pycnocline (Hay and Brock, 1992). Ventilation by the AAIW is considered to be directly proportional to the strength of the BC (Berger et al., 1998).

South of 15°S Ekman motion combined with south and southeast zonal trade winds from the African continent, cause waters to be upwelled over the inner shelf and shelf break (Holmes et al., 1997). This takes place within a narrow band not more than 100 km wide, mostly within the 300 m isobath (Hay and Brock, 1992). Although occurring as a year-round phenomenon in this region (Lalli and Parsons, 1997), upwelling is subject to variation resulting from the seasonal north-south migrations of the trade winds. During austral winter, the Intertropical Convergence Zone (ITCZ) shifts northward, pulling the subtropical high-pressure zone toward the equator (Diester-Haass et al., 1992). As a result, modern upwelling in the Benguela system is at its maximum during austral spring from December to April (Hay and Brock, 1992). Upwelling varies alongshore in eight cells (Fennel, 1999), extending from 15° to 35°S, with the strongest signal found at Lüderitz, Namibia (27°S) (Lange et al., 1999).

Estimates of the depth from which waters are upwelled range from 150 m (Holmes et al., 1998) to 330 m (Hay and Brock, 1992). The depth of the pycnocline off Walvis Bay is considered to be only 250 m below the surface (Hay and Brock, 1992), enabling cold, nutrient-rich waters (from within the PUC) to be vertically advected into the photic zone. This results in high nearshore nutrient concentrations (Holmes et al., 1996) with coastal surface water nitrate and phosphate levels frequently reaching 30 and 2.5-8 µM/L, respectively (Holmes et al., 1998). Consequently, productivity levels are high, with carbon fixation rates reaching 125-180 g C/m² per yr (Berger, 1989; Holmes et al., 1996).

The BC system was previously drilled by the Deep Sea Drilling Project (DSDP) (Legs 40, 74, and 75); however, most of these sites were considered too far offshore to provide a direct record of upwelling (Berger et al., 1998). Instead, they were interpreted to contain records of productivity from eddies that had been transported westward by the BC (Diester-Haass et al., 1992). Additionally, the cores were rotary drilled and considered too disturbed to permit high-resolution reconstructions (Berger et al., 1998). Nevertheless, a fairly detailed study of productivity was completed by Dowsett and Willard (1996) using sediments drilled from DSDP Site 532.

Sediments used in this study were taken from Site 1083 (20°53.6481´S, 11°13.0720´E), drilled during ODP Leg 175 during the summer of 1997. Site 1083 is situated near to the edge of the continental shelf and, together with Sites 1081 and 1082, and DSDP Sites 532 and 362, forms a rough north-south transect within the Walvis Ridge/Walvis Basin area. Four holes were drilled at the site, and sediments from Hole 1083A were used in this study.

Lying close to the major upwelling centers off southwest Africa, which at present maintain a year-round activity, these sites should directly record paleointensity fluctuations in coastal upwelling. Site 1083 is located the farthest from the shore, and is therefore expected to have the best representation of pelagic signals. It was also the deepest, drilled in a water depth of 2178 m (Berger et al., 1998). The lithology of the site throughout the time period investigated consisted of one single unit composed of moderately bioturbated clayey nannofossil ooze (Shipboard Scientific Party, 1998b).

This study attempts to reconstruct productivity within the BC upwelling system during the time of the expansion of the Laurentide Ice Sheet in the Northern Hemisphere, at ~2.54 Ma (Maslin et al., 1998, and references therein), principally using diatoms and nitrogen isotopes (¹⁵N). However total organic carbon (TOC) is also used to strengthen the interpretation. The relationship between productivity and upwelling is investigated through the "hard" isothermal remnant magnetism (HIRM) record, which serves as a proxy for wind strength, obtained from the same sediment.

Attention is focused on the time frame 2.40-2.65 Ma, not only to investigate the period of rapid ice-volume expansion but also to reconstruct activity both prior to and immediately following the event. The reconstruction of a high-resolution record is achieved through a sampling interval of 20 cm (~2 k.y.). This will permit inferences to be made concerning the relationship between the Northern and Southern Hemispheres at times of ice-volume expansion.