This preliminary description of planktonic foraminiferal distribution at Site 1087 during the last four climatic cycles suggests that the surface hydrology of the southern Cape Basin off southern Africa was affected by a complex series of both short- and long-term drastic variations. The studied site lies at the crossroads of three distinct oceanographic realms, namely, the Southern Ocean, the Benguela coastal upwelling system, and the southwest Indian Ocean, which presently influence to a large extent the nature of the surface mixed layer in the southern Benguela region (Shannon and Nelson, 1996). Consequently, we argue that proxy data measured at Site 1087 can hardly provide on their own firm answers about the mechanisms that drive the hydrographic changes recorded in this study and that subsequent studies will have to rely deeply upon information gained from previous or ongoing paleoceanographic studies conducted in the above-mentioned nearby realms. Of particular relevance in this respect are investigations of Ocean Drilling Program (ODP) Site 704 (Hodell, 1993) and ODP Leg 177 (Shipboard Scientific Party, 1999) in the Atlantic part of the Southern Ocean, studies by Flores et al. (1999), Acheson et al. (1999), Rau et al. (1999) in sediment cores from the Agulhas retroflection region, and the outcomes from investigations of Leg 175 material in the Benguela coastal upwelling region (Shipboard Scientific Party, 1998, and this volume).

Among the various questions raised by this preliminary investigation, future works should particularly address the following aspects. First, the downcore distributional trend of the dominant planktonic foraminiferal species responds roughly in phase with global climate change down to MIS 8 but shows an ill-defined relationship with the glacial-interglacial cycles from ~250 k.y. to the end of the studied interval of Site 1087 (i.e., MIS 12). This shift in distributional trend might be at least in part viewed as a local response of the southern Benguela productivity and circulation pattern to the so-called mid-Brunhes event (Jansen et al., 1986), a global climate oscillation of unknown origin, the manifestation of which included major changes in the whole-ocean carbon inventory and in the nature of intermediate and deep water masses (Howard and Prell, 1994).

Second, a single pulse of newly upwelled waters over Site 1087 is recorded during the early phase of MIS 9. This unique event has recently been documented with a similar micropaleontological signature (peak occurrence of N. pachyderma (s) in a nearby piston core (MD 962085) collected on the continental slope off the Orange River, ~2 north of Site 1087 (M.-T. Chen, pers. comm., 1999), confirming the reality of this signal in terms of surface-water changes (i.e., the observed micropaleontological signature is not related to processes of downslope transfer of shelf material). Equatorward advection of cold, G. pachyderma (s)-rich polar waters along southwest Africa can hardly be inferred as a mechanism for this anomalous cooling at Site 1087. Previous paleoceanographic investigations conducted in the Southern Ocean indicate that, together with MIS 11, MIS 9 was one the warmest climatic stages of the last 500 k.y. These two interglacials are characterized by extreme poleward shifts of the Polar Frontal Zone and Subtropical Convergence (Howard and Prell, 1992; Niebler, 1995), a process that acts against the presence of polar waters at the latitude of the studied site. Possible mechanisms of the recorded MIS 9 upwelling pulse at Site 1087 might include an extreme southward position of the South Atlantic high-pressure cell and associated southeasterlies and/or drastic changes in the physico-chemical nature of the upwelled waters. Detailed mapping of hydrological and atmospheric conditions during MIS 9 along the whole margin of southwest Africa presently affected by coastal upwelling will greatly improve our understanding of this short-term event.

Finally, we show that input of warm, salty Indian Ocean thermocline waters to the South East Atlantic was most effective at glacial terminations, the maximum pulses of interocean exchange being associated with reseeding of G. menardii in the tropical Atlantic Ocean during Terminations I and II, as well as during the upper part of MIS 12. Whether the input of Indian Ocean thermocline water is controlled mainly by the latitudinal position of the Subtropical Convergence, by changes in volume transport of the Agulhas Current along southeast Africa, which is thought to control the zonal position of the Agulhas retroflection regime (Shannon et al., 1990), or by a combination of both mechanisms is still a matter of debate. Understanding the late Quaternary timing and causes of the observed variability of Indian water inflow will therefore benefit from additional studies looking both at the dynamics of hydrological fronts in the Southern Ocean and at the circulation changes along Southeast Africa (velocity of the Agulhas Current and associated atmospheric circulation in the Southwest Indian Ocean).