DISCUSSION

Dissolution in the Water Column and at the Sediment-Water Interface

Biosiliceous particles produced in ocean surface waters partly or completely dissolve while they settle through the water column because of the undersaturation of seawater with respect to silica. Dissolution predominantly takes place within the ocean surface layer and at the sediment water interface (e.g., Nelson and Gordon, 1982; van Bennekom et al., 1988). In order to assess the possible effects of these dissolution processes on the opal record in the study area, we would need detailed analyses regarding production, dissolution, and accumulation of opal in surface sediments, which are not available for the Bellingshausen Sea, but are for the Ross Sea and the Weddell Sea (DeMaster et al., 1996; Nelson et al., 1996; Schlüter et al., 1998). These marginal seas of the Southern Ocean represent environmental and depositional settings comparable to those in the Bellingshausen Sea because the biological productivity in the surface waters is mainly controlled by the sea-ice cover, the slope and rise sediments predominantly consist of lithogenic components, and the Holocene sedimentation rates vary between 1 and 10 cm/k.y. (DeMaster et al., 1996; Bonn et al., 1998; Schlüter et al., 1998). In the Ross Sea and the Weddell Sea, the concentrations of dissolved silica in the bottom waters bathing the continental slope and rise are typically 120-135 µmol/L (Jacobs, 1989; Rutgers van der Loeff and van Bennekom, 1989). These values are consistent with the dissolved silica concentration of 120 µmol/L measured on a bottom-water sample above Site 1096 (J. Schuffert, pers. comm., 2000), indicating that at least recently the WSDW in the Bellingshausen Sea is not more corrosive to opal than bottom waters in the Ross and Weddell Seas. Thus, we can assume that opal dissolution processes in the water column and the surface sediments of the Bellingshausen Sea are similar to those in the Ross Sea and the Weddell Sea.

At the continental margins in the Ross Sea and the Weddell Sea, the ratio of opal accumulation in the surface sediments to primarily produced opal in the surface water was estimated to be 0.1%-10% (DeMaster et al., 1996; Nelson et al., 1996; Schlüter et al., 1998). For the surface sediment sample at Site 1096, we computed a MAR_opal of 2.8 g/m² per yr, assuming a Holocene sedimentation rate of 4 cm/k.y. (Pudsey and Camerlenghi, 1998). According to Ragueneau et al. (2000) the biogenic silica production rate in the seasonal ice-covered Pacific sector of the Southern Ocean is 408 mg opal/m² per day. On the condition that the sea-ice limited productivity season above Site 1096 lasts 90 days per yr, the modern ratio of opal accumulation to opal production is ~8% in the Bellingshausen Sea. This value is consistent with the ratios given for the Ross and Weddell Seas.

Dissolution within the Sediment Column

Diagenetic dissolution of opal within the sediment column predominantly takes place in the surficial sediments (e.g., Schlüter et al., 1998), as also revealed by dissolved silica concentrations in pore water profiles at Sites 1095, 1096, and 1101 (Barker, Camerlenghi, Acton, et al., 1999). Late diagenetic dissolution apparently did not significantly bias the primary opal fluctuations in the older, deeper buried sediments at Sites 1095, 1096, and 1101 because the dissolved silica concentrations conform with contents in siliceous microfossils and, thus, the opal concentrations (Barker, Camerlenghi, Acton, et al., 1999). A bottom-simulating reflector (BSR) observed in seismic profiles across Drift 7 was interpreted to be caused by the diagenetic transformation of opal-A to opal-CT (Rebesco et al., 1997). This BSR was reached by drilling at the base of Site 1095 (Camerlenghi et al., 2001). Therefore, we consider that the opal concentrations in the two lowermost samples at Site 1095 may have been biased by opal diagenesis.

Comparison of the Opal Record with Other Paleoproductivity Proxies

In spite of the various dissolution processes, opal records in upper Quaternary sediments from the Bellingshausen Sea were interpreted in terms of a proxy for climatic-induced productivity changes in the surface waters (Pudsey and Camerlenghi, 1998; Hillenbrand, 2000; Pudsey, 2000). This assertion is based on the close correspondence between the records of opal and biogenic barium. The biogenic barium signal is assumed to be resistant to dissolution processes in the Antarctic continental rise setting, where oxic conditions prevail in the sediment column (Shimmield et al., 1994; Bonn et al., 1998). At gravity core site PS1565, located between Drifts 3 and 4 (Fig. F1), both contents (Ba_bio) and accumulation rates (MAR_{Ba bio}) of biogenic barium in sediments deposited during the last 130 k.y. positively correlate with opal contents and MAR_opal, whereas contents of TOC and CaCO₃ as well as accumulation rates of TOC (MAR_TOC) and CaCO₃ (MAR_CaCO3) show remarkable discrepancies in respect to MAR_{Ba bio} (Figs. F5, F6). These findings suggest that the opal signal is a more reliable proxy for paleoproductivity in respect to TOC and CaCO₃.

Opal Contents vs. Opal Accumulation Rates

In general, MAR_opal is controlled by the deposition and the preservation of opal, whereas the opal content is additionally influenced by the dilution with other sedimentary components. In the drift sediments of the Bellingshausen Sea, lithogenic components chiefly operate as diluting agents. No significant discrepancies between the records of opal contents and MAR_opal are observed in the upper Quaternary sediments at site PS1565, where LSR exhibited hardly any fluctuations (Figs. F5, F6). Also at Sites 1096 and 1101, the long-term patterns of opal concentrations and MAR_opal show only minor discrepancies (Figs. F3, F4), suggesting that the record of opal contents in the drift-crest sediments was not masked significantly by the dilution with lithogenous particles. At Site 1095, however, no connection between the long-term trends of opal contents and MAR_opal is observed. MAR_opal generally declines throughout the last 9.5 m.y., thereby exhibiting spikes during time intervals with high LSR (Fig. F4), whereas the opal concentrations show different long-term variations (e.g., a prominent maximum between 5.3 and 3.0 Ma) (Fig. F3). This disconnection seems to point to a major influence of lithogenic dilution on the opal contents at drift-flank Site 1095.

Effects of Lateral Sediment Supply on Opal Accumulation Rates

The relationship between MAR_opal in the sediments and original depositional rates of siliceous particles onto the seabed is strongly affected by the supply of other sedimentary constituents, so that changes of LSR play a major role for opal preservation (e.g., Archer et al., 1993; Ragueneau et al., 2000). For the Antarctic continental margin in the Ross Sea, DeMaster et al. (1996) found a positive correlation between the LSR and the "opal seabed preservation efficiency" (that is, the ratio of MAR_opal in the surface sediment to the original opal deposition on the seabed) for LSR varying from 0 to 20 cm/k.y. (Fig. F7). Moreover, MAR_opal is often affected by lateral redistribution of biosiliceous particles by bottom currents (e.g., Frank et al., 2000; Pondaven et al., 2000). Accumulation rates of biogenic components in late Quaternary continental margin deposits in the Ross and Weddell Seas were shown to be influenced by bottom current-induced lateral supply of biogenic particles ("focusing"), which may contribute up to two times more biogenic matter than vertical particle rain (Frank et al., 1995; Ceccaroni et al., 1998).

The bottom-current flow on the Antarctic Peninsula continental rise probably acted as a focusing mechanism, too. It has to be stated here that the high sedimentation rates at the drift sites in the Bellingshausen Sea gave the reason for choosing them for deep-sea drilling. We suppose, however, that the direct lateral supply of opal to Sites 1095, 1096, and 1101 was not as important for the MAR_opal at these locations as the better opal preservation caused by the embedding with laterally delivered lithogenic detritus. The particles entrained in the bottom current are assumed to derive from vertical settling through the water column during interglacial periods and from turbidity currents during glacial episodes (e.g., Barker Camerlenghi, Acton et al., 1999; Pudsey, 2000). The vertical settling matter probably comprised a significant amount of biosiliceous particles, but we expect that its current-controlled redistribution to the drift sites has accentuated the interglacial productivity signal delivered from the overlying water column even more.

In contrast, the turbidity currents originated from unstable locations at the Antarctic Peninsula continental slope, which are unfavorable for vast accumulation of biosiliceous particles. Therefore, the resuspended sediments should predominantly consist of lithogenic components. This assumption is supported by the low opal contents in distal turbidites deposited at the drift sites (Fig. F2). Nevertheless, the extensive occurrence of reworked diatoms at Sites 1095, 1096, and 1101 points to a redistribution of biosiliceous matter from the shelf down to the continental rise, even though its proportion never biased the microfossil assemblages deposited from the overlying water column (Barker, Camerlenghi, Acton et al., 1999). Also, the accumulation of biosiliceous matter at gravity core site PS1565 throughout the glacial marine isotope Stages 4-2 (Fig. F6) may indicate the supply of reworked opal because we would not expect any significant opal deposition at least for the last glacial maximum, when permanent sea-ice coverage prevailed above site PS1565 (CLIMAP Project Members, 1981). Therefore, the MAR_opal computed for Sites 1095, 1096, and 1101 probably comprises a lateral supply of opal, but its contribution is supposed not to have crucially altered the original opal productivity signal.

The effect of better opal preservation caused by lateral supply of lithogenic detritus is clearly documented in the MAR_opal at Sites 1095 and 1096. Because of the proximity of both locations (Fig. F1), the production of opal in the surface waters, its flux to the seabed, and the opal accumulation in the sediments should be comparable. In contrast, major discrepancies in MAR_opal occur between Sites 1095 and 1096 throughout the last 4.6 m.y., especially during time intervals when LSR differed markedly (Fig. F4). Between 4.6 and 2.6 Ma, MAR_opal at Site 1096 was about five times higher than at Site 1095, and between 1.7 and 0 Ma, MAR_opal at Site 1096 was about one-third higher than at Site 1095. Thus, we argue that the continuously higher MAR_opal at Site 1096 with respect to Site 1095 chiefly reflects a better preservation of biosiliceous particles caused by the higher LSR at the crest of Drift 7 with regard to its distal flank. Moreover, we conclude that at Sites 1095, 1096, and 1101, where the LSR varied markedly (Fig. F4), the effect of lateral sediment supply on opal preservation has to be considered, whereas it may be neglected at site PS1565, where LSR exhibited hardly any fluctuation (Fig. F6).

Estimation of Opal Depositional Rates for Sites 1095, 1096, and 1101

We used the close relationship between the opal preservation efficiency and the LSR found for continental margin sediments in the Ross Sea (DeMaster et al., 1996) to decipher the signal of the original opal depositional rate on the seabed from the MAR_opal record at Sites 1095, 1096, and 1101. The validity to adopt this approach for the investigated drift sediments in the Bellingshausen Sea arises from the similarity of the depositional settings and the fact that the LSRs at Sites 1095, 1096, and 1101 were <20 cm/k.y. throughout most of the Neogene and Quaternary (Fig. F4). The linear regression shown in Figure F7 can be expressed by the following equation:

Opal depositional rate [g/m²/yr] =
MAR_opal [g/m²/yr]/(-0.0264 + 0.0379 [k.y./cm] x LSR [cm/k.y.])

The resulting opal depositional rates for Sites 1095, 1096, and 1101, including mean values for 200-k.y. time intervals, are shown in Figure F8. For LSRs much higher than 20 cm/k.y. the regression curve between opal preservation efficiency and LSR can be no longer linear and an asymptotic approach of the regression curve to a value of 100% preservation efficiency is expected instead. This is documented by an opal preservation efficiency of 86% for settings with a LSR of 250 cm/k.y. (DeMaster et al., 1996). Therefore, at Site 1095 we equated the opal depositional rates with the MAR_opal for the time intervals between 9.308 and 9.230 and between 8.651 and 8.635 ka, when LSRs exceeded 30 cm/k.y. (Fig. F4).

At Site 1095, an extremely high opal depositional rate of 200 g/m² per yr was inferred for the time period from 4.890 to 4.620 ka, when the LSR was 1.03 cm/k.y. Our approach should be valid for this relatively low LSR because the linear correlation of opal preservation efficiency with LSR was established mainly on LSRs in this order (Fig. F7), but such a high opal depositional rate is only typical for deep basins on the inner Antarctic shelf (DeMaster et al., 1996). The nonrealistic high value obtained for Site 1095 may indicate that the opal concentration preserved in the corresponding sample is too high for the supposed LSR. We consider that only one sample was analyzed for opal from the time interval between 4.890 to 4.620 ka and that the assumption of a constant LSR may be wrong for the whole period of time (cf. Froelich et al., 1991). On the other hand, a hiatus suggested to span the time interval from 6.140 to 5.040 ka at Site 1095 (M. Iwai, pers. comm., 2001) would also affect the LSR between 4.890 and 4.620 ka and ultimately result in continuously realistic values for the opal depositional rate. Therefore, the nonrealistic high value was ignored in the calculation of the long-term trend of opal depositional rates at Site 1095.

During the late Miocene, estimated opal depositional rates at Site 1095 predominantly varied between 30 and 50 g/m² per yr with relative minima centered at ~7.8, ~6.6, and ~5.6 Ma (Fig. F8). Throughout the Pliocene and Quaternary, the long-term trends of opal depositional rates at Sites 1095, 1096, and 1101 exhibit a much better correspondence than MAR_opal (Figs. F4, F8). Highest opal depositional rates (45 to 60 g/m² per yr) occurred between 5.2 and 3.1 Ma at Sites 1095 and 1096. Today, comparable opal depositional rates are reported from the Scotia Sea, south of the Polar Front, and from the eastern ACC (Schlüter et al., 1998). During the Quaternary, opal depositional rates in the Bellingshausen Sea varied between 20 g/m² per yr at Site 1095 and 11 g/m² per yr at Site 1096. For comparison, on the continental margin in the Weddell Sea, between 18 and 30 g/m² per yr opal are deposited during the present interglacial (Schlüter et al., 1998). In general, the long-term patterns of the estimated opal depositional rates at Sites 1095, 1096, and 1101 reflect those of the opal contents (Fig. F3). This finding suggests that the opal concentrations represent a more reliable proxy for paleoproductivity than the MAR_opal in the drift setting of the Bellingshausen Sea.

Biological productivity in the surface waters of the polar Southern Ocean is mainly controlled by the availability of light, which is controlled by sea-ice cover. In general, (sub-)polar conditions seem to have prevailed in the Bellingshausen Sea throughout the Neogene and Quaternary. The microfossil assemblages found in the drift sediments at Sites 1095, 1096, and 1101 show no evidence of significantly warmer surface-water temperatures than today, even though the occurrence of particular diatom species and calcareous nannofossils in some intervals might indicate a more subantarctic setting (Barker, Camerlenghi, Acton, et al., 1999; M. Iwai, pers. comm., 1999). Therefore, we assume that sea-ice coverage was the main factor in limiting paleoproductivity. At present, the duration of sea-ice coverage in the Bellingshausen Sea is controlled by atmospheric heat supply and by upwelling of warm CDW (Hofmann et al., 1996; Jacobs and Comiso, 1997).

We expect past changes in sea-ice coverage to have been linked to regional climate in the Antarctic Peninsula area, which should have influenced the air temperatures, the upwelling rate of CDW, and/or the geographic position of the Polar Front. Another global aspect concerns the heat budget of the CDW, which depends on interhemispheric thermohaline circulation, particularly on the contribution of deep waters from northern hemispheric sources to the Southern Ocean. The heat transfer by this Northern Component Water (NCW) may have influenced sea-ice extent in the Bellingshausen Sea. Of course, we expect both positive and negative feedbacks between regional climate and global thermohaline circulation, raising "chicken and egg problems." For example, regional warming in Antarctica and the adjacent Southern Ocean might trigger a southward shift of the ACC, resulting in enhanced upwelling and a stronger thermohaline overturn, which might cause more inflow of NCW (Toggweiler and Samuels, 1995; Billups et al., 1998; Kim and Crowley, 2000). In turn, heating of the Southern Ocean by NCW injection might result in higher precipitation and, thus, ice buildup in Antarctica, leading to regional cooling (Prentice and Matthews, 1991; Billups et al., 1998). Moreover, the waxing and waning of sea-ice coverage in the Southern Ocean is expected to influence global climate via the atmosphere. Sea-ice expansion not only enhances the global albedo, but also reduces the atmospheric CO₂ level because sea-ice coverage restricts CO₂ release from the ocean to the atmosphere (Hanna, 1996; Elderfield and Rickaby, 2000).

Late Miocene (9.5-5.23 Ma)

The occurrence of IRD in upper Miocene sediments from the Bellingshausen Sea recovered at Site 1095, ODP Leg 178 (Barker, Camerlenghi, Acton, et al., 1999) and at Site 325, Deep Sea Drilling Project Leg 35 (Hollister, Craddock, et al., 1976), as well as clay mineralogical and sedimentological evidence from continental margin deposits recovered in the Weddell Sea during ODP Leg 113 (Kennett and Barker, 1990), suggest that major cooling and glaciation had affected the Antarctic Peninsula region at least since the early late Miocene. A major eustatic sea level drop during the middle Miocene coinciding with a significant increase in global ice-volume (Lear et al., 2000) was interpreted to indicate ice sheet buildup in West Antarctica at ~15 Ma (Abreu and Anderson, 1998). At ~10 Ma the production of NCW similar to modern NADW had started in the North Atlantic (Woodruff and Savin, 1989; Wright et al., 1991; King et al., 1997). Concomitantly, opal sedimentation in the Southern Ocean, comparable to the present situation, became widespread (Ciesielski and Weaver, 1983; Wise et al., 1985). As indicated by oxygen isotope data measured on benthic and planktonic foraminifers in the subantarctic South Atlantic, relatively warm and stable oceanographic conditions prevailed in the Southern Ocean between ~11 and 6.6 Ma (Müller et al., 1991; Warnke et al., 1992).

The estimated opal depositional rates at Site 1095 were generally higher during the late Miocene than during the late Quaternary (Figs. F8, F9), suggesting reduced sea-ice coverage in the Bellingshausen Sea. The flow of NCW was well established but weaker than present (King et al., 1997). Therefore, we conclude that regional climate was warmer than during the Quaternary and was mainly responsible for higher productivity offshore from the Antarctic Peninsula.

Based on low-resolution benthic ¹³C records, Wright et al. (1991) and Wright and Miller (1996) reported distinct long-term changes in the NCW flow to the Southern Ocean throughout the late Miocene. We do not recognize possible effects of such circulation changes at Site 1095 because the relative minima of opal depositional rates centered at ~7.8, ~6.6, and ~5.6 Ma (Figs. F8, F9) are not synchronous with the supposed changes in NCW flow and might therefore also represent consequences of regional climatic trends. High-resolution records of carbonate fluxes at the Ceara Rise in the equatorial West Atlantic, however, indicate a more uniform NCW flow (King et al., 1997), which would be consistent with the opal record in the Bellingshausen Sea. We consider that the minimum of opal deposition recorded at ~5.6 Ma may be linked to the Messinian Event, when the Mediterranean Sea had desiccated (Krijgsman et al., 2000). Because Mediterranean Overflow Water (MOW) is supposed to act as an important source water for the formation of deep water in the Nowegian-Greenland Sea (Reid, 1979), the cessation of MOW outflow during the Messinian stage likely caused strong decrease in NCW production (e.g., Woodruff and Savin, 1989; Müller et al., 1991). A concomitant reduction of heat injection into the CDW during the Messinian stage may have triggered enhanced sea-ice coverage and low productivity in the Bellingshausen Sea, but stratigraphy is uncertain for this period of time at Site 1095.

Early Pliocene (5.23-3.58 Ma)

During the early Pliocene, the climate over Antarctica and the adjacent Southern Ocean was warmer than today (e.g., Abelmann et al., 1990; Hodell and Venz, 1992; Barker, 1995; Burckle et al., 1996; Barker et al., 1999). In the deep western equatorial Atlantic, a significant lowering of the calcite-lysocline at 5.1 Ma is interpreted to indicate the onset of a strengthened southward NCW flow analogous to modern NADW flow (King et al., 1997). Additionally, vertical gradients in benthic foraminiferal ¹⁸O from that area showed the NCW to be slightly warmer than present-day NADW between 4.2 and 3.7 Ma (Billups et al., 1998). Foraminiferal oxygen isotope data and high carbonate accumulation in the subantarctic South Atlantic (Müller et al., 1991; Froelich et al., 1991; Hodell and Venz, 1992), as well as siliceous microfossil assemblages in the Weddell Sea (Abelmann et al., 1990) indicate elevated surface water temperatures prevailing in the Southern Ocean, at least during particular early Pliocene time intervals. These findings agree with a paleoceanographic scenario suggested by Berger and Wefer (1996), which assumes a stronger thermohaline deep-water circulation in the Atlantic Ocean with maximum formation and convection of NCW during the early Pliocene in response to the closure of the Panama Isthmus.

Throughout the early Pliocene, the opal depositional rates at Sites 1095 and 1096 were much higher than during the late Quaternary (Figs. F8, F9), pointing to a strong reduction in annual sea-ice extent or even open ocean conditions in the Bellingshausen Sea. We consider, however, that also during that time interval short-term advances of ice masses across the Antarctic Peninsula shelf occurred, as documented by clay mineral fluctuations at Sites 1095 and 1096 (Hillenbrand and Ehrmann, Chap. 8, this volume) and by glacial unconformities on the shelf (Bart, 2001). Nevertheless, a general early Pliocene warming in the Bellingshausen Sea agrees well with both warm atmospheric conditions over Antarctica and with an intensification of meridional heat transfer into the Southern Ocean due to enhanced thermohaline circulation in the Atlantic Ocean.

A crucial point, however, is the exact timing of the climate change. The estimated opal depositional rates at Site 1095 exceeded the mean late Miocene level already at ~5.2 Ma (Figs. F8, F9), synchronously with strengthened NCW flow recorded at the Ceara Rise (King et al., 1997). Considering the possible hiatus spanning the time interval between 6.1 and 5.0 Ma at Site 1095, the early Pliocene opal depositional rates would have reached maximum values at ~4.8 Ma. In both cases, the increase of productivity in the Bellingshausen Sea precedes the closure of the Panama Isthmus starting, at 4.6 Ma (Haug and Tiedemann, 1998). Moreover, Warnke et al. (1992) reported particular warm conditions to have dominated in the subantarctic South Atlantic between 5.2 and 4.6 Ma. Thus, we conclude an early onset of Pliocene warming in the Antarctic realm, which might have enabled an intensification of thermohaline deep-water circulation in the Atlantic Ocean at ~5 Ma via a positive feedback mechanism, as suggested by Toggweiler and Samuels (1995) and Kim and Crowley (2000). On the other hand, the formation of NCW, which was warmer and more saline than present NADW, took place at 4.2 Ma (Billups et al., 1998) and might therefore be caused exclusively by the closure of the Panama Seaways (Berger and Wefer, 1996).

Late Pliocene (3.58-1.77 Ma)

Intense cooling of the Northern Hemisphere started at ~3.2 Ma and triggered a stepwise decrease in glacial NCW production, probably due to enhanced sea-ice formation in the Nordic Seas (e.g., Raymo et al., 1990, 1992; Berger and Wefer, 1996; Wright and Miller, 1996). Enhanced glacial carbonate dissolution recorded in the deep western equatorial Atlantic also reflects the decrease in NCW flux in response to the northward expansion of southern source deep waters between 3.3 and 2.6 Ma (Tiedemann and Franz, 1997). During that time interval, significant cooling affected the Atlantic Sector of the Southern Ocean, as indicated by a decrease of surface- and bottom-water temperatures, a northward shift of the Polar Frontal Zone, expansion of perennial sea-ice coverage, and decreasing ventilation rates in the deep waters caused by reduced NCW injection into the CDW at ~2.8 Ma (Kennett and Barker, 1990; Abelmann et al., 1990; Hodell and Warnke, 1991; Hodell and Venz, 1992; Warnke et al., 1992). It was speculated that with the onset of northern hemisphere glaciation, the Antarctic continent was affected by an episode of major cooling in combination with increased ice buildup (Kennett and Barker, 1990).

Late Pliocene opal depositional rates at Sites 1095, 1096, and 1101 did not decrease significantly until ~3.1 Ma (Fig. F8). Since that time, biological productivity in the Bellingshausen Sea declined continuously toward the Pliocene-Pleistocene transition (Figs. F8, F9). The decrease in NCW formation and flow throughout the late Pliocene likely caused the annual sea-ice coverage in the Bellingshausen Sea to expand, thus reducing opal depositional rates. Intensification of deep- and bottom-water circulation at the same time may have resulted in nondeposition or sediment erosion near the Matuyama/Gauss boundary at Site 1095. A similar scenario was suggested to explain a hiatus between 2.7 and 2.4 Ma on the Meteor Rise in the subantarctic South Atlantic (Hodell and Venz, 1992).

During the Pliocene and Quaternary, sedimentation rates at Site 1095 on the distal flank of Drift 7 were generally much lower than at Site 1096 near the crest (Fig. F4), suggesting that at Site 1095 the bottom-currents around the drift held more particles in suspension than at Site 1096. Therefore, a stronger bottom-current flow on the Antarctic Peninsula continental rise, possibly induced by enhanced WSDW formation in the Weddell Sea, would interrupt deposition or remove sediments at Site 1095 rather than at Site 1096.

Quaternary (1.77-0 Ma)

Benthic ¹³C records in the North Atlantic and in the western equatorial Atlantic were interpreted to reflect a significantly reduced NADW flow during the Quaternary glacials compared to the late Pliocene (Raymo et al., 1990; Bickert et al., 1997). In the Atlantic sector of the Southern Ocean, early Quaternary climatic conditions were found to be warmer than at present (e.g., Abelmann et al., 1990; Hodell and Venz, 1992). Distinct glacial-interglacial cyclicity, which was superimposed on a general cooler climate with widespread occurrence of sea ice (Abelmann et al., 1990), characterized the late Quaternary Southern Ocean (e.g., Charles et al., 1991; Grobe and Mackensen, 1992; Howard and Prell, 1994; Hodell et al., 2000).

The long-term trend of opal depositional rates recorded at Sites 1095, 1096, and 1101 shows no significant changes of biological productivity in the Bellingshausen Sea throughout the Quaternary (Figs. F8, F9). However, mean NADW contribution to the Southern Ocean should have rather weakened during that time interval because of the stronger suppression during glacials. Therefore, we conclude that the observed constant level in productivity may be related to relatively stable climatic conditions in the Antarctic Peninsula area. In the Bellingshausen Sea, distinct short-term fluctuations of productivity linked to orbital-forced glacial-interglacial cyclicity on Milankovitch timescales are documented at least for the late Quaternary (e.g., at gravity core site PS1565) (Figs. F5, F6) (Pudsey and Camerlenghi, 1998; Hillenbrand, 2000; Pudsey, 2000).