A comparison of the records plotted to age for the Norwegian Sea and the North Atlantic sites is shown in Figure 7, Figure 8, Figure 9. An abrupt decrease in the carbonate contents at Site 985 at 3.05 Ma indicates a distinct change toward severe glacial conditions, probably with a relatively continuous sea-ice cover. This pattern may possibly be interpreted as the culmination of a gradual long-term cooling trend in the Northern Hemisphere with significant intensification of glaciation since ~6 Ma (Jansen and Sjøholm, 1991; Larsen et al., 1994; Fronval and Jansen, 1996). The intensification of the glacial regime occurred almost synchronously in the Iceland Sea (Baumann et al., 1996; Fronval and Jansen, 1996) and possibly documents the influence of an earlier and/or larger Greenland Ice Sheet. However, >125-µm IRD at Site 985 occurs only in relatively small quantities until ~2.5 Ma (and subsequently). The IRD, however, forms nearly 100% of this size fraction, and the amount of silt-sized particles is >90% throughout most of this interval (Fig. 7, Fig. 9). This may have resulted from weak ice rafting and little continental erosion. The major onset of the Scandinavian glaciation occurred somewhat later in comparison with the Greenland Ice Sheet. The Vøring Plateau, which is strongly influenced by the Norwegian Current, initially shows a significant increase in IRD input at 2.75-2.6 Ma (Jansen et al., 1988; Henrich et al., 1989; Wolf and Thiede, 1991). Correspondingly, a sudden and strong decrease in carbonate contents at the Vøring Plateau also occurred at ~2.75 Ma (Baumann et al., 1996).
In comparison, indications of the increased glaciation do not occur in the North Atlantic until 2.8 Ma, the time of a first and significant increase in IRD (Fig. 7). The North Atlantic Drift would probably have turned straight eastward without entering the Norwegian Sea, but heat flux transport to the Rockall Plateau area was not reduced. Besides, between 2.8 and 2.5 Ma, glacial intervals became progressively more drastic as reflected by short but extreme decreases in carbonate contents and numbers of coccoliths (Fig. 7, Fig. 8). These findings are in good accord with results of isotope and IRD studies of the North Atlantic (Shackleton et al., 1984; Shackleton and Hall, 1984; Keigwin, 1987; Jansen et al., 1988; Raymo et al., 1989; Raymo, 1994). There is strong evidence for this progressive cooling from the increase in the percentage of cooler living planktonic foraminifers (Loubere and Moss, 1986). Also, Backman et al. (1986) and Backman and Pestiaux (1987) reported both marked variability of warm-adapted discoaster abundances and significantly decreasing accumulation rates of discoasters in Deep Sea Drilling Project (DSDP) Hole 552A before 2.5 Ma. Sea-surface temperature estimates, however, are not easy to reconstruct because the assemblages are not completely analogous to modern faunas (Ruddiman and Raymo, 1988).
In addition, the carbonate curve from Site 982 looks quite similar to those of Sites 552, 607, and 609 (Shackleton et al., 1984; Raymo et al., 1989; Ruddiman et al., 1989). In particular, the high values in the late Gauss interrupted by small carbonate minima, which are caused by IRD input, can be well correlated between the sites. These short-termed carbonate decreases are not well documented at Sites 552 and 609, probably because significant amounts of sediment are missing at core breaks at these sites (Raymo et al., 1989). Correlation is thus best with Site 607 as well as in the Pleistocene sequences of all sites. These short but synchronous intervals, however, probably reflect progressively increased glacial conditions during the late Gauss.
The causes of the initiation and intensification of the Northern Hemisphere glaciation remain essentially unresolved. This is a very controversial topic and not a basic part of this study. A variety of causes were discussed. These included changes in orbital forcing as well as tectonic explanations, such as the emergence of the Panama Isthmus (e.g., Keigwin, 1982; Keller et al., 1989) or the progressive uplift of the Tibetan-Himalayan regions (Ruddiman and Raymo, 1988; Ruddiman et al., 1989; Ruddiman and Kutzbach, 1991).
In the time interval from ~2.8 to 1.1 Ma, very low to zero carbonate production is suggested for the Norwegian Sea. Thus, siliciclastic fine-grained sediments (constantly >90%) and coarse IRD (Fig. 9) characterize most of the sediments of this period. Similar record patterns have also been found at Site 907 on the Iceland Plateau (Baumann et al., 1996; Fronval and Jansen, 1996) although the pattern at Site 644 farther east is slightly different (Baumann et al., 1996). Here, at least periodic intrusions of warmer waters into the easternmost Norwegian Sea were reported by Henrich and Baumann (1994) and Baumann et al. (1996). Nonetheless, only sparse evidence for relatively warm surface-water intrusions can be found at Site 985. Low carbonate contents and numbers of planktonic foraminifers are characteristic for the whole interval. Only at 1.9 and 1.45 Ma do carbonate contents and the total numbers of coccoliths reach higher values. Another indication for Atlantic water inflow during this interval possibly comes from the IRD record. Strong terrigenous input is reported to be restricted to the flanks of the isotope curves, indicating that the occurrence of IRD displays rapid melting after warm periods (Wolf and Thiede, 1991; Baumann et al., 1996). High IRD contents in the period until ~1.1 Ma therefore indicate at least temporary open-water conditions before melting and deposition of ice-rafted material in the Norwegian Sea.
In the North Atlantic, the high carbonate values of the Pliocene are relieved by highly fluctuating carbonate contents since 2.5 Ma and are accompanied by relatively strong IRD peaks. Thus, the cold glacial phases had diminished surface-water productivity but also increased IRD supply. In addition, especially in the interval 2.5-1.65 Ma, colder sea surface-water temperatures in the North Atlantic are indicated by the coccolith assemblage. The numbers of total coccoliths significantly decreased after 2.5 Ma and, contemporaneously, the cold water-adapted species C. pelagicus reached very high abundances. The occurrence of C. pelagicus could also be controlled by different ecological factors and not by temperature alone. Nonetheless, these findings indicate somewhat similar conditions in the surface waters of the North Atlantic in the interval 2.5-1.65 Ma compared to the sea-surface conditions in the Norwegian Sea during interglacials of the late Pleistocene. This holds true both for numbers of total coccoliths and for relative abundances of C. pelagicus. Further evidence comes from biometric analyses of C. pelagicus placoliths (Baumann, 1995; Baumann and Meggers, 1996; K.-H. Baumann, unpubl. data). Today, this species is dominant with small-sized placoliths (7-12 µm) in plankton and surface sediments of the Greenland and Iceland Seas (Samtleben et al., 1995). Large placoliths (>10 µm), however, are observed in the northern North Atlantic, which probably indicate unfavorable conditions for C. pelagicus (Baumann, 1995). However, during the interval 1.6-2.5 Ma, small-sized placoliths of C. pelagicus were also observed in sediments of the northern North Atlantic and of the Labrador Sea (Baumann and Meggers, 1996). Thus, the preferred ecological requirements of this species seem to be stable through the Matuyama and Gauss Chrons.
Colder
surface-water conditions in the North Atlantic significantly
reduced the potential heat export to the Norwegian Sea.
Consequently, carbonate sedimentation in the Norwegian Sea was
extremely reduced. The gradient in sea-surface temperature must
have been extremely strong, however, and probably no Atlantic
water entered the Norwegian Sea. Hence, the circulation type was
different from that of the present-day ocean. Jansen et al.
(1988) has already proposed a stronger thermal gradient than at
present caused by a more zonal circulation system with deflected
North Atlantic surface-water currents. In addition, benthic
foraminifer 
13C
values (Raymo et al., 1990, 1992; Sikes et al., 1991) show a
constantly present but significantly reduced production of
northern source deep water. Reduced deep-water formation in the
Norwegian Sea must have been coupled to decreased advection and
weakened northward flow of Atlantic surface waters. Since the
Atlantic surface waters were strongly cooled (Sikes et al.,
1991), the resulting low carbonate production in the Norwegian
Sea and the persistent glacial conditions are not surprising.
This interval is also characterized by consistently present IRD. Visual inspection of the >500-µm fraction additionally shows the presence of larger dropstones between 2.4 and 2.1 Ma. In the period from 2.5 to 2.1 Ma, Jansen et al. (1989) and Sikes et al. (1991) found surprisingly low-amplitude fluctuations of oxygen isotope records after the glaciation events. Hence, Sikes et al. (1991) concluded that the magnitude of glacial events from 2.5 to 2.1 Ma was no more than one-half or two-thirds of the last Pleistocene glacial. The ice volume was estimated to be less than one-half of Pleistocene glaciation. Despite generally low ice volume, ice rafting as far as into the southern North Atlantic (as documented at Site 607 by Raymo et al., 1989) could be explained by surface waters that have cooled much more in response to Pliocene ice volume changes than in the late Pleistocene (Sikes et al., 1991).
After ~1.65 Ma, the relative IRD input to the North Atlantic decreased significantly (Fig. 9). In addition, total numbers of coccoliths increase, whereas the dominance of the cold water-adapted C. pelagicus declines (Fig. 7, Fig. 8). Thus, a considerable warming of the surface waters in the North Atlantic during interglacials is indicated after 1.65 Ma. In the easternmost Norwegian Sea, an appreciable warming occurred synchronously during short interglacial-like phases (Henrich and Baumann, 1994; Baumann et al., 1996). As shown by these data, however, the interval from 1.65 to 1.1 Ma seems to be glacial-dominated north of the Iceland-Faeroe Ridge, in strong contrast to the high potential heat export of the North Atlantic recorded in high carbonate contents during interglacials. Jansen et al. (1988) and Henrich and Baumann (1994) explained low carbonate contents in the Norwegian Sea in this interval with dissolution as a result of low carbonate productivity and decreased deep-water ventilation. More recent investigations at Iceland and Norwegian Sea Sites 907 and 644 (Baumann et al., 1996) have revealed high carbonate contents and a relatively high abundance of planktonic foraminifers on the Vøring Plateau (Site 644) in contrast to significantly diminished carbonate production in the Iceland Sea (Site 907). Such an extreme gradient could possibly be explained by extensive sea-ice cover and an extremely eastward-situated polar front between Sites 644 and 643, as earlier proposed by Henrich and Baumann (1994). On the other hand, ice-rafted material on the Iceland Plateau (Baumann et al., 1996) and at Site 985 reflects occasional melting and thus warming.
An explanation for these extreme south-north and east-west gradients in carbonate sedimentation could be given by faunal data. In the North Atlantic, N. pachyderma sin. is described to have first appeared 1.8 m.y. ago. The observed decrease in subpolar foraminifer numbers at Site 982 between 1.8 and 1.2 Ma, thus, would indicate cooling in contrast to the previously explained trend toward warmer surface water conditions, as reflected by coccolith assemblages. Raymo et al. (1987), however, showed that N. pachyderma sin. did not take on a distinct "cold-indicator" role in the North Atlantic until 1.7 Ma. This role was abandoned by N. pachyderma sin. from 1.3 to 1.2 Ma and recurred at 1.2 to 1.1 Ma with the mid-Pleistocene transition (Ruddiman et al., 1986). Other authors, however, suggested that N. pachyderma sin. did not attain its role as a cold-water indicator before 0.9 Ma (Ruddiman et al., 1986; Jansen et al., 1988). Biometric analysis of this species clearly indicated an increase in the maximum diameter of N. pachyderma sin. after 1.1 Ma, which is most likely related to a better adaptation to a polar habitat (Baumann and Meggers, 1996). Additionally, factor analysis showed that N. pachyderma sin. was first a polar indicator at 1.1 Ma (Meggers, 1996; Meggers and Baumann, 1997). In conclusion, this might be an important reason for the little faunal response to a possible sea-surface heat export from the North Atlantic to the Norwegian Sea, especially between 1.85 and 1.35 Ma. Except for the cold water-adapted C. pelagicus, there was no other polar-adapted carbonate secreting species and thus only limited (coccolith) carbonate accumulation.
For the last 1.1 m.y., a strong reaction of the Norwegian Sea carbonate production to heat export from the North Atlantic is obvious roughly each 100 k.y. During this time, the periodicity in the variation of the ice-sheet volume increased from a 41-k.y. to a 100-k.y. periodicity (Ruddiman et al., 1986, 1989; Raymo, 1992; Imbrie et al., 1993; Berger and Jansen, 1994). The gradient between the two observed sites, however, is strong. A comparison of the carbonate maxima of the two sites shows differences of 30-55 wt%. North Atlantic carbonate records are strongly influenced by varying productivity and IRD dilution (Ruddiman et al., 1986; Raymo et al., 1989), caused by ice-sheet control on sea-surface temperature. High carbonate productivity fluctuations in the Norwegian Sea were probably caused by variations in the intensity and extension of warm Atlantic water inflow (Henrich, 1989). In general, a somehow synchronous development of carbonate sedimentation between the Norwegian Sea and the North Atlantic can be observed during the last 1.1 m.y., which became more pronounced during the last 0.65 m.y. Strong gradients in surface-water temperatures existed, however, reflected by much lower carbonate contents and lower percentages of subpolar planktonic foraminifer species at Site 985 compared with Site 982.
A striking difference between the Norwegian Sea and the North Atlantic sites can also be observed in the composition of the planktonic foraminifer assemblages. At Site 985, the total abundance of planktonic foraminifers generally mirrors the shape of the carbonate record; this does not happen at Site 982 (Fig. 10). This difference may be explained by the fact that the cold water-adapted N. pachyderma sin. contributes as much as 80% of the total planktonic foraminifer assemblage in the North Atlantic during glacials. As a result, many planktonic foraminifers appear in glacials as well as in interglacials, whereas in the Norwegian Sea planktonic foraminifers reach high numbers only during interglacials. High abundances of polar foraminifers in the North Atlantic, however, should be restricted to glacials, where they found optimal environmental conditions. This is partly confirmed by the strong relationship between the carbonate record and subpolar foraminifer abundance in the youngest section until 0.65 Ma, when a highly significant correlation (r2 = 0.62) occurs (Fig. 10). In contrast, in the interval older than 0.6 Ma this relationship is strongly weakened, and the correlation coefficient is close to zero (r2 = 0.001). This also supports the previously discussed ecological meaning of N. pachyderma sin., which did not have a clear cold-water preference and cannot be considered as a polar-adapted species before ~0.65-0.8 Ma. In addition, these data confirm the general conclusion of Meggers and Baumann (1997) that a change in the life habitat of N. pachyderma sin. occurred. The optimum adaptation of the species to the polar environment, however, seems to have happened after ~1.1 Ma. As indicated by regression analysis, N. pachyderma sin. reached its optimum adaptation at ~0.65 Ma.
During the past 0.65 Ma, highest amplitude oscillations in biogenic carbonate records can be seen in the Norwegian Sea. This indicates that short but warm interglacials with presumably intensive deep-water formation occurred between relatively long-lasting glacials characterized by high inputs of IRD. It is known that the rates of ice decay are clearly faster than the rates of ice growth during the last 0.65 m.y. (Ruddiman et al., 1986; Raymo, 1992). Significant Atlantic water intrusions, especially during interglacials, reached far up into the Fram Strait (Baumann et al., 1996; Hevrøy et al., 1996). Hence, the surface circulation pattern probably occurred with only minor variations during the interglacials of the late Pleistocene. In addition, planktonic isotope data document the fact that glacial periods became colder and more pronounced (Fronval and Jansen, 1996). Carbonate preservation further improved, during both glacials and interglacials (Henrich and Baumann, 1994). Hence, continuous formation of North Atlantic deep water may have also occurred during glacials. A major precondition is sufficient salt supply by temperate Atlantic surface-water intrusions (e.g., Hebbeln et al., 1994).
