tion of the true rates of faunal change at the site, unmasked by bioturbation.3
The comparison of high-resolution faunal results for Troll 3.1 with other climate proxy records leads to the following conclusions concerning the variability and climatic impact of surface circulation during deglaciation:
In addition to initial deglacial warming from approximately 13,500 to 13,100 yr BP (20 to 18 m depth in core) and the Younger Dryas cold oscillation between about 11,200 and 10,500 yr BP (6 to 4 m), both of which have been recognized and dated previously in several cores from the northern North Atlantic, additional warm-cold-warm oscillations are documented at about 12,600 to 12,400 yr BP (15 to 10 m), 11,700 to 11,500 yr BP (8.5 to 7 m), and 9,700 yr BP (3.8 m) (see Figure 9). Although none of these oscillations was as severe as the Younger Dryas, they indicate that the surface limb of the conveyor was more variable in strength than previously imagined.
Estimates of the duration of single warm-to-cold and cold-to-warm transitions based on interpolation between 14C dates indicate that several of these transitions (marked by arrows in Figure 8) occurred in 40 years or less. As these transitions span the full or nearly the full range of conditions recorded by N. pachyderma (sin), they represent swings between near-present-day levels and near-absence of Atlantic water within the basin, that is, between near-extreme states of the conveyor circulation. These changes appear to occur with an abruptness similar to that of the isotopic shifts recorded in Greenland ice cores (Dansgaard et al., 1989). (The transition to warmer water conditions at the close of the Younger Dryas appears to have taken around 100 years, but since deposition rates at the site of Troll 3.1 drop to about 1 m/kyr by this time, the apparently slower transition may be an artifact of the decrease in stratigraphic resolution. Nonetheless, absolute rates of SST change at this time may have been just as great as earlier ones. As we shall see below, this transition may have involved a warming of as much as 9°C; see, e.g., Figure 10.)
On the basis of the observed range of temperature sensitivity of N. pachyderma (sin), Lehman and Keigwin (1992) estimated that larger faunal changes in Troll 3.1 corresponded to changes in SST of 5°C or more. A new, independently dated, high-resolution record of SST changes in the southeastern Norwegian Sea based on diatom assemblages, which have a much broader range of temperature
sensitivity than N. pachyderma (sin), confirms the earlier estimates from Troll 3.1 (Koc-Karpuz and Jansen, 1992). The comparison in Figure 10 shows that the range of SST that might be inferred from the percentage of N. pachyderma (sin) was truncated primarily at the warm end, as might be expected on the basis of present knowledge of its modern distribution (e.g., Bé and Tolderlund, 1971). The duration of major SST changes evident in Troll 3.1 and the magnitude of change confirmed by the diatom record suggest that the rates of SST change in the southeastern Norwegian Sea were approximately 1°C per decade, which is directly comparable to the rates of air-temperature change inferred from ice-core studies.
When plotted by age, the percentages of N. pachyderma (sin) variations in Troll 3.1 correspond closely with the history of mean annual air temperatures over the Greenland ice sheet deduced from changes in d18O of ice in Dye 3, which has the highest resolution of the long Greenland ice cores (Figure 11). This relationship is consistent with earlier suggestions that sudden changes in circum-Atlantic temperatures were governed by changes in strength of the conveyor circulation (Oeschger et al., 1984; Broecker et al., 1985). The strong similarity of temperature proxy records from Greenland, Britain, and Switzerland (Figure 1) indicate that conveyor circulation also affected air temperatures downwind of the northern North Atlantic at least as far as central Europe, in accordance with model predictions (e.g., Rind et al., 1986; see Figure 4).