Main

We used an established deepwater proxy parameter — the carbon-isotope composition (δ13C) of the epifaunal benthic foraminifera Cibicidoides wuellerstorfi5,6 — to evaluate Holocene deepwater variability. Variations in the δ13C of the total amount of CO2 in bottom waters, which is accurately recorded by C. wuellerstorfi, can be used to monitor variations in the contribution of high-δ13C North Atlantic Deep Water (NADW) relative to low-δ13C Southern Ocean Water (SOW) to a site. We worked on sediment from Ocean Drilling Project site 980 (55° N, 15° W; depth, 2,179 m), on the Feni Drift in the subpolar northeastern Atlantic.

Holocene δ13C values (Fig. 1, and see supplementary information) show marked millennial oscillations around the modern value of about 1‰ (ref. 7). Values range from a high of 1.3‰ to values lower than 0.6‰, indicating times of enhanced and reduced NADW contribution, respectively. The largest reductions in the relative NADW contribution occurred around 9,300 years ago (9.3 kyr), and 8.0, 5.0 and 2.8 kyr ago. Smaller events occurred more frequently.

Figure 1: Holocene climate records, top to bottom: benthic δ18O and δ13C (North Atlantic deepwater contribution) records from site 980 (ref. 12, and see supplementary information); GISP2 sea-salt sodium flux8 (increasing, calm to windier), and percentage of haematite-stained grains (%HSG) in core V29-191 (ref. 2).
figure 1

Triangles, dated levels in sediment cores; arrow marks the onset and intensificatoin of the 5-kyr event. Accelerator mass-spectrometer radiocarbon dates converted to calendar age13 from site 980 provided the chronology for the Holocene (see supplementary information). Accumulation rates average about 25 cm kyr−1 in the Holocene, roughly double that in nearby core V29-191. The average interval between samples is about 100 yr from 9.7 to 1.2 kyr. Shading shows low-δ13C events (see text) and possible correlative events in the other records; a more tenuous correlation is denoted by the question mark. Rectangles, extreme winter-like conditions deduced from statistical analysis of the full set of palaeochemical indicators from GISP2 (ref. 8).

Full size image

The most pronounced feature of the Holocene is a trend of decreasing relative NADW contribution that began at about 6.5 kyr and culminated with a minimum at around 5 kyr. Statistical analysis of the full set of chemical records from the Greenland Ice Sheet Project 2 (GISP2) ice core indicates that meteorological conditions from 6.1 to 5.0 kyr were especially winter-like (characterized by an expanded polar vortex) at high latitudes; this is consistent with high values of sea-salt sodium (Fig. 1) during this millennial-scale event8.

A high relative abundance of haematite-stained grains in a nearby core indicates a large proportion of cold, fresh, ice-bearing surface water from north of Iceland2. All three records indicate that the 5-kyr event was the most severe climate event of our Holocene study interval. Apparently contemporary variations in benthic δ13C and sea-salt sodium levels indicate that linked, century-scale variability may have occurred during this event.

The GISP2 palaeochemical records indicate enhanced winter-like conditions from 2.4 to 3.1 kyr, and from 8.8 to 7.8 kyr, although to a lesser extent than in the 5-kyr event8. The more recent of these cold intervals corresponds precisely to the reduced NADW contribution at around 2.8 kyr. The 8-kyr δ13C minimum occurs late in the 8.8–7.8-kyr cold interval, but corresponds to a brief maximum in sea-salt sodium. The age of the 9.3-kyr event is poorly constrained, but it may correspond to one of several younger or older maxima in sea-salt sodium flux. Three of the four low-δ13C events correspond to maxima in the percentage of haematite-stained grains (increased incursion of rock-bearing ice from north of Iceland), suggesting a possible surface–deepwater linkage.

Prior geochemical evidence for deepwater reduction in the Holocene was limited to events outside our study interval. Low δ13C values during the Little Ice Age4, the most recent of the Holocene millennial cold events8, and during an early Holocene event at ∼10.3 kyr (ref. 9) hint at a linkage between millennial climate and deep water. Downcore variations in sedimentological indices also indicate this possibility10,11.

The new North Atlantic benthic δ13C record unambiguously demonstrates that NADW varied on centennial–millennial time scales during the Holocene. The most significant Holocene event of reduced NADW contribution, which occurred at 5 kyr, was correlated to winter-like atmospheric conditions at high northern latitudes8, and to incursion of sea ice from north of Iceland2. Similarly, a more recent cold event (at 3.1–2.4 kyr) was associated with a reduced NADW contribution.

Evidence for a climate–deepwater linkage during the earlier events is weaker, and may indicate an increasing sensitivity of deepwater to surface forcing from the Early to the Late Holocene. Further well-dated deepwater proxy records are needed to test this possibility.

Our study raises other important issues that can be addressed through data collection and modelling. The amplitude of the largest Holocene δ13C fluctuation near 5 kyr, for example, is similar to those of δ13C oscillations during earlier glacial and deglacial events12. This suggests that significant variations in the relative NADW contribution can occur in the absence of forcing by large ice sheets, and that NADW may be more sensitive to surface forcing than was previously imagined.