High-resolution data on ocean circulation during the last glacial cycle suggest that the formation of North Atlantic Deep Water and associated heat transport may be more stable than previously thought. See Letter p.73
Data from ice cores drilled on Greenland in the 1980s revealed alternating warm (interstadial) and cold (stadial) intervals embedded within the last glacial period1, which occurred from about 70,000 to 17,000 years ago. Subsequent studies suggested that these millennial-scale climate variations were linked to changes in the formation and export of deep water in the North Atlantic2, which is associated with ocean transport of heat to high northern latitudes. Three modes of circulation of deep water were identified and correlated to climate regimes3. These modes include: formation and flow of North Atlantic Deep Water, the dominant water mass in the Atlantic, during interstadials; a slower, shallower circulation system during stadial events; and a shutdown of North Atlantic Deep Water formation during extreme cooling and ice-rafting events known as Heinrich stadials (Fig. 1). On page 73 of this issue, Böhm et al.4 present high-resolution data sets of ocean palaeocirculation that call into question the existence of three distinct modes and challenge our understanding of the coupling between climate and ocean circulation on millennial timescales.
The concept of multiple modes of ocean circulation during the last glacial period developed from nutrient-based proxy data, such as carbon isotopes in ancient marine sediments, which indicate that deep waters in the North Atlantic were young, and therefore locally sourced, during interstadial events. This young water was confined to shallower depths during cooler, stadial events, and was replaced by older water, presumably sourced from the Southern Ocean, during Heinrich stadials. Although nutrient proxies provide information about the age of a water mass, they can be altered by biological processes and cannot be used to identify water-mass sources or the strength of circulation. By contrast, the ratio of isotopes of neodymium (143Nd/144Nd) — reported as εNd, the deviation of this ratio from a mean chemical composition, known as the Chondritic Uniform Reservoir, in parts per 10,000 — are quasi-conservative tracers of water mass that clearly distinguish between North Atlantic Deep Water (εNd about −13.5; ref. 5) and water sourced from the Southern Ocean (εNd about −9; ref. 6). In addition, the ratio of protactinium to thorium isotopes (213Pa/230Th) monitors the strength of circulation, such that higher ratios indicate slower circulation.
Both these proxies have been applied to studies in the North Atlantic for the last glacial maximum (the peak of the last glacial period, around 20,000 years ago) and into the Holocene epoch (11,700 years ago to today), illustrating the expected patterns for warm, cold and shutdown (or 'off') circulation modes7,8. The high-quality, high-resolution data sets produced by Böhm et al., from a site on Bermuda Rise in the northeastern Sargasso Sea, extend these records back through the last glacial and interglacial periods into the penultimate glacial maximum (140,000 years ago). The data allow the authors to study ocean circulation and its relationship to climate change throughout the last glacial cycle.
Böhm and colleagues' data reveal two surprising findings. First, the cold-mode circulation — southern-sourced deep water accompanied by relatively rapid, northern-sourced flow in a shallower ocean layer that delivers less heat to high northern latitudes — existed only during glacial maxima that occurred at the end of each glacial period (see Fig. 1b of ref. 4). Second, the off mode of circulation, expected during all Heinrich stadials and defined by the increased presence of southern-sourced deep waters and sluggish circulation rates, occurred only during Heinrich stadials that, again, coincided with maximum ice extent (see Fig. 1c of ref. 4). Contrary to expectation, most of the cool stadials and Heinrich stadials coincided with intermediate values of εNd and 213Pa/230Th, suggesting that they occurred during gradual transitions between warm interstadial and off modes, rather than as a distinct cold mode of circulation and oceanic heat transport as previously thought.
The significance of this transitional configuration is the implication that the North Atlantic continued to produce deep waters despite evidence that the region of modern deep-water formation was covered by ice, and, more surprisingly, when massive iceberg discharge is expected to have freshened surface waters in the deep-water source regions. Such freshening is thought to reduce ocean-density contrasts that lead to transport of North Atlantic surface waters to depth. Böhm et al. point out that the continued strength of North Atlantic Deep Water formation and its flow throughout climate variations indicate that this mode of circulation may be more stable in response to freshwater inputs than models imply9. On the basis of this stability, the authors suggest that North Atlantic Deep Water formation may also be less susceptible to shutdown in response to freshening by glacial meltwaters during modern global warming, a process that, if carried to the extreme, could eliminate oceanic northward heat transport and produce icy scenarios similar to that in the Hollywood film The Day After Tomorrow.
One word of caution, however, is that the climate-system response to freshening in a cold climate may be very different from that in a warm climate. The view of continuous North Atlantic Deep Water formation and flow during the last glacial period presented by Böhm et al. requires a re-evaluation of the relationship between ocean circulation and climate over millennial timescales. This may ultimately help us to understand the fundamental causes of short-term climate variability.
Finally, another intriguing observation that emerges from this study is that, during past interglacials and interstadials, the εNd value of North Atlantic Deep Water was often much lower than that today (−15 to −18, compared with −13.5 today). Modern North Atlantic Deep Water is composed of a mixture of North Atlantic water masses, with the relative proportions of each responding to local climate impacts. Most of these waters have εNd values higher than −13.5, but water from the Labrador Sea region has lower values and thus explains why North Atlantic Deep Water is the major water mass with the lowest εNd value5; a low εNd indicates weathering contributions of old continental material to the source regions.
The past composition of North Atlantic Deep Water is crucial for reconstructing changes in the relative proportions of northern- and southern-sourced waters through time. Previous studies10,11 have argued for a stable North Atlantic Deep Water Nd isotopic composition across intervals of glacial climate variability. The εNd values identified by Böhm and co-workers for past interglacials and interstadials suggest a variable, rather than a stable, North Atlantic Deep Water Nd isotopic composition, prompting a revision of the relative proportions of northern- and southern-sourced waters through time and their impact on heat transport.