Establishing what caused Earth’s climate to undergo large and sustained changes in the past requires that we know both the forcing mechanism(s) and the regional responses to that forcing. This represents a substantial challenge because our ability to precisely resolve the temporal relationships derived from archival climate records is typically not good enough to fingerprint the phasing relationships. With respect to the oceans, the response time may be substantially delayed because of the slow rate at which ocean currents communicate a climate signal over the vast reaches of the ocean basins.

Among the largest climatic changes that Earth has experienced in the past are the transitions from glacials to warm interglacials, such as the one within which human society has flourished. These dramatic climate changes have recurred at approximately 100 000 year intervals over the past 1.5 million years of Earth’s history (Fig. 1). But interglacial warmth is fleeting. The Earth has spent much more time locked in a cold climatic state over the past 1.5 million years than in a warm state. It appears that even during inter-glacials the climate can oscillate between warm and cold, and yet the warm intervals are short-lived. We know from observational data derived from ice cores and marine proxies that the transitions between cold climates and warm climates occur much more rapidly than the slow slide into a cold or glacial state. Indeed, it appears that longer-term cooling and relatively abrupt, short-term warming has been a recurrent climate pattern on planet Earth at all time scales (Fig. 2). Why should this be? Solar forcing does not vary in the same temporal fashion as the glacial/interglacial transitions. Instead, the abrupt nature of the glacial to interglacial transition points to a dynamic coupling between parts of the Earth’s ocean/atmosphere system that store and transport energy. And despite over half a century of extensive investigation, the climate science community still lacks a satisfactory explanation for dramatic transitions between cold and warm climates.

We have learned from polar ice cores that the concentration of atmospheric greenhouse gases, including CO₂ and CH₄, increased and decreased in close temporal association with cold and warm climate transitions (Blunier et al. 1998; Chappellaz et al. 1997; Indermuhle et al. 2000; Jouzel et al. 1987; Kawamura et al. 2007; Petit et al. 1999). In fact, it appears from ice core records that each warm interval was associated with higher CO₂ concentrations. This has led some to suggest that this close temporal correspondence represents a causal relationship between greenhouse gas forcing and the transition from cold to warm climatic conditions. Yet, while a rise in greenhouse gases increases the radiative forcing, we do not adequately understand what caused CO₂ levels to rise and fall; therefore, atmospheric CO₂ variability may be as much a response as it has been a forcing.

We have also learned that some of Earth’s abrupt climate switches were associated with changes in ocean circulation that disrupted the production and transport of warm, salty, deep currents within the Atlantic basin (Boyle 1992; Broecker and Denton 1989; Lynch-Stieglitz et al. 2006; Marchitto et al. 2002; McManus et al. 1999; Piotrowski et al. 2005). Less well documented, however, are ocean circulation changes that occurred in other parts of the globe, particularly the Pacific Ocean. This has important implications for understanding the inter-related influences of ocean circulation, atmospheric greenhouse gas, and corresponding changes in climate. The Pacific Ocean covers nearly 1/3 of the Earth’s surface and averages 4200 m in depth; it contains more than twice the volume of water than the Atlantic. Consequently, from an energy balance standpoint, the Pacific Ocean dominates the Earth’s radiative heat capacity. Therefore, the timing, magnitude and volumetric distribution of temperature change in the Pacific Ocean play an important role in the overall energy balance for the planet. Furthermore, we know from modern observations that the ocean – atmospheric dynamic centered in the tropical Pacific can impart a strong influence on extratropical climates at inter-annual time scales. For example, evidence from oceanic and terrestrial archives suggests that the El Niño Southern Oscillation has varied in the past, both in frequency and in magnitude (Graham et al. 2007); hence, the Pacific has the capacity to influence Earth’s climate at a variety of time scales through its impact on Earth’s energy balance and the transmittal of latent heat and sensible heat.

Paleoclimatologists have traditionally used the distinctive glacial – interglacial δ¹⁸O changes recorded in the fossil shells of benthic foraminifera (higher values in glacial and lower values in interglacials) to temporally align climate records, assuming that large δ¹⁸O changes represent synchronous change in the ¹⁸O/¹⁶O of sea water in response to both temperature and ice volume variability (Lisiecki and Raymo 2005; Fig. 2). However, this assumption can now be tested using ¹⁴C radioisotopic dating of biogenic carbonates in high-resolution deep sea cores (i.e. cores displaying a higher sediment accumulation rate). Both the temperature and the ¹⁸O/¹⁶O of seawater, which determine the δ¹⁸O value of benthic foraminiferal calcite, are conservative properties of seawater that are acquired at the surface of the ocean at high latitudes where deep-water masses form (Fig. 3). The δ¹⁸O of benthic foraminifers is therefore a record of high-latitude surface ocean change, and that change need not have occurred at the same time at high northern and at high southern latitudes (Blunier and Brook 2001). Taking into account transit time of ocean water between high latitude ‘source’ regions and distal deep-sea sites, a deep-sea δ¹⁸O signal is literally hundreds of years old by the time it is recorded in the shells of benthic foraminifera (Fig. 4). In the case of the largest climatic changes during glacial terminations, evidence shows that the deep Pacific warmed well in advance of the initial deglacial warming in the North Atlantic region (Stott et al. 2007; Fig. 5). A reassessment of deep-sea temperature records from the Pacific, for depths between 1000 and 3500 m, indicates that virtually all of the deglacial deep-sea warming in the Pacific occurred prior to the beginning of substantial ice sheet retreat and sea-level rise. Deep-sea temperatures warmed by ca. 2°C between 19 and 17 000 years before the present (BP), leading the rise in atmospheric CO₂ and tropical-surface-ocean warming by ca. 1000 years (Stott et al. 2007).

The cause of this deglacial deep-water warming does not lie within the tropics, nor can its early onset between 19 and 17 ky BP (Fig. 5) be attributed to CO₂ forcing. Increasing austral-spring insolation combined with sea-ice albedo feedback appear to be the key factors responsible for the warming. One potentially important implication of this result is that from an energy balance standpoint, much of the climate work was accomplished, not through summer-time insolation over high northern latitudes, but rather through ocean – atmosphere dynamics centered at high southern latitudes (Timmer-mann et al. 2009).

In summary, it appears that much of the deep-sea warming within the Pacific preceded the rise in atmospheric CO₂ during the last deglaciation. This is an important observation because deep-sea warming may have been a trigger that contributed to the release of CO₂ stored in deep-sea reservoirs during glacial periods, although this remains an open question.