High resolution multiproxy analysis (microcharcoal, pollen, organic carbon, Neogloboquadrina pachyderma (s), ice rafted debris) of the deep-sea record MD04-2845 (Bay of Biscay) provides new insights for understanding mechanisms of fire regime variability of the last glacial period in western France. Fire regime of western France closely follows Dansgaard–Oeschger climatic variability and presents the same pattern than that of southwestern Iberia, namely low fire regime associated with open vegetation during stadials including Heinrich events, and high fire regime associated with open forest during interstadials. This supports a regional climatic control on fire regime for western Europe through fuel availability for the last glacial period. Additionally, each of Heinrich events 6, 5 and 4 is characterised by three episodes of fire regime, with a high regime bracketed by lower fire regime episodes, related to vegetational succession and complex environmental condition changes.

Hypotheses and inferences concerning the nature of abrupt climate change, exemplified by the Dansgaard–Oeschger (D–O) events, are reviewed. There is little concrete evidence that these events are more than a regional Greenland phenomenon. The partial coherence of ice core δ18O and CH4 is a possible exception. Claims, however, of D–O presence in most remote locations cannot be distinguished from the hypothesis that many regions are just exhibiting temporal variability in climate proxies with approximately similar frequency content. Further suggestions that D–O events in Greenland are generated by shifts in the North Atlantic ocean circulation seem highly implausible, given the weak contribution of the high latitude ocean to the meridional flux of heat. A more likely scenario is that changes in the ocean circulation are a consequence of wind shifts. The disappearance of D–O events in the Holocene coincides with the disappearance also of the Laurentide and Fennoscandian ice sheets. It is thus suggested that D–O events are a consequence of interactions of the windfield with the continental ice sheets and that better understanding of the wind field in the glacial periods is the highest priority. Wind fields are capable of great volatility and very rapid global-scale teleconnections, and they are efficient generators of oceanic circulation changes and (more speculatively) of multiple states relative to great ice sheets. Connection of D–O events to the possibility of modern abrupt climate change rests on a very weak chain of assumptions.

Data from multiple ice and sediment cores in the North Atlantic show that Marine Oxygen Isotope Stage 3 (MIS 3) was characterized by recurring millennial-scale variations in climate, but the periodic behavior of the well-known millennial-scale variations, Heinrich events and Dansgaard–Oeschger events, is uncertain. We use oxygen isotope values from the Greenland Ice Sheet Project 2 (GISP2) and North Greenland Ice Core Project (NGRIP) ice cores and estimated sea-surface temperature derived from a Bermuda Rise marine sediment core as climate proxies to assess the periodic behavior of Heinrich events and Dansgaard–Oeschger events using Lomb–Scargle spectral decomposition and continuous time autoregressive models. We find that continuous time autoregressive models produce less variable estimates of periodicity for Heinrich events than Lomb–Scargle methods. Heinrich events during MIS 3 are periodic with an estimated periodicity of 6.29–6.49 ka in the GISP 2 ice core, 6.71–6.76 ka in the marine sediment core, and 7.89–8.23 ka in the NGRIP core. There is insufficient evidence from these data to conclude that Dansgaard–Oeschger events exhibit a single periodicity during MIS 3. We also find that the periodic behavior of millennial-scale variations depends on the observational time frame.