<p>During the Pleistocene (approximately 2.6 Ma to present) glacial to interglacial climate variability evolved from dominantly 40 kyr cyclicity (Early Pleistocene) to 100 kyr cyclicity (Late Pleistocene to present). Three aspects of this period remain poorly understood: Why did the dominant frequency of climate oscillation change, given that no major changes in orbital forcing occurred? Why are the longer glacial cycles of the Late Pleistocene characterised by a more asymmetric form with abrupt terminations? And how can the Late Pleistocene climate be controlled by 100 kyr cyclicity when astronomical forcings of this frequency are so much weaker than those operating on shorter periods? Here we show that the decreasing frequency and increasing asymmetry that characterise Late Pleistocene ice age cycles both emerge naturally in dynamical systems in response to increasing system complexity, with collapse events (terminations) occuring only once a critical state has been reached. Using insights from network theory we propose that evolution to a state of criticality involves progressive coupling between climate system 'nodes', which ultimately allows any component of the climate system to trigger a globally synchronous termination. We propose that the climate state is synchronised at the 100 kyr frequency, rather than at shorter periods, because eccentricity-driven insolation variability controls mean temperature change globally, whereas shorter-period astronomical forcings only affect the spatial pattern of thermal forcing and thus do not favour global synchronisation. This dynamical systems framework extends and complements existing theories by accomodating the differing mechanistic interpretations of previous studies without conflict.</p>
Persistent millennial‐scale climate variability in the eastern tropical North Pacific over the last two glacial cycles
