Links between eccentricity forcing and the 100,000-year glacial cycle

Variations in the eccentricity (100,000 yr), obliquity (41,000 yr) and precession (23,000 yr) of Earth’s orbit have been linked to glacial–interglacial climate cycles. It is generally thought that the 100,000-yr glacial cycles of the past 800,000 yr are a result of orbital eccentricity1–4. However, the eccentricity cycle produces negligible 100-kyr power in seasonal or mean annual insolation, although it does modulate the amplitude of the precession cycle. Alternatively, it has been suggested that the recent glacial cycles are driven purely by the obliquity cycle5–7. Here I use statistical analyses of insolation and the climate of the past five million years to characterize the link between eccentricity and the 100,000-yr glacial cycles. Using cross-wavelet phase analysis, I show that the relative phase of eccentricity and glacial cycles has been stable since 1.2 Myr ago, supporting the hypothesis that 100,000-yr glacial cycles are paced8–10 by eccentricity4,11. However, I find that the time-dependent 100,000-yr power of eccentricity has been anticorrelated with that of climate since 5 Myr ago, with strong eccentricity forcing associated with weaker power in the 100,000-yr glacial cycle. I propose that the anticorrelation arises from the strong precession forcing associated with strong eccentricity forcing, which disrupts the internal climate feedbacks that drive the 100,000-yr glacial cycle. This supports the hypothesis that internally driven climate feedbacks are the source of the 100,000-yr climate variations12. The defining aspect of the 100,000-yr (100-kyr) problem is (1) the lack of significant external forcing at that frequency2. Other aspects of the 100-kyr problem are2 (2) the lack of 400-kyr power in late Pleistocene ice volume despite its presence in eccentricity; (3) an amplitude mismatch between 100-kyr cycles in eccentricity and climate13, particularly during marine isotope stage (MIS) 11; (4) uncertainty about why 100-kyr power increased at the midPleistocene transition (MPT) approximately 0.8Myr ago; and (5) why glacial cycles show a small (but consistent) phase relative to eccentricity4. Furthermore, (6) too few 100-kyr glacial cycles occur in the late Pleistocene to distinguish between many possible causal mechanisms5,14,15. Glacial cycles are commonly characterized using the δ18O of foraminiferal calcite1–6, which measures global ice volume and ocean temperature. Previous studies have observed that the 100-kyr phase of late Pleistocene δ18O is qualitatively consistent with eccentricity1,3,4, but a statistical test found that the phases of glacial terminations and eccentricity are not correlated at the 5% significance level from 0.7 to 0Myr ago5 or 1 to 0Myr ago6. As termination phases were found to correlate with obliquity forcing, these studies proposed that ∼100-kyr cycles result from quantized bundles of 41-kyr obliquity cycles5,6. Here I present a modified version of that statistical test5,6 applied to a different δ18O record, the LR04 benthic stack16. To