Abstract
Recurrent shifts in Holocene climate define the range of natural variability to which the signatures of human interference with the Earth system should be compared. Characterization of Holocene climate variability at the global scale becomes increasingly accessible due to a growing amount of paleoclimate records for the last 9 000{11 000 years. Here, we integrate 124 proxy time series of different types (e.g.,δ18O, lithic composition) and apply a modified Lomb-Scargle spectral analysis. After bootstrapping the data in moving time windows we observe an increased probability for generation or loss of periodic modes at the mid-Holocene. Spatial autocorrelation of spectral changes robustly reveals that this (in)activation of modes was organized in regional clusters of subcontinental size. Within these clusters, changes in spectral properties are unexpectedly homogeneous, despite different underlying climatological variables. Oscillations in the climate system were amplified especially at the upwelling areas and dampened in the North Atlantic. We cross-checked the spectral analysis by counting events in the time series and tested against possible dating errors in individual records or against an overestimation of singular events. A combination of different mechanisms may have affected the coupling intensity between climate subsystems, turning these more or less prone to oscillations.
DOI Link to Publication