Milankovitch frequencies in tephra records at volcanic arcs: The relation of kyr-scale cyclic variations in volcanism to global climate changes

Abstract

The increase in volcanic activity after the last glacial maximum observed on Iceland has led to one of the most fascinating hypothesis in science in the last decades: that deglaciation may force volcanism. Consequently, tephrostratigraphic records of sufficient length that cover multiple glacial cycles have been used to test whether such relationships hold systematically through the Quaternary. Here we review such tephra records that have been linked with climate proxy records such as δ18O in marine sediments, which is a measure of sea-level change and which is thought to be orbitally forced, as it exhibits the characteristic Milankovitch periodicities of precession (∼23 kyr), obliquity (∼41 kyr) and eccentricity (∼100 kyr).Statistical analyses have identified these periodicities also in long tephra records from different latitudes and geotectonic settings, as well as in compiled semi-global records. These studies detect Milankovitch periods in their tephra record, and also a phase shift relative to the δ18O record in such that periods of increased eruption frequencies coincide with the deglaciation period at the glacial/interglacial transition when ice and water loads on the lithosphere change most rapidly. However, there are also disparities in results and interpretations, which may be attributable to the different methods of analysis applied by the studies.We have therefore re-analyzed the four best-characterized tephra records by the same methods. We distinguish between analysis in the frequency domain, a novel approach, and analysis in the time domain, which has been used in previous studies. Analysis in the frequency domain identifies harmonic frequencies that arise from the binary nature of the tephra records and complicate the identification of primary frequencies. However, we show that all four records show spectral density peaks near the main Milankovitch periodicities of 41 and 100 kyr, and that they produce meaningful and significant statistical correlations with each other and the global δ18O record but not with random time series. Although the time-domain correlations with δ18O roughly confirm phase shifts implying peak volcanism during deglaciation, correlation coefficients arising from very noisy records are generally too low for precise constraints on the relative timing.These deficiencies presently hamper the recognition of the physical mechanisms through which global climate changes affect volcanism at both, high-latitude glaciated regions and low-latitude non-glaciated regions.