Unmixing deep-sea paleoclimate records: A study on bioturbation effects through convolution and deconvolution

Abstract

Marine sediments are typically subject to bioturbation from benthic macrofauna, which disrupt sediments and can have a substantial impact on the amplitude and timing of climate proxy signals recorded in marine sediment cores. This study explores a computational approach to quantify and remove these post-depositional effects from marine climate proxy records. The bioturbation process is modeled as a linear time-invariant filter building upon prior work (Guinasso and Schinck, 1975). Using forward modeling we find that, given modeled mixing layer depths ≥ 5 cm, simulated centennial scale climate variability is generally not preserved even when the sedimentation rate in our model is above 15 cm/kyr. On the scale of ice ages (104-105 yr), the observed effects of bioturbation have the smallest impact, considering the event scale and typical sedimentation rates and mixed layer depth in the deep sea. For millennial scale events, the signal attenuation strongly depends on the event scale and specific bioturbation and sedimentation parameters. To account for the bioturbation effect on a given climate proxy series, a deconvolution method is proposed. We apply the approach to three individual benthic foraminifera oxygen isotope records to reconstruct δ18O across the last interglacial (Marine Isotope Stage 5e; MIS 5e), and to assess the magnitude of the deglacial shift between MIS 6 and MIS 5e. In the highest resolution records (Site MD952042, Site 1012), the deconvolution results reveal an increase in the amplitude of the recovered deglacial δ18O shift by approximately 0.1‰, with implications for reconstructed global average sea level variations and temperature estimates from benthic foraminifera δ18O. The recovered signal at Site 677 suggests high amplitude δ18O variability during MIS 5e, but exhibits larger uncertainty due to a lower sampling frequency, lower sedimentation rate, and an inferred higher noise level. While uncertainties on the deconvolution results can be large, the results provide a new view of the potential impact of bioturbation on paleoclimate reconstruction. Future work can reduce the uncertainty in the deconvolution estimates, through quantitative integration of additional constraints, provided by our knowledge of the climate and depositional systems.