Diagenesis Scorecard for Reducing Uncertainty in Holocene Coral Reconstructions

Coral oxygen isotopes (δ¹⁸O) are a powerful tool for studying past climate variability, however these records can be significantly affected by diagenetic alteration and intercolony variability (associated with unresolved biomineralization processes coral calcification) that obscure the temperature and δ¹⁸O seawater signals recorded by coral δ¹⁸O. Here we present a new framework for quantifying the impacts of diagenesis on coral δ¹⁸O records and improving the signal-to-noise ratio in coral-based mean climate reconstructions. We develop a numerical diagenesis rating system based on the surface area of secondary aragonite, calcite cements, and dissolution in scanning electron microscope (SEM) images. In addition, we characterize the distribution of point-to-point variations in mm-scale coral δ¹⁸O measurements, which provide a useful way to identify unusual characteristics in the records, such as those associated with diagenesis and coral spawning events. We construct a partial least square regression model using these metrics as predictor variables to explain a substantial fraction of the variability in mean coral δ¹⁸O. This method was tested on a large compilation of published coral δ¹⁸O records spanning the last 6,000 years from Kiritimati Atoll. A total of 42 coral δ¹⁸O records were compiled and randomly divided into a training set and a validation set. Using our predictive model, the effects of diagenesis and point-to-point variability on mean coral δ¹⁸O were substantially reduced, resulting in a >20% decrease in the variability of mean coral δ¹⁸O across the Holocene dataset. This method has the potential to serve as a community-wide standard for reducing uncertainty in ensembles of coral δ¹⁸O climate reconstructions, as it utilizes a simple model and metrics that are routinely measured in the coral paleoclimate community. We are currently investigating the suitability of this method to reduce the uncertainty in other types of coral δ¹⁸O reconstructions, including reconstructions of the annual cycle and ENSO variability.