Statistical and analytical innovations for high-resolution paleoclimate reconstructions using corals and bivalves
Hughes, H.P. (2025). Statistical and analytical innovations for high-resolution paleoclimate reconstructions using corals and bivalves. PhD Thesis. University of North Carolina at Chapel Hill: Chapel Hill. 200 pp. https://dx.doi.org/10.17615/2yjf-dv81
|  |
| Available in | Author |
|
Document type: Dissertation
|
| Keywords |
Chemistry > Geochemistry
Climate change
Corals
|
| Author keywords |
Paleoclimate science; high-resolution; multivariate statistics; bivalves; analytical geochemistry; Climate change; paleoclimate
|
| Abstract |
High-resolution marine archives (e.g., corals and bivalves) offer critical insight into past climate variability. However, many potential archives are either inaccessible due to slow growth rates, or limited in interpretability due to non-climate-related variability in their skeletal geochemistry. This dissertation seeks to advance both analytical and statistical techniques to improve the fidelity, resolution, and reproducibility of paleoclimate reconstructions from two key high-resolution marine archives: the high latitude bivalve Astarte borealis and tropical Porites corals. Chapter 2 explores the use of secondary ion mass spectrometry (SIMS) to generate ultra-high-resolution δ18O records from A. borealis collected in the southern Baltic Sea. These records are used to assess the sensitivity of shell δ18O values to local and regional environmental variability, demonstrating strong correlations with salinity, temperature, and large-scale climate modes such as the North Atlantic Oscillation. The results establish A. borealis as a viable high-latitude proxy archive capable of recording hydroclimatic variability over seasonal to decadal scales. Chapter 3 introduces a novel analytical framework for converting two-dimensional LA-ICP-MS maps from Porites astreoides into one-dimensional time series, enabling high-resolution reconstructions of sea surface temperature (SST) and seawater pH. This chapter also presents the Scleractinian Multivariate Isotope and Trace Element (SMITE) method, a multivariate calibration method that leverages geochemical covariance to improve reconstruction skill. SMITE is shown to outperform traditional univariate methods in terms of accuracy, robustness to noise, and reproducibility across synthetic experiments and short coral datasets. Chapter 4 evaluates the performance and reproducibility of SMITE, Sr/Ca, and Li/Mg temperature estimates using 15 coral records from the Great Barrier Reef and 2 records from the Galápagos Archipelago. Synthetic climate field reconstructions reveal tradeoffs between statistical performance and signal recoverability, with SMITE providing more reproducible calibrations at the expense of variance suppression. Together, these chapters demonstrate a flexible and scalable approach for enhancing both analytical access and statistical rigor in high-resolution marine paleoclimatology. By extending proxy coverage into high-latitude environments and establishing a multivariate statistical framework for tropical coral reconstructions, this work lays the foundation for next-generation paleoclimate proxy networks and climate data assimilation efforts. |
|
