Reconstructing Climate From Coral Skeletal Banding: Statistical Effects of Replicating Cores and Observers

Abstract

Corals archive past ocean‐climate variability in their annual skeletal density bands, but the reliability of these records is dependent on precise and reproducible measurements. We assessed how variability among observers and cores affects coral growth measurements and their relationship with climate indices. The width of each annual density‐band couplet records linear extension, so climate‐related changes in coral growth can appear as wider or narrower annual bands. Ten Porites cores collected from Dongsha Atoll in the South China Sea were computed tomography‐scanned and each interpreted by 12 different observers to identify annual density bands. By comparing combinations of observers and cores, we evaluated inter‐series correlation (ISC) as a measure of consistency in the time series. Our findings show that individual observers can produce twofold differences in linear extension measurements from the same core based on variations in the annual band‐boundary placement. This subjectivity can alter the inferred year‐to‐year trajectory of annual growth rates, including the identification of relative increases and decreases between successive years. Including more observers increased ISC and reduced sensitivity to individual observer interpretations. Analyzing a greater number of cores reduced the spread of correlations between linear extension rates and regional sea‐surface temperature and the Niño3.4 Index. Observer replication yielded faster improvements in chronology agreement than adding more core samples. Overall, observer and core replication improved the reproducibility of coral‐growth chronologies. This multi‐observer and multi‐core strategy increases confidence in measuring year‐to‐year growth variability and therefore bolsters our ability to discern the long‐term climatic drivers behind those patterns.