Reconstructing twentieth-century sea surface temperature variability in the southwest Pacific: A replication study using multiple coral Sr/Ca records from New Caledonia

Coral-based climate reconstructions typically have not used multiple cores from a region to capture and replicate a climate signal largely because of concerns of coral conservation, analytical expense, and time constraints. Coral Sr/Ca reproducibility through the twentieth century was investigated using three intracolony and three intercolony coral records from the reefs offshore of Amedee Island, New Caledonia. Different sampling resolutions were examined in coral Sr/Ca (fortnightly and monthly) and delta O-18 (fortnightly, monthly, and seasonally) as well as similar scale subsampling of the daily in situ sea surface temperature (SST) record. The mean coral Sr/Ca, delta O-18, and SST values do not change as a function of sampling resolution. The coral Sr/Ca signal is highly reproducible; the average absolute offset between coeval monthly Sr/Ca determinations between any two coral time series is 0.035 +/- 0.026 mmol/ mol (1 sigma) (similar to 0.65 degrees C), which is less than twice the analytical precision of the coral Sr/Ca measurements. The stack average of the monthly coral Sr/Ca variations and monthly anomalies are significantly correlated with monthly in situ SST (1967-1992; r = -0.96 and -0.64, respectively; p < 0.05; and n = 302) and 1 degrees grid monthly SST data product (1900 - 1999; r = -0.95 and -0.56, respectively; p < 0.05; and n = 1198). The coral Sr/Ca - SST reconstruction exhibits interannual and decadal- timescale fluctuations that exceed those observed in the gridded SST record, which may reflect true differences between SST at a shallow reef site and those averaged over a 1 degrees grid box or inadequacies in the methodology used to create the gridded SST product when few observations are available. A warming trend of similar to 0.6 degrees C is observed in the twentieth century coral Sr/Ca-SST record.