Reproducibility of Geochemical and Climate Signatures in Montastrea annularis

Geochemical variations in modern and fossil coral skeletons (Sr/Ca, Mg/Ca, δ ¹⁸O, δ ¹³C) are increasingly being used to reconstruct climate variability in tropical ocean-atmosphere system on interannual to centennial timescales. Coral-based climate studies are usually carried out using a single coral core from a reef, or a collection of cores covering different time intervals from the same reef site. Cross checking or replication - the generation of multiple climate time series from a single locality - is not a standard operating procedure in coral paleoclimatology because of the expense of generating additional geochemical time series. Quantitative assessment of intra-reef geochemical variability in multiple coral cores from the same reef is needed to verify the fidelity of climate reconstructions based on geochemical variations from a single coral. We perform such an assessment using the geochemical signals derived from two Montastrea annularis corals (LK1 and LK23) recovered in 2002 from Looe Key reef, Florida USA (24.5° N, 81.4° W). Looe Key is located in the central portion of the Florida Keys and has an hourly in situ seawater temperature record extending back to 1990. We extended this in situ SST record further back in time by splicing in HadISST 1.1 data extracted from the appropriate 1° by 1° grid point. Paired geochemical measurements of Sr/Ca, δ ¹⁸O and δ ¹³C were made by sampling coral LK1 and LK23 along their respective major growth axes. Geochemical variations versus depth were converted to monthly resolved time series extending from 2002-1966. The two ∼37 year coral time series replicate well in terms of both phasing and mean perspective: (δ ¹⁸O LK1, -3.90+/-0.28 ‰ , 1σ ; LK23, -3.93+/-0.31 ‰ , 1σ ), δ ¹³C (LK1, -0.80+/-0.68 ‰ , 1σ ; LK23, -0.72+/-0.59 ‰ , 1σ ) and Sr/Ca (LK1, 9.208+/-0.080 mmol/mol, 1σ ; LK23, 9.226+/-0.082 mmol/mol, 1σ ). Coral Sr/Ca-SST estimates of mean SST over the period 2002-1966 for LK1 (27.25+/-2.00° C) and LK23 (26.71+/-1.80° C) agree to within 0.3° C of observed mean SST in the instrumental record (26.94+/-2.12° C). Coral δ ¹⁸O-SST estimates of mean SST (cf, LK1, 26.83+/-1.33; LK23, 27.05+/-1.79) also match well the mean SST of the instrumental record. Stacking or averaging the LK1 and LK23 records to create composite Sr/Ca-SST and δ ¹⁸O-SST records improves the estimate of the proxy-based mean SST relative to the instrumental record. Month-to-month comparisons between predicted and observed SST indicate that the maximum misfit for coral δ ¹⁸O-SST of ∼1° C occurs in March-April and Aug-Oct; whereas for coral Sr/Ca-SST the maximum misfit of ∼0.5° C peaks in September. In summary, Sr/Ca and δ ¹⁸O records from two M. annularis coral heads from Looe Key, Florida replicate quite well, which provides confidence that accurate records of climate variability in the tropical Atlantic can be reconstructed from this scleractinian coral.