Evaluation of Sr/Ca-based paleoclimate reconstructions in modern and Medieval Diploria strigosa corals in the northeastern Caribbean

Tropical sea surface temperature (SST) has been implicated as a driver of climate changes during the Medieval Climate Anomaly (MCA, 950-1300 A.D.) but little data exists from the tropical oceans during this time period. We collected three modern and seven sub-fossil Diploria strigosa coral colonies from an overwash deposit on Anegada, British Virgin Islands (18.73 °N, 63.33 °W) in order to reconstruct climate in the northeastern Caribbean and Tropical North Atlantic during the MCA. The first step in our reconstruction was to verify the climate signal from this species at this site. We sub-sampled the modern corals along thecal walls with an average sampling resolution of 11-13 samples per year. Sr/Ca ratios measured in the sub-samples were calibrated to temperature using three different calibration techniques (ordinary least squares, reduced major axis, and weighted least squares (WLS)) on the monthly data that includes the seasonal cycles and on the monthly anomaly data. WLS regression accounts for unequal errors in the x and y terms, so we consider it the most robust technique. The WLS regression slope between gridded SST and coral Sr/Ca is similar to the previous two calibrations of this species. Mean Sr/Ca for each of the three modern corals is 8.993 × 0.004 mmol/mol, 9.127 × 0.003 mmol/mol, and 8.960 × 0.007 mmol/mol. These straddle the mean Diploria strigosa Sr/Ca found by Giry et al., (2010), 9.080 mmol/mol, at a site with nearly the same mean SST as Anegada (27.4 °C vs. 27.5 °C). The climatological seasonal cycles for SST derived from the modern corals are statistically indistinguishable from the seasonal cycles in the instrumental SST data. The coral-based seasonal cycles have ranges of 2.70 × 0.31 °C, 2.65 × 0.08 °C and 2.71 × 0.53 °C. These results indicate that this calibration can be applied to our sub-fossil coral data. We applied the WLS calibration to monthly-resolution Sr/Ca data from multiple sub-fossil corals dating to the medieval period with initial U-series dates near the top of the cores ranging from 1277 × 5 A.D. to 1327 × 5 A.D. Initial Sr/Ca results from the first sub-fossil coral have a seasonal range of 2.65 × 0.27 °C when converted to temperature units with our modern calibration, indicating no significant change from modern times. However, the mean Sr/Ca for this coral is very high (9.388 mmol/mol) compared to the modern corals. We explore the potential causes for this discrepancy in our study. Because reconstructing the mean SST during the Medieval Climate Anomaly may be difficult without temporal overlap with modern corals, our focus is on interannual variability. The coral Sr/Ca based monthly SST anomalies for both modern and sub-fossil corals have larger interannual variances than the instrumental record. One explanation for this is that the SSTs derived from sub-fossil corals are local data for which one expects larger variances than the instrumental data averaged over a 2 x 2 ° grid. This species shows great promise for future paleoclimate reconstructions.