Improving coral-base paleoclimate reconstructions by replicating 350 years of coral Sr/Ca variations

Coral-based climate reconstructions are typically based on a single record and this lack of replication leads to questions in regards to chronology accuracy and reliability of the inclusive geochemical variations to record climate viability. Here we present two multi-century coral Sr/Ca records recovered from a Porites lutea colony offshore of Amédée Island, New Caledonia (22°28.8′S, 166°27.9′E). The chronology was developed by cross dating the coral Sr/Ca time series and verifying the chronology with high precision absolute 230Th dating. We identify chronological discrepancies of − 2.3 and + 3.7 years century− 1 for reconstructions based on a single core with uncertainty increasing after ~ 250 years. We assess the impact of Porites skeletal architecture on coral geochemistry by characterizing centimeter-scale architectural structures with respect to sampling. Optimal sampling paths are those on the slab surface parallel to the growth direction of individual corallites along the central axis of an actively extending corallite fan. Coral Sr/Ca time series extracted from optimal skeletal structures are highly reproducible with a mean absolute difference of 0.021 mmol mol− 1 or 0.39 °C for monthly determinations. Suboptimal skeletal architecture is characterized by corallites extending through the slab surface, and the coral Sr/Ca determinations derived from suboptimal paths tend to produce a warm bias that varies between + 0.04 and + 2.30 °C. Disorganized skeletal architecture is characterized by small, terminating, or unclear corallite fans that produce a cold bias of − 0.11 to − 2.45 °C in coral Sr/Ca determinations. These problematic architecture types also produce biases in coral δ18O determinations, but to a lesser extent. We assess the impact of sampling a coral colony along paths that vary from vertical to horizontal in large and small Porites colonies with extension rates > 6 mm year− 1 and we determine there is no significant difference in the coral Sr/Ca records.