Geochemistry of slow-growing corals : reconstructing sea surface temperature, salinity and the North Atlantic Oscillation

Abstract

A 225-year old coral from the south shore of Bermuda (64°W, 32 0 N) provides a record of decadal-to-centennial scale climate variability. The coral was collected live, and sub-annual density bands seen in x-radiographs delineate cold and warm seasons allowing for precise dating. Coral skeletons incorporate strontium (Sr) and calcium (Ca) in relative proportions inversely to the sea surface temperature (SST) in which the skeleton is secreted. 6180 of the coral skeleton changes based on both temperature and the 6180 of sea water ( 6 Ow), and 6Ow is proportional to sea surface salinity (SSS).Understanding long-term climate variability requires the reconstruction of key climate parameters, such as sea surface temperature (SST) and salinity, in records extending beyond the relatively short instrumental period. The high accretion rates, longevity, and skeletal growth bands found in coral skeletons make them an ideal resource for well-dated, seasonal climate reconstructions. Growing between 2 and 6 mm/year and reaching more than im in length, slowgrowing corals provide multi-century records from one colony. Additionally, unlike the fast growing (10-20 mm/year) species Porites, slow-growing species are generally found in both tropical and sub-tropical locations greatly expanding the geographical location of these records.A high resolution record (HRR, -11 samples per year) was drilled for the entire length of the coral record (218 years). Samples were split and Sr/Ca, 6180, and 61 3 C were measured for each sample. Sr/Ca was used to reconstruct winter time and mean-annual SST. Oxygen isotopic measurements were used to determine directional salinity changes, in conjunction with Sr/Ca based SST reconstructions. Winter-time and mean annual SSTs show SSTs -1.5 'C colder during the end of the Little Ice Age (LIA) relative to today. Simultaneously, SSS is fresher during that time.Sr/Ca based climate reconstructions from coral skeletons have been met with some skepticism because some reconstructions show temperature changes back in time that are 2-4 times greater than the reconstructions of other marine proxies. In this study, we show that when using bulk-sampled, slow-growing corals, two steps are critical to producing accurate reconstructions: 1) incorporating growth rate into multi-variant regressions with SST and Sr/Ca and 2) using multiple colonies that grew at the same time with varying average growth rates and Sr/Ca. Application of these novel methods over the period of the instrumental record from Hydrostation S (monthly since 1954, 32°10'N, 64°30'W) reduces the root mean square of the residuals between the reconstructed SST and the instrumental SST by as much as 1.52'C to 0.46 0 C for three coral colonies. Winter-time SSTs at Bermuda are correlated to phases of the North Atlantic Oscillation (NAO), a meridional oscillation in atmospheric mass. Much uncertainty remains about the relationship between the NAO and the ocean, and one critical outstanding question is whether anthropogenic changes are perturbing the ...

Sea surface temperature and salinity variability at Bermuda during the end of the Little Ice Age

“…We reconstruct mean annual SST from the 218‐year Sr/Ca record by inverting the regression () to find SST as a function of Sr/Ca and growth rate (Figures 2a–2c). Standard methods of error propagation [ Bevington , 1969], including the slope and intercept errors shown with and slope and intercept covariance values (Table 1), are used to estimate the error in the reconstructed SST, as detailed by Goodkin [2007]. Error for the Sr/Ca data was based on the standard error of the regression (0.031 mmol/mol), which is somewhat larger than the measurement error (0.0244 mmol/mol), providing a more statistically conservative result.…”

“…In addition, a strong LIA cooling is seen in both Sr/Ca–generated SST records, whereas the mean annual and wintertime δ O c show no (or minimal) indication of this cooling. The relationship between δ O w and SSS within the thermocline and within this region has been found to be 0.49‰/salinity unit [ Goodkin , 2007; G. A. Schmidt, Global Seawater Oxygen‐18 Database, 1999, available at http://data.giss.nasa.gov/o18data/]. Incorporating this relationship into and solving for a change in SSS for the 150 years from 1840 to 1990 estimates a salinity increase on the order of 0.6 psu from the end of the LIA to the end of the twentieth century.…”

“…Repeat measurements on the external coral standard (run at variable concentrations matching the range of unknowns) analyzed over more than a year showed good reproducibility (standard deviation (SD) = 0.0244 mmol/mol, residual SD (or coefficient of variation) = 0.27%, n = 1,493). A small long‐term drift (over several months) in the homogenous coral standard was observed on the order of 0.011 mmol/mol, and corrections were made to the unknowns as described by Goodkin [2007].…”

“…These corrections were up to 0.2‰ for a sample/standard voltage of 2.5, the maximum correction. Details of these corrections are described by Goodkin [2007].…”

“…Age models were developed using density banding visible in the X radiographs to identify the calendar year. X radiographs for these corals have been previously published [ Goodkin , 2007; Goodkin et al , 2007, 2005]. For the instrumental period, subannual age models were then generated by correlating Sr/Ca to monthly averaged SSTs at maxima, minima and inflection points.…”