Seawater nutrient and carbonate ion concentrations recorded as P/Ca, Ba/Ca, and U/Ca in the deep-sea coral Desmophyllum dianthus

As paleoceanographic archives, deep sea coral skeletons offer the potential for high temporal resolution and precise absolute dating, but have not been fully investigated for geochemical reconstructions of past ocean conditions. Here we assess the utility of skeletal P/Ca, Ba/Ca and U/Ca in the deep sea coral D. dianthus as proxies of dissolved phosphate (remineralized at shallow depths), dissolved barium (trace element with silicate-type distribution) and carbonate ion concentrations, respectively. Measurements of these proxies in globally distributed D. dianthus specimens show clear dependence on corresponding seawater properties. Linear regression fits of mean coral Element/Ca ratios against seawater properties yield the equations: P/Ca_coral (μmol/mol) = (0.6 ± 0.1) P/Ca_sw(μmol/mol) - (23 ± 18), R² = 0.6, n = 16 and Ba/Ca_coral(μmol/mol) = (1.4 ± 0.3) Ba/Ca_sw(μmol/mol) + (0 ± 2), R² = 0.6, n = 17; no significant relationship is observed between the residuals of each regression and seawater temperature, salinity, pressure, pH or carbonate ion concentrations, suggesting that these variables were not significant secondary dependencies of these proxies. Four D. dianthus specimens growing at locations with Ω_arag ? 0.6 displayed markedly depleted P/Ca compared to the regression based on the remaining samples, a behavior attributed to an undersaturation effect. These corals were excluded from the calibration. Coral U/Ca correlates with seawater carbonate ion: U/Ca_coral(μmol/mol) = (−0.016 ± 0.003) (μmol/kg) + (3.2 ± 0.3), R² = 0.6, n = 17. The residuals of the U/Ca calibration are not significantly related to temperature, salinity, or pressure. Scatter about the linear calibration lines is attributed to imperfect spatial-temporal matches between the selected globally distributed specimens and available water column chemical data, and potentially to unresolved additional effects. The uncertainties of these initial proxy calibration regressions predict that dissolved phosphate could be reconstructed to ±0.4 μmol/kg (for 1.3-1.9 μmol/kg phosphate), and dissolved Ba to ±19 nmol/kg (for 41-82 nmol/kg Ba_sw). Carbonate ion concentration derived from U/Ca has an uncertainty of ±31μmol/kg (for ). The effect of microskeletal variability on P/Ca, Ba/Ca, and U/Ca was also assessed, with emphasis on centers of calcification, Fe-Mn phases, and external contaminants. Overall, the results show strong potential for reconstructing aspects of water mass mixing and biogeochemical processes in intermediate and deep waters using fossil deep-sea corals.