A refined method for calculating paleotemperatures from linear correlations in bamboo coral carbon and oxygen isotopes

Bamboo corals represent an emerging paleoclimate archive with the potential to record variability at intermediate depths throughout much of the global ocean. Realizing this potential has been complicated by biologically mediated vital effects, which are evident in linear correlations of skeletal carbon (C-13) and oxygen (O-18) isotope composition. Previous efforts to develop a bamboo coral O-18 paleothermometer by accounting for such vital effects have not been completely successful as they still rely on empirical calibrations that are offset from the temperature dependence of abiogenic experiments. Here we describe an approach that better corrects for bamboo coral vital effects and allows paleotemperatures to be calculated directly from the abiogenic temperature dependence. The success of the method lies in calculating apparent equilibrium carbon and oxygen isotope fractionation at the temperature, pH, and growth rate of each coral, as well as in the use of model II regressions. Rigorous propagation of uncertainty suggests typical errors of 2-3 degrees C, but in select cases errors as low as 0.65 degrees C can be achieved for densely sampled and strongly correlated data sets. This lower limit approaches the value attributed to uncertainty in pH and growth rate estimates alone, as predicted by a series of pseudoproxy experiments. The incorporation of isotopically light metabolic CO2 appears to be negligible in most Pacific corals, but may be significant in Atlantic specimens, potentially requiring an additional correction. The success of the method therefore hinges on how well complex environmental systems and biomineralization strategies are constrained, with the most reliable temperatures occurring when calcifying fluid pH, growth rate, and incorporation of metabolic carbon into skeletal calcite are constrained using multiple geochemical proxies.