We test for trace element proxies in the high-magnesium calcite fraction of bamboo coral internodes by comparing environmental conditions and growth rates to the specimen-mean compositions of 73 corals that were live-caught at depths ranging from 3 to 3950 m and collected from habitats ranging from tropical coral reefs to the Antarctic slope.Comparisons were done at a large geographic scale (LGS) and for a well sampled area south of Australia, across depths at a single site, in order to help separate the effects of environmental variables that co-vary at one spatial scale, but not the other.Thirty-seven trace elements were measured using solution-based Sector Field ICP-MS, of which seventeen were significantly detected in more than a third of the specimens.Only eight element/calcium ratios correlated significantly with any environmental variable at the large geographic scale, and only four did so at the local level.At the LGS, the highest correlation was between ambient temperature and Mg/Ca, which accounted for 89% of the variance across specimens, spanned all four Isidid sub-families and was independently significant in the two best sampled sub-families.The predictive (geometric mean) relationship is T(°C)=-23.9(±2.46)+0.34(±0.25)Mg/Ca(mmol/mol)T(°C)=-23.9(±2.46)+0.34(±0.25)Mg/Ca(mmol/mol) spanning a temperature range of -1.9 to 26.8 °C,Mg/Ca ratios from 58.6 to 155.1 mmol/mol, and an uncertainty (RMS) of 2.78 °C.The numbers in parentheses are 95% CIs.The slope of the regression does not differ significantly from that of abiotic high-Mg calcites, which suggests that the temperature-dependent incorporation of Mg into the carbonate results from kinetic reactions at the crystal surface.Analysis at the SH scale for the sub-set of specimens for which we had data suggests the ratio is also affected by growth rates.There were no obvious trace element correlates at either spatial scale of salinity, pH or oxygen levels that could not be accounted for by covariance between these environmental parameters and, in most cases, temperature.Single and multiple correlation analyses also confirm previous suggestions that Ba/Ca in bamboo coral calcite is a proxy for seawater barium and hence refractory nutrients, suggest that Sr/Ca is influenced by specimen-mean Mg/Ca ratios and water temperature as well as possibly seawater Sr/Ca, and falsify for bamboo corals P/Ca (as well as P/Cd and Cd/Ca) as a proxy for seawater phosphate levels.The predictive relationship between Isidid skeletal-mean Ba/Ca and seawater silicate concentrations appears to be linear, and is given by silicate($μ$molkg-1)=-56.7(±20.8)+9217(±1632)Ba/Ca(mmol/mol)silicate($μ$molkg-1)=-56.7(±20.8)+9217(±1632)Ba/Ca(mmol/mol) spanning a silicate range of 0.5 to 120 $μ$mol kg-1, a Ba/Ca range of 0.0042 to 0.0195 mmol/mol, and with an uncertainty (RMS) of 33.1 $μ$mol kg-1.Mn/Ca differences among specimens and sites are highly significant and appear to reflect seawater Mn, suggesting a proxy for this micronutrient.The compilation of growth rate data across 34 specimens indicates a wide range of growth rates even among con-familial specimens from within a single habitat, and suggests both ambient temperature and food availability underlie at least part of this variability.
