Correlated trace element “vital effects” in tropical corals: A new geochemical tool for probing biomineralization

Abstract

Micron-scale variations in the trace-element (TE) composition of tropical coral skeletons were measured using laser-ablation ICP-MS (LA-ICP-MS) as part of an investigation into the chemical processes underlying paleoenvrionmental proxy reconstructions. Fluctuations in B, Mg, Sr, Ba and U were measured at high spatial resolution in two Porites corals from the Great Barrier Reef (Australia), and the fine-scale fluctuations (< ∼1.0 mm) were compared with seasonal TE cycles in a third coral. Fine-scale TE variations were found to have a large amplitude over distances corresponding to less than 1 month growth. Variations were quasi-periodic and appeared to have characteristic wavelengths on weekly (6–7 d) and monthly (28 d) scales, although periodicity was not continuous and variations could not be matched either within or between individual corallites. Fine-scale variations between Mg, Sr and U were significantly correlated with each other (Sr and U are positively correlated, but negatively correlated with Mg). This 3D correlation “vector” has the same slope as the seasonal-scale Mg, Sr and U correlations, suggesting that the same chemical/biologic biomineralization process mediates trace element variations at both timescales. Importantly, the fine-scale variations are too large to be caused directly by daily to monthly fluctuations in sea-surface temperature. This means that seasonal variations in these elements cannot reflect purely inorganic temperature-dependent coprecipitation. Models of physicochemical calcification were developed to test whether changes in calcification rate could explain the trace-element correlations. The calculations show that increases in calcification rate will result in correlated decreases in all TE/Ca ratios. The models reproduce the Sr partition coefficient, trace-element correlation slopes, and amplitude of fine-scale variations for an average calcifying pH of 8.5, varying by ±0.2 pH units. The models, however, predict U partition coefficients which are too low, and cannot reproduce the negative correlation between Mg and the other trace elements, which may be caused by crystallographic factors.