Factors Controlling Coral Skeletal U/Ca Ratios with Implications for their Use as a Proxy for Past Ocean Conditions

Abstract

Seawater temperature, salinity and carbonate chemistry have been shown to influence the uranium/calcium (U/Ca) ratios of scleractinian coral skeletons. This apparent sensitivity of U/Ca to multiple environmental parameters calls into question whether there is one environmental variable that most strongly controls coral U/Ca, and whether U/Ca can be straightforwardly applied as a paleoenvironmental proxy due to the tendency of environmental variables to covary in space and time. In this study, uranium concentration data from an existing compilation of tropical scleractinian coral U-series measurements is paired with environmental data from the World Ocean Atlas (WOA) and the Global Ocean Data Analysis Project (GLODAP) to examine the sensitivity of coral skeletal U/Ca to multiple seawater properties including temperature, salinity, pH, and saturation state. First, univariate linear regressions and multiple linear regressions were used to compare relationships between uranium and environmental parameters in the dataset with relationships observed in previous studies. Next, principal component analysis and regularized regression were used to identify the most likely predictors of coral U/Ca in order to create a multiple linear regression model. Results indicate that pH, Ω, alkalinity, and temperature are all significant predictors of uranium concentrations in coral. The magnitude and strength of relationships between U/Ca and environmental variables also differ across different genera. Seawater properties with strong correlations and small ranges make interpretation of these results difficult. However, results of these analyses indicate that U/Ca is dependent on multiple environmental parameters and that previously developed univariate regressions may be insufficient to characterize the full range of variables that influence coral [238U]. KEYWORDS: Coral; Paleoceanography; Proxy Calibration; Uranium; Multiple Linear Regression; Regularized Regression; Environmental Change; Oceanographic Databases