Paired foraminiferal Magnesium/Calcium ratio and oxygen isotope analyses are widely used to estimate surface ocean delta18Oseawater, a robust proxy for surface salinity. We assess the fidelity of shell-derived delta18Oseawater estimates for the surface-dwelling foraminifer Globigerinoides ruber (white) using an Atlantic meridional coretop transect spanning basin-scale temperature and salinity gradients. Shell-derived and observed delta18Oseawater values are well correlated (r2 = 0.77), but a large systematic bias is observed. Shell Mg/Ca ratios are significantly elevated above values expected from observed and isotopic calcification temperatures in the saline subtropical gyres of both hemispheres. This temperature-independent shell Mg/Ca ratio variability, termed ``excess Mg/Ca'', is highly correlated with surface salinity (r2 = 0.77), and the observed salinity dependence (27 ± 4%) is much higher than indicated by culture studies (6 ± 2%). Our coretop data are used to develop new Atlantic Basin temperature and salinity calibration equations that are accurate (± 1.1 °C and ± 0.20, respectively), precise (r2 = 0.82 and 0.81, respectively), and verifiable using previously published data. These results are valid for the relatively high salinities of the subtropical Atlantic (35.5-37.3). We discover that inclusion of other published data from lower salinity regions (< 35) indicates little or no excess Mg/Ca. Taken together, these results point to a strongly non-linear, positive salinity effect on shell Mg/Ca ratios that significantly affects the accuracy of SST and delta18Oseawater estimates in high salinity settings (> 35).