Mg/Ca ratios of planktonic foraminiferal tests are important tools for reconstructing past ocean temperatures at different levels of the upper water column. Yet numerous studies suggest a significant influence of calcite dissolution on Mg/Ca ratios lowering their initial signal recorded within a planktonic foraminiferal habitat. To determine the effect of dissolution, this study presents Mg/Ca ratios of eight planktonic foraminiferal species from the South China Sea sediment surface. Continuously decreasing with increasing water depth, the Mg/Ca ratios also decrease with calcite-saturation states close to and below saturation (bottom water Δ[CO 2− 3 ] <30 μmol kg −1 ) but are stable in well calcite-supersaturated bottom waters (>40 μmol kg −1 ). This preservation pattern compares well with examples of Mg/Ca dissolution from the tropical Atlantic Ocean and is independent of the foraminiferal species. Merging a global data set by separate normalization of 79 Mg/Ca data sets from the Pacific, Atlantic, and Indian Oceans, which removes thermal differences between the ocean regions and foraminiferal species, enabled us to quantify a global decrease in planktonic foraminiferal Mg/Ca ratios of 0.054 ±0.019 μmol mol −1 per μmol kg −1 below a critical threshold for dissolution of 21.3 ±6.6 μmol kg −1 . The absolute decline in Mg/Ca ratios, which is similar for all species, affects temperature estimates from (sub-)thermocline species more strongly than those from shallow dwellers. The water depth of this critical threshold in the global oceans shoals from >3.5 km in the North Atlantic to <0.5 km in the North Pacific based on calculations of the global calcite-saturation state from 6321 hydrographic stations. Above this critical threshold Mg/Ca ratios are well preserved, and paleotemperature estimates are broadly unaffected by dissolution.

<p><strong>Abstract.</strong> In ocean biogeochemical models pelagic CaCO₃ dissolution is usually calculated as <i>R</i> = <i>k</i> * S^<i>n</i>, where <i>k</i> is the dissolution rate constant transforming <i>S</i>, the degree of (under-) saturation of seawater with respect to CaCO₃, into a time dependent rate <i>R</i>, and n is the reaction rate order. Generally, there are two ways to define the saturation state of seawater with respect to CaCO₃: (1) Δ[CO₃^2&amp;minus;], which reflects the difference between the in-situ carbonate ion concentration and the saturation concentration, and (2) Ω, which is approximated by the ratio of in-situ carbonate ion concentration over the saturation concentration. Although describing the same phenomenon, the deviation from equilibrium, both expressions are not equally applicable for the calculation of CaCO₃ dissolution in the ocean across pressure gradients, as they differ in their sensitivity to ocean acidification (change of [CO₃^2&amp;minus;]) over depth. In the present study we use a marine biogeochemical model to test the sensitivity of pelagic CaCO₃ dissolution to ocean acidification (1–4 &amp;times; CO₂ + stabilization), exploring the possible parameter space for CaCO₃ dissolution kinetics as given in the literature. We find that at the millennial time scale there is a wide range of CaCO₃ particle flux attenuation into the ocean interior (e.g. a reduction of −55 to −85% at 1000 m depth), which means that there are significant differences in the impact on particle ballasting, depending on the kinetic expression applied.</p>