Description: Recently Hathorne et al. (2009) documented large intratest trace element (TE) variations in planktonic foraminifera from a single sediment trap sample that could not be explained by variations in water column properties. The laser ablation ICP-MS depth profiles of trace elements through the test walls revealed strong positive correlations between Li, Mg, Mn and Ba resulting from the mixing of a lower TE outer calcite with a higher TE inner calcite. In contrast Sr/Ca ratios remained relatively constant throughout the test wall. These intratest TE variations likely result from biomineralization processes and therefore should be explained by any valid biomineralization model. However, changes in calcite precipitation rate, crystal structure, or the chemical composition of the internal calcification reservoir could not, by themselves, fully account for the pattern of cation intratest variability. Here I expand on this work and investigate if a model of coral biomineralization by Sinclair and Risk (2006) can be adapted to explain the pattern of intratest TE variability in foraminifera. It is clear that the low Mg calcite secreting foraminifera must reduce the Mg/Ca ratio of the calcifying solution by at least a factor of 10 (e.g. Hathorne et al., 2009) and it has been suggested this is achieved by active removal of Mg from the calcification reservoir, although the actual mechanisms remain debatable (e.g. Bentov and Erez, 2006). However, a recent study of the calcification of a low Mg calcite species in the laboratory found a large shortcoming in the amount of Ca potentially provided by seawater transported to the site of calcification in vacuoles compared to a conservative estimate of the amount required to form the new calcite wall (de Nooijer et al., 2009a). This suggests active Ca transport to the site of calcification is required to provide sufficient Ca. If Ca specific, this Ca addition would effectively dilute the TE content (including Mg) of the calcification reservoir to varying degrees and potentially cause the positive TE correlations seen across the test wall. Sinclair and Risk (2006) used this dilution model to successfully explain some TE correlations in coral skeletons. This model can be effectively adapted to foraminifera as it accounts for recent observations of foraminiferal calcification including the transport of seawater by liquid endocytosis to the calcification site and an elevated pH at the site of calcification (Bentov et al., 2009; de Nooijer et al., 2009a, 2009b). This model therefore provides a powerful tool with which to integrate constraints from experimental observation with those from micro-analytical measurements to improve the accuracy, precision and scale of the palaepalaeoceanographic application of foraminiferal geochemistry. Bentov and Erez (2006) Geochem. Geophys. Gepsyst. 7, Q01P08. Bentov et al. (2009) PNAS 106, 21500. de Nooijer et al. (2009a) Biogeosciences 6, 2669. de Nooijer et al. (2009b) PNAS 106, 15374. Hathorne et al. (2009) Paleoceanography 24, PA4204. Sinclair and Risk (2006) Geochim. Cosmochim. Acta 70, 3855.

Description: We present Mg/Ca analyses performed via a Flow Through sequential dissolution device connected to an ICP-OES on the planktonic foraminifer Globorotalia inflata. The aim of the study is to explore the possibility to reconstruct the thermal gradient in the water column by separating non-crusted and crusted calcite phases in the tests of G. inflata using the difference between their Mg/Ca ratios as a measure of the thermal gradient. An important assumption is that the non-crusted part of the tests is calcified in shallow, warmer water than the crusted part. For analyses a range of different preparation steps were used to determine the ideal way of separating the phases. Foraminifer tests were (not) cleaned, (not) crushed, and (not) pulverized before online analysis with the FT device. To analyze samples with a FT device the foraminifer tests are placed on a filter with a mesh of 0.45 μm preventing clay minerals to wash through. A sequential dissolution protocol first rinses the samples with buffered Seralpur water before QD HNO3 is added in small steps to create a ramp of increasing acid strength. As acid is kept constant at each concentration for several minutes, dissolution of a specific calcite phase can take place. Initial results show that it is most effective to slightly crush the tests without applying standard cleaning procedures, but rather analyze them without cleaning. Samples were selected from the South Atlantic (core tops and specific downcore samples) and the Mediteterranean (plankton tows). All samples were chosen based on previous work on them to provide comparison with routinely analysed Mg/Ca ratios. The South Atlantic samples have been analyzed extensively as bulk samples separated in difference size fractions and crusted vs. non-crusted (Groeneveld and Chiessi). The Mediterranean samples were not only analyzed as bulk samples but also by Laser Ablation ICP-MS (von Raden et al.). Results show that bulk analyses are reliably reproduced by the FT method, especially for samples which are dominated by crusted calcite. Samples which were uncrusted often gave much higher Mg/Ca ratios than the bulk analyses. These higher Mg/Ca ratios mainly occur in the plankton tow samples and were also identified with Laser Ablation ICP-MS. A possible reason for this could be the presence of a high Mg amorphous calcite layer on the outside of foraminifer tests which have not completed their calcification yet as was recently also pointed out in several other studies. Identification of the crusted and uncrusted phases, and therewith a thermal gradient, seems to give the expected differences but a more rigorous statistical treatment is needed to pinpoint singular dissolution phases.