For the two experiments in which Ca- and/or K-concentrations were varied, specimens of A. lessonii were isolated from coral rubble retrieved from the tropical aquarium of Burgers’ Zoo [Arnhem, the Netherlands; (Ernst et al., 2011)]. Approximately 100 living specimens were selected and transferred to Petri dishes and placed at 26°C with a light/dark cycle of 12h/12h to stimulate asexual reproduction. Within 2 weeks, between 80 and 90 juveniles were released from several adults that were incubated in pre-made culture media.

For these two sets of experiments, the amounts of salts were added following the recipes of (Kester, 1967) and (Wit et al., 2013), however changing the concentrations of CaCl₂ and KCl ( Supplementary Tables S4 , S5 ). We mixed the artificial seawater with filtered (< 2 μm) North Sea Water, using a ratio of 1 to 5 (20% North Sea Water and 80% artificial seawater) and we measured and controlled the salinity (within +/-0.1 salinity units) at the start and at the end of the experiment using a conductometer.

The prepared media were divided over culture flasks (De Goeyse et al., 2021) that were placed in a cabinet with a controlled atmospheric pCO₂ of 600 ppm and a light/dark cycle of 12h/12h. Evaporation of the culture media was minimized by saturating the atmosphere inside the cabinet with water vapor. The culture media were subsampled for DIC and TA by filling replicate 5mL vials, pre-poisoned with HgCl₂, and storage at 4°C until analyses using a Continuous Segmented Flow Analyzer (QuAAtro). For DIC the method was the same as the one used in the temperature experiment (2.1.1) but for TA an adapted method from (Sarazin et al., 1999) was used, changing the organic acid from formic acid to Phthalic acid, keeping the same indicator Bromo Phenol Blue which is blue at a pH of 3.7 and giving a green color at more alkaline phases having an absorbance at 590nm ( Supplementary Tables S1 , S2 ). A. lessonii was grown under these pre-set conditions for six weeks. Almost all replicates had sufficient material to have one or two analysis per replicate for K/Ca_cc using ICP-MS ( Supplementary Tables S1 , S2 ). Every batch of foraminifera measured contained between 10-25 individuals per condition. The specimens that were selected for analysis were fully grown (i.e. consisted of more than 15 chambers) to minimize the contribution to the chemical signal by the first 1 or 2 chambers that were built before the experimental incubation a correlation with weight of Ca against K/Ca_cc was plotted to ensure there was no correlation and hence no effect of growth rate when K and Na incorporated into the calcite ( Supplementary Table S12 , Supplementary Figure S2 ) The mortality rate during the experiments was derived from the temperature experiment and equaled 22% (i.e. 78% of the incubated specimens were alive at the end of the experiment).

The procedure for selecting specimens, feeding, and subsampling of the culture media were identical to that in the first two sets of experiments. Juveniles derived from asexual reproduction events were divided into 4 equal groups, with at least 20 juveniles per group, which were transferred to Petri dishes with filtered (< 2 μm) North Sea water. At this stage, the juveniles had not more than seven chambers. Three or four replicates (depending on the number of juveniles available per condition) were subsequently incubated at each of the four different temperatures: 18°C, 22°C, 26°C, and 28°C ( Table 1 ). All incubations used a light/dark cycle of 12h/12h and a light intensity of approximately 180 μmol photons m^-2 s^-1 for a total of eight weeks. During the experiment, the foraminifera were observed weekly to monitor how many of the incubated specimens were alive. Every week, culture media were replaced and the foraminifera were fed freeze-dried algae Dunialiella salina (Van Dijk et al., 2017a) suspended in precleaned water (cleaned by rinsing the Falcon tubes with Milli-Q water three times).

To monitor potential changes in inorganic carbon chemistry, culture media were subsampled for dissolved inorganic carbon (DIC) and total alkalinity (TA) ( Supplementary Table S3 ). Vials for these sub-samples were pre-poisoned with HgCl₂ and stored at 4°C until they were analyzed. DIC samples were acidified, and the CO₂ was dialyzed to reduce the phenolphthalein indicator and spectrophotometrically measured at 550nm using Continuous Segmented Flow Analyzer (QuAAtro); TA was measured by using an acid buffered solution of formic acid and was measured spectrophotometrically (Sarazin et al., 1999) using a QuAAtro. For the temperature experiments salinity of the culture media was monitored using a refractometer and was always 36 +/- 0.5. Because we used the same culture water for the different temperature experiments, we do not expect offsets between these experiments. Measured values for DIC and TA together with the temperature and salinity were used to calculate pH, pCO₂, [HCO₃ ^-], [CO₃ ^2-] and the saturation state with respect to calcium (Ω_Ca) using PyCO2SYS (Humphreys et al., 2022) ( Supplementary Table S3 ).

To analyze (potential) correlations between seawater temperature/chemistry and foraminiferal calcite composition, a bootstrap method was applied. This allows calculating confidence intervals for potential correlations through iterative resampling of the dataset generating distinct simulated sub-sets of samples (in this case 1000 iterations were used). The results of these regressions are only shown when p-values< 0.05. Confidence intervals displayed in the figures were taken from the means and standard deviations obtained from the bootstrap analysis. Analyses were performed using a standard package available for Python: scipy.stats (Virtanen et al., 2020).

The incorporation of elements is often presented as the El/Ca (where El is the element analyzed) or D_EL, known as the partition coefficient:

When D > 1 the element is more concentrated in the shell than in the seawater and when D< 1 the element is excluded from the shell (Lea, 2003). As it is explained by (Lea, 2003) this relation is not strictly following thermodynamic properties, but also affected by biology. Nevertheless, expressing element incorporation into the calcite in terms of D allows better comparison between different ocean chemistries and also between experiments.

In the first experiment, both [K⁺]_sw and [Na⁺]_sw were kept constant and therefore variability in foraminiferal calcite composition must be related to changes in [Ca²⁺]_sw ( Supplementary Table S7 ). The K/Ca_cc values of the different conditions used are statistically the same. The K/Ca_cc at the lowest K/Ca_sw appear higher than at higher K/Ca_sw, but this is likely caused by the contribution of pre-existing calcite (see also discussion, 4.1). Na/Ca_cc, however, increases with increasing Na/Ca_sw values (and hence decreasing [Ca²⁺]_sw) with a significant offset between the lower and the higher Na/Ca_cc conditions used in the experiments ( Figure 1B ; bootstrap analysis, p<0.05). Also clear from these figures is that the partitioning coefficients does not remain constant with changing [Ca²⁺]_sw, for both K and Na incorporation.

When varying both [K⁺]_sw and [Ca²⁺]_sw and keeping K/Ca_sw constant, K/Ca_cc increases linearly ( Figure 2A ). In the same cultured foraminifera, the Na/Ca_cc also increases significantly with higher seawater Na/Ca ratios ( Figure 2B ). When directly comparing these results with those from the first experimental set, it is clear that K-incorporation depends not on [Ca²⁺] but does respond when both [K⁺] and [Ca²⁺] are varied ( Figure 2C ). Na-incorporation, on the other hand, does react to changes in [Ca²⁺]_sw, irrespective of changes in [K⁺] ( Figure 2 , Supplementary Table S8 ).

Both K/Ca_cc and Na/Ca_cc of the cultured A. lessonii do not show consistent changes with varying temperature ( Figures 3A, B ), although the average K/Ca_cc was slightly higher at the lowest temperature (0.21 +/- 0.06 mmol/mol) ( Supplementary Table S9 ). Variability observed for K/Ca_cc was higher than that of Na/Ca_cc but does not change consistently between the four temperature conditions using a regression analysis giving for both plots ( Figure 3 ) p-values > 0.05.

Combining all data, Na/Ca_cc and K/Ca_cc of the cultured A. lessonii do not correlate with each other ( Figure 4 ) when varying temperature or changing only [Ca²⁺]. However, when considering the experiments in which both [K⁺] and [Ca²⁺] varied with a constant K/Ca_sw K- and Na-incorporation were negatively correlated, this correlation is due to the difference in [Ca²⁺]_sw that changes the Na incorporated, and not because of a Na⁺ and K⁺ compete for the same spot in the calcite crystal lattice during incorporation.

The K/Ca_cc values obtained for A. lessonii in all experiments ranged between 0.05 and 0.26 mmol/mol ( Supplementary Table S9 ), without a noticeable effect of temperature ( Figure 3A ). The total amount of calcite added per treatment, either when changing [Ca²⁺]_sw or temperature did not correlate to K/Ca_cc ( Supplementary Figures S2 and S3 , respectively). This means that the average K/Ca_cc does not depend on growth rate and changes therein reflect directly the effects of temperature or seawater chemistry. The relatively high variability observed for K/Ca_cc at the lowest temperature ( Figure 3A ) may be explained by uneven growth of foraminiferal specimens across temperatures. At the lowest temperature (18°C) on average fewer chambers were added per specimen than at higher temperatures ( Supplementary Tables S13 , S14 ). As this increases the relative contribution of pre-existing carbonate (i.e. from before the actual experiment) this would add to the variability in the data. The contribution of chambers formed before start of the incubation may have also biased average K/Ca and Na/Ca values, with a larger impact for specimens that added less chambers during the experiment.

The potassium-to-calcium ratio in A. lessonii (0.10 – 0.26; Figure 1A ) overlap with those established for inorganically precipitated calcites (Ishikawa and Ichikuni, 1984); K/Ca_cc: 0.01-0.1 and (Okumura and Kitano, 1986). Still, it is challenging to compare these coefficients directly as the overall composition of the culture media (artificial seawater) in our experiment differed considerably from that used in the inorganic precipitation experiments, which only contained few elements.

The only other report on foraminiferal K/Ca_cc indicated similar ratios for the benthic O. ammonoides (0.12-0.30; Nambiar et al., 2023), as those reported here ( Supplementary Tables S10 , S11 ). The small offset in the average ratios could be caused by differences in culturing conditions or by inter-species differences in El/Ca values as observed for many elements (Bentov and Erez, 2006). Although we used a slightly different analytical approach (higher mass resolution) this is unlikely to have resulted in appreciable offsets between studies (methods and Supplementary Figure S1 ).

The average and range in Na/Ca_cc of cultured specimens ( Supplementary Tables S9 - S11 ) are similar to ratios reported previously for this genus (Geerken et al., 2019; Levi et al., 2019). The Na/Ca_cc in specimens cultured from 18 to 28°C vary between 8.8 and 11 mmol/mol ( Supplementary Table S9 ) and are not correlated to temperature, which is in line with previous reports (Allen et al., 2016; Geerken et al., 2019). In the experiment with temperatures ranging from 18 to 28°C ( Supplementary Table S9 , Figure 3 ), K/Ca_cc values show no significant differences between different temperature conditions, agreeing with the field data from (Nambiar et al., 2023). The absence of an effect of temperature on either K/Ca_cc or Na/Ca_cc underscores the potential for applying these ratios in reconstructing (long-term) changes in seawater composition.

Element/Ca ratios vary greatly between genera and species and generally show elevated ratios for multiple elements in the same taxon (Evans et al., 2015; Mewes et al., 2015a; Hauzer et al., 2018; Van Dijk et al., 2019; Dämmer et al., 2021). The magnitude by which the different elements vary among species, however, is not constant. For K/Ca_cc, ratios are on average slightly higher in O. ammonoides (Nambiar et al., 2023) than in A. lessonii. ( Figures 1 – 3 ), which is in line with the observed difference in Mg/Ca_cc between these species. Mg/Ca_cc is approximately 140 mmol/mol for O. ammonoides (Evans et al., 2015) and varies between 25 and 30 mmol/mol for Amphistegina sp (De Nooijer et al., 2017). Similarly, Na/Ca_cc in A. lessonii varies between 7 and 11 mmol/mol ( Figures 1B , 2B and 3B ; (Van Dijk et al., 2017c) and around 20 - 25 mmol/mol in O. ammonoides (Hauzer et al., 2018). In contrast, interspecific differences in Sr/Ca are relatively small [~1.7 versus ~2.6 mmol/mol for Amphistegina and Operculina, respectively; (Geerken et al., 2019; Hauzer et al., 2021)], but still, as for the other elements, they are all higher in Operculina than in Amphistegina. The consistent offsets, albeit with different magnitudes, suggest that differences in biomineralization pathways, e.g. Ca²⁺ pumping rate, between genera and species are reflected in element incorporation (De Nooijer et al., 2023). These inter-specific differences in element uptake are consistent with the reported relatively low K/Ca_cc values for planktonic foraminiferal species [0.06-0.12 mmol/mol; (Li et al., 2021)], as these are also known to have low values for Mg/Ca (1-5 mmol/mol; (Nürnberg et al., 1996; Anand et al., 2003) compared to Amphistegina and O. ammonoides ( Figure 5 ). More recently (Nambiar et al., 2023), however, showed ratios (about 0.25 mmol/mol) for G. ruber comparable to those reported here and for A. lessonii (Nambiar et al., 2023).

Recently, foraminiferal Na/Ca_cc was suggested to act as a proxy for seawater [Ca²⁺] (Hauzer et al., 2018) in addition to its potential to reconstruct salinity (Wit et al., 2013; Mezger et al., 2019). Our results confirm that foraminiferal Na/Ca_cc is responsive to changes in [Ca²⁺] ( Figures 1 , 2 ) and the salinity range in Tables 2 and 4 in this case, show no significant effect in the Na or K incorporation, due to [Ca²⁺]_sw differences act as the main driver where they are incorporated. Here we show that foraminiferal K/Ca_cc may be used to reconstruct past seawater [K⁺], adding to our understanding of the long-term cycling of the oceans’ major ions as was suggested by Nambiar et al., 2023). For seawater potassium concentrations, such a reconstruction may indicate rates of continental weathering (Kronberg, 1985; Sun et al., 2016) and indirectly, may reflect long-term changes in total alkalinity. Alternatively, seawater potassium concentrations may reflect rates of oceanic crust formation due to the difference between seawater and the mantle (Bloch and Bischoff, 1979).

K/Ca_cc in A. lessonii does not vary consistently with [Ca²⁺]_sw (and hence K/Ca_sw; Figure 1A ). Conversely, K/Ca_cc increases when [Ca²⁺] and [K⁺] both increase (and K/Ca_sw remains similar; Figure 2A ). The absence of an effect of seawater Ca²⁺ and a positive effect of the combined K⁺ and Ca²⁺ elevation suggests that seawater [K⁺] is the main driver of K-incorporation into foraminiferal calcite ( Figure 2C ), whereas it is not affected by [Ca²⁺]. This resembles results from inorganic experiments in which K-incorporation was found to depend on solution [K⁺] (Ishikawa and Ichikuni, 1984; Okumura and Kitano, 1986). Such a dependency is hence fundamentally different from that observed for many divalent cations like Ba²⁺, Mn²⁺, Mg²⁺, or Sr²⁺, for which the concentration in seawater relative to [Ca²⁺]_sw directly translates into a calcitic El/Ca ratio (Okumura and Kitano, 1986; Alkhatib et al., 2022). Such an incorporation pattern is caused by the competition of these ions with Ca²⁺ for a place in the calcite crystal lattice: for the monovalent cation incorporation is not governed by competition with Ca²⁺. However, incorporation of monovalent ions (Li⁺, Na⁺, K⁺, etc.) may be affected by different processes such as crystal surface sorption, speciation, and crystal growth rate, which complicates a straightforward interpretation of the environmental controls on element partitioning during calcification. Most likely a combination of all these effects impacts the uptake of monovalent cations.

More precisely, incorporation of an element into foraminiferal calcite is at least a two-step process ( Figure 6 ; Bentov and Erez, 2006). There is the biological activity that configures the size and shape of the calcifying space and sets the composition of the fluid from which CaCO₃ precipitates (Erez, 2003; Bentov and Erez, 2006; Bentov et al., 2009; De Nooijer et al., 2014). The exact controls are only partly characterized and may vary between species, but include selective ion transport (Nehrke et al., 2013; Toyofuku et al., 2017), pH manipulation (Bentov et al., 2009; De Nooijer et al., 2009; Glas et al., 2012) and production of organic templates (Branson et al., 2016; Tyszka et al., 2019). Secondly, there are various (physico-) chemical processes that operate within the calcifying space that determine how much of each ion in the fluid is incorporated into the shell. Relevant processes include Rayleigh fractionation (Elderfield et al., 1996; Evans et al., 2018), kinetics (Uchikawa and Zeebe, 2012; Devriendt et al., 2021), CaCO₃ phase-transformations (Gray and Evans, 2019), chemical speciation (Van Dijk et al., 2017b) and configuration of ions within the crystal lattice (Branson et al., 2015) ( Figure 6 ). For K one would expect a different trend as K uptake does not depend on Ca concentration ( Figures 1 , 2 ), but should reflect K concentration at the site of calcification (SOC). Nambiar et al. (2023), however, suggested an effect of [Ca²⁺] on K-incorporation, as was observed in corals (Ram and Erez, 2021). The within chamber wall K profiles of Geerken et al. (2019) actually show a gradual increase towards the end of a low concentration band (i.e. towards the outer surface of the shell wall). Other studies, however, have shown a tight correlation between the alternating high- and low-K bands with those of other elements (e.g. Mg) in Amphistegina lobifera and A. lessonii (Levi et al., 2019), which hints at a similarity in their incorporation dynamics. In addition, phosphorus is present in relatively narrow bands, similar to those of potassium, which hints to a coupling in their incorporation mode.

The observed relation between a solution’s El/Ca and the calcitic El/Ca is likely affected by both biological and ‘chemical’ steps during foraminiferal biomineralization. Transport of K-ions over a cell membrane is not likely facilitated by Ca-transporters, while an occasional Mg-ion may well pass a Ca-pump (e.g. Dudev and Lim, 2013). For the precipitation dynamics, the monovalent K ion may occupy the interstitial sites in the calcite lattice (Ishikawa and Ichikuni, 1984), in the organic phases, or be incorporated in amorphous K₂CO₃ (Li et al., 2021). It is likely that all or most of these processes together determine the eventual K/Ca_cc and Na/Ca_cc of the foraminiferal calcite ( Figure 6 ).

Our results show that the seawater’s [K⁺], rather than [Ca²⁺] is the main driver for K partitioning. This is in line with free K⁺ being the main species present in seawater (Johnson and Pytkowicz, 1979), with K sorption at the crystal lattice being the primary way K is incorporated. It should be noted however, that the chemical speciation of potassium in the calcifying fluid may differ from that in seawater due to the foraminfer’s biological control on that fluid’s composition (Erez, 2003; Bentov and Erez, 2006; De Nooijer et al., 2014) In contrast, for Na-incorporation, our results, and the dependence of Na/Ca_cc on the solution’s Na/Ca ( Figure 2 ) suggest that sodium and calcium compete for the same place in the crystal lattice. This would imply that sodium incorporation substitutes by replacement of two Na ions for one Ca ion (Devriendt et al., 2021) in the calcite and form a quadratic relationship. Since we do not observe this, alternatives including lattice vacancies can be filled by other divalent elements such as Mg²⁺ or Sr²⁺ may also play a role in the calcite’s El/Ca.

Although our results cannot exclude that K-incorporation at interstitial sites is responsible for a (minor) part of the foraminiferal K/Ca_cc, (Nambiar et al., 2023) showed an increasing K/Ca_cc with K/Ca_sw for O. ammonoides with similar conditions as the one show in the [Ca²⁺]_sw experiment from this study. This may indicate that the crystallographic orientation of potassium differs between foraminiferal species but may also indicate that biological uptake of potassium ions varies among species. The presence of specialized K-transporters in addition to Ca-transporters (Toyofuku et al., 2017) promoting K-incorporation into the shell would imply an advantage of higher K/Ca, which is not immediately apparent. Alternatively, inward K-transport into the site of calcification may result from the organismal need to accompany inward transport of a negatively charged ions and/or to counter the outward transport of another positively charged ion in the calcifying fluid. Major anions are SO₄ ^2-, Cl^-, but both do not seem to be enriched in foraminiferal calcite compared to seawater and inorganically precipitated calcite (Kontrec et al., 2004; Van Dijk et al., 2017b; Roepert et al., 2020). Cations that are pumped out during calcification include H⁺ (Glas et al., 2012) and are hypothesized to be coupled to an inward Ca²⁺ pumping (De Nooijer et al., 2017). The outward pumping of H⁺ may sometimes also affect Na⁺- and/or K⁺-transport. With it is considerably larger ionic radius K⁺ (152 pm) is less likely to be ‘accidentally’ transported through a transporter designed for H⁺ (ionic radius of 66 pm) compared to Na⁺ (116 pm).

When pumping of K⁺ by the foraminifer is somehow affecting the K/Ca at the site of calcification incorporation of potassium in foraminiferal calcite may be impacted by an intracellular role of [K⁺]. Potassium is often used as an agent to regulate osmosis and is found in cells at [K⁺]/[Na⁺] ratios in the 0.1–1.0 range (Dibrova et al., 2015). Sodium ions form relatively strong hydrogen bonds, therefore attract water, and disrupt intracellular processes. For this reason, many (marine) cells have [K⁺]/[Na⁺] antiporters (Nakamura et al., 1992; Shabala et al., 2009). Such a transport mechanism may (partly) explain the observed difference in Na/Ca and K/Ca in foraminifera when compared to inorganically precipitated calcite. The partition coefficients for potassium are relatively similar between inorganic and foraminifera calcite, while the partition coefficients for sodium are much lower in inorganically precipitated calcite (compared to foraminiferal shells). Overall, this makes K uptake in foraminiferal calcite a much more reliable proxy for reconstructing past sea water major ion composition.