Sciences Sciences Discussions

Salinity and temperature determine seawater density, and differences in both thereby control global thermohaline circulation. Whereas numerous proxies have been calibrated and applied to reconstruct temperature, a direct and independent proxy for salinity is still missing. Ideally, a new proxy for salinity should target one of the direct constituents of dissolved salt, such as [Na p ] or [Cl − ]. This study investigates the impact of salinity on foraminiferal Na/Ca values by laser ablation ICP-MS analyses of specimens of the benthic foraminifer Ammonia tepida cultured at a range of salinities (30.0–38.6). Foraminifera at lower salinities (30.0 and 32.5) added more chambers (10–11) to their test over the course of the culturing experiment than those maintained at higher salinities (36.1, 7–8 chambers, and 38.6, 6–7 chambers), suggesting that growth rates in this species are promoted by lower salinities. The Na/Ca of cultured specimens correlates significantly with seawater salinity (Na/Ca = 0.22S–0.75, R 2 = 0.96, p D Na vary between 5.17 and 9.29 mmol mol −1 and 0.12–0.16 × 10 −3 , which are similar to values from inorganic precipitation experiments. The significant correlation between test size and Na/Ca results from co-variation with salinity. This implies that foraminiferal Na/Ca could serve as a robust and independent proxy for salinity, enabling salinity reconstructions independent of calcitic δ 18 O.


Introduction
Temperature and salinity are among the most relevant parameters when studying past ocean circulation, as they together control ocean water density and thus thermohaline circulation.A large number of proxies for reconstructing seawater temperature established and widely applied (e.g.Epstein et al., 1953;O'Neil et al., 1969;Shackleton, 1974;Prahl and Wakeham, 1987;N ürnberg et al., 1996;Bemis et al., 1998;Hastings et al., 1998;Elderfield and Ganssen, 2000;Lear et al., 2000;Anand et al., 2003).Independent quantitative tools for the reconstruction of past salinities are, however, lacking.Until now salinity can be reconstructed using dinoflagellate and diatom species composition (Zonneveld et al., 2001), process length of dinoflagellates (Mertens et al., 2009), under some conditions using the bivalve strontium isotope composition (Israelson and Buchardt, 1999) and Ba concentration of foraminiferal calcite (Weldeab et al., 2007).These proxies, however, are either indirect, through changes in ecology, or strongly dependent on regional oceanography (i.e.river water input).Other indirect approaches involve using the correlation between seawater oxygen and hydrogen isotope composition and salinity (Gat, 1996).Foraminiferal δ 18 O (after correcting for the effect of temperature using Mg/Ca or U k 37 ) or hydrogen isotope composition (δD) of alkenones can be used to infer salinity (Elderfield and Ganssen, 2000;Schouten et al., 2006;Van der Meer et al., 2007).Spatial and temporal variations in the relationship between water stable isotopes and salinity, however, limit the accuracy of this approach.At best, the relation between salinity and seawater isotope composition can be modeled, involving assumptions on precipitation, evaporation and/or river runoff, introducing relatively large uncertainties in reconstructed salinities (Rohling and Bigg, 1998).
Ideally, a new proxy for seawater salinity should target one of the major constituents of sea salt, rather than using indirect relations between the isotopic composition of seawater and salinity.By far the largest components of seawater salinity are Na + and Cl − .
Incorporation of sodium into calcium carbonate has been shown to vary with salinity in the Atlantic Oyster (Crassastrea virginica: Rucker and Valentine, 1961), barnacle shells (Gordon et al., 1970), and inorganically precipitated calcium carbonate (Kitano et al., 1975;Ishikawa and Ichikuni, 1984).Application in paleoceanographic reconstructions requires, however a more general proxy signal carrier.Here, we investigate the impact of salinity on foraminiferal Na incorporation by laser ablation ICP-MS analyses of Introduction

Conclusions References
Tables Figures

Back Close
Full specimens of the benthic foraminifer Ammonia tepida cultured at controlled conditions at different salinities (30.0-38.6).Since seawater pH, alkalinity and total dissolved inorganic carbon (DIC) were kept constant to avoid a possible impact by inorganic carbon chemistry (as observed for, e.g.Mg;N ürnberg et al., 1996;Kisak ürek et al., 2008;Due ñas-Boh órquez et al., 2009), changes can be interpreted in terms of salinity only.

Methods
The benthic, cosmopolitan foraminifer Ammonia tepida inhabits a wide range of environments, and is one of the few species found in habitats with highly variable temperatures and salinities (Murray, 1968;Hayward et al., 2004).This adaptation makes it a suitable species to culture at a wide range of conditions (e.g.De Nooijer et al., 2007;Dissard et al., 2010;Due ñas-Boh órquez et al., 2011a).Surface (0-2 cm) sediments were collected from an intertidal mudflat at the Wadden Sea near Den Oever, the Netherlands.Upon return in the laboratory, individual specimens were isolated from this material and placed in filtered seawater with a salinity of 35, at 20 • C.These specimens were subsequently monitored for reproduction.Shortly after an asexual reproduction event, megalospheric juvenile specimens consisting of 2-3 chambers were handpicked and transferred to the culture set up.Culture media consisted for 50 % of artificially prepared and 50 % natural seawater (Mediterranean Sea, filtered and diluted to a salinity of 35) using the recipe of Kester et al. (1967).Culture media with different salinities were obtained by varying the amount of salts added to each solution (Table 1).Amounts of added NaHCO 3 and B(OH) 3 were kept constant for each batch of culture media in order to obtain media with similar alkalinity and dissolved inorganic carbon (DIC) concentration.Subsequently, all prepared artificial seawater batches were mixed with equal amounts of natural seawater with a salinity of 35.The resulting media thus had salinities of 30.0, 32.5, 36.1 and 38.6, covering a large part of the range in salinities found in the open ocean.Temperature, salinity, alkalinity and DIC were monitored over the course of each experiment (Table 2).Introduction

Conclusions References
Tables Figures

Back Close
Full Element composition of the culture media was determined by ICP-OES (Spectro Arcos).Measured alkalinity and DIC were used to calculate the other parameters of the inorganic carbon system of the culture media, using CO2SYS (Lewis and Wallace, 1998) (Table 2).Speciation and activities of free elements in the culturing experiments was modeled using PHREEQC (LLNL database, Parkhurst and Appelo, 1999).
Experiments ran for 6-8 weeks to ensure that sufficient foraminiferal chambers were added for elemental analysis.Specimens were harvested and sieved over a screen with a 125 µm mesh size.Size of individual foraminifera was determined using a calibrated microscope camera and computer software.Since the spiral growth mode of A. tepida produces outlines that are not perfectly circular, diameter of every individual foraminifer was measured 4 times to calculate average diameter (Tables 3 and 4).
Foraminiferal Na/Ca and Mg/Ca were determined using Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS), performed at Utrecht University.Single chamber element composition was determined using 23 Na, 24 Mg, 26 Mg, 27 Al, 43 Ca, 44 Ca, 55 Mn and 88 Sr and their relative natural abundances (Reichart et al., 2003;Wit et al., 2010).All laser spots were 80 µm in diameter, repetition rate was 7 Hz and laser energy density was set at ∼ 1 J cm −2 .Time resolved signals were selected for integration, background subtracted, internally standardized to 43 Ca, and calibrated against a glass standard (NIST SRM610; elemental concentrations from Jochum et al., 2011) that was ablated at a higher energy density (∼ 5 J cm 2 ).Using different ablation energies for glass and calcite was previously shown not to affect the analyses (Wit et al., 2010;Due ñas-Boh órquez et al., 2011).Integration windows that separate the calcitic signal from background and detection of any contaminants at the test surface was done using designated software (Glitter).Elemental ratios with respect to Ca were based on the average of each ablation profile.Introduction

Conclusions References
Tables Figures

Back Close
Full

Results
Culture solution parameters were stable, for each of the four experimental conditions and, with the exception of salinity, similar between experiments.Seawater Mg/Ca values increased slightly with salinity (Table 2).Ammonia tepida at lower salinities (30.0 and 32.5) added more chambers (10-11) to their test over the course of the experiment than foraminifera cultured under higher salinity (36.1, 7-8 chambers, and 38.6, 6-7 chambers), implying somewhat higher growth rates at lower salinities.Foraminiferal survival rates were similar across experiments (44-47 %), while test size significantly increased with salinity (R 2 = 0.97, p < 0.01) (Table 3, Fig. 1).
The Na/Ca of cultured specimens correlated significantly (p < 0.01) with seawater salinity (Fig. 2) and size (R 2 = 0.96, p < 0.01, Fig. 1).Foraminiferal Na/Ca values ranged from 5.17 to 9.29 mmol mol −1 over all experiments.Inter-individual variability within each experiment was between 9 and 17 % (Table 3) and of the same order in single chamber ablation profiles.Values for Mg/Ca measured on the calcite test correlated positively with salinity (R 2 = 0.99, p < 0.01) (Table 4, Fig. 2).Foraminiferal Mg/Ca ranged from 0.98 to 2.57 mmol mol −1 over all experiments.Inter-individual variability in Mg/Ca within each salinity treatment varied between 13-16 % (relative standard deviation) and was similar between the different experiments.

Na/Ca and salinity
Results indicated a significant (p < 0.01), positive correlation between seawater salinity and calcite Na/Ca (Fig. 2).Inter-individual variability for Na/Ca values (relative standard deviation of 9-17 %) was similar as previously recorded for some other elements (e.g.Mg) measured on single chambers (Sadekov et al., 2008;Wit et al., 2010Wit et al., , 2012 carbonate of some species (Erez, 2003) and changes in the relative contributions of high and low Na and Mg bands could explain the observed inter chamber variability.
To facilitate comparison of our foraminiferal Na/Ca values to ratios obtained in other studies, calcitic Na/Ca values are henceforth expressed as partition coefficients (D Na ).The D Na is the ratio of sodium over calcium in foraminiferal calcite divided by the ratio of these elements in the culture medium.The D Na for foraminifera reported here (0.12-0.16 × 10 −3 ) is similar to that found in inorganically precipitated calcites (0.07-0.20 × 10 −3 ; Kitano et al., 1975;Ishikawa and Ichikuni, 1984).This similarity suggests that the general biological control on Na incorporation in A. tepida is likely to be minor.Moreover, D Na for planktonic foraminiferal species is within the same range (0.11-0.17 × 10 −3 ) (Delaney et al., 1985;Lea et al., 1999), suggesting that incorporation of Na is similar across Rotaliid foraminifera.Still, the observed Na banding (Erez, 2003) indicates that some biological control must exist.Previous studies reporting partition coefficients for Na in biogenic and inorganically precipitated calcium carbonates could not distinguish between lattice-bound Na and that present in microscopic seawater inclusions (Rucker and Valentine, 1961;Gordon et al., 1970, Kitano et al., 1975;Ishikawa and Ichikuni, 1984).Although chloride is also incorporated in calcium carbonate, this occurs at 20-40 times lower concentrations than Na (Kitano et al., 1975).The much lower Cl concentration excludes primary incorporation of Na in fluid inclusions and hence suggests that Na is structurally bound in the calcite lattice.This is supported by the lack of so-called hotspots in the Na profiles.
Some cations (e.g.Mg 2+ and Sr 2+ ) are incorporated into calcite by substituting for calcium ions (Morse et al., 2007).For these elements, seawater element/Ca values impact calcitic element/Ca values (e.g.De Nooijer et al., 2007;Segev and Erez, 2006).Since Na + is not directly substituting for Ca 2+ , due the charge difference, incorporation of Na in inorganically precipitated calcium carbonate does not depend on seawater Ca 2+ concentration and is therefore not necessarily impacted by seawater Na/Ca (Ishikawa and Ichikuni, 1984).Instead, Na incorporation depends primarily on the activity of Na in seawater, which is a function of its concentration and, to a lesser extent, its

Conclusions References
Tables Figures

Back Close
Full activity coefficient (Ishikawa and Ichikuni, 1984).Increasing salinity (and hence [Na + ]) increases the activity of Na, while the associated decrease in its activity coefficient, because of the higher salinity, might explain the offset from a 1 : 1 relation between (foraminiferal) Na/Ca and salinity (Fig. 2) (Zeebe and Wolf-Gladrow, 2001).Since the activity coefficient of Na in seawater is only slightly affected by temperature over the relevant range, the effect of temperature on Na incorporation is negligible (Ishikawa and Ichikuni, 1984;Delaney et al., 1985;Lea et al., 1999;Zeebe and Wolf-Gladrow, 2001).This implies that foraminiferal Na/Ca may thus primarily reflect seawater [Na + ] and can thus be used to directly reflect seawater salinity Incorporation of certain elements into the calcite of some foraminiferal species is correlated to ontogeny, i.e. depends on test size (N ürnberg et al., 1996;Wit et al., 2010;Due ñas-Boh órquez et al., 2011b).Since test size of the cultured foraminifera varied between salinities (Fig. 1, Table 3), the relation between foraminiferal Na/Ca and salinity may potentially have been affected by differences in test sizes.To test the potential impact of test size on the relation between Na/Ca and salinity, results for the four salinity treatments were analyzed separately.Foraminifera from experiments at salinities 30.0, 32.5, 36.1 and 38.6 show a significant correlation between size and Na/Ca (R 2 = 0.33, p < 0.05, R 2 = 0.23, p < 0.05, R 2 = 0.59, p < 0.01 and R 2 = 0.29, p < 0.05, respectively Fig. 3), indicating a potential ontogenetic effect.However, if the relation between salinity and foraminiferal Na/Ca would be exclusively caused by ontogeny, no relation between foraminiferal Na/Ca of a single chamber number (whorl position) and salinity would be expected.Chambers 3 through 11 (Fig. 4) all showed a positive correlation between Na/Ca and salinity.The correlation between foraminiferal Na/Ca and test size is, therefore, primarily caused by co-variation of both parameters with salinity.Still, a small size effect may be present due to variability in the relation between salinity and single chamber Na/Ca (Fig. 3).However, this does not effect the calibration significantly when measuring multiple chambers per individual, as done here.a general control of salinity on the incorporation of Mg into foraminiferal calcite.The relation between salinity and foraminiferal Mg/Ca could, therefore, be related purely to a-biotic differences in element speciation.This hypothesis was tested by modeling the speciation and activities of (free) elements in the culture media using PHREEQC (Parkhurst and Appelo, 1999).
Model results showed that the activities of free Mg and Ca increase linearly with salinity, and are an order of magnitude higher for free Mg 2+ than for free Ca 2+ (Fig. 6).
So, despite the fact that we kept the Mg/Ca concentration ratio constant with salinity, the free Mg to free Ca activity ratios of the culture media increased with increasing salinity, providing a mechanistic link between salinity and foraminiferal Mg/Ca.
Incorporation of Mg might also be affected by ontogeny, since test size of the cultured foraminifera varies significantly with salinity (Fig. 1, Table 4).Foraminifera from experiments at salinities 30.0 and 32.5 do not show a significant correlation between size Introduction

Conclusions References
Tables Figures

Back Close
Full and Mg/Ca (R 2 = 0.02, p > 0.10 and R 2 = 0.03, p > 0.10 respectively, Fig. 3).At higher salinities (36.1 and 38.6), however, a small but significant effect of size on foraminiferal Mg/Ca is present (R 2 = 0.22, p < 0.05 and R 2 = 0.45, p < 0.01 respectively).Chambers 3 through 14 (Fig. 5) all show a positive correlation between Mg/Ca and salinity, indicating that a minor ontogenetic effect is to be expected, if multiple chambers per individual foraminifer are measured, following the same reasoning as for the Na/Ca.The impact of salinity on foraminiferal Mg/Ca as measured here may impact reconstructed temperatures by 0.25-1.5 • C per salinity unit, depending on the sensitivity of Mg incorporation to temperature (N ürnberg et al., 1996;Kisak ürek et al., 2008;Due ñas-Boh órquez et al., 2009;Dissard et al., 2010;Wit et al., 2012).However, in addition to independent and direct reconstructions of seawater salinity, foraminiferal Na/Ca values can also be used to correct the effect of salinity on Mg/Ca-temperature reconstructions.As an example, the relation established between Ammonia tepida Mg/Ca values and salinity was combined with the relation between temperature and Mg/Ca for Ammonia beccarii (Toyofuku et al., 2011).A multiple regression combining the response of Mg/Ca as a function of salinity (Fig. 2) and temperature (Toyofuku et al., 2011) resulted in Eq. ( 1), in which foraminiferal Mg/Ca is depending on both temperature (T ) and salinity (S).
Within Eq. ( 1), Mg/Ca is now depending on both temperature and salinity.Since salinity can now be independently reconstructed using foraminiferal Na/Ca values (Fig. 2), paired analyses of Na/Ca and Mg/Ca allows correcting Mg/Ca-based temperatures for changes in salinity.

Conclusions
Foraminiferal Na/Ca and Mg/Ca as measured on cultured benthic foraminifer A. tepida correlate significantly with salinity.This positive correlation is related to the increase in Introduction

Conclusions References
Tables Figures

Back Close
Full the activity of free Mg/Ca and free Na + with increasing salinity.Although size effects might play a role in these calibrations, their effect is insignificant compared to the effect of salinity, and negligible (within the experimental error) when multiple chambers in different individuals are analyzed.Foraminiferal Na/Ca appears to be a robust and independent proxy for salinity, one of the most sought after paleoceanographic proxies, enabling a whole new range of independent salinity reconstructions.Furthermore, in combination with foraminiferal Mg/Ca, it allows a direct correction for one of the most used paleo-thermometers (i.e.Mg/Ca).

BGD Figures Back Close
Full    Full have been developed (e.g.carbonate δ 18 O and Mg/Ca), some of which are now well Figures Back Close Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | ; Due ñas-Boh órquez et al., 2011b).Both Mg and Na showed distinct banding in the test Introduction Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Due ñas-Boh órquez, A., da Rocha, R. E., Kuroyanagi, A., Bijma, J., and Reichart, G. J.: Effect of salinity and seawater calcite saturation state on Mg and Sr incorporation in cultured planktonic foraminifera, Mar.Micropal., 73, 178-189, 2009.Due ñas-Boh órquez, A., Raitzsch, M., De Nooijer, L. J., and Reichart, G. J.: Independent impacts of calcium and carbonate ion concentration on Mg and Sr incorporation in cultured Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

2 Correcting Mg/Ca based temperatures for salinity
(Kisak ürek et al., 2008;isak ürek et al., 2008;rbonate isDissard et al., 2010) bDissard et al., 2010)et al., 1996;Kisak ürek et al., 2008; Due ñas-Boh órquez et al., 2009;Dissard et al., 2010)in addition to temperature and paired Mg/Ca and Na/Ca ratios, measured on the same individuals, now offer the opportunity to correct for such a salinity effect.The D Mg of the cultured A. tepida correlated positively with salinity (R 2 = 0.99, p < 0.01) (Table4, Fig.1), indicating that the effect of seawater Mg/Ca on foraminiferal Mg/Ca did not influence our results.Within previous experiments(Kisak ürek et al., 2008; Due ñas-Boh órquez et al., 2009;Dissard et al., 2010), saturation state of the seawater (Ω) was kept constant, but alkalinity and DIC varied with salinity.Assuming Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |4.limited differences between species in their calcification pathways, the similarity in response of Mg/Ca with salinity (where alkalinity and DIC were kept constant) indicates that alkalinity and DIC have a minor effect on foraminiferal Mg/Ca over this range.The similarity between the relation between Mg/Ca and salinity, for such a variety of independent culture experiments, analytical approaches and species, furthermore hints at

Table 1 .
Kester et al. (1967) for 1 L of artificial seawater following the recipe ofKester et al. (1967).Artificial seawater is mixed with one liter of natural seawater at salinity of 35.

Table 2 .
(Lewis and Wallace, 1998)data.Carbonate ion concentration and Ω calcite are calculated using CO2SYS and alkalinity and DIC as system parameters(Lewis and Wallace, 1998).Standard deviations are based on all measurements over the course of each experiment.

Table 4 .
Mg/Ca and size data for the individual foraminifera of each salinity experiment.The uncertainty (±) in the individual measurements and the average (bold) are based on the standard error of the mean (σ/ √ n).Mg/Ca values of individual foraminifers are based on the ablation profiles of 2-4 chambers.