Zinc incorporation in the miliolid foraminifer Pseudotriloculina rotunda under laboratory conditions

a Department of Life and Environmental Sciences (DiSVA), Politechnic University of Marche, Ancona, Italy b UMR CNRS 6112 LPG-BIAF, University of Angers, Angers, France c Department of Chemical and Geological Sciences, University of Modena and Reggio Emilia, Modena, Italy d Centre de Recherche sur la Paléobiodiversité et les Paléoenvironnements, UMR 7207 CNRS MNHN UPMC, Muséum National d'Histoire Naturelle, Paris Cedex 05, France


Introduction
The recent worldwide legislation aims to restore the "pre-anthropic impacts" status in marine environments (e.g., WFD, 2000/60/EC andMSFD, 2008/56/EC in Europe). Information about the pristine faunas like those in pre-industrial times, however, are often impossible to obtain because of the scarcity (or lack) of reference stations that could still represent unimpacted present-day conditions. Fossilizing organisms represent an excellent historical archive of environmental conditions. Foraminifera are distributed worldwide in many different habitats, from brackish to marine, and their fossil records create an excellent historical archive which can be used as proxies for the reconstruction of past environments, such as pre-industrial ecological conditions (Schönfeld et al., 2012). A new approach involves the use of foraminiferal test geochemistry to assess the evolution of pollutant (i.e., metals) concentrations through time. Incorporation rates of trace elements are widely used as specific proxies in paleoceanography and paleoecology (e.g., Eggins et al., 2003;Hönisch and Hemming, 2005;Levi et al., 2007;Katz et al., 2010;Sabbatini et al., 2011), despite the possible bias linked to the biological influence on calcification processes (i.e., vital effects). The need to calibrate these proxies through culturing experiments was highlighted in the last decade by several authors (e.g., de Nooijer et al., 2007). This approach offers the advantage of changing one single variable (while all the others are kept constant) in order to better evaluate the vital effect. These biological aspects could be even more important for the incorporation rates of chemicals whose concentrations exceed natural baselines due to human activity, and that could be potentially used as pollution markers.
In this study we investigate the incorporation rates of Zn in the shell of the benthic miliolid foraminifer Pseudotriloculina rotunda (Schlumberger, 1893). Among foraminiferal species miliolids showed contradictory responses to heavy metal pollution in different studies. For example, decreasing miliolid relative abundances in polluted (by both organic and inorganic chemicals) coastal zones are reported and interpreted by some authors as a sensitivity index (e.g., Ferraro et al., 2006;Frontalini and Coccioni, 2008). Other studies, on the other hand, suggest a strong tolerance of several miliolid species to pollution, both in situ and under laboratory conditions (e.g., Samir and El-Din, 2001;Romano et al., 2008;Cherchi et al., 2009;Foster et al., 2012;Nardelli et al., 2013).
The aim of the present study is to calibrate incorporation rates of Zn in miliolid foraminiferal shells and thus evaluate the usefulness of the Zn incorporation rate as an environmental proxy. Zn is, in fact, one of the most common pollutants associated with human activities (e.g., Callender and Rice, 2000;Wuana and Okieimen, 2011), that can be toxic for biological systems when its concentration exceeds a threshold value (e.g., Haase et al., 2001;Valko et al., 2005;Díaz et al., 2006;Formigari et al., 2007). Nardelli et al. (2013) showed that inorganic Zn Marine Micropaleontology 126 (2016) 42-49 at concentrations higher than 0.1 mg/L can cause biological stress in Pseudotriloculina rotunda, causing delay in calcification rates.
In this regard, our study also aims to test the hypothesis that the biological stress caused by high Zn concentrations may influence metal incorporation rates as well. Zn incorporation was investigated using Laser Ablation Inductively Coupled Plasma Mass Spectrometer (LA-ICP-MS) analysis. Moreover, morphological observation using the Environmental Scanning Electron Microscope (ESEM) was employed to check for abnormalities in the organization/distribution of single shell crystallites. Nardelli et al. (2013) previously suggested that Zn does not cause macroscopic test deformations in miliolid foraminiferal shells, on the base of their observation of coiling patterns and chamber shapes of P. rotunda specimens grown at increasing zinc concentrations, under the binocular microscope. The aim of our ESEM analyses was to deepen these observations and check for calcite anomalies at the crystallite level.

Experimental set up
All the analyses were performed on specimens of Pseudotriloculina rotunda that grew at least one chamber during 70-days exposure to six different Zn concentrations under laboratory conditions. Znenriched solutions were prepared adding respectively 0.01, 0.1, 1.0, 10, and 100 mg/L of Zn to natural seawater (sea) from an unpolluted site (Portonovo, Adriatic Sea); Zn and Ca concentrations in natural seawater were measured using an Inductively Coupled Plasma Mass Spectrometer (ICP-MS). Although precautions were taken during manipulations, the ICP-MS analyses revealed that the zinc concentration of the natural seawater used for the experiment was higher than natural background. In fact the zinc concentration measured on culture waters before the addition of zinc was 0.149 ± 0.01 mg/L, while the zinc concentration of seawater from the sampling site was originally 0.237 ± 0.01 μg/L. However, as the same water was used to prepare all the zinc solutions for the different treatments and, considering the very high concentrations of added zinc, we believe that this contamination does not compromise the dataset. But, of course, this means that the lowest tested seawater zinc concentration of the experiment (named "sea" hereafter) cannot be considered as a control representative of unpolluted seawater conditions. Temperature (15.0 ± 0.5°C), salinity (38.0 ± 0.001) and pH (8.0 ± 0.1) were kept constant during the experiment. Refer to Nardelli et al. (2013) for further details on culture settings and preparation of Zn solutions.
As reported in Nardelli et al. (2013), none of the specimens produced new chambers at the highest tested Zn concentration (sea + 100 mg/L), therefore only specimens coming from culture sets from treatments sea, sea + 0.01, sea + 0.1, sea + 1 and sea + 10 mg/L were investigated. Moreover, two samples were treated for a "passive Zn incorporation test": empty tests of P. rotunda were exposed to sea + 10 mg/L Zn concentrations for two weeks in order to measure Zn passively adsorbed to calcite, without involving cellular-mediated mineralization processes.

ESEM and LA-ICP-MS sample preparation
The analyzed foraminiferal tests (n = 41) were washed with Millipore water and dried at 40°C. The same cleaning procedure was performed on the two empty tests (n = 2) used for the passive incorporation test. To perform both ESEM observations and LA-ICP-MS analyses, samples were fixed to aluminium stubs using conductive carbon adhesive discs. All samples analyzed with an LA-ICP-MS were photographed under the ESEM before and after the analyses to check for the success of ablations (i.e., the correct chamber, no multiple chamber sampling, and no breakage of chamberssee Appendix, Fig. A.1a-c).

LA-ICP MS analysis: analytical protocol optimization
This study represents the first LA-ICP-MS investigation on miliolid foraminifera. Chemical composition, micro-structure and chamber arrangements of miliolids strongly differ from other benthic foraminifera more commonly analyzed with the LA-ICP-MS (i.e., Rotaliidae, e.g., de Nooijer et al., 2007;Munsel et al., 2010;Dissard et al., 2010;or Buliminidae, e.g., Hintz et al., 2006;Barras et al., 2010). Miliolid foraminifera have a calcareous non-lamellar imperforate test consisting of calcite needles randomly oriented in an organic matrix. They also possess a smoothly finished outermost layer of well crystallized calcite, with rhombohedral crystal faces arranged parallel to the surface (Debenay et al., 1998). Moreover, Pseudotriloculina rotunda creates chambers, each one-half coil in length, adding the new ones in planes oriented at 120°, with only three final externally visible chambers (Loeblich and Tappan, 1964).
For this reason it was necessary to optimize the existing protocols and to obtain the best ablation setting to be applied to our specimens. In particular, it was compulsory to prevent the laser from ablating the innermost (older) chambers. Several trials on foraminiferal tests were thus performed to find the most suitable combination of the instrument setting parameters. A detailed description of the followed procedures is given in Appendix A, and the values of the optimized parameter used for measurements are reported in Table 1. An example of successful sampling is shown in Fig. 1.

Analytical standards preparation and instrument calibration
A mass spectrometer, like any measurement device, requires a suitable calibration procedure. When laser ablation is employed, the interaction between laser and solid sample is complex and the response is dependent on the sample matrix. For this reason, two forms of calibration are mandatory: i) a reference (internal standard) is required to compensate for changes in the quantity of ablated mass, even when the concentration remains constant; ii) matrix-matched solid standards (frequently referred to as "external standards") are necessary to calibrate laser ablation processes and the instrument response. In fact, a relative measure of ablated mass can be achieved by simultaneously measuring emission from the analyte and a common matrix element (internal standard). For absolute calibration of the LA-ICP-MS conditions, standards made of the same matrix as the samples would be required, but are seldom available (e.g., Darke and Tyson, 1994;Raith et al., 1996;Hathorne et al., 2003). Table 1 Optimized laser parameters. Linear ablation, carried out on the flatter part of the last chamber proceeding always from the center to the aperture of the chamber, was preferred to spot ablation. See Hathorne et al. (2003) for comparison and SI-1 for further information. A pre-ablation step was introduced to clean the surface from potential contaminations before data acquisition. Previous studies on perforate calcareous species of foraminifera (or other organisms with calcareous compounds) used NIST610-611 or NIST612-613 as external standards, together with internal standards (generally 44 Ca isotope) and "in-house made standards" (e.g., Hathorne et al., 2003;Eggins et al., 2003;Montagna et al., 2007;Rathmann and Kuhnert, 2008;Munsel et al., 2010). The same standards were also used in our study in order to test their possible employment also for miliolids. In-house made standards were obtained as hereafter described. Stock solutions with defined Zn concentration were prepared using Zn ICP-standard solutions [1 mg/mL in 2 (vol.%) HNO 3 ] and Millipore water. A proper amount of each solution was added to a mixture composed by 400 mg of ultrapure CaCO 3 powder (particle size less than 1 μm) and cellulose; to prevent CaCO 3 dissolution, the pH of each solution was adjusted to 7.5 ± 0.1 using an appropriate amount of ammonia solution. Each suspension was mixed and homogenized in an agate mortar and then dried at 30°C for 12 h. The resulting powder was then re-homogenized in the agate mortar and pressed at 12 tons into tablets of 12 mm diameter. Such "standard tablets" at different Zn concentrations were then checked via LA-ICP-MS using ablation lines to verify whether Zn distribution was homogeneous. As shown in Fig. 2, the spectra resulting from these analyses revealed a fairly smooth plateau, confirming a homogeneous distribution of the element into the standard tablets.
According to the literature (see above), NIST standards could be used as well. However, the difference in composition of the matrix (i.e., silica glass in NIST standards, Ca carbonate in miliolids) does not match the second requirement mentioned above. However, several measurements using NIST610 were performed in order to test the possibility to use a certified and easily acquired analytical standard. In particular, the same laser parameters (i.e., the same ablation conditions) were applied to both NIST610 and our samples (miliolids) but the response on the internal standard ( 44 Ca) was not satisfactory (see next paragraph for more details).
In the light of these results, and also in agreement with Hathorne et al. (2003), only in house-made tablets were used as external standards for measurements here reported. In detail, nine in house made standards were used with Zn concentrations ranging from 0 (blank, CaCO 3 + cellulose + Millipore) to 1050 ppm.

LA-ICP-MS: analyses of samples and data elaboration
Eleven specimens from "sea", five from sea + 0.01 mg/L, six from sea + 0.10 mg/L, ten from sea + 1.0 mg/L, nine from sea + 10 mg/L, were analyzed by the LA-ICP-MS. Moreover, the two specimens for the "passive sorption tests" were investigated as well. In a few cases more than one ICP-MS analysis was carried out on the same chamber, otherwise one linear ablation per chamber was generally realized. Standards were ablated using exactly the same laser setting parameters used for foraminifera, and their concentrations were regularly measured during sampling procedure, in order to correct instrumental deviation. 66 Zn and 68 Zn isotopes, and 44 Ca were measured. Collision mode using a mixture of helium (95%) and hydrogen (5%) was employed to discriminate the analyte ions from any interfering polyatomic ions on the basis of Kinetic Energy Discrimination (KED). Both for calibration and for sample measurements, Ca and Zn concentrations were calculated integrating over a time-interval of 1.00E + 05 ms in the flattest region of each spectrum of each individual ablation profile using PlasmaLab™ Software Package 2007. For data analyses the relation between Zn/Ca ratios in calcite and seawater was observed. Partition coefficients for Zn were calculated for each set of Zn enriched cultures following the formula D Zn = (Zn/ Ca calcite )/(Zn/Ca seawater ).

ESEM observations
Results from ESEM observations on structure and crystal organization of Zn-exposed foraminiferal tests did not reveal any obvious anomalies. All specimens showed the typical structure of miliolid tests ( Fig. 3a-d), characterized by organized crystals on the external test surface and disorganized crystals in an organic matrix in the inner part of the wall (Hay et al., 1963;Towe and Cifelli, 1967;Haake, 1971;Debenay et al., 1998).
ESEM observations were also performed to adjust laser ablation parameters (see Appendix A) and to check samples for correct ablation after LA-ICP-MS measurements.

LA-ICP-MS measurements
The main results of the LA-ICP-MS analysis are given in Fig. 4(a, b). The Zn seawater concentrations are given in Zn/Ca mmol/mol to facilitate the comparison with Zn/Ca ratios in the calcite. The different treatments (i.e., Zn concentrations) are indicated by the colors. The LA-ICP-MS analysis revealed that Zn/Ca ratio increases linearly with water concentrations at least up to sea + 1.0 mg/L (Fig. 4b). The Zn/Ca shell concentrations vary between 1.11 ± 0.02 mmol/mol for the "sea" treatment and 1.77 ± 0.04 mmol/mol of the treatment sea + 1.0 mg/L ( Table 2; see also Table A.2 in Appendix A for the complete dataset). The measurements obtained on calcite produced at the highest tested Zn concentrations (sea + 10 mg/L), however, with average Zn/Ca values  The passive sorption test on empty shells incubated at sea + 10 mg/ L conditions showed very low Zn concentrations (Zn/Ca calcite = 1.17 ± 0.01 mmol/mol), comparable to the ones of samples from the "sea" treatment) (Zn/Ca calcite = 1.12 ± 0.02 mmol/mol), and more than 3 times lower than living samples from highest tested zinc concentrations (sea + 10 mg/L) (Zn/Ca calcite = 3.83 ± 0.17 mmol/mol) ( Fig. 4 and Table A.2 of Appendix A).
The calculated Zn partition coefficients (D Zn ) for each Zn treatment (indicated by the colors) are given in Fig. 5a. The average D Zn (±standard deviation) is given in numbers. Because of non-homogeneous  variance, the Kruskal-Wallis test and Mann-Whitney pairwise comparisons post-hoc test were performed to determine whether the difference between partition coefficients was significant among treatments (p-value b 0.001). The results showed that the Zn partition coefficients were all significantly different among treatments. The obtained average values of D Zn varied between 4.03 ± 0.06 at "sea" conditions and 0.2 ± 0.01 at sea + 10 mg/L Zn concentration (Fig. 5) (total average D Zn = 2.22 ± 1.56). The observed decrease of D Zn at increasing seawater Zn concentration is well described by a power function (R 2 = 0.997; pvalue b 0.001; see Fig. 5a). However, due to the difficulty to assess whether the results obtained for the highest tested concentration (sea + 10 mg/L) are affected by neglected precipitation of zinc oxides/ hydroxides (see discussion paragraph 4.2) we reported in Fig. 5b only the D Zn obtained at Zn treatments lower than this concentration. Even without the last concentration, the data are still well described by a power function (R 2 = 0.998; p-value b 0.001) and the regression equation slightly change from y = 1.5552x −0.713 to y = 1.4826x −0.775 .

ESEM structural observation of tests
Miliolids are often regarded as pollution sensitive organisms. Several studies on test deformity induced by environmental contamination suggest that miliolids are more easily affected by test deformities than other foraminifera. For example, Samir and El-Din (2001) found that the majority of deformed tests collected in the El-Mex Bay (Egypt) were miliolid shells. Also Sharifi et al. (1991) reported that deformed foraminiferal tests from the Southampton coastal area contained much higher Cu and Zn concentrations than non-deformed specimens, suggesting again a responsibility of these metals for test deformations. However, Nardelli et al. (2013) tested in laboratory conditions the biological effects of several Zn concentrations on P. rotunda and, despite the fact that some specimens calcified new chambers at Zn concentrations up to 10 mg/L, no obvious deformations due to Zn exposure were observed, in terms of chamber arrangements or general shape of the foraminiferal shells. Our ESEM observations confirm these results because the calcite produced during Zn exposure showed the typical aspect and arrangement known for the species (as described by Debenay et al., 1998). So, Zn exposure does not appear to be, by itself, a cause of abnormal calcification for P. rotunda, either at the macro or micro scale. A possible explanation for the fact that Zn is often found in anomalous tests is its possible covariance with other pollutants (for example other metals) or environmental parameters, which can be alone the real cause of test deformation or have synergic behaviors to zinc and enhance its toxicity (as demonstrated, for example, for Co and Zn by Bresson et al., 2013). In fact, even if Zn is one of the most cited heavy metals, potentially responsible for test deformation, many other pollutants and anthropogenic events were related to test deformities, e.g. Cd, Cr and Ti (Yanko et al., 1998), Cu (Le Cadre andDebenay, 2006), hydrocarbons and oil spills (e.g., Vénec-Peyré, 1981;Vénec-Peyré et al., 2010), rapid changes of salinity and hypersalinity (e.g., Eichler-Coelho et al., 1996;Sousa et al., 1997;Stouff et al., 1999;Geslin et al., 2002). The absence of visible test anomalies can be coherent with the hypothesis that Zn incorporation into biomineralized calcite may occur, as already observed on non-biogenic calcites (e.g., Elzinga and Reeder, 2000;Temmam et al., 2000), after substitution of Ca or Mg to form isomorphic Zn carbonates, as suggested by Madkour and Ali (2009).

Zinc incorporation
Despite the absence of deformations or visible abnormalities in the specimens analyzed in this study, high Zn/Ca contents were detected in the internal wall of the last chamber of the test by LA-ICP-MS analysis. The incorporation rates of Zn into the calcite do not seem to be constant at increasing Zn concentrations. Zn incorporation, in fact, linearly increases at lower concentrations (from sea to sea + 1.0 mg/L, Fig. 4b), but the measures obtained at Zn concentrations higher than sea + 1.0 mg/L, even if represented by the results from only one Zn treatment, suggest that at higher concentrations (i.e., sea + 10 mg/L) the trend determined at low concentrations (up to sea + 1.0 mg/L) is no longer respected (Fig. 4a). The decreasing Zn incorporation rate in the shell is reflected by the partition coefficients that are lower for higher Zn seawater treatments (Fig. 5a, b). Considering that we worked at high zinc concentrations we cannot avoid taking into account that zinc oxides/hydroxides precipitation could have been important at the highest zinc concentration we tested (sea + 10 mg/L). In fact, if we roughly estimate the expected Zn 2+ saturation based on the pH of our cultures (8.0 ± 0.1) and neglect all potential influences of kinetics, we obtain a concentration of saturation of 4.37E-05 M, which is 3 times lower than the concentration of zinc we added to this treatment (see Table A.2 in Appendix A). Zinc oxides/hydroxides precipitation could therefore at least partially explain the decrease of Zn incorporation at sea + 10 mg/L condition.
However, this is not the first time that a similar decreasing trend for metal incorporation is observed on benthic foraminifera and this could be also partially due to biological reasons. For example Munsel et al. (2010) report similar observations for incorporated Ni in Ammonia tepida's calcite. The incorporation of the metal is linear at low experimental concentrations and drops down when foraminifera are exposed to Ni concentrations 20-fold higher than natural seawater. The authors suggest that this could be due to biological effects. In fact, according to them, Ni at high concentrations could have toxic effects on foraminiferal cells and inhibit calcification. Toxic potential of some highly concentrated metals could trigger their cellular expulsion or blocking mechanism (for example metallothioneins-mediated) and prevent their incorporation in proportion to their sea water concentrations. This hypothesis is also valid for our study. Nardelli et al. (2013) reported delayed growth rates (i.e., calcification) of Pseudotriloculina rotunda exposed to Zn concentrations higher than sea + 1.0 mg/L and ascribed the reduced calcification rate to biological effects of high zinc concentrations on foraminiferal cell, invoking similar processes. Thus we think that the potential biological influence on the Zn incorporation observed at sea + 10 mg/L in the present study cannot be excluded. Some other possible explanations are hypothesized by Mewes et al. (2014) who also observed an exponential decreasing partition coefficient for Mg/Ca at increasing seawater metal concentrations following a power regression for two rotaliid foraminifera (Ammonia aomoriensis and Amphistegina lessonii): (a) presence of two different CaCO 3 layers, (b) involvement of two different biomineralization pathways. The first hypothesis depends on the possibility, shown for some foraminifera, to precipitate two different calcite phases, with two different Mg/Ca calcite ratios, independently of a migration of the foraminifera through different chemical environments (in the water column or in the sediment). This pathway was described by Bentov and Erez (2005) and particularly for A. lessonii and Orbulina universa by Branson et al. (2013). However, as far as we know, this biomineralization mechanism has not been described and does not fit with the existing calcification models for miliolid foraminifera. In fact the existing biomineralization models for miliolid foraminifera propose two main kinds of calcification mechanisms: miliolid foraminifera form bundles of oriented crystals into intracellular vesicles (e.g., Berthold, 1976;Angell, 1980) thanks to locally increased intracellular pH (de Nooijer et al., 2009). Each crystal is surrounded by an organic matrix, and bundles are passed through the cell membrane by exocytosis (e.g., Berthold, 1976;Hemleben et al., 1986). After exocytosis these crystals are accreted onto extracellular surfaces to form a packed wall structure (Hemleben et al., 1986).
The second hypothesis considers the involvement of different biomineralization pathways of Mg incorporation during calcification. This hypothesis derives from the biomineralization model proposed by Nehrke et al. (2013) for rotaliids foraminifera, but there is no evidence that the same processes occur in miliolid foraminifera. Therefore also this hypothesis does not seem to validly explain our results. Thus, the oversaturation of the highest tested zinc solution and/or the biological effect of potentially toxic zinc concentrations on the biomineralization process remain, to us, the more realistic hypothesis to explain the low zinc incorporation at sea + 10 mg/L treatment.
Apart from the exponential decrease of partition coefficient, another interesting point highlighted by our results is the apparent positive intercept of D Zn (Zn/Ca calcite vs Zn/Ca seawater ) regression (Fig. 4b). This result suggests a difference in Zn incorporation for this miliolid species compared to all the rotaliid foraminifera for which D Zn was previously estimated (e.g. Marchitto et al., 2000;Bryan and Marchitto, 2010). In fact, despite the highly variable D Zn estimated for the different species, Marchitto et al. (2000) and Bryan and Marchitto (2010), report a Zn/ Ca calcite vs Zn/Ca seawater regression line passing from the origin of the axes for all the measured rotaliids foraminifera. Similar results, showing a positive intercept on the y-axis, reported in Mewes et al. (2014) for Mg/Ca of two rotaliid foraminifera (Ammonia aomoriensis and Amphistegina lessonii), were hypothesized by the authors to be possibly due to purely chemical reasons: sorption of Mg 2+ to mineral surfaces would be stronger than sorption of Ca 2+ (Mucci and Morse, 1983) and then increased Mg 2+ (adsorbed on calcite) would locally increase Mg/Ca, and then explain the positive y-axis intercept, especially at low seawater Mg/Ca. Due to its divalent nature and its ionic radii smaller than Ca 2+ , this kind of mechanism is also possible for Zn 2+ (Elzinga and Reeder, 2000). In this regard, the results of our sorption test (Fig.  4) seems to suggest that, at least at the highest tested concentration (sea + 10 mg/L), Zn sorption is negligible compared to the fraction incorporated into the calcite of living foraminifera incubated at the same Zn concentrations. This suggests that this kind of mechanism could have low influence on Zn enrichment of calcite. However, the test was performed only at the highest Zn treatment and it is not enough to draw any definitive conclusion about this aspect.
Another hypothesis, invoked by Langer et al. (2006) to explain the enriched Sr/Ca ratio in the calcite of the coccolithophore Emiliania huxleyi, could also explain the Zn/Ca enrichment we observe for Pseudotriloculina rotunda. Similarly to miliolid foraminifera, coccolithophores precipitate calcite into specific intracellular vesicles. According to the authors, the intracellular calcite precipitation yields an accumulation of Sr in the coccolith vesicle until Sr steady state is achieved. Then, the observed Sr/Ca of the coccolith calcite would be an integral value that arises from the sum of Sr/Ca over the time required for the formation of one coccolith. This is suggested to explain the enriched Sr/Ca ratios found in the coccolithophore compared to theorethical ones expected by purely chemical precipitation. According to us, and taking into account the results of our sorption test, this hypothesis seems more convincing to explain the apparent positive yaxis intercept of Fig. 4.

Implications for reconstructing past seawater zinc concentrations
As zinc is one of the most diffused pollutants related to the industrial activity (e.g., Callender and Rice, 2000;Wuana and Okieimen, 2011) a proxy for its concentrations in historical times could represent an important tool to study the evolution of marine ecosystems under anthropic stress. This is particularly interesting in view of the animated debates on the possible boundary for the recently proposed new epoch of the Anthropocene (e.g., Crutzen, 2002;Zalasiewicz et al., 2011). The accelerated increase of some industrial-derived pollutants is, in fact, one of the proposed points to establish the start of this epoch. Our results suggest a possible reliable identification and interpretation of one of these pollutants in the historical sediment record. In fact, except for the D Zn estimated for the sea + 10 mg/L that could have been influenced by Zn oversaturation problems (as discussed above), we are confident that the D Zn estimated for Pseudotriloculina rotunda in this study can be applied for past seawater concentrations reconstructions. The partition coefficients we obtained are different from the ones previously estimated for other foraminiferal species (mainly rotaliids), but rely on the variability range of values found for different species of rotaliid foraminifera (e.g.~3 for U. peregrina to~22 for C. pachyderma reported by Bryan and Marchitto, 2010) and they represent the Zn/Ca of the seawater where the calcite was biomineralized.
Zn/Ca contents into miliolid calcite could therefore allow reconstruction of concentrations of this metal in the past seawater.
The obtained D Zn for each tested zinc concentrations (Fig. 5a) show statistically significant differences. For this reason we would suggest not to average all the values but rather using the equations given in Figs. 4b and 5b for Zn seawater reconstructions. For the highest tested zinc concentration, for which the estimated D Zn could have been highly driven by zinc oxides/hydroxides precipitation and physiological cell problems due to potential toxic effects of zinc, our results are more difficult to apply and need further investigation to be confirmed. For the use of this proxy in potentially contaminated environments, the possible toxic effect of Zn at high concentrations on P. rotunda could be an important boundary constraint. A possible solution to overcome this problem could be the use of multiple species, with different thresholds of tolerance to zinc.

Concluding remarks
Zn/Ca ratios in calcite of Pseudotriloculina rotunda are a linear function of Zn/Ca in seawater at lower Zn/Ca seawater and appear to turn into a power function at Zn seawater concentrations higher than sea + 1.0 mg/L. Further studies are needed to quantify the possible influence of zinc oxides/hydroxides precipitation on the results we obtained at this higher experimental condition. However, even if we omit the last point from the dataset, the results show that Zn is incorporated in equilibrium with seawater concentrations and that partition coefficients significantly decrease at increasing seawater zinc concentrations following a power function. The result of passive sorption test performed on dead specimens incubated in water with high Zn concentration suggests that most of the zinc incorporation is mediated by the cell and that chemical passive sorption is putatively negligible (Table A.2). These results suggest the suitability of Zn/Ca into foraminiferal calcite as an environmental proxy. Zinc contents of calcite can be in fact considered representative of the concentrations in seawater at the time of their calcification, and relatively independent of post-mortem conditions occurring in the depositional area. The analysis of Zn/Ca contents into the calcite could therefore allow the reconstruction of past zinc concentrations in seawater. Then, present-day knowledge about tolerance limits of foraminifera (e.g., de Freita Prazeres et al., 2011;Nardelli et al., 2013) and other organisms to Zn seawater concentrations would allow us to study potential ecological effects induced by high concentrations of this metal in environmental systems in the past. However further studies would be needed for a more precise calibration of Zn/Ca as a proxy. In particular the study of potential synergic, additive or antagonistic effects of several chemicals and/or of environmental variables (e.g., salinity, temperature, alkalinity) on Zn incorporation should be deeply investigated. Moreover, further calibration will be very useful in the future to fill the gap for concentrations between 1 and 10 mg/L and between 0.1 mg/L and natural unpolluted seawater zinc concentrations (i.e., one order of magnitude lower).