Boron Incorporation in the Foraminifer Amphistegina Lessonii under a Decoupled Carbonate Chemistry

A number of studies have shown that the boron isotopic composition (δ 11 B) and the B / Ca ratio of bio-genic carbonates (mostly foraminifers) can serve as proxies for two parameters of the ocean's carbonate chemistry, rendering it possible to calculate the entire carbonate system. However, the B incorporation mechanism into marine carbonates is still not fully understood and analyses of field samples show species-specific and hydrographic effects on the B proxies complicating their application. Identifying the carbonate system parameter influencing boron incorporation is difficult due to the co-variation of pH, CO 2− 3 and B(OH) − 4. To shed light on the question which parameter of the carbonate system is related to the boron incorporation , we performed culture experiments with the ben-thic symbiont-bearing foraminifer Amphistegina lessonii using a decoupled pH–CO 2− 3 chemistry. The determination of the δ 11 B and B / Ca ratios was performed simultaneously by means of a new in situ technique combining optical emission spectroscopy and laser ablation MC-ICP-MS. The boron isotopic composition in the tests gets heavier with increasing pH and B / Ca increases with increasing B(OH) − 4 / HCO − 3 of the culture media. The latter indicates that boron uptake of A. lessonii features a competition between B(OH) − 4 and HCO − 3. Furthermore, the simultaneous determination of B / Ca and δ 11 B on single specimens allows for assessing the relative variability of these parameters. Among different treatments the B / Ca shows an increasing variability with increasing boron concentration in the test whereas the variability in the isotope distribution is constant.

1. Writing can be improved.For example, the first sentence in the Abstract should be deleted/revised, because it has been shown that, even we can get two parameters from d11B and B/Ca, it is still difficult to define the seawaterCO2 system.This has been well demonstrated previously by Yu et al. (2010) and Rae et al.(2011) both published in EPSL.
The interpretation of proxy data always has to take into account several interfering processes and is never straight forward.Therefore, we wrote "… can serve as proxies for two parameters …" .Nevertheless, the general potential of B based proxies is well demonstrated in many studies.Furthermore, we want to point out that the present study is based on samples from culture experiments and therefore we do not have to deal with unknown "interfering processes".However, the suggestion of the reviewer has been taken into account by adding the following sentence to the abstract: "However, the B incorporation mechanism into marine carbonates is still not fully understood and analyses of field samples show species specific and hydrographic effects on the B proxies complicating their application."2. Another issue is that discussions of literature data are mixed for benthic and planktonic forams.It would be nice to make a clear separation of these two.Relevant publications should be cited, but are missing at present.
As requested by the referee we made a clear separation in the discussion between planktonic and benthic foraminifers.Planktonic species are discussed in lines 296 -341 and benthic species in lines 342 -348 (revised manuscript with track changes).
To the best of our knowledge we cited all relevant publications, but we would be happy to include any paper we might have accidentally omitted.
3. Line 332-334: the reason is not due to symbionts, it is due to the lower pH in deep waters.
The content of lines 332 -334 was corrected according to the referee's comment and the following added "In benthic foraminifers without symbionts (Neogloboquadrina dutertrei, Cibicidoides mundulus, Cibicidoides wuellerstorfi) studied so far a lighter δ 11 B is observed than for planktonic species (Foster, 2008;Rae et al., 2011) due to a lower pH of the growth habitant of benthic foraminifers in deeper waters" in the lines to 342 -345 (revised manuscript with track changes).The following discussion on the possible causes for positive Δδ 11 B values in section 3.2.1.("The role of B(OH) 3 ",) was added."The incorporation of B(OH) 3 could modify foraminiferal δ 11 B (Klochko et al., 2009).This B species always has a heavier isotopic composition than B(OH) 4 -.Therefore, additional incorporation of B(OH) 3 would result in heavier δ 11 B of the foraminifers.Assuming that B(OH) 3 incorporation is positively correlated to B(OH) 3 concentration of seawater, the foraminifers from the pH 8.6 treatment should display the lightest δ 11 B. Contrariwise, this treatment features the heaviest δ 11 B. Therefore, incorporation of B(OH) 3 appears to be unlikely", lines 349 -356 (revised manuscript with track changes).Regression lines and R 2 are added as requested due to the co-variation of pH, CO 3 2-, and B(OH) 4 -.To shed light on the question which parameter of the carbonate system is related to the boron incorporation, we performed culture experiments with the benthic symbiont-bearing foraminifer Amphistegina lessonii using a decoupled pH -CO 3 2-chemistry.
The determination of the δ 11 B and B/Ca ratios was performed simultaneously by means of a new in situ technique combining optical emission spectroscopy and laser ablation MC-ICP-MS.The boron isotopic composition in the tests gets heavier with increasing pH and B/Ca increases with increasing B(OH) 4 -/HCO 3 -of the culture media.The latter indicates that boron uptake of A. lessonii features a competition between B(OH) 4 -and HCO 3 -.Furthermore, the simultaneous determination of B/Ca and δ 11 B on single specimens allows for assessing the relative variability of these parameters.Among different treatments the B/Ca shows an increasing variability with increasing boron concentration in the test whereas the variability in the isotope distribution is constant.The δ 11 B signature of A. lessonii

INTRODUCTION
The oceans carbonate system comprises six co-varying parameters ( total alkalinity (TA), and dissolved inorganic carbon (DIC)).Changes of the carbonate system caused by past changes in the atmospheric pCO 2 can be reconstructed if at least two of these parameters are known.A number of studies have shown that the boron isotopic composition (δ 11 B) and the B/Ca ratio of biogenic carbonates (mostly foraminifers) may serve as proxies that can provide these two parameters.
In seawater boron (B) mainly exists as boric acid (B(OH) 3 ) and borate (B(OH) 4 -).The isotopic composition and concentration of both species are pH dependent (Fig. 1).Since the B isotopic composition of biogenic carbonates precipitated at a certain pH value is similar to that of B(OH) 4 -, Hemming & Hanson (1992) concluded that only B(OH) 4 -is incorporated into biogenic carbonates.
Therewith, the B isotopic composition can be used as a proxy to infer the pH that prevailed during the formation of the biogenic carbonate.However, several studies show a deviation between the B isotopic composition of the biogenic carbonates and B(OH) 4 -of the sea water (Sanyal et al., 1996;Sanyal et al., 2001;Foster, 2008;Rae et al., 2011).This deviation is often explained by physiological processes like photosynthesis and respiration of symbionts (e.g.dinoflagellates) which modify the pH in the micro-environment around the foraminifera (Zeebe et al., 2003) leading to shifts in the B equilibria.Yet another explanation for the observed deviation is that not only B(OH) 4 -is incorporated during the formation of calcium carbonate but to some extend also the isotopically heavier B(OH) 3 (Klochko et al., 2009).To account for physiological effects, species specific calibration experiments have been carried out to be able to apply this proxy and reliably reconstruct seawater pH (Sanyal et al., 2001;Hönisch et al., 2003;Henehan et al., 2013).
While the B isotope composition of biogenic carbonates is used to reconstruct past seawater pH, the B/Ca of foraminiferal calcite is often used to infer past seawater CO 3 2-concentrations e.g.(Yu et al., 2007;Brown et al., 2011).Inherent to all field studies and most experimental studies is that pH and CO 3 2-concentration of natural seawater are correlated.It is therefore impossible to determine which parameter of the carbonate system is in control of B/Ca.Not surprisingly, correlations between B/Ca and pH were described in addition to B/Ca and CO 3 2-concentration (Yu et al., 2007;Tripati et al., 2011).The latter studies are based on field samples, but experimental studies suffer from the same ambiguity if the experimental setup uses a classical carbonate system manipulation, i.e. either DIC or TA manipulation.To identify the parameter of the carbonate system responsible for foraminiferal B/Ca, it is necessary to decouple pH and CO 3 2-concentration.Such an experimental setup will allow for excluding up to five out of the six parameters of the carbonate system.In an experimental study on the relationship between B/Ca and the seawater carbonate system Allen et al. (2012) showed "a competition between aqueous boron and carbonate species for inclusion into the calcite lattice" for Orbulina universa, Globiberinoides ruber, and Globigerinoides sacculifer.In this study we cultured A.
lessonii under conditions in which pH and CO 3 2-concentration were decoupled in order to assess the controlling carbonate system parameter for B incorporation.The simultaneous determination of δ 11 B and B/Ca on single specimens by means of a newly developed technique (based on a femto second laser ablation MC-ICP-MS connected to a fiber optic spectrometer) allows for the first time the determination of the elemental and isotope B variability among single specimens.

Culturing and experimental setup
Live specimens of the benthic symbiont-bearing foraminifer A. lessonii were obtained from a coral reef aquarium at the Burgers Zoo (Arnheim, The Netherlands).SCUBA divers collected approximately 1 kg of sediment containing different species of foraminifers (Ernst et al., 2011).The sediment was transported to the Alfred Wegener Institute (Bremerhaven, Germany) immediately and transferred into a small aquarium (5L) filled with filtered (0.2 µm pore-size) North Sea seawater (NSW).The aquarium was equipped with a circulation pump to supply air and a time switched light source providing a light/dark cycle (12 h/12 h).About 100 specimens of A. lessonii were transferred to well plates containing NSW and placed in a temperature controlled room at 25°C (again exposed to a 12 h/12 h light/dark cycle).After two weeks ~20% of the specimens had asexually reproduced, yielding 10-30 juveniles per specimen.Subsequently, juvenile foraminifers were transferred into petri dishes containing NSW with a dedicated carbonate system (see 2.2.Preparation of culture media).
Each petri dish was placed into one of six boxes each receiving a concentration of pCO 2 that was in equilibrium with the corresponding carbonate chemistry of the prepared NSW media.The supply of pCO 2 was realized by a gas-mixing system producing a constant gas flow of 40 L per hour for each box.Concentration of CO 2 was logged using CO 2 sensors (type FY0D00CO2B10 Ahlborn) and did not deviate by more than 25 µatm from the target-value.In order to avoid evaporation of culture media in the petri dishes, the gas was saturated with water by bubbling it through a fritted wash bottle filled with de-ionized water.The complete experimental setup was placed in a temperature-controlled (25°C) room.Because of heat produced by the lamps the temperature within the boxes containing the petri dishes increased by up to 2°C during the light cycle.Since this holds for all treatments, it did not impair the interpretation of results.Light intensity was 100-150 µmol photons m -2 s -1 .Every third day the culture media was replaced by a freshly opened aliquot from the corresponding batch of culture media, which was stored without headspace at ~3°C.Approximately 24 hours before the culture media was replaced it was filled in a petri dish and placed in the corresponding gas box to equilibrate.Each time when the culture media was replaced, foraminifers were fed with concentrated and sterilized algae Dunaliella salina (20000 cells/ml).Before feeding algae were centrifuged to minimize dilution of the culture media, and exposed to 90°C for 20 minutes after centrifugation in order to reduce bacterial activity in the culture media.Foraminifers grew for three months.Afterwards specimens were harvested, bleached in NaOCl (active chlorine: 4.6%) for six hours, rinsed four times using de-ionized water, and dried for 12 hours at 50°C.For laser ablation analysis specimens were mounted on a glass slide using double sided adhesive tape.
Single, juvenile specimens of a clone were distributed equally between the different treatments to verify whether specimen specific effects on δ 11 B would occur which, however, were not observed after the B analysis.The size of all foraminifers ranged between 400 and 900 µm before specimens were harvested.The morphology of the tests was indistinguishable from the one of specimens grown in the natural habitat.

Preparation of culture media
Six treatments of manipulated NSW were prepared: treatments 1 to 4 had a constant pH but The csys.m file was modified to allow calculations of borate concentrations different from the natural concentration of seawater.The equilibrium constants of Mehrbach (for K1 and K2) and the total scale for pH were chosen.Temperature was set to 27°C, salinity to 32.Calculating the whole carbonate system chemistry requires at least two of its parameters.The input parameters for the pH constant treatments (pH_8.1*)were pH and pCO 2 , for the [CO 3 2-] constant treatments (pH_8.6 640 + pH_8.1 640 and pH_7.9 260 + pH_8.1 260 ) [CO 3 2-] and pCO 2 .The basis for the different culture media was sterile filtered (0.2 µm pore size) NSW enriched in B (using B(OH) 3, chemical purity: > 99.5% ) to a final concentration of ~4 mmol/kg, which is ~10 times the B concentration of natural seawater.The enrichment with B was done to obtain a higher concentration within the test for better B analysis.For each treatment two litres of culture media were prepared and filled without headspace into 50 ml (for the replacement of culture media) and 200 ml (for chemical analysis) gastight, boron free, silicate flasks and stored at ~3°C.

Analysis of the culture media
Since the amount of culture media in the petri dishes containing the foraminifers (which was replaced all three days) was not sufficient for all chemical analysis, approximately 200 ml of each batch of culture media were filled in polypropylene beakers and placed into the corresponding CO 2 box to equilibrate.Even though determining the chemical parameters once would have been sufficient, we performed this procedure bi-weekly to verify that all conditions stayed constant during the experimental period.After ~24 hours salinity and pH of these solutions were measured at in situ conditions and samples were taken for Ca, B, DIC, and TA analysis.Salinity measurements were performed using a conductivity meter (WTW Multi 340i) interfaced with a TetraCon 325 sensor.
Measurements of pH were carried out by means of a combined pH glass electrode (Ectotrode Plus,Metrohm) interfaced to a Radiometer pH-Meter (PHM240).Repeated measurements of buffers show a reproducibility of 0.05 pH units.After calibration (NBS buffer) the conversion to total scale was performed by measuring a Tris/Tris-HCl seawater buffer prepared in accordance with the recipe described in (Dickson et al., 2007).Calcium and B concentrations were determined by a Thermo Elemental (TJA) IRIS Intrepid ICP-OES Spectrometer using Merck 4 (multi element standard) as reference material.The average external error as estimated by multiple measurements of the reference material was ± 3.5%.Total alkalinity was calculated from linear Gran plots (Gran, 1952) after triplicate potentiometric titration (Bradshaw et al., 1981)  Boron isotopic composition of the culture media were analyzed by means of a Thermo ® Element XR, a single collector, sector field, high-resolution inductively coupled plasma mass spectrometer, fitted with a high sensitivity interface pump (Jet pump) as described in Misra et al. (2014).Boron isotopic composition is reported as per mil (‰) deviation from NIST SRM 951a ( 11 B/ 10 B = 4.04362 ± 0.00137) (Catanzaro et al., 1970) where:  (Vogl & Rosner, 2012).Information about sample preparation for analysis can be found in the supplement.

Simultaneous determination of B isotopic composition and B concentration of single tests
For the simultaneous determination of the B isotopic composition and B concentration a Fiber Optics Spectrometer (Maya2000 Pro, Ocean Optics) was connected to the torch of a Thermo Finnigan Neptune multiple-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at the Leibniz University of Hannover.Laser ablation on reference material, NISTSRM 610, and samples was performed by an inhouse build UV-femtosecond laser ablation system based on a regenerative one box femtosecond laser (SolsticSpectra Physics).A detailed description of this methodology can be found in Kaczmarek et al. (accepted).A brief summary of the method is given in the supplement.
The measured intensity for B in a standard is related to of a reference material corresponds to its known B concentration.Based on this relationship the unknown B concentration of a sample can be calculated.However, in our case measurements of the reference material (NISTSRM 610) and samples have not been performed at the same laser repetition rate, hence their B ratio is not proportional.The correction for different laser repetition rates can be realized using an optical spectrometer by the  Longerich et al. (1996).A detailed description of this methodology can be found in Kaczmarek et al. (2015).A brief summary of the method is given in the supplement.For simultaneous determination of B isotopic composition and B concentration a Fiber Optics Spectrometer (Maya2000 Pro, Ocean Optics) was connected to the torch of a Thermo Finnigan Neptune multiple-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at the Leibniz University of Hannover.Laser ablation on reference material, NISTSRM 610, and samples was performed by an inhouse build UV-femtosecond laser ablation system based on a regenerative one box femtosecond laser (SolsticSpectra Physics).A detailed description of this methodology can be found in Kaczmarek et al. (accepted).A brief summary of the method is given in the supplement.

Carbonate system
The determination of pH, TA and DIC of the culture media yielded three parameters of the carbonate system.In theory, any two of these parameters can be used to calculate the entire carbonate system.However, it has been shown that the results can differ depending on the choice of input parameters (Hoppe et al., 2012).To evaluate in how far the choice of input parameters (pH/DIC, DIC/TA, and pH/TA) would affect the calculated carbonate system within the same treatment, calculations have been performed with all three combinations of input parameters.As can be seen from Table S1 for this study the choice of input parameters does not result in significant differences.
Therefore, further discussions and plots are based on the carbonate system calculated from the input parameters pH and DIC.

The B isotopic signature of A. lessoniis' tests
The measured boron isotopic composition of the foraminiferal tests is given in Table 1 (mean values calculated from single measurements of all foraminifers within one treatment) and Table S2 (single measurements of each foraminifer).For the treatments pH_8.1* the boron isotopic composition is identical (~ -32‰) while treatment pH_8.6 640 shows an increase of the boron isotopic composition by 8.5‰.The boron isotopic composition determined for treatment pH_7.9 260 shows a decrease of 3.4‰ compared to the values determined for the treatments pH_8.1*.The results show that the boron isotopic signature is clearly related to pH and independent of the CO 3 2-concentration (Fig. 2). to 1.97‰, see Table 1). in all treatments defined as tThe standard deviation (SD =   Walker (2004) showed a linear increase between size and symbionts in A.lessonii.If larger foraminifers accommodate more symbionts, smaller foraminifers experience less symbiotic activity, which might lead to lighter δ 11 B. However, in our study we do not observe either a correlation between the size of foraminifers and δ 11 B or a correlation between growth rate and δ 11 B (Fig. 4).In our experiment specimens grew for three months reaching a size between 400 and 900µm.Although we observed different growth rates within each treatment, we do not see a correlation between the test size and the boron isotopic composition.If such an effect really exists in A.lessonii, it is very small and not reflected in the boron isotopic composition.Various values for α are reported in the literature.The first theoretical estimate of 1.01194 at 25°C was given by Kakihana et al. (1977) based on reduced partition function ratio calculations using data on molecular vibrations obtained from spectroscopic measurements.Zeebe (2005) showed that the calculation of α is sensitive to the choice of the theoretical methods used to calculate the forces in the molecule and to molecular vibration frequencies.Based on these observations α is suggested to vary between 1.02 and 1.05 (at 25°C).An experimental study was performed by Klochko et al. (2006) using spectrometric pH measurements in order to determine α from differences in the pK B of

δ
From Figure 4 it can be seen that δ 11 B of the foraminifers is lighter than δ 11 B of B(OH) 4 except for treatment pH_8.6 640 .Since the boron isotopic composition of the culture medium (δ 11 B sw ) used in our experiments was measured regularly (every two weeks) and did not change significantly during the experiment, an offset in δ 11 B caused by a changing isotopic composition of the culture media can be excluded.In the following we will discuss various reasons which may explain the offset assuming that α is 1.0272.

The Boron standards
One factor that could explain the offset between δ 11 B foram and δ 11 B B(OH)4-is the usage of two different standards.From equation 3 it can be seen that δ 11 B of the culture media is needed to calculate δ 11 B of B(OH) 4 -.We used NIST 610 for the determination of δ 11 B foram and NIST 951 for the determination of δ 11 B culture media .In order to compare δ 11 B foram and δ 11 B B(OH)4-, the boron isotopic difference in these standards have to be taken into account.NIST 610 and NIST 951 were compared by several studies (Kasemann et al., 2001;le Roux et al., 2004;Fietzke et al., 2010).The results of these studies show that both standards are on average within errors isotopically equal.Therefore, the usage of them cannot explain the deviation between δ 11 B foram and δ 11 B B(OH)4-seen in the pH_8.1*treatments.

Vital effects
The most widely discussed reason for the δ 11 B offset between foraminifers and B(OH) 4 -are the physiological processes involved in the calcification process, the so called vital effects.
In For planktonic foraminifers symbiont activity strongly influences the pH near the surfacein their micro-environment of the foraminifers (Rink et al., 1998;Zeebe et al., 2003) and impacts affecting the δ 11 B signature of the test.The photosynthetic activity of symbionts consumes CO 2 leading to a pH increase while symbionts' respiration generates CO 2 leading to a pH decrease within the micro environment around the foraminifer.In theory, acidification of the microenvironment due to respiration and calcification would result in lighter δ 11 B of the test whereas consumption of CO 2 by photosynthesis leads to heavier δ 11 B. The net impact of these different processes depends on their respective rates (Zeebe et al., 2003).The effect of photosynthesis on δ 11 B in two planktonic species of foraminifers was studied by Hönisch et al. (2003).Based on a comparison between the field grown, symbiont-bearing species Orbulina universa and the symbiont-barren Globigerina bulloides Hönisch conditions.The impact of photosynthesis on δ 11 B was also studied by Zeebe et al. (2003) based on a model approach which also includes the data of Hönisch et al. (2003).The diffusion-reaction model of Zeebe et al. (2003)  O.universa with a shell diameter of 554 µm at 700 µmol photons m -2 s -1 .The photosynthesis data for A.lessonii in the study of Walker (2004) is normalized to the surface area and is ~ 3.5 nmol O 2 s -1 mm 2 at 700 µmol photons m -2 s -1 (Fig. 19 in the study of Walker (2004)).Based on the round shape (sphere) of O.universa we first calculated the surface area (A = 4πr 2 ) of the sphere using a shell diameter of 554 µm and then normalize the photosynthesis rate to second per mm 2 as performed by Walker (2004).
The comparison between the photosynthesis of O.universa (32557 nmol O 2 s -1 mm 2 ) and A.lessonni shows that symbiont O 2 production and therefore photosynthesis is lower for A.lessonii.Walker (2004) showed that in A.lessonii photosynthesis reaches its maximum at 170 µmol photon m -2 s -1 .We used 120 µmol photons m -2 s -1 which might have led to weak light limitation, further decreasing O 2 production.Thus it is likely that O 2 production in our A. lessonii specimens was at least three orders of magnitude lower than in the O. universa specimens analysed by Hönisch et al. (2003) and Zeebe et al. (2003).We hypothesize that respiration and calcification (counteracting photosynthesis) are of relative greater importance in A. lessonii than in O. universa.The latter assumption explains why δ 11 B values of A. lessonii are closer to symbiont-barren species than the ones of O. universa.
In benthic foraminifers without symbionts (Neogloboquadrina dutertrei, Cibicidoides mundulus, Cibicidoides wuellerstorfi) studied so far a lighter δ 11 B is observed than for planktonic species (Foster, 2008;Rae et al., 2011) due to a lower pH of the growth habitant of benthic foraminifers in deeper waters.Respiration and calcification of benthic foraminifers are the dominant processes leading to an acidification in the micro environment.In support of this inference Glas et al.
(2012) showed that the micro-environment pH of the symbiont-barren benthic species Ammonia spec.

The role of B(OH) 3
Another possible contribution of shifting foraminiferal δ 11 B is t The incorporation of B(OH) 3 could modify foraminiferal δ 11 B (Klochko et al., 2009).This B species always has a heavier isotopic composition and than B(OH) 4 -.Therefore, additional incorporation of B(OH) 3 would result in heavier δ 11 B of the foraminifers.Assuming that B(OH) 3 incorporation is positively correlated to B(OH) 3 concentration of seawater, the foraminifers from the pH 8.6 treatment should display the lightest δ 11 B.
Contrariwise, this treatment features the heaviest δ 11 B. Therefore, incorporation of B(OH) 3 appears to be unlikely.

The variability in δ 11 B
A significant variability in δ 11 B between specimens from the same treatment was reported by Rollion-Bard & Erez (2010).The authors used a different appproach to evaluate the natural variation variability in δ 11 B within the test of Amphistegina lobifera.Instead of the standard deviations they calculated the difference between the heaviest and lightest δ 11 B value (Δδ 11 B), a method which overvalues data points outside the confidence interval.Using the latter approach Rollion-Bard & Erez (2010) These authors described Δδ 11 B (the difference between the heaviest and lightest δ 11 B values) to be pH dependent in Amphistegina lobifera.In their study the Δδ 11 B increased from 4.7‰ for foraminifers cultured at a pH of 8.45 to 12.2‰ for foraminifers cultured at a pH of 7.9.This variability is explained in terms of a calcification mechanism based on sea water vacuolization.It should be noted that the spot size of the analytical method they used to measure the δ 11 B of the test (secondary ion mass spectrometry (SIMS)) was ~30 µm.This would require that areas, of at least this size, exist within the test, which are formed from vacuoles of the same pH.The latter is unlikely since the authors suggest themselves that the vacuoles cover a pH range starting at the bulk pH and ending with pH 9. Since in their study only a small portion of the test was grown under experimental conditions, the question arises whether the determined Δδ 11 B would be the same if the whole test had been grown under experimental conditions.Furthermore, the hypothesis that seawater vacuolization is the only source for calcification in foraminifers is controversially discussed (Nehrke et al., 2013).We calculated Δδ 11 B from our data as done in the study of Rollion-Bard & Erez (2010) We discussed above several mechanisms that could cause the offset of δ 11 B of A.lessonii from the theoretical value expected under the assumption that only B(OH) 4 -is taken up into the test.Even though a combination of these mechanisms could explain the observed offset, they would have to operate with different magnitudes in different specimens (even for specimens from exactly the same treatment) to be in accordance with the observed variability.The latter is very unlikely and therewith no explanation on the observed offset can be given at this point.However, it is interesting to notice, that for all experimental conditions the same variability between specimens is observed.Variability between specimens is documented for the uptake of other elements like e.g.Mg.This points towards a mechanism inherent to the biomineralization process itself, which is responsible for the observed variability.
In our study the δ 11 B values of the benthic A. lessonii are lighter than the δ 11 B values of B(OH) 4 -.
Thus, an incorporation of B(OH) 3 at pH 8.1 cannot explain the measured values and seems unlikely.
Even though, there is on average no offset at pH 8.6 incorporation of B(OH) 3 seems unlikely, too, as with increasing pH the concentration of B(OH) 3 decreases.

Boron enrichment of the culture media
We increased the boron concentration of all treatments ten times compared to the natural concentration of seawater.Borate has got a strong buffer capacity and its concentration is nearly doubled in the pH_8.6 640 treatment (compared to pH_8.1*).This leads to a stronger compensation of the acidification in the microenvironment (Zeebe et al., 2003).This, in turn, could explain why foraminiferal samples from the pH_8.6 640 treatment show on average a lower offset between δ 11 B foram and δ 11 B B(OH)4-compared to samples from the pH_8.1*and pH_7.9 260 treatments.

The B/Ca of A. lessonii
The B/Ca data of the foraminiferal tests plotted against pH and [CO 3 2-] of the culture media is shown in Figure 5.No correlation between the plotted parameters is observed.In a culture study of Allen et al. (2011) it was shown that the pH of culture media and B/Ca of foraminiferal tests are positively correlated.An increase of pH is associated with changes in the carbonate system: The concentrations of CO 3 2-and B(OH) 4 -increase with increasing pH while the concentration of HCO 3 decreases.Because of these coupled processes it is, in the framework of a classical carbonate system perturbation study like the one of Allen and co-workers (2011), not possible to identify the causal agent.In a second study Allen and co-workers (2012) suggested based on data from a culture study on three different planktonic foraminiferal species using a decoupled carbonate chemistry a "competition between aqueous boron and carbon species for inclusion into the calcite lattice".To further elaborate on this hypothesis we plot our B/Ca data against several possible candidates (B(OH) 4 -/CO 3 2-, B(OH) 4 -/HCO 3 -, and B(OH) 4 -/DIC) .The best correlation is given when B/Ca is plotted against B(OH) 4 -/HCO 3 -(Fig.6).This is in good agreement with the data shown in the publication of Allen and co-workers

Further observations
At this point we would like to draw the attention of the reader to two interesting observations within our data which cannot be elaborated further within the framework of this study, but that represent an interesting basis for further investigations.1) Since both parameters (δ 11 B and B/Ca) were determined simultaneously, the question arises whether a correlation between both parameters can be identified.As can be seen from Figure 7 no preference for the incorporation of the lighter or heavier B isotope as a function of the B concentration in the tests is observed.2) It could be observed that the standard deviation for B/Ca does show a significant increase with increasing B incorporation (Fig. 6).

CONCLUSION
Culture experiments based on a decoupled pH and CO -is incorporated into the shell and a strong inter-specimen variability is observed.We evaluated potential processes responsible for these observations such as test size, vital effects, and incorporation of B(OH) 3 .However, we found that none of these processes, or a combination of them, can explain the observed variability in the offset between specimens.

TABLES Table 1
Mean values of the B isotopic composition and B/Ca of A. lessonii.Errors are expressed as SD (SD = ∑ ̅ information about N is given in the supplement).Also listed are the calculated isotopic composition of B(OH) 4 -(using eq. 3 and based on a calculated carbonate system using pH and DIC as input parameters) and the offset between the isotopic composition of foraminifers and B(OH) 4 -(Δδ 11 B).(Klochko et al., 2006)), and ɛ = (α -1) x 1000.In order to calculate Δδ 11 B the isotopic difference between NIST 610 (reference material to determine δ 11 B foram ) and SRM 951 (reference material to determine δ 11 B CM ) has to be taken into account.As shown by several studies (Kasemann et al., 2001;le Roux et al., 2004;Fietzke et al., 2010) both standards are within analytical uncertainty isotopically equal.

Figure 43
Size and growth rate (defined as final size divided by the number of days in culture) versus B isotopic compositions of foraminifers.If a specimen was measured several times the mean δ 11 B is presented here.Error bars of single δ 11 B values represent 2RSE and were calculated according to eq.S3.

4.
Figures require some further work.For example, Fig 2b & 5b, the unit for CO32-should be umol/kg; 4. Units in figures 2b and 5b are corrected.

Fig
Fig. 6b, c: add regression lines and R2 values.
using a TitroLine alpha plus auto sampler (Schott Instruments).Culture media samples were calibrated against an in-house standard (NSW) which is calibrated regularly against certified reference material batch No. 54 of Dickson (Scripps Institution of Oceanography).The average reproducibility is ±10 µmol/kg.Determination of DIC was performed photometrically in triplicates with a TRAACS CS800 QuaAAtro autoanalyzer with an average reproducibility of ±10 µmol/l based on calibrations of an in-house standard (NSW) calibrated against Certified Reference Material Batch No. 54 of Dickson (Scripps Institution of Oceanography).
δ 11 B data between treatments Under the general assumption that B(OH) 4 -is the only species incorporated into the test of foraminifers, δ 11 B of the test should equal the δ 11 B of B(OH) 4 -.Therefore, theoretically the offset between both (δ 11 B foram -δ 11 B B(OH)4-) should be zero.FFigure 3 shows the offset from the theoretical δ 11 B of B(OH) 4 -for each specimen shows the difference in δ 11 B (defined as Δδ 11 B: δ 11 B foramδ 11 B B(OH)4-) between each specimen and B(OH) 4 -and the inter-specimen variability in δ 11 B foram .It can be seen that most foraminifers grown at a pH of 7.9 and 8.1 show an offset towards more negative δ 11 B values.Foraminifers grown at a pH of 8.6 are shifted towards more positive δ 11 B values.The inter-specimen variability in δ 11 B variabilityspans a range of ~ 7‰ in δ 11 B Foraminifers from for foraminifers within the same treatment (the standard deviation for one foraminifera ranges from 1.20 is given in the supplement)show is a nearly constant and is on average ~ 1.4 ‰.variation in their δ 11 B values.The standard deviation (SD = ∑ ̅ information about N is given in the supplement) of the measured foraminiferal δ 11 B reflects the natural variation in the δ 11 B for the different treatments.The standard deviations of the foraminiferal δ 11 B values for the different treatments are nearly identical (~1.4‰) (see

-
11 B of the test versus δ 11 B of B(OH) 4 If the assumption that only B(OH) 4 -is incorporated into marine carbonates is correct, than the δ 11 B of the foraminifers (δ 11 B foram ) should equal that of B(OH) 4 -of the culture media (δ 11 B B(OH)4-).The comparison between δ 11 B foram and δ 11 B B(OH)4-is shown in Figure 4.One parameter needed to calculate δ 11 B B(OH)4-is the boron fractionation factor α, which is defined as: photosynthesis and respiration are the major processes causing deviations in foraminiferal δ 11 B, foraminifers with high symbiont activity (like O. universa) should record heavier δ 11 B values whereas symbiont-barren foraminifers (like G. bulloides) should record lighter δ 11 B values.In the same study Hönisch et al. (2003) also investigated the impact of symbionts on δ 11 B within one species.From culture experiments with O. universa (using culture media with a similar B concentration as used in this study) the authors report δ 11 B values to be 1.5 ‰ heavier under high light than under low light

( 2012 )
for cultured G. sacculifer, G.ruber, and O. universa.To summarise: if pH and subsequently [B(OH) 4 -] increase in the culture media, [HCO 3 -] decreases resulting in less competition for B(OH) 4 for uptake into the foraminifer's test.In a natural system the competition between B(OH) 4 -and HCO 3 -support the underlying concept of the B/Ca proxy: the observed linearity of foraminiferal B/Ca and [CO 3 2-] can be inferred from the inverse correlated relationship between [B( 3 2-chemistry indicate that the δ 11 B of the test of A. lessonii is related to pH whereas the B/Ca of the foraminiferal shells show a positive correlation with B(OH) 4 -/HCO 3 -.The latter observation suggests a competition between B(OH) 4 -and HCO 3 -of the culture media for B uptake into the test.Furthermore, we observe a natural variability in δ 11 B of ~5‰ (Δδ 11 B) in the tests which seems to be independent of the carbonate chemistry.The B isotopic composition of the tests is lighter than the one of B(OH) 4-at pH 8.1.We conclude that the effects of calcification and respiration on δ 11 B dominating over the effects of photosynthesis are responsible for the offset between δ 11 B foram and δ 11 B B(OH)4-.The δ 11 B values determined on single tests of foraminifers show an offset from the values expected if only B(OH) 4 FIGURE CAPTIONS

Figure 2 ( a )
Figure 2 (a) Boron isotopic composition versus pH of the culture media for all treatments.δ 11 B data represent mean values obtained from single measurements within one treatment.Error bars for δ 11 B represent SD.(b) Calculated carbonate ion concentration versus pH.

Figure 3
Figure 3 Difference between measured δ 11 B in foraminifers and calculated δ 11 B of B(OH) 4 -(y-axis) plotted against measured foraminiferal δ 11 B. The solid black line represents the B isotopic composition of B(OH) 4 -.Error bars of single δ 11 B values represent 2RSE and were calculated according to eq.S3.The δ 11 B of B(OH) 4-was calculated by(Zeebe & Wolf -Gladrow, 2001):δ B δ B B ε B OH B OH α B OHWhere δ 11 B CM and [B CM ] are the δ 11 B and B concentration of the culture media, α B is the B isotope fractionation factor between B(OH) 3 and B(OH) 4 -(α B = 1,0272(Klochko et al., 2006)), and ɛ = (α -1) x 1000.In order to calculate Δδ 11 B the isotopic difference between NIST 610 (reference material to determine δ 11 B foram ) and SRM 951 (reference material to determine δ 11 B CM ) has to be taken into account.As shown by several studies(Kasemann et al., 2001; le Roux et al., 2004;Fietzke et al., 2010) both standards are within analytical uncertainty isotopically equal.

Figure 5 (
Figure 5 (a) B/Ca plotted against pH of culture media and (b) B/Ca plotted against [CO 3 2-] of culture media.Both graphs show no correlation neither with pH nor with [CO 3 2-].B/Ca data represents mean values of all measurements of foraminifers.Error bars are expressed as SD.

Figure 7
Figure 7 collection of Ca on the two high intensity first order emission lines of Ca II at 393.48 and 396.86 nm in cps.The detection of Ca intensities of NISTSRM 610 and samples (whose Ca concentrations are known: [Ca] of NISTSRM 610 is 8.45%, [Ca] of CaCO 3 is 40%) allows to correct for different laser repetition rates as described in Table 1).offset difference of the δ 11 B foram from the δ 11 B of B(OH) 4 -inthe pH_8.6 treatment is shifted towards higher values.In the following we address two questions: (1) what causes the differenceoffset between theof δ 11 B fo ram of foraminifers from the Hönisch & Hemming (2004)at the δ 11 B of foraminifers is related to its test size.isanadditional factor influencing the δ 11 B of foraminifers.Hönisch & Hemming (2004)report heavier δ 11 B by 2.1 to 2.3‰ for individuals of Globigerinoides sacculifer in the sieve size class 515 -865 µm than for shells in the 250 -380 µm size class.This observation is explained by a reduced photosynthetic activity in smaller specimens at greater depth.A study by Zeebe et al. (2003)3)the carbonate chemistry and B equilibrium caused by vital effects in the micro-environment of O.universa.Based on this model changes in δ 11 B due to different symbiont activities (as observed for high light and low light in the culture study ofHönisch et al. (2003)) can be calculated.In general, the calculated changes in δ 11 B are in good agreement with the changes observed in the cultured O.universa.Furthermore, the model showed that the δ 11 B of O. universa cultured at high light is heavier than the δ 11 B of B(OH) 4 -in the culture media, whereas at low light the opposite is reported.Amphistegina lessonii is a symbiont-bearing species.The δ 11 B values of the this species are lighter than those of B(OH) 4 -, a fact which is seemingly at odds with the conclusions ofHönisch et al. (2003)andZeebe et al. (2003).In order to shed light on the question whether symbiont activity may explain the lighter δ 11 B values in our study (as opposed to O. universa) we compare photosynthesis rates (nmol O 2 h -1 foraminifer -1 ) of O. universa (Rink et al., 1998) and A.lessonii (Walker 2004).Rink et al. (1998) reported a net photosynthesis of 8.72 nmol O 2 h -1 for Δδ 11 B ) and pH_8.6 640 treatments exhibits a shift of δ 11 B by only 0.79 ‰ and is lower than the error of a single foraminiferal measurement (2RSE, formula S3, supplement).Based on the Δδ 11 B in our treatments (see above) we do not observe a correlation between Δδ 11 B and pH in A. lessonii.