Single chamber Mg/Ca analyses of Globigerinoides ruber for paleo-proxy calibration using femtosecond LA-ICP-MS

Mg/Ca is an independent proxy in paleoceanography to reconstruct past seawater temperature. Femtosecond Laser Ablation Inductively Coupled Plasma Mass Spectrometry (fs-LA-ICP-MS) was employed to determine the Mg/Ca composition of tests (shells) of the planktic foraminifer species Globigerinoides ruber albus (white chromotype) and G. ruber ruber (red/pink chromotype) sampled alive from the temperate to subtropical eastern North Atlantic with the research sailing yacht Eugen Seibold. Mg/Ca data are compared to (i) the measured in-situ temperature of ambient seawater, (ii) average mixed layer temperature, and (iii) sea surface temperature (SST). The pooled mean chamber Mg/Ca from each plankton tow site exhibits a positive relationship with SST. Two chamber-specific calibrations are derived, which are consistent with previous calibration equations for comparable paleo-archives. The results confirm fs-LA-ICP-MS as reliable method for determining Mg/Ca in G. ruber, and both the penultimate and antepenultimate chambers of adult specimens may provide comprehensible Mg/Ca temperatures of the surface ocean.

proxies of seawater of the low-latitude surface oceans in addition to wet-chemical methods applied to entire tests, as proposed by Nürnberg et al. 12 .
Foraminifers are major marine archives for climate reconstruction and paleoceanography, owing to the high fossilization potential of their calcareous shell and omnipresence in calcareous sediments of the ocean basins over the past 100 Myrs.The chemical composition of their chambered test is affected by the chemical composition and physical parameters of the environment they live in 1 .These marine protozoans live in the surface to sub-thermocline layer of the open ocean and deep marginal seas 1 .Their shell usually consists of low-Mg calcite layers orientated in a radial structure (Arns et al., 2022, and references therein) 19 .
The temperature dependence of the partitioning of Mg into ontogenetic foraminifer calcite is the basic principle of Mg/Ca palaeothermometry 20 .Mg/Ca is affected by temperature and other factors such as salinity, pH, and ∆CO 3 2− 12, 21,22   .Mg/Ca tends to increase with increasing salinity 12,21 and decreasing pH 3,21 .In addition, alternating high and low Mg/Ca bands within the chamber walls of single specimens are assumed to result from diurnal changes in Mg uptake 23,24 .
Traditionally, Mg/Ca thermometry relies on the bulk analysis of samples comprising 10-30 foraminifer tests of the same species providing averaged temperature signals incorporated at different habitat depths 5,14,16,25 .
Recently new techniques such as LA-ICP-MS were developed and facilitate investigation of the underlying mechanism of differences in Mg/Ca composition by profiling Mg/Ca distribution across individual chambers of the same test over a month-long (on average) life cycle [26][27][28] .
In this study, fs-LA-ICP-MS was employed to determine and compare the Mg/Ca composition of different chambers of the spinose, symbiont-bearing planktic foraminifer G. ruber at high resolution across the walls of the last three chambers (final, F0; penultimate, F-1 (final minus one); and antepenultimate, F-2) of individual foraminifer tests.Due to the shallow dwelling depth, G. ruber constitutes an excellent archive for studying past hydrological changes in tropical and subtropical surface oceans 1,5,15 .Globigerinoides ruber were collected onboard S/Y Eugen Seibold during three research cruises in 2019 in the Madeira Basin.Finally, new chamber-specific equations for the relationship between Mg/Ca and ambient water temperature are presented and compared to existing Mg/Ca-temperature equations.

Methods
Sampling and sample processing.Samples were collected in the subtropical eastern North Atlantic with the S/Y Eugen Seibold during cruises ES19C08, ES19C12, and ES19C14 in 2019 (Table 1) with a bongonet and a multinet (both Hydrobios ® , Germany) from different water depth intervals (Table 2).The samples were preserved in a hexamine buffered 3% formalin solution at pH > 8.2 immediately after sampling.Ambient temperature was measured using Temperature-Depth (TD48, Sea&Sun Technology ® , Germany) and Conductivity-Temperature- Depth (CTD75M, Sea&Sun Technology ® , Germany) probes attached to the bongonet and multinet, respectively (Table 2).
In the micropaleontology laboratory at the Max Planck Institute for Chemistry (MPIC, Mainz), the plankton net samples were rinsed with one liter of tap water.The planktic foraminifers were picked from the solution with a glass pipette, dried at room temperature, and identified at the species level according to Schiebel and  Hemleben (2017).For Mg/Ca analyses, 21 clean individuals of the planktic foraminifer G. ruber were selected from the >100 μm size fraction.Additional chemical (reductive and oxidative) cleaning procedures were not applied to the tests to not alter the original and pristine shell material.The foraminifer tests were glued (Tylopur MOBS 4000, ShinEtsu, 1:100 additive-free methylhydroxypropylcellulose (MHPC), Wiesbaden, Germany) onto a glass slide with the spiral side, with the umbilical side facing up and exposing the final three chambers for LA-ICP-MS analyses 29 .
Laser ablation, mass spectrometry, and Mg/Ca temperature calculation.Magnesium and calcium of 19 specimens of G. ruber albus and two G. ruber ruber tests were measured at high-resolution on either two or three chambers of each test using a 200 nm wavelength NWR femtosecond laser ablation system (NWRFemto) from Electro Scientific Industries (ESI, New Wave Research Division, Portland, USA) combined with a ThermoFinnigan high-resolution sector-field ICP-MS Element2 mass spectrometer 27,30,31 .
Femtosecond LA-ICP-MS (fs-LA-ICP-MS) analyses were performed on 45 μm diameter laser spots on each chamber with a pulse repetition rate (PRR) of 1 Hz at low fluence of 0.1 J/cm 2 27 .MACS-3 was used as reference material for calibration at the beginning of analyses of the objects and every 2 hours in the following, i.e., after each 27 to 30 measurements of the analyzed objects 32 .MACS-3 is a homogeneous pressed powder pellet consisting of synthetic calcium carbonate powder provided by the United States Geological Survey (USGS), used for calibration with the mass fractions of Mg = 1756 μg/g and Ca = 37.69% m/m 33 .
The ablated material was analyzed with a single collector sector-field SF mass spectrometer (ThermoFinnigan Element2), with electric and magnetic fields, operated at low mass resolution mode.Employing fast (<0.001 s) electric jump, the double-charged 44 Ca 2+ , which is within a similar mass range (m/2e = 22) as 25 Mg + (m/e = 25) was measured 27 .This adjustment shortens measurement time and improves the measurement precision.The disadvantage of this technique is the difference between the masses of interest, which must not be more than 30% apart 27 .Accordingly, 44 Ca and 25 Mg measurements were carried out on the less abundant double-charged 44 Ca rather than on the single charged Ca ions 27 .
Raw fs-LA-ICP-MS data were evaluated with an automated Microsoft Excel application, at a minimum of 44 Ca 2+ and 25 Mg + count rates per second 27 .The washout time of the ESI Large Format Cell is one second, thus the individual peaks were separated due to their shorter lengths of about 0.9 seconds.Data calibration was conducted on the MACS-3 reference material during each session.Obvious outliers and unusually high trace element abundances at the beginning of the ablation (1-3 seconds) were identified and rejected by the Microsoft Excel routine aiming at excluding data from potential surface contamination 27,34 .Ablation profiles (see Supplement) were optically inspected for verification of the different ablation intervals and to only interpret data from the ontogenetic calcite.In order to correct for interferences, background data were subtracted from the individual ion intensities.The background-corrected count rates of 25 Mg were divided by the background-corrected count rates of 44 Ca for each scan 34 .The trace element concentration C El (μg g −1 ) is given by El El,uncorr with C El,uncorr being the apparent (uncorrected) concentration, and RSF being the relative sensitivity factor 34 .C El,uncorr is determined by: with C IS being the concentration (μg g −1 ) of an internal standard element (IS), R ik being the ratio of the ion intensities of the isotope i of the element of interest (EL), and of isotope k of the internal standard element.A i and A k are the isotopic abundances of isotope i and k, respectively 34 .M EL and M IS are the relative atomic masses of the element of interest and the internal standard element, respectively 34 .The RSF is determined by: El,uncorr El,true with C El,uncorr being the uncorrected concentration of the element El obtained by reference to the internal standard element, and C El,true being the 'true' concentration in a reference material 34 , with values (mass fractions in mg kg −1 ) for MACS-3 of Mg = 1756, and Ca = 376900 33 .
The measurement precision (1 relative standard deviation in percent, 1RSD, 1 s) of the Mg/Ca, is determined by repeated measurements of homogeneous calcium carbonate reference materials.Six to 12 independent analyses of the pressed power pellets of MACS-3 yield uncertainties (1 RSD) in Mg/Ca between 4.96% and 7.34%.
The 3-standard deviation and the mean of the blank estimates the limit of detection (LOD) 27 .For calcareous samples, such as foraminifer tests, the Ca content is high and uniform at about 40% m/m 27 .Therefore, the LOD of Mg/Ca mainly depends on the LOD of Mg ranging at 0.4-1.2mg kg −1 23 .with a = 0.38 and b = 0.09 for G. ruber 5,15 .

Data Records
The raw (MgCa_per_laser_spot.tab) and averaged (MgCa_avg.tab)Mg/Ca data are available at https://doi.org/10.17617/3.D9FSSN 35 on the Edmond Open Research Data Repository of the Max Planck Society, Germany.All column headings of the metadata (sample ID, latitude, longitude, time of sampling), measured Mg/Ca data, and statistical data, i.e., 1σ (1RSD, 'Error') are detailed in the headers of the data files.Temperature (°C) of ambient seawater at Fig. 3 Mg/Ca of the antepenultimate chamber (F-2) calcite of G. ruber per measured temperature.Relationship between measured Mg/Ca (ln scale) of the antepenultimate chamber (F-2) and SST, surface mixed layer (MLD, mixed layer depth), and in-situ temperatures.Error bars are in 1σ.Sensitivities of Mg/Ca vs. SST, average surface mixed layer temperature, and in-situ temperature are 11 ± 4%, 10 ± 4%, and 4 ± 7% per 1 °C, respectively.Fig. 4 Calculated Mg/Ca temperatures of the penultimate chamber (F-1) G. ruber per measured temperatures.Comparison of calculated Mg/Ca temperature (°C) after Dekens et al. 15 and Anand et al. 5 versus measured temperatures, SST, surface mixed layer (MLD), and in-situ.Black line marks 1:1 relationship.Fig. 5 Calculated Mg/Ca temperatures of the antepenultimate chamber (F-2) G. ruber per measured temperatures.Comparison of calculated Mg/Ca temperature (°C) after Dekens et al. 15 and Anand et al. 5 versus measured temperatures, SST, surface mixed layer (MLD), and in-situ.Black line 1:1 relationship.Fig. 6 Paleothermometers developed in the present study compared to other Mg/Ca-paleothermometers. Two thermometers (Lines 7 and 8) derived here from Mg/Ca data of the penultimate chamber and SST data (7) as well as from the antepenultimate chamber (F-2) and mixed layer temperature data (8) of G. ruber, compared to the findings of ( 1 the time of sampling (also Table 2, T in-situ) is given for comparison with the calculated temperature of the measured Mg/Ca of the shell carbonate of the respective planktic foraminifer species G. ruber (white) or G. ruber (red).
Sample identification (ID) in the data tables numbers provide information per expedition of the sailing yacht Eugen Seibold (ES), and year of sampling (e.g., ES19 in 2019).Cruise numbers are consecutive and start with cruise number one each year (e.g., C08 for cruise ES19C08).The cruise number is followed by the number for station, cast, and sample with a consistent format.Station numbers have three digits (e.g., 008 in ES19C08_008).Cast labels have two digits (e.g., 01 as in ES19C08_008_01).Sample labels have one digit (e.g., 1 as in ES19C08_008_01_1).Hence, the label ES19C08_008_01_1 identifies a sample taken onboard the S/Y Eugen Seibold, in 2019, on Cruise 8, at Station 8 in the same year, obtained with the first cast at station, and subsampled from the first multinet employed at the same station (Table 2).

technical Validation
Relationship of Mg/Ca data between the different chamber F0, F-1, and F-2 of G. ruber.The majority of Mg/Ca data (MgCa_per_laser_spot.tab) range at 1-6 mmol/mol and are considered reasonable (Fig. 1).The high Mg/Ca values above 20 mmol/mol of the three chambers of one G. ruber (pink) individual from ES19C12_012_06 (Table 2) are considered not realistic and may not be used for further analyses.F-test and t-test show that the remaining Mg/Ca data from the F-1 (x = 3.74 ± 0.58, n = 16) and F-2 (x = 3.79 ± 1.00, n = 17) chambers are statistically similar, and significantly different from F0 (x = 2.71 ± 0.99, n = 17).
Relationship between single chamber-Mg/Ca values of G. ruber and measured water temperature.The Mg/Ca of the penultimate (F-1; Fig. 2) and antepenultimate chambers (F-2; Fig. 3) of G. ruber increases with the SST, temperature of the surface mixed layer, and in-situ temperature.The Mg/Ca of the penultimate (F-1) and antepenultimate chambers (F-2) show statistically significant relationships with SST (Figs. 2, 3).In comparison, the relationships of Mg/Ca of both penultimate (F-1) and antepenultimate (F-2) chambers with the mixed layer temperatures are rather weak (Figs. 2, 3).The Mg/Ca relationships to in-situ temperatures are weakest among the three different temperature measures (Figs. 2, 3).
The relatively poor numerical relationship of the Mg/Ca of the antepenultimate chamber (F-2) with the in-situ temperature may be explained by the temporal distance of chamber formation and sampling date of the foraminifer shell in the respective water body, i.e., typically a couple of days to weeks between the formation of the F-2 and the final chamber [36][37][38] .The question remains, why -on average-SST does best explain the Mg/Ca of the penultimate and antepenultimate chambers (Figs. 2, 3).The fact that the Mg/Ca of final chamber (F0) shows weak relationships with any measure of seawater temperature recorded at the time of sampling may be explained by the potentially incomplete calcification of the chamber wall 1,39 .
Relationship between single chamber-Mg/Ca temperature of G. ruber and measured water temperature.Mean Mg/Ca from each chamber of the analyzed G. ruber (Table 2) are used to assess the relationship with the sea surface temperature (SST), ambient water temperature (in-situ), and average mixed layer temperature, ranging from the sea surface to the upper limit of the thermocline, by applying the relationship formulated by Dekens et al. 15 and Anand et al. 5 (Table 2).
Comparing the calculated temperatures (°C) of penultimate chamber (F-1) to the measured SST a statistically significant correlation is revealed at r = 0.83, and p < 0.02 (Fig. 4).With the average mixed layer temperature and in-situ temperature being statistically not significant, only positive trends can be detected with r = 0.59 and r = 0.49, respectively.
Comparing the calculated temperatures (°C) of antepenultimate chamber (F-2) to the measured temperatures, no statistically significant relationship exists with SST (r = 0.64) and average mixed layer temperature (r = 0.71).Calculated Mg/Ca temperatures show no relationship (r = 0.22) or trends in-situ temperatures (Fig. 5).
Finally, the Mg/Ca temperatures produced here range at the upper limit of the Mg/Ca temperature of G. ruber of earlier analyses, which have analyzed various types of samples and species with different methodologies (Fig. 6, and references therein).In particular, absolute values and slope of the regression of the data from the penultimate chambers (Fig. 6 39 , and the salinity and pH corrected values of Gray et al. 21, whereas the slope of the regression of the data from the antepenultimate chambers (Fig. 6, Line 8) is steeper than in Gray et al. 21.The high-resolution data set provided here is based on a limited number of individual G. ruber representative of the 19-22 °C temperature range, and may be used together with other data of similar nature and quality for a broader application of the Mg/Ca thermometer at the basin to global scale.

Fig. 1
Fig. 1 Mg/Ca data of planktic foraminifer shell calcite from fs-LA-ICP-MS analyses.Comparison of the Mg/Ca values obtained for different chambers in 20 G. ruber tests.The final chamber (F0), penultimate chamber (F-1), and antepenultimate chamber (F-2) are shown by red dots, green triangles, and blue squares, respectively.Data sorted according to Mg/Ca in F-1.Error bars are in 1σ.
, Line 7) are close to the values of Bolton et al. (2011; LA-ICP-MS of single chambers)

Table 1 .
Scientific expeditions of S/Y Eugen Seibold in 2019.7-digit campaign labels give Eugen Seibold (ES) Year Year (19) Cruise (C), and Digit1 Digit2 (D1D2) for the consecutive expedition number, reading ESyyCD1D2.Time is given as UTC.CSS is Central Sampling Site.Negative longitudes indicate degrees western hemisphere.Latitudes and longitudes give the targeted sampling sites, and actual samples were obtained adrift within the same water body, and over some distance from the starting point.PT is Portugal, ES is Spain, ESTOC* is near the European Station for Time-Series in the Ocean Canary Islands (ESTOC).

Table 2 .
35/Ca of Globigerinoides ruber.F0, F-1, and F-2 are the final, penultimate, and antepenultimate chambers, respectively.SST is sea surface temperature, T MLD is average mixed layer temperature, and T in-situ is ambient temperature per sampled water depth interval.Temperatures are in degrees Celsius (°C).n.a. is not available, n.app. is not applicable because settling foraminifer tests were sampled from the subsurface water column below their live habitat.For metadata on the sampling locations and sample numbers see Table1.MN is multinet, BN is bongonet.For primary data see files MgCa_avg.tabandMgCa_per_laser_spot.tab in https://doi.org/10.17617/3.D9FSSN35.
To test data reliability, Mg/Ca values were converted into ambient water temperature estimates according to