This study measured Mg/Ca values in cross sections of the benthic foraminifer Planoglabratella opercularis (d'Orbigny) using an electron probe microanalyzer (EPMA) (spot size, 2 μm; >1000 analyses). We cultured clonal juveniles of this species under controlled temperatures. This species precipitates high-magnesium calcite as its test material. Mean Mg/Ca values increased linearly with the culture temperature; therefore Mg/Ca values in walls of subsequently formed chambers reflected changes in temperature during the experiment. Foraminiferal specimens yielded variable Mg/Ca values within their test walls, which resulted from the pattern of chamber formation and from minor variations of Mg/Ca within one chamber wall.
 Many researchers have investigated the relationship between Mg/Ca and temperature in low-Mg planktonic and benthic foraminifera. However, few have probed the relationship between high-Mg species and environmental factors [Toyofuku et al., 2000; Toler et al., 2001], even though many foraminiferal species precipitate high-Mg calcite as a test material [Blackmon and Todd, 1959]. Mg/Ca ratios in low-Mg species range from 0.5 to 10 mmol/mol, whereas typical high-Mg species show Mg/Ca ratios higher than 100 mmol/mol [Lea, 1999; Toyofuku et al., 2000; Toler et al., 2001]. Mg contents differ by as many as two digits between the low-Mg and high-Mg species. Therefore the Mg-uptake mechanism must be different between high-Mg species and others. Thus studies about Mg/Ca of high-Mg species will improve the understanding of foraminiferal Mg/Ca and biomineralization mechanisms.
 Foraminifera precipitate a chamber wall over periods ranging from several to 24 h [Angell, 1967; Hemleben et al., 1977; Bé et al., 1979]. Foraminifera form a bilamellar test deposit calcite crystals on both sides of the organic membrane in Buliminida, Rotaliida and Globigerinida species [Hansen, 1999]. Foraminiferal calcifications begin at many small sites on the membrane. After a few hours, patchy calcite crystals interconnect and then thicken the chamber wall. At this time, the elemental distribution of the chamber wall will show partial and temporal variability of both chemical composition and microenvironments during chamber formation. Therefore we attempted to elucidate the Mg/Ca distribution with high spatial resolution by employing an electron probe microanalyzer (EPMA).
 This study uses EPMA to investigate Mg/Ca variations and micromap their two-dimensional distribution (∼2-μm resolution) in chamber walls of the high-Mg benthic foraminiferal species Planoglabratella opercularis (d'Orbigny) following culture under controlled temperatures.
2.1. Foraminiferal Species
 We selected the shallow-water benthic foraminiferal species Planoglabratella opercularis (d'Orbigny) for our experiments because its biology has been well studied. The species is common along the rocky shores of the Japanese islands, living attached to the fronds of coralline algae [Kitazato, 1994]. Its life cycle is rather short compared to that of other species [Tsuchiya et al., 1994]. In culture experiments, a haploid gamont took from 15 days at 20°C to 25 days at 10°C to grow to adult size. A diploid agamont grew to adult size in about 40–50 days. We therefore infer that an ideal sexual–asexual life cycle of this species is about 55–75 days under optimal conditions. Previously, we showed that this species has a hyaline calcareous test of high-Mg calcite, and we documented the correlation between Mg/Ca values and calcification temperature for the species [Toyofuku et al., 2000].
2.2. Culture Experiments
 The culture experiments followed procedures described by Kitazato . Specimens of naturally occurring agamonts of Planoglabratella opercularis were picked from coralline algae and put into Petri dishes. The specimens were kept alive and fed at laboratory room temperature until asexual reproduction.
 We used juvenile gamont specimens of a single clonal population produced during asexual reproduction, which thus should display very little genetic variability. The tests of this population are two-chambered at formation. The clonal specimens were placed into Petri dishes a few days after they were produced. During those days, the specimens added several chambers, which calcified at laboratory room temperatures.
 The Petri dishes with foraminiferal specimens were then placed in temperature-controlled incubators at temperatures of 10.4°C, 14.7°C, and 23.1°C during the experimental period of 60 days. Almost all specimens survived during the complete experimental period.
 The maximum temperature fluctuations and standard deviations were <0.5°C and 0.2°C, respectively. To prevent changes in salinity resulting from evaporation, the culture water was replaced every day using filtered open-ocean surface water. The salinity was kept at 35.0‰, and the variation in salinity was <1‰. The process of chamber formation was observed in the living specimens using an inverted microscope.
 The specimens were rinsed several times with distilled water using a brush. Cleaned specimens were dried for a few hours at 50°C and embedded in epoxy resin on glass slides. After the epoxy resin had filled the chamber cavities, the embedded specimens were polished using carborundum (silicon carbide) and aluminum oxide powder so that the internal, transverse chamber walls were exposed. The specimens were coated with a 20-nm-thick carbon film by vacuum evaporator.
 The elemental microanalyses were carried out by electron probe microanalyzer (EPMA, JEOL JXA-8900RL, JAMSTEC) on seven cultured specimens: three cultured at 10.4°C (specimens A–C); two cultured at 14.7°C (specimens D and E); and two cultured at 23.1°C (specimens F and G). The magnesium and calcium contents were measured at an accelerating voltage of 10 kV and a beam current of 0.75 ± 0.02 nA. Periclase and wollastonite crystals were used as standards for magnesium and calcium, respectively. The duration was 15 s per element peak, and 15 s for the background before and after the peak measurement. The diameter of the electron probe spot was ∼2 μm.
 All data were expressed in millimolar units of magnesium per molar unit of calcium (mmol magnesium per mol calcium). Standard deviations (2σ) were 1.2% for magnesium and 1.6% for calcium on the standard materials using three measurements. Analytical variability was expected to be <5 mmol/mol relative to the measured foraminiferal Mg/Ca ratio that ranged from 74.2 to 142.4 mmol/mol in the chamber walls in this study. We measured the SiO2 and Al2O3 content of the chamber wall within the electron probe spots to check for contamination from polishing materials.
 To compare the values of the two-dimensional Mg/Ca distribution in the wall structure, we observed the microstructure of the foraminiferal chamber wall with a scanning electron microscope (SEM). The specimens used for SEM observations were selected randomly from the cultured population, and were not the specimens analyzed by EPMA. The sectioned specimens were weakly etched with NaOH-buffered 0.1 M ethylenediaminetetraacetic acid (EDTA) to observe the primary organic membrane on which foraminiferal calcite is precipitated within the chamber wall.
3.1. Measurement of Contaminants
 Some electron probe spots showed high concentrations (>1%) of Al2O3 and SiO2, indicating that some of the polishing powder penetrated into the foraminiferal tests. Analytical data for spots that contained >1 wt% of either Al2O3 or SiO2 were discarded. Foraminiferal Mg/Ca values did not correlate significantly with the measured contents of Al2O3 and SiO2 (Figure 1), indicating no effect of the presence of polishing powders on the Mg/Ca measurements.
3.2. Test Structure and the Calcification Process
 Microscope observations documented how calcification of new chambers proceeds in Planoglabratella opercularis. First, the protoplasm arranges itself in the shape of the new chamber (Figure 2, part 1), and an organic membrane forms on its outside. This membrane is thought to correspond to an organic layer in the later-formed chamber wall. The micrograph in Figure 3 shows that calcite crystals are precipitated on both sides of the organic layer. The precipitation of calcite does not occur evenly over the organic membrane of the developing chamber. It commences as isolated crystals (Figure 2, part 2), which serve as loci for laterally expanding calcification patches (Figure 2, part 3) and cover the whole membrane after a few hours (Figure 2, part 4).
3.3. Mg/Ca Variations With Temperature
 Mean Mg/Ca values of seven cultured specimens show a close relationship with temperature (see Figure 4 and Table 1).
Table 1. Mg/Ca Values Found in Seven Specimens of Planoglabratella opercularis (d'Orbigny) Cultured at Three Temperaturesa
T = 10.4°C
T = 14.7°C
T = 23.1°C
The unit of the Mg/Ca values is mmol/mol. T, temperature.
Number of spots
 The linear regression equation is
where T is the temperature and r is the correlation coefficient.
 Temperature dependency in this study is similar to that of our previous study; however, both slope and intercept are smaller than in the previous equation: Mg/Ca (mmol/mol) = 2.22T (°C) + 89.7 [Toyofuku et al., 2000]. The regression line of this study fell below that of the previous one.
3.4. Mg/Ca Variations Within a Test
3.4.1. Intrachamber Variability
 Electron probe measurements revealed the Mg/Ca variability within a single chamber (Figures 5 and 6; Table 1). The maximum Mg/Ca variation was 45.4 mmol/mol in the penultimate chamber of specimen B at 10.4°C. The intrachamber variation was not random, but showed a weak planar distributional pattern. For instance, the Mg/Ca values tended to decrease from one end of the chamber (indicated as i) to the other end (v) in the ultimate chamber of specimen A (Figure 7; Table 2). However, the Mg/Ca spatial patterns differed among the cultured specimens, although the individuals belonged to a single clone.
Table 2. Mg/Ca Variation Within the Ultimate Chamber of Specimen A
3.4.2. Interchamber Variability
 Small Mg/Ca variations existed between neighboring chambers even though both chambers were calcified under identical temperatures (Figure 5; Table 1). The maximum difference was 7.4 mmol/mol observed in specimen F at 23.1°C. In specimen B at 10.4°C and specimen F at 23.1°C, mean Mg/Ca decreased from the older chambers to the newer chambers. In contrast, mean Mg/Ca increased from the older chambers to the newer chambers in specimen A at 10.4°C. These differences are not systematic with ontogeny.
 The two newer chambers yielded significantly lower Mg/Ca values than the older ones (Mann-Whitney U-test, p < 0.001) (Figure 6). Similarly low Mg/Ca values also occurred on the outside of earlier formed chambers of tests (Figure 6, locations a–c). The specimen was part of a clonal population formed at room temperature; the first through seventh chambers developed in the first few days after reproduction, when juveniles form one chamber each day. These first seven chambers, which possessed relatively high Mg/Ca values, thus formed at room temperature, whereas the two later chambers displaying significantly lower Mg/Ca values formed during culture when the specimen was kept at 10.4°C. The outer wall of the seven earlier chambers yielded lower Mg/Ca values because this wall was covered by calcite during the formation of the eighth and ninth chambers at the lower temperature (Figure 6, locations a–c).
4.1. Previous Studies
 As in our previous study [Toyofuku et al., 2000], these observations revealed that mean Mg/Ca values increase systematically with increasing temperature (Figure 4). These results again demonstrated that Mg concentrations of high-Mg foraminiferal species are strongly controlled by temperatures, but the regression lines differ between the studies. Both the slope and intercept of this study were smaller than the respective parameters of the previous study. We expected the opposite because the previous study measured the average test chemistry of several specimens. For instance, at a higher temperature the test should include chambers that were calcified at a lower temperature; thus average Mg/Ca values should be lower, and vice versa. If this study perhaps missed an Mg-enriched part of a chamber wall within a test, the mean Mg/Ca should have yielded a lower value than the Mg/Ca of the average measurement.
 Some studies have reported partial Mg enrichment in foraminiferal test [Nürnberg et al., 1996; Eggins et al., 2003; Erez, 2003]. However, we did not observe such an Mg-enriched layer in our specimens because this study measured Mg/Ca values in the horizontal plane. We speculate that an Mg-enriched layer might be discovered on vertical sections as previous studies.
4.2. Mg/Ca Variations Within a Single Chamber Wall
 Contrary to our expectations, we observed variations in the Mg/Ca values within a single chamber of each specimen, even though all parts of this chamber wall calcified at the same temperature (Figures 56–7; Table 1). Our electron probe data revealed that the Mg/Ca values fluctuated over a range of about 40 mmol/mol within a single chamber wall, despite limiting the temperature fluctuation to ±0.5°C. According to equation (1), a temperature fluctuation of 0.5°C is equivalent to less than 1 mmol/mol change of Mg/Ca. Thus this temperature variation is too small to explain the 40 mmol/mol of Mg/Ca variability.
 Recent Mg/Ca studies of planktonic species have shown that carbonate ion concentration is also an important factor influencing foraminiferal Mg/Ca [Lea et al., 1999; Russell et al., 2004]. Our experiment controlled and monitored temperature and salinity, but not pH or carbonate chemistry. In our other similar experiment, we monitored the pH of the procedure's seawater in Petri dishes as in this study. In the other experiment, water pH fluctuated between 8.1 and 8.5. We used the program of Lewis and Wallace  to calculate carbonate ion concentration ([CO32−]) from pH, temperature, and estimated total alkalinity. Total alkalinity was 2338 μmol/kg, which was extrapolated from Pacific Ocean values observed in waters near the islands of central Japan (Japan Oceanographic Data Center: http://www.jodc.go.jp/), which is the origin of the seawater used in our experiments.
 We assumed that the total alkalinity would not fluctuate much. The culture water was changed daily; therefore the elemental concentration should have been almost stable. The estimated [CO32−] varied between 130 and 318 in this study. According to results of a recent study of planktonic species [Russell et al., 2004], fluctuations of this concentration exert minimal influence on foraminiferal Mg/Ca. However, more investigations are necessary to evaluate Mg/Ca variation resulting from [CO32−] fluctuation in high-Mg species.
 We observed that Planoglabratella opercularis did not precipitate calcium carbonate evenly over the whole chamber wall (Figure 3), but rather in patches. Mg/Ca varied weakly within a chamber (Figure 7). One would expect Mg/Ca values of foraminiferal cells to have varied during calcification if they are influenced by such microenvironmental changes as seawater chemistry within foraminifera. Seawater chemistry within foraminiferal vacuoles should be influenced by the consumption and uptake of elements during calcification. Both the calcification rate and the elemental uptake rate should strongly correlate with foraminiferal activity.
 Therefore Mg/Ca variation between neighbor chambers can perhaps be explained by differences of metabolic activity during foraminiferal calcification (Figure 5). However, no direct proof exists concerning the influence of seawater chemistry on foraminiferal vacuoles within cells. Further observations such as intraforaminiferal seawater chemistry and the influence of calcification rates would be useful in quantifying these currently unexplained variables.
 Our observations on the distribution of Mg/Ca values within the tests of high-magnesium benthic species cultured under controlled conditions showed that mean Mg/Ca increased with increasing temperature. This result reconfirms that foraminiferal Mg/Ca are strongly controlled by calcification temperature. These observations also showed that the significant variability of these values between test chambers reflected precipitation at different temperatures. However, the observed Mg/Ca values varied within a single chamber wall, possibly as the result of fluctuations in microenvironmental conditions (e.g., Mg/Ca in foraminifera, pH, calcification rates) during calcification.
 We thank Katsuyuki Uematsu for his advice and assistance during the electron probe measurements of this study. The authors also extend sincere thanks to Ellen Thomas and Simon Johnson for fruitful discussions and English-language proofreading. We also wish to thank Steve Eggins, Pamela Martin, and an anonymous reviewer for their excellent suggestions on this manuscript. This study was supported by a Grant-in-Aid for Scientific Research (T.T.: 16740294; H.K.: 14340156) from the Japan Ministry of Education, Culture, Sports, Science, and Technology.