Foraminifers and water sampling

Samples were collected in Sagami Bay, Kanagawa, Japan (35°05′N, 139°28′E; water depth, 920 m) on 11 October 2011 (Figure 1). The planktic foraminifers were collected with a closing plankton net (aperture opening, 30 cm; 100 μm mesh; Marukawa Closing Plankton Net, Rigosha, Tokyo, Japan). The net was towed vertically through the water over specific intervals between 0 and 600 m (i.e., 0–20, 0–33, 0–40, 0–50, 0–100, 0–200, 40–400, 0–400, and 0–600 m). Seawater samples were collected at depths of 0, 10, 20, 33, 40, 50, 100, 200, 400, and 600 m. Seawater samples for analyses of δ¹³C_DIC and δ¹⁸O_sw were collected in 100 mL glass vials and sealed onboard without bubbles. To prevent the biological consumption of carbon, mercuric chloride was added to the vials of water for δ¹³C_DIC analysis before they were sealed. Other water samples for nutrient analysis were filtered with a cellulose acetate filter (pore size of 0.45 μm) immediately after collection. These aliquots were kept frozen until analysis (see below for details). The vertical distributions of temperature, salinity, and chlorophyll fluorescence were measured with a conductivity-temperature-depth sensor (Compact-CTD, JFE Advantech, Hyogo, Japan) in the water column at depths of <600 m.

Foraminiferal analyses

The net-collected samples were washed with tap water and then soaked in 12% sodium hypochlorite solution overnight to decompose the organic matter. The samples were then rinsed through a 63 μm sieve with tap water. Three species of planktic foraminifers with different ecologies were analyzed: Globigerinoides sacculifer (n = 5), Neogloboquadrina dutertrei (n = 4), and Globorotalia inflata (n = 3). Globigerinoides sacculifer was selected as representative of shallow-dwelling species with sym biotic algae (e.g. Bé, 1977; Erez and Honjo, 1981; Birch et al., 2013). This is one of the best-studied species of planktic foraminifers, especially in terms of its ecology and the isotope characteristics of its test (e.g. Bé et al., 1981; Caron et al., 1981; Spero and Lea, 1993). Neogloboquadrina dutertrei was selected as a general intermediate-dwelling species (e.g. Bé, 1977; Erez and Honjo, 1981). Globorotalia inflata was selected as representative of deep-dwelling Globorotalia species without symbionts (e.g. Bé, 1977; Erez and Honjo, 1981). The specimens of these species were picked under a stereoscopic microscope and selected specimens were dissected into chambers with ophthalmic surgical microblades (Feather Safeshield Incision Scalpel 15° Angle, Feather, Osaka, Japan) (Figure 2). The dissection of the chambers was continued until the remaining test became too small to dissect (Figure 2d; see Takagi et al., 2015 for details). The δ¹³C and δ¹⁸O values for each chamber were analyzed with a customized continuous-flow isotope ratio mass spectrometry system at the Geological Survey of Japan, Advanced Industrial Science and Technology (AIST) (for details of the system, see Ishimura et al., 2004, 2008). Each dissected chamber was reacted with anhydrous phosphoric acid at 25°C. To calculate the CaCO₃ mass of the sample, the pressure of the manually purified CO₂ gas in the fixed volume of the vacuum line was measured. The CO₂ gas was then introduced to the isotopic ratio mass spectrometer (Isoprime, Isoprime Ltd., Stockport, UK) with helium carrier gas. The system determines both δ¹³C and δ¹⁸O values for microvolumes of carbonate samples as small as 1.0 μg, with external precision better than ±0.1‰. The NBS-19 carbonate standard was used for calibration of the Vienna Peedee belemnite (VPDB) scale.

Figure 1.

a, Map showing the position of the Kuroshio Current on the sampling day (11 October 2011), and the study area (rectangle), which is enlarged in panel b. b, Map showing the bathymetry of Sagami Bay. The open star shows the location of the sampling site (35°05′N, 139°28′E; water depth, 920 m).

Analyses of water samples

To determine δ¹³C_DIC and δ¹⁸O_sw, the CO₂ gas generated from and equilibrated with each seawater sample was analyzed. An aliquot (20 mL) of each seawater sample was injected into an evacuated glass vial containing amidosulfonic acid. All the DIC in the seawater was converted to CO₂. The vials were shaken in a water bath at 25°C overnight to equilibrate the δ¹⁸O of the seawater with that of the CO₂ gas generated. The dual inlet system of the mass spectrometer (Isoprime) at AIST was used for the analysis, which determines the δ¹³C and δ¹⁸O with analytical error of ±0.1‰. The δ¹⁸O_sw values were reported relative to Vienna Standard Mean Ocean Water (VSMOW).

The concentrations of nitrate (NO₃), nitrite (NO₂), phosphate (PO₄), and silicate (SiO₄) were determined with a colorimetric analysis in a continuous flow system at Ehime University (AutoAnalyzer 3, Bran + Luebbe, Norderstedt, Germany). The water samples for chlorophyll a analysis (1 L) were filtered through Whatman GF/F glass fiber filters within 5 h of collection, and the filters were then immediately soaked in 10 mL of N,Ndimethylformamide (DMF) in dark glass vials for more than 24 h. The chlorophyll a concentration in the DMF extract was measured with a fluorometer at Ehime University (10AU, Turner Design, California, USA).

Estimation of calcification depth

Based on the δ¹⁸O_sw values determined and the temperatures measured in the water column, we calculated the expected oxygen isotope values for the inorganic calcite precipitated in equilibrium with seawater (δ¹⁸O_eq). We used the equation of Kim and O'Neil (1997), which was established for inorganic calcite precipitated in a temperature range of 10–40°C:

Figure 2.

Scanning electron micrographs of the species used in this study, and an example of a series of dissected chambers. a, Globigerinoides sacculifer; b, Neogloboquadrina dutertrei; c, Globorotalia inflata; d, Dissected chambers of Neogloboquadrina dutertrei (specimen dut2). f, f-1, f-2, … indicate the chamber positions counting backward from the final chamber (f). Scale bars are 200 µm.

Many authors have proposed species-specific relationships among calcification temperature and δ₁₈O_sw, and the δ₁₈O value of the foraminiferal test, indicating that there is a small vital effect on the δ₁₈O of foraminiferal calcite (Erez and Luz, 1983; Bouvier-Soumagnac and Duplessy, 1985; Bemis et al., 1998; Mulitza et al., 2003; Spero et al., 2003). For comparison, we also calculated the species-specific δ₁₈O of the foraminifers using equations that were most recently established:

Figure 3.

Physical and chemical parameters of the water column at the study site on the sampling day (35°05′N, 139°28′E, 11 October 2011). a, Vertical distributions of chlorophyll a content (Chl; gray triangles), temperature (T; black circles), and salinity (S; white squares); b, vertical distributions of nitrate and nitrite (N; black squares), silicate (Si; white circles), and phosphate (P; gray inverse triangles); c, vertical distributions of the oxygen isotope ratio of seawater (δ¹⁸O_sw; white triangles) and the carbon isotope ratio of dissolved inorganic carbon (δ¹³C_DIC; black diamonds); d, vertical distribution of expected oxygen isotope ratio in foraminiferal calcite. δ¹⁸O_eq, equilibrium calcite (bold line with black squares); δ¹⁸OG_ss , Gs. sacculifer-specific fractionation (dashed line with gray circles); δ¹⁸O_Nd, N. dutertrei-specific fractionation (fine line with black circles); δ¹⁸O_Grm, Gr. menardii-specific fractionation (broken line with white circles) (see text for detail). Gs., Globigerinoides; N., Neogloboquadrina; Gr., Globorotalia

Water column profiles

The water temperature was almost constant (23.0–23.1°C) in the upper 20 m. It decreased sharply with depth to 50 m (to 18.0°C), and decreased gradually below ca. 70 m (Figure 3a). Salinity increased rather gradually from the surface to 70 m (to 34.5), showing two sharply increasing intervals at around 18 and 37 m. Salinity decreased gradually below 70 m. The concentration of chlorophyll a showed two peaks, the shallower at 18 m (0.33 μg L^-1) and the deeper at 33 m (0.37 μg L^-1). These depths nearly corresponded to those of the increases in salinity. The chlorophyll a concentration then decreased sharply with depth to 45 m (to 0.06 μg L^-1), and any significant signal was lost below 75 m. The concentrations of nutrients (NO₃ + NO₂, PO₄, and SiO₄) were significantly lower in the upper 20 m, increased rapidly to 50 m, and then increased gradually with depth below 50 m (Figure 3b). δ¹⁸O_sw remained constant in the upper 20 m (-0.1‰), increased to 50 m (+0.2‰), and gradually decreased below 50 m (Figure 3c). δ¹³C_DIC was almost constant in the upper 20 m (ca. +0.8‰), then decreased sharply between 20 and 50 m, and decreased gradually with depth below 50 m.

Ontogenetic stable isotope records for foraminifers

Because our foraminiferal samples were obtained from a vertically towed natural population, the ontogenetic stages of the individuals may have varied within the same species. Therefore, for the purpose of interindividual comparisons, we used the cumulative test mass as an index of growth (Table 1). The δ¹³C and δ¹⁸O results for each individual are shown in Figure 4. For each individual, the most ¹⁸O-depleted δ¹⁸O value was always recorded at the smallest chambers (Table 1; Figure 4a). The most ¹⁸O-depleted δ¹⁸O values for each individual ranged from -2.6‰ to -1.9‰ for Globigerinoides sacculifer (test mass 2.4–6.1 μg) and from -2.1‰ to -2.0‰ for Neogloboquadrina dutertrei (test mass 1.7–2.4 μg). The most 18O-depleted δ¹⁸O value for Globorotalia inflata was higher than those for the other two species, ranging from -1.4‰ to -0.7‰ (test mass 1.5–4.3 μg). δ¹⁸O values tended to become more positive throughout ontogeny in all species (Figure 4a), and this 18O-enrichment was rather rapid in the earlier growth stage (cumulative test mass < ca. 8 μg). The δ¹⁸O values for N. dutertrei and Gr. inflata appeared to converge around −1.0‰ and 0.0‰, respectively, at the latest ontogenetic stage observed for each species. For Gs. sacculifer, the δ¹⁸O value for the largest chamber in each individual ranged from −1.5‰ to −1.1‰ (test mass 21.7–35.0 μg), representing the most ¹⁸O-depleted values among the three species.

Table 1.

Results of the chamber-by-chamber analysis of δ¹³C and δ¹⁸O, and the estimated calcification depth for each chamber. Numbers in parentheses in the “chamber calcification depth” column indicate error intervals for the depth estimation calculated from the ±0.1‰ analytical error of δ¹⁸O for each chamber. The presence or absence of a sac-like final chamber is noted in the Gs. sacculifer specimens. f, f-1, f-2, … indicate the chamber positions, counting backward from the final chamber (f). The error of measurement for the test length, calculated from quintuplicate measurements, was ±10 μm.

Figure 4.

Chamber-by-chamber δ¹⁸O (a) and δ¹³C (b) versus the cumulative test mass. Data obtained from the same individual are connected in ontogenetic order. The same shade is used for symbols of the same species (gray: Globigerinoides sacculifer; black: Neogloboquadrina dutertrei; and white: Globorotalia inflata). Error bars represent analytical errors (±0.1‰).

The most ¹³C-depleted values of δ¹³C for each individual corresponded to the smallest chambers, as for ¹⁸O-depleted values of δ¹⁸O (Table 1; Figure 4b). The most ¹³C-depleted δ¹³C values were +0.3‰ to +0.6‰ for Gs. sacculifer (test mass 2.4–6.1 μg). However, they were much more negative for N. dutertrei and Gr. inflata relative to that of Gs. sacculifer (by at least 0.8‰), ranging from −1.1‰ to −0.6‰ for N. dutertrei (test mass 1.7–2.4 μg) and from −1.4‰ to −0.5‰ for Gr. inflata (test mass 1.5–4.3 μg). The δ¹³C values of the Gs. sacculifer specimens became more positive up to +0.9‰ to +1.4‰ (test mass 21.7–35.0 μg), although the ontogenetic trajectories varied to some extent among individuals. The δ¹³C values for specimens sac2, sac3, and sac5 tended to become more positive gradually throughout test growth, whereas those for sac1 and sac4 showed large positive shifts in the last chamber (by 0.4‰ and 0.7‰, respectively) (Figure 4b). The δ¹³C values for N. dutertrei and Gr. inflata became more positive rapidly in the earlier growth stage, up to a cumulative test mass of ca. 12 μg, followed by a slight ¹³C-depletion. In the largest specimen, inf1 (final test mass 29.3 μg), the δ¹³C value in the final growth stage was −0.2‰.

Expected δ¹⁸O in foraminiferal calcite through the water column

δ¹⁸O_eq, δ¹⁸O_Gss, δ¹⁸O_Nd, and δ¹⁸O_Grm were calculated from δ¹⁸O_sw and temperature. Because the variation in δ¹⁸O_sw in the water column was small (<0.3‰; Figure 3c), the expected δ¹⁸O_eq, δ¹⁸O_Gss, δ¹⁸O_Nd, and δ¹⁸O_Grm values were controlled mainly by the variation in temperature. They showed almost constant values above 20 m (−2.1‰), became more positive rapidly to 50 m, and then gradually below 50 m (Figure 3d).

Effect of lamellar structure

Chamber walls of planktic foraminifers usually form a lamellar structure (Reiss, 1958; Bé and Lott, 1964), i.e., a chamber is over-layered by the outer lamella(e) of the subsequently formed chamber(s) (cf. Figure 5). This means that except for the final chamber, the isotopic signal initially recorded in a calcite wall during the chamber formation is contaminated by those in the later ontogenetic stages. Since the thickness of the over-layering cannot be directly evaluated from the materials utilized in this study, we here modeled three scenarios of relative thickness of the over-layers, which enable us to assess its effect on the bulk-chamber δ¹⁸O values. For calculating the hypothetical mass-balanced δ¹⁸O values, we assumed the ratio between the mass of the initially precipitated wall of a given chamber and that of the over-layering lamellae as 3:1 (25% over-layer), 2:1 (33% over-layer), and 1:1 (50% over-layer) (Figure 5).

Figure 5.

Hypothetical mass-balanced δ¹⁸O of each chamber at the time the chamber was secreted as the final chamber (a1, b1, c1) and the difference from the δ¹⁸O actually measured (a2, b2, c2). a, Globigerinoides sacculifer; b, Neogloboquadrina dutertrei; c, Globorotalia inflata. The schematic illustration of the lamellar structure is shown in bottom right. The hypothetical mass-balanced values were calculated at three different scenarios by assuming the ratio of the mass of an existing chamber wall versus that of overlying lamellae of the subsequent chamber for 3:1 (25%), 2:1 (33%), and 1:1 (50%).

The variability in each specimen's profile is amplified as the proportion of the over-layer becomes larger (Figure 5a1, b1, c1). With regard to 25%- and 33%- layering models, the δ¹⁸O differences between the hypothetical mass-balanced values and that of the measured bulk-chambers are small (less than 0.5‰ in most cases). For the 50%-model, however, the modeled δ¹⁸O values greatly differ from the measured ones especially for earlier chambers (maximum 2.8‰ difference) (Figure 5a2, b2, c2). On the basis of our visual observation, the thickness of the uncovered original wall (septa) of a given chamber is always significantly thicker than 50% of the total thickness of the original wall and the over-layer of the succeeding chamber. Considering the actual wall thickness of the chambers of Neogloboquadrina dutertrei we observed under the sample preparation (shown in Figure 2d), it is hard to assume that penultimate chamber has 50% of over-layering. If we assume a given chamber has 50% of over-layering, f-2 and f-3 chambers should have ca. 75% and 87.5% of over-layering, which is impractical (Figure 2d). Even in an extreme example of the chamber wall thickness with crust in N. dutertrei reported in Steinhardt et al. (2015b), the total wall thickness of the f-1 chamber was less than the double to the f chamber. Additionally, since a septum of a chamber is isolated from the surface, over-layering of the succeeding chamber never covers all the area of a given chamber. According to these assumptions, we infer that 50% of over-layering is quite unlikely. Therefore, we hereafter exclude the 50%-model, and at most a 33%-model was considered.

Migration histories through ontogeny

The calcification depth of each chamber was estimated based on the assumption that the foraminiferal calcite analyzed was precipitated under isotopic equilibrium (δ¹⁸O_eq) or by species-specific fractionation (δ¹⁸O_Gss, δ¹⁸O_Nd, and δ¹⁸O_Grm) (Table 1; Figure 5a, b).

For Globigerinoides sacculifer, the calcification depth estimated from δ¹⁸O_eq was slightly shallower than that estimated from δ¹⁸O_Gss. At the smallest ontogenetic stage analyzed (<5 μg), all the Gs. sacculifer specimens calcified their tests within the upper part of the thermocline (shallower than 30 m). With the hypothetical modeled value for 33% over-layer, the calcification depth calculated is equivalent to the mixed layer depth (0–20 m) (Figures 3, 5). Therefore, we estimate that Gs. sacculifer analyzed calcified its test within or very close to the mixed layer in their very early stage of ontogeny. Globigerinoides sacculifer finally migrated to ca. 50 m, which was near the steep decline in chlorophyll a content. Among the five specimens analyzed, sac1 had a saclike chamber in the final chamber position (see Table 1). Because a sac-like shape is an indication of the very-lastformed chamber in this species, the calcification depth of the final chamber of sac1 represents the habitat depth during the terminal stage of its ontogeny. The best estimate of the calcification depth of that sac-like chamber was 42 m (equilibrium estimate) or 55 m (Gs. sacculiferspecific estimate). Since the difference between the hypothetical model and bulk δ¹⁸O is almost negligible in the later ontogenetic stage, deeper migration with ontogeny is a robust finding from our records (Figure 5). Our results for this species are consistent with the deeper migration ecology through ontogeny reconstructed by single chamber Mg/Ca studies using laser ablation techniques (Eggins et al., 2003).

Globigerinoides sacculifer is known to harbor dinoflagellates (Pelagodinium béii) in its cytoplasm as obligate symbionts (Hemleben et al., 1989; Shaked and de Vargas, 2006; Siano et al., 2010). Therefore, because its photosynthetic symbionts require light, the vertical distribution of Gs. sacculifer is limited by light attenuation. Based on the relationship between light and the chlorophyll a concentration in the water column, the bottom depth of the euphotic zone (the depth at which irradiance is 1% of the surface level) can be estimated with the following equation (Morel and Berthon, 1989; Lee et al., 2007):

Since the net-collected specimens we analyzed still possessed intact spines (see Figure 2a), they are adultsized but before reproduction. Once foraminifers reach reproductive maturation, they should sink within the water column because of loss of buoyancy due to shedding of spines (e.g. Hemleben et al., 1989). Thus it is likely that the terminal descending pathway in each individual is not recorded in our net-collected specimens. Therefore, the deepest habitat depth of matured Gs. sacculifer should be much deeper than that estimated from our isotopic records. A previous study using depthdiscrete plankton tows (0–200 and 200–600 m) at the Gulf of Aqaba/Eilat, Red Sea, suggested that the release of gametes of Gs. sacculifer may occur at 200 m or deeper (Erez et al., 1991). This would also indicate that the very young foraminifers ascend from the deep water to the photic zone. In this respect, this ascending pathway in the beginning of ontogeny after the gametes fusion was not recorded in our results, if very little test carbonate would be precipitated during this ascending pathway. Alternatively, it was masked by the larger mass of the juvenile test precipitated at shallow depth, even if they calcified a small amount of test material. It should be noted that isotopic/geochemical proxies recorded on calcareous tests represent environmental parameters only at the calcification depth. As such, the interpretation of chamber-by-chamber isotopes needs to be done cautionly due to the possibility that the very last and initial phase, i.e., the depth of their reproduction and initial calcification of the new generation, may not be recorded. Since the final calcification, so-called gametogenic calcification (Bé, 1980), may record the depth just before reproduction, analyses of specimens after reproduction obtained by sediment traps and seafloor sediments can possibly provide the habitat depth of their final life stage.

Because of the difference between the equilibrium precipitation equation and the Neogloboquadrina dutertreispecific equation, the estimated calcification depth of N. dutertrei showed relatively large variations in colder water. However, the estimated equilibrium calcification depth of N. dutertrei was within the thermocline throughout the ontogenetic interval. In contrast, when the N. dutertrei-specific equation was used, dut3 was estimated to descend as deep as 90 m and to continue to calcify in the deeper water mass until it formed its final chamber. However, dut3 was collected from the vertical tow of the upper 50 m of water (Table 1). This may indicate that the species-specific equation for N. dutertrei produced a toodeep (cold) estimate, at least in our samples, although we cannot exclude the possibility that dut3 ascended within the water column after the calcification of its last chamber. In contrast, the calcification depths of the chambers at a test mass of <3 μg showed well constrained results in the range of 20–26 m, regardless of the equations applied. This depth corresponds to the uppermost part of the thermocline. If we assume 33% over-layer of the subsequent chamber lamellae, the depth in the earlier growth stage (younger than f-3 or f-4) is correlated with the mixed layer. In either case, this species migrated to deeper waters with growth. Compared with the migration profile of Gs. sacculifer, the migration toward deeper water is restricted to the earlier stages of its ontogeny. In most cases, if not all, it migrated to the bottom of the level of chlorophyll maximum at a test mass of >10 μg. A number of studies using depth-discrete plankton tows (Fairbanks and Wiebe, 1980; Fairbanks et al., 1982; Kemle-von Mücke and Oberhänsli, 1999) and sediment traps (Erez and Honjo, 1981; Tedesco et al., 2007) have suggested that N. dutertrei prefers a water depth near the chlorophyll maximum, where phytoplankton prey is abundant. Our results are consistent with those of previous field studies. These results indicate that the habitat depth of N. dutertrei is restricted to within the photic zone throughout its ontogeny, except for the deeper estimate for specimen dut3. In contrast to the obligate symbiosis of Gs. sacculifer, which has been studied in detail, the symbiotic nature of N. dutertrei is not well known. The ecology of N. dutertrei is usually described as “facultative symbiosis” (Hemleben et al., 1989; Spero et al., 2003), which implies that symbionts are not mandatory for this species. Our habitat depth estimate in the euphotic zone does not exclude an obligate photosymbiotic lifestyle for the specimens analyzed. In either case, we conclude that N. dutertrei primarily inhabits the thermocline where the chlorophyll a concentrations are high.

In the previous reports by Eggins et al. (2003), Steinhardt et al. (2014, 2015a), an increase of Mg/Ca toward the final chamber was recognized. The profile, in general, implies ontogenetic ascending pathways that contradict the deeper migration suggested from our study and many plankton-tow studies so far conducted (Hemleben et al., 1989). Jonkers et al. (2012) also reported an ontogenetic increase in Mg/Ca both in inner layers (ontogenetically precipitated layers) and in the crust of N. dutertrei. However, they reported that low Mn/Ca, indicating a deep-water habitat, corresponds to higher Mg/Ca, implying warmer temperature, which is apparently a contradiction. They suggested that the higher Mg/Ca in later ontogenetic stages does not reflect upward migrating pathways. Since the interpretation of Mg/Ca of N. dutertrei cannot be straightforward, we refrain from comparison of these studies to our results in terms of its migration history. Future study of singlechamber δ¹⁸O of cultured specimens under controlled temperature, combined with their Mg/Ca profile is required.

The sizes of the Globorotalia inflata specimens analyzed (n = 3) differed markedly (Table 1). However, all the specimens descended to deep water. The smallest specimen, inf3, calcified its smallest chambers in the deeper half of the thermocline, then migrated to just below the thermocline (ca. 60 m) at its largest stage (test mass 5.8 μg). The depth trajectory of specimen inf2 also started from the deeper half of the thermocline, and reached its deepest habitat at ca. 130 m, with a test mass of >10 μg. The largest specimen, inf1, reached its deepest habitat at ca. 200 m, where it calcified its final chamber (at a test mass of 29 μg). The profiles of these three differently sized specimens showed the overall ontogenetic migration profile for this species (Figure 6). Generally, we can say that Gr. inflata migrated a great distance from near the bottom of the thermocline to a depth of around 100–200 m, where it remained to calcify several chambers. Thus, the habitat depth of Gr. inflata changed from a photic, phytoplankton-rich environment to an aphotic, nutrient-rich environment throughout its life. This type of migration, from shallower to deeper water, in deep-dwelling globorotaliids has also been suggested by depth-discrete plankton tows and the isotopic characteristics of a series of size-fractionated tests (Lončarić et al., 2006; Wilke et al., 2006; Birch et al., 2013). Moreover, the Mg/Ca study of this species also showed the deepening of the calcification depth (van Raden et al., 2011). Our results reinforced that this profile can be identified even within an individual test.

Figure 6.

Estimated changes in the calcification depth of foraminifers and comparison with chlorophyll a content in the water column. a, Calcification depths of foraminiferal chambers estimated with equilibrium precipitation (Kim and O'Neil, 1997). b, Calcification depths of foraminiferal chambers estimated with species-specific equations (see text for detail). Data obtained from the same individual are connected in ontogenetic order. The same shade is used for symbols of the same species (gray, Globigerinoides sacculifer; black, Neogloboquadrina dutertrei; and white, Globorotalia inflata). Error bars represent error intervals in the depth estimates calculated from the ±0.1‰ analytical error of the δ¹⁸O of each chamber. All estimated depths are listed in Table 1. c, Chlorophyll a content shown on a monochrome scale. Note that the vertical axis (depth) is on a logarithmic scale to show the upper water column in detail.

δ¹³C deviation: relationship to temperature

We calculated the deviations of δ¹³C for the foraminifers from that of DIC (i.e., Δδ¹³C_foram-DIC) at the estimated calcification depth of each chamber (Table 1). It is noteworthy that the Δδ¹³C_foram−DIC values for all the chambers of Globorotalia inflata and almost all those of Neogloboquadrina dutertrei were negative, ranging between −2.1‰ and −0.1‰ and between −1.8‰ and +0.1‰, respectively. In contrast, Globigerinoides sacculifer showed much higher values of between −0.4‰ and +1.0‰ (Figure 7). To examine the possible mechanisms underlying these variations in Δδ¹³C, we considered the Δδ¹³C from δ¹³C of the equilibrium inorganic calcite rather than from δ¹³C of DIC, because the equilibrium δ¹³C value for inorganically precipitated calcite (δ¹³C_eq) is not equal to that of DIC (δ¹³C_eq = δ¹³C_DIC + 1.0‰; Romanek et al., 1992). Figure 7 shows that all the species had negative deviations from the estimated δ¹³C_eq values (see the rightmost scale in Figure 7). Therefore, some process(es) must be generating the negative shift in the δ¹³C of foraminiferal calcite. Three major candidate factors potentially explain these negative deviations from equilibrium: foraminiferal respiration (e.g. Spero and Lea, 1996), kinetic fractionation related to the fast calcification rate (growth rate) (McConnaughey, 1989a, 1989b), and the carbonate chemistry of the calcifying seawater (Spero et al., 1997). The rates of vital activities, such as respiration and growth, are usually a function of temperature. In fact, we found a significant negative correlation between temperature and Δδ¹³C in all three species, but the distributions in the plots were species specific (Figure 7). Because the δ¹³C of inorganically precipitated calcite is not dependent on temperature (Romanek et al., 1992), the negative correlation between Δδ¹³C and temperature may be attributable to the possible temperature dependence of biological activities; i.e. growth rate and/or respiration rate. It is known that the optimum growth rate shows little temperature dependence over a wide range of temperatures in many foraminiferal species (Caron et al., 1987; Lombard et al., 2009). In contrast, the respiration rates of organisms vary exponentially with temperature, both in theory and practice (e.g. Brown et al., 2004; López-Urrutia et al., 2006). Such a common mechanism is also expected to occur in planktic foraminifers, and a decrease in δ¹³C in foraminiferal shells has been explained by the greater incorporation of respired 12CO2 at a higher rate of respiration (Berger et al., 1978; Bijma et al., 1990; Spero et al., 1991; Spero and Lea, 1996; Ortiz et al., 1996). Although the mechanism by which the respired CO₂ is finally incorporated into the test is still unclear, the observed negative correlation between temperature and Δδ¹³C in our results is consistent with the explanation above. Therefore, we assume that the major contribution to the generation of a temperature-dependent Δδ¹³C might be the respiration rate.

Figure 7.

Relationships between the δ¹³C deviation of foraminiferal calcite from δ¹³C_DIC and δ¹³C_eq (Δδ¹³C_foram-DIC/Δδ¹³C_foram-eq) and the estimated calcification temperature. Data are based on the equilibrium estimation of temperature (a), and the species-specific estimation of temperature (b). The same shade is used for symbols of the same species (gray, Globigerinoides sacculifer; black, Neogloboquadrina dutertrei; and white, Globorotalia inflata). Δδ¹³C_foram-DIC is scaled on the leftmost axis, and Δδ¹³C_foram-eq is scaled on the opposite axis. A significant negative correlation between temperature and Δδ¹³C was seen in all species (p < 0.05). Lines are reduced major axis regressions. The correlation coefficient (R) is shown next to each line.

δ¹³C deviation: Interspecific differences and ontogenetic trends

As well as the intraspecific changes in δ¹³C, interspecific differences were observed (Figure 7). This interspecific segregation cannot be explained by a temperature effect, because the δ¹³C values for each species sharing the same temperature differed significantly. Instead, the species distribution in the plot shown in Figure 7 represents the opposite relationship to temperature; i.e. warmerwater species, such as Globigerinoides sacculifer, showed a smaller Δδ¹³C deviation than deeper-dwelling Globorotalia inflata. This indicates that the difference in Δδ¹³C among species largely reflects their speciesspecific ecology rather than their temperature-dependent physiological activity. In fact, Gs. sacculifer showed higher δ¹³C than Neogloboquadrina dutertrei, even though they shared almost the same depth habitat range (temperature range) in the water column (Figures 6, 7).

Figure 8.

Deviation of the δ¹³C of foraminiferal calcite from δ¹³C_DIC and δ¹³C_eq at the estimated calcification depth (Δδ¹³C_foram−DIC/ Δδ¹³C_foram−eq) plotted against the cumulative test mass. Δδ¹³C was calculated based on the equilibrium estimation of calcification depth (corresponding to that shown in Figure 7a) (a1–a3), and on the species-specific estimation of calcification depth (corresponding to that shown in Figure 6b) (b1–b3). a1, b1, Globigerinoides sacculifer; a2, b2, Neogloboquadrina dutertrei; a3, b3, Globorotalia inflata. Horizontal broken lines indicate zero deviation from δ¹³C_DIC. Δδ¹³C_foram−DIC is scaled on the leftmost axis, and Δδ¹³C_foram−eq is scaled on the opposite side.

Of several biological mechanisms that affect δ¹³C of the test, the only effect that makes δ¹³C value more ¹³Cenriched is symbiont photosynthetic activity (Spero and Lea, 1993; Bijma et al., 1999; Schiebel and Hemleben, 2005), which only affects symbiotic species. Therefore, as expected from their known symbiotic ecology, δ¹³C value was most ¹³C-enriched in symbiont-bearing Gs. sacculifer and most ¹³C-depleted in asymbiotic Gr. inflata (Figure 7). The photosynthetic fixation of carbon by the enzyme ribulose 1,5-biphosphate carboxylaseoxygenase (rubisco) of symbiotic algae preferentially utilizes ¹²C (O'Leary, 1981). This causes the relative enrichment of ¹³C in the remaining DIC pool near the foraminiferal test (e.g. Spero et al., 1991; Spero and Lea, 1993; Wolf-Gladrow et al., 1999). This photosynthetic fractionation in the calcifying microenvironments of foraminifers explains the differences in δ¹³C observed among species. The Δδ¹³C values for N. dutertrei were between those for Gs. sacculifer and Gr. inflata. If the specimens of N. dutertrei harbored symbionts, their photosynthetic activity might also have affected the test δ¹³C. However, the symbiont effect for N. dutertrei, if any, might work in a slightly different way from that proposed for Gs. sacculifer. Globigerinoides sacculifer radially distributes its symbionts along its densely developed spines, forming a so-called “symbiont halo” (e.g. Hemleben et al., 1989; Bijma et al., 1999; Wolf-Gladrow et al., 1999). The photosynthetic alteration of δ¹³C_DIC within this microenvironment ultimately affects the δ¹³C composition of the test (Spero et al., 1991; Spero and Lea, 1993; Zeebe et al., 1999). In contrast, N. dutertrei has no spines to form such a symbiont halo, and its symbionts do not occur outside the test, where the new chamber is formed.

Instead of the above photosynthetic fractionation to enrich the ¹³C in DIC in the symbiont halo, Bijma et al. (1999) proposed another mechanism to produce the relative enrichment of ¹³C in DIC by symbiont photosynthesis. Their scenario is, in short, the reduction of the diffusive flux of the respired CO₂ from a foraminifer to the calcifying microenvironment. They proposed that the “scavenging” or utilization of the ¹³C-depleted respired CO₂ by symbiont photosynthesis can reduce the flux of the respired CO₂ that affects δ¹³C_DIC. This is based on the idea that the respired CO₂ commonly affects the foraminiferal calcite to some extent (Berger et al., 1978; Spero et al., 1991; Spero and Lea, 1996). If N. dutertrei contains endosymbionts, the higher δ¹³C of N. dutertrei relative to that of Gr. inflata can be explained by the effective ¹²CO₂ scavenging of its symbionts. Moreover, the much more ¹³C-enriched δ¹³C values of Gs. sacculifer can be explained by the additional effect of the microenvironmental change in δ¹³C attributable to the photosynthetic fractionation within the symbiont halo.

To explain this deviation in accordance with the foraminiferal ontogeny, the Δδ¹³C values are shown against the cumulative test masses (Figure 8). All three species showed negative Δδ¹³C_foram−DIC values in the earlier ontogenetic stages (<10 μg), and the largest negative Δδ¹³C values were observed in the earliest ontogenetic stage in all species. This reflects the higher respiration rate in the relatively warmer habitat the foraminifers experience during early ontogeny. Neogloboquadrina dutertrei and Gr. inflata showed asymptotic ¹³Cenrichment in δ¹³C values to near-zero deviation from δ¹³C_DIC during their later ontogeny (>10 μg) (Figure 8). We assume that this saturation in Δδ¹³C reflects the decline in metabolic activity as the foraminifers descended to colder waters. Such metabolic deterioration is also a basic principle of organismal ontogeny (Reiss, 1989; West et al., 2001). Thus, these common ontogenetic changes in metabolism also affected the observed asymptotic pattern of Δδ¹³C_foram−DIC. However, the Δδ¹³C_foram−DIC of Gs. sacculifer became positive and continued increasing (Figure 8). This profile probably reflects its photosymbiotic activity, which increases as the host grows (Spero and Parker, 1985; Spero et al., 1991; Spero and Lea, 1993; Wolf-Gladrow et al., 1999).

Interestingly, for N. dutertrei and Gr. inflata, the point of apparent saturation of Δδ¹³C can be assigned to a cumulative test mass of around 10 μg. At that point, the Δδ¹³C_foram−DIC of all three species was around 0‰. Recently, Birch et al. (2013) showed that the disequilibrium effect in δ¹³C becomes minimum when foraminiferal specimens of 212–355 μm in test size are used for analyses. Thus, they concluded that this test size window is preferred to reconstruct the past δ¹³C_DIC (Pearson, 2012; Birch et al., 2013). In this study, the cumulative test mass cannot be directly compared to the mesh-size range proposed in the study above. However, the important fact is that the individual test size, in this case 10 μg in test mass, reflects the δ¹³C_DIC of the water column well, regardless of habitat depths or symbiotic ecology. Thus, our results confirm and reinforce the conclusion of Birch et al. (2013) that a certain test size range or mass range can reconstruct past δ¹³C_DIC more accurately over other choices.