Effects of photosymbiosis and related processes on planktic foraminifera-bound nitrogen isotopes in South Atlantic sediments

Foraminifera often form symbiotic relationships with photosynthetic algae, providing a host environment and inorganic nutrients in exchange for photosynthetic organic matter from the algal symbiont. To date, the history of this relationship has been studied in paleoceanographic records using the oxygen and carbon stable isotopes of foraminiferal calcite. More recently, photosymbiotic activity has been observed to impact the nitrogen isotope ratio (δ¹⁵N) of foraminiferal tissue and the organic matter incorporated into foraminiferal tests. Dinoflagellate symbiont-bearing species appear to be lower in δ¹⁵N than symbiont-barren species and more similar to their feeding sources, likely due to their retention of low-δ¹⁵N metabolic ammonium and thus a weaker amplitude for the “trophic enrichment factor”, the δ¹⁵N increase per trophic level that is widely observed in food webs. We report new glacial–interglacial foraminifera-bound δ¹⁵N (FB-δ¹⁵N) data from Deep Sea Drilling Program Site 516, located in the subtropical South Atlantic gyre, which contains multiple foraminifera species at adequately high abundance for interspecies comparison of foraminiferal nitrogen, carbon, and oxygen isotopes over a full glacial cycle. Our data show a conserved δ¹⁵N difference of 3 ‰–5 ‰ between dinoflagellate-bearing species and the other species, qualitatively consistent with, but greater in amplitude than, the δ¹⁵N difference observed in previous modern ocean and core-top studies. We propose that this greater amplitude at Site 516 is the result of the lateral transport of symbiont-barren species into the South Atlantic subtropical gyre, which appears to represent a small region of low thermocline nitrate δ¹⁵N surrounded by regions with higher thermocline nitrate δ¹⁵N.

We demonstrate that FB-δ¹⁵N provides a clear signal of dinoflagellate endosymbiosis and that it may be able to identify other, weaker endosymbioses (e.g., with chrysophytes or pelagophytes). However, the data also suggest caution in regions with strong gradients, where species from contrasting environments may occur in a single sediment sample.

1 Introduction

Biologically available nitrogen (or “fixed N”) is an essential nutrient and thus central to ocean productivity and biogeochemical cycling (Sarmiento and Gruber, 2006), with links to atmospheric CO₂ and climate (Broecker, 1982; Broecker and Henderson, 1998; Falkowski, 1997; McElroy, 1983). The ocean's fixed N reservoir size is dominantly controlled by the balance between N₂ fixation and denitrification (Gruber and Sarmiento, 1997), which can be reconstructed by N isotopes (¹⁵N/¹⁴N, or δ¹⁵N) measured on organic matter in marine sediments (Brandes and Devol, 2002; Deutsch et al., 2004; Galbraith et al., 2013). In addition, variation in the degree of nitrate consumption in ocean surface waters can also be investigated with N isotopes in marine sediments (Altabet and Francois, 1994).

The isotopic composition of bulk N in marine sediments is variably influenced by diagenesis on the seabed and exogenous N inputs (e.g., Meckler et al., 2011; Robinson et al., 2012; Möbius, 2013; Schubert and Calvert, 2001). Largely to address these concerns, over the past decades, fossil-bound nitrogen isotope methods have been developed (Kast et al., 2022; Lueders-Dumont et al., 2018; Sigman et al., 1999; Wang et al., 2015; Wang et al., 2014; Brunelle et al., 2007; Robinson et al., 2004), including foraminifera-bound (FB-) δ¹⁵N (Ren et al., 2009). In contrast to bulk sediment δ¹⁵N, fossil-bound organic matter represents a specific N pool in the sediment archive that is physically protected from bacterial/chemical diagenesis, exogenous N contamination, and thermal stress by the biomineral matrix, and it appears to be minimally affected by partial calcite dissolution (Martínez-García et al., 2022). The first applications of the FB-δ¹⁵N proxy focused on reconstructing past changes in the N cycle in the North Atlantic (Ren et al., 2009; Straub et al., 2013), Pacific (Ren et al., 2012a, 2017, 2015), and Southern Ocean (Martínez-García et al., 2014) over Pleistocene glacial–interglacial cycles. More recent studies have shown that FB-δ¹⁵N can provide reliable information about changes in the N cycle through the Cenozoic (Auderset et al., 2022; Hess et al., 2023; Kast et al., 2019; Wang et al., 2022).

Studies in the oligotrophic ocean, where surface water nitrate is fully consumed, have established that FB-δ¹⁵N in surface sediments reflects the δ¹⁵N of shallow subsurface nitrate (i.e., the nitrate that is supplied to and consumed in the euphotic zone) (Ren et al., 2012b, 2009; Schiebel et al., 2018). Water column studies indicate seasonal changes in FB-δ¹⁵N that parallel the changes in the δ¹⁵N of bulk filtered particles (>0.7 µm) and/or net-tow-collected material (>150 µm) in surface waters (Smart et al., 2020, 2018), as is consistent with the role of planktic foraminifera as heterotrophic zooplankton that feed variously on phytoplankton, other heterotrophs, and other types of particulate N (PN) (e.g., Anderson et al., 1979; LeKieffre et al., 2020; Spindler et al., 1984; Takagi et al., 2019). Phytoplankton assimilate the nitrate supplied from below and provide the food to zooplankton such as foraminifera. The metabolism of zooplankton and other heterotrophs releases low-δ¹⁵N regenerated N (e.g., ammonium), which is avidly reassimilated by phytoplankton. The net result is that phytoplankton biomass, on average, is lower in δ¹⁵N than the nitrate supply (Altabet, 1988; Fawcett et al., 2011; Knapp et al., 2005), while herbivorous zooplankton are similar in δ¹⁵N to the nitrate supply (Montoya et al., 2002). Foraminifera, which feed on both zooplankton and phytoplankton, thus have a δ¹⁵N that varies with the δ¹⁵N of the nitrate supplied and consumed in the euphotic zone (Ren et al., 2009, 2012b; Smart et al., 2018; Schiebel et al., 2018).

Aside from the δ¹⁵N of their diet, some foraminifera species have a special aspect of their physiology that influences their δ¹⁵N: most extant spinose planktic foraminifera species host algal symbionts (Hemleben et al., 1989). A typical heterotrophic organism puts most of its dietary organic carbon toward energy generation, releasing the metabolized carbon as CO₂. If the C : N : P of the organism is not vastly different from that of its food, this will leave the organism with an excess of N (as well as P). The N must be removed, typically being lost by deamination and excretion of the resulting ammonium (or of urea or uric acid that is produced from the ammonium). Foraminifera, like other symbiont-bearing organisms (e.g., scleractinian corals), take advantage of their metabolic N and P waste to obtain additional organic carbon: they provide these nutrients to symbiotic algae, which use them to fix additional organic carbon, with a portion of this organic matter then being returned to the host for its nutrition. This strategy is particularly prominent in low-nutrient marine systems, where environmental nutrient concentrations are low and light is abundant, such as in the oligotrophic subtropical gyres.

Because decisive steps in N catabolism break an N-containing bond, which involves substantial isotopic fractionation, the catabolically produced ammonium is low in δ¹⁵N (Silfer et al., 1992). In normal heterotrophic organisms, the loss of this low-δ¹⁵N ammonium raises the δ¹⁵N of the organism, paraphrased by the isotope ecologist's rule of thumb: “You are what you eat plus a few (permille)” (DeNiro, 1976; DeNiro and Epstein, 1981). In dinoflagellate-bearing foraminifera, however, the dinoflagellates are thought to assimilate the low-δ¹⁵N ammonium, recycling it back into the host–symbiont system and thus reducing the δ¹⁵N elevation of the foraminifera relative to its food (Ren et al., 2009; Smart et al., 2018). In summary, because N excretion to the environment can be greatly reduced by photosynthetic endosymbiosis, foraminifera that host symbionts are expected to have a δ¹⁵N closer to that of their food source and lower than “pure” heterotrophs with the same food source.

The role of symbiosis in ancient organisms and ecosystems is of great interest. In low- to midlatitude surface waters of the ocean, limitation of autotrophic biomass by the availability of nutrients (especially N and P) is common, and changes in ocean nutrient conditions through time may have played a major role in evolution (Hohmann-Marriott and Blankenship, 2011). Photosynthetic endosymbiosis – hosted by organisms such as scleractinian corals and planktic foraminifera – has been described as an adaptation to settings of N- and P-limited photosynthetic growth, with the heterotrophic host sharing its metabolic N and P with the autotrophic symbionts in exchange for organic carbon (C) (e.g., Hallock, 1981). Thus, the origin and history of these symbioses may reflect changes in the ocean's nutrient inventories and cycling. In turn, these symbioses may affect nutrient cycling and the availability of nutrients, especially in systems such as shallow reefs.

Photosymbiosis in modern foraminifera species can be investigated by microscopic observations of intracellular algae (Anderson and Be, 1976; Gastrich, 1987; Lee et al., 1965), pulse-chase experiments with isotope labels (Bird et al., 2020; LeKieffre et al., 2020), active chlorophyll fluorometry (Takagi et al., 2019), or detection of algal DNA (Gast and Caron, 1996; Gast et al., 2000). However, these methods may be complicated by foraminiferal feeding on algal cells, making it difficult to distinguish endosymbionts from prey. More to the point, identifying and/or quantifying photosymbiotic activity from fossil foraminifera is even more challenging, particularly for extinct species. The carbon isotopic composition (δ¹³C) of the CaCO₃ of planktic foraminifera tests may be affected by photosymbiosis. Symbiotic photosynthetic organisms have a preference for taking up ¹²C, resulting in local dissolved inorganic carbon (DIC) becoming enriched in ¹³C, which is subsequently integrated into the calcium carbonate tests of foraminifera (Spero and DeNiro, 1987). In larger foraminifera specimens, which host a greater number of photosymbionts, there is a relatively higher enrichment of ¹³C than in smaller individuals (Spero et al., 1991). Thus, the relationship between the δ¹³C and test size of symbiont-bearing foraminifera has been proposed and applied as a metric by which to identify symbiosis in ancient foraminifera (D'Hondt et al., 1994; Edgar et al., 2013; Ezard et al., 2015; Norris, 1998). The oxygen isotopic composition (δ¹⁸O) of the CaCO₃ can be used to infer the depth habitat of different foraminifera species, with higher δ¹⁸O indicating deeper and colder water masses (e.g., Mulitza et al., 1997; Rashid and Boyle, 2007, and references therein). Thus, it has also been proposed that, due to persistent shallower depth habitats throughout the ontogeny of symbiont-bearing foraminifera, their size-specific δ¹⁸O relationship should be minimal in comparison to symbiont-barren foraminifera species, which have a larger potential range of depth habitats (Rebotim et al., 2017). However, there are potential pitfalls to these approaches: the foraminiferal test size can be influenced by environmental conditions such as surface water stratification, with smaller test sizes reported during periods of upwelling (Schmidt et al., 2004). Furthermore, sediment mixing preferentially affects finer fractions and thus impacts the size fraction vs. δ¹³C and δ¹⁸O relationships (Hupp et al., 2019). Both proxies, their biological mechanisms, and their pitfalls require further investigation.

This study investigates foraminifera-bound nitrogen isotopes (FB-δ¹⁵N) as a new tool for the reconstruction of past symbiosis in planktic foraminifera. We measure shell-bound δ¹⁵N in five foraminifera species from a South Atlantic sediment core over a full glacial cycle. We also analyze the same samples for test δ¹⁸O and δ¹³C. In addition, in a subset of samples, we measure δ¹⁸O and δ¹³C in different test size fractions.

Figure 1 Oceanographic context for Deep Sea Drilling Program (DSDP) Site 516. (a) Nitrate map (in µmol kg⁻¹) at 100 m water depth with the sample location of DSDP Site 516 for FB-δ¹⁵N (this study) and nitrate δ¹⁵N profiles from CLIMODE (CLIvar MOde water Dynamics Experiment; this study), BATS (Fawcett et al., 2015), and A13.5 (this study). Main surface currents and oceanographic fronts modified from Peterson and Stramma (1991), Wefer et al. (1996), and Schmitz and McCartney (1993). (b) Mean annual temperature cross section. Map and cross section generated with Ocean Data View and the data set from the World Ocean Atlas (Garcia et al., 2014; Schlitzer, 2015).

2 Material and methods

2.1 Core site and local hydrography

We analyzed samples from sediment core DSDP (Deep Sea Drilling Program) Site 516, located on the Rio Grande Rise (30°16^′ S, 35°17^′ W; 1313 m water depth) near the core of the oligotrophic subtropical gyre in the South Atlantic Ocean (Fig. 1). The local hydrography is dominated by the wind-driven anticyclonic South Atlantic gyre. The subtropical gyre is associated with net subduction, deepened isopycnals, slow nutrient supply from the subsurface to the euphotic zone, and N and P impoverishment in the euphotic zone (Cullen et al., 2002). The nutrient supply to the euphotic zone at Site 516 can be augmented by eddy mixing with the nutrient-bearing waters of the Antarctic Circumpolar Current, which flow east-northeastward along the southern boundary of the gyre (Reid et al., 1977; Schmid et al., 2000). The Brazil Current to the west may also contribute nutrients. However, Site 516 is near the core of the gyre and is relatively isolated from nutrient inputs at the margins. Accordingly, the annual average nitrate concentration ranges from 0.2 µmol kg⁻¹ to below the limit of detection, and chlorophyll a concentrations are <0.7 mg m⁻³ (Siccha et al., 2018). These conditions are comparable to those observed in the oligotrophic Sargasso Sea in the North Atlantic, as established for the Bermuda Atlantic Time-series Study (Lomas et al., 2013; Steinberg et al., 2001). The eastern boundary of the subtropical gyre is marked by the Benguela Current, where the persistent southerly and southeasterly winds drive surface water offshore and facilitate the upwelling of cold and nutrient-rich subsurface water (Cole and Villacastin, 2000). The Benguela Current flows northward and turns into the southern branch of the South Equatorial Current, which forms the northern edge of the subtropical gyre (Stramma, 1991).

2.2 Nitrogen isotope analysis

2.2.1 Foraminifer-bound nitrogen isotopes

Sediment samples were wet sieved through a 63 µm mesh and dried overnight in a clean oven at 40 °C. Between 600 and 800 individual foraminifera tests were manually picked from the 250–400 µm size fraction. We analyzed dinoflagellate-bearing Trilobatus sacculifer, the two chromotypes Globigerionoides ruber ruber and G. ruber albus (pink and white, respectively), chrysophyte-bearing Globigerinella siphonifera (synonym G. aequilateralis), and symbiont-barren Globigerina bulloides and Globorotalia truncatulinoides. The detailed protocol used here for measuring FB-δ¹⁵N can be found in Auderset et al. (2022) and Moretti et al. (2024). In short, about 5–7 mg of foraminifera tests were gently crushed and chemically treated to remove external organic matter, clay, and manganese coatings. Subsequently, 3–5 mg of the cleaned material was weighed out per sample for N content determination and, ultimately, N isotope analysis. The test fragments were first dissolved in hydrochloric acid (50 µL of 3N HCl), and the organic N was converted into nitrate (NO${}_{\mathrm{3}}^{-}$) by persulfate oxidation following a slightly modified protocol first used for foraminifera by Ren et al. (2009). The NO${}_{\mathrm{3}}^{-}$ was converted to nitrous oxide (N₂O) using denitrifying bacteria Pseudomonas chlororaphis, and its δ¹⁵N was measured by an automated, custom-built N₂O extraction system and inlet by continuous helium carrier flow to an isotope ratio mass spectrometer (Thermo MAT 253) (Casciotti et al., 2002; Sigman et al., 2001; Weigand et al., 2016). To quantify the precision and accuracy of the corrected isotope values, for each series of 30 samples, a total of three different in-house (Max Planck Institute for Chemistry – MPIC) calcite and aragonite laboratory standards were analyzed in triplicate, namely a coral standard from the taxon Porites (PO-1) with δ¹⁵N of 6.2±0.3 ‰, a coral standard from the taxon Lophelia (LO-1) with δ¹⁵N of 10.1±0.4 ‰, and a mixed foraminifera standard (MF-1) (63–315 µm size fraction) from the North Atlantic (MSM58-17-1; Repschläger et al., 2018) with δ¹⁵N of 5.92±0.28 ‰ (Moretti et al., 2024). After correcting for blank isotope composition, the precision of the multi-analysis average from each run is better than 0.2 ‰ (± 1 SD).

2.2.2 Seawater nitrate nitrogen isotopes

Seawater samples from the South Atlantic A13.5 section were collected from full-water-column hydrocasts, spanning 32.0 to 41.5° S. Acid-washed 60 mL high-density polyethylene bottles (Nalgene) were rinsed at least three times with sample water prior to filling and immediately frozen at −20 °C until analysis for NO${}_{\mathrm{3}}^{-}$ concentration and δ¹⁵N. For comparison, we also measured seawater from the upper 1000 m of the water column in the subtropical North Atlantic gyre, collected during the CLIMODE (CLIvar MOde water Dynamics Experiment) campaign in 2006 (Hutto et al., 2006). Sampling stations are located across the northern boundary of the subtropical gyre between roughly 36.45 and 38.6° N, which covers the sharp transition in water properties associated with the northern edge of the subtropical gyre (Fig. 1).

As with FB-δ¹⁵N analysis, the δ¹⁵N (vs. air) and δ¹⁸O (vs. Vienna Standard Mean Ocean Water – VSMOW) of NO${}_{\mathrm{3}}^{-}$ were determined using the denitrifier method (Casciotti et al., 2002; Sigman et al., 2001). For seawater, denitrifying bacteria Pseudomonas chlororaphis ssp. aureofaciens (ATCC 13985, Manassas, VA, USA) were used to quantitatively convert the NO${}_{\mathrm{3}}^{-}$ and nitrite (NO${}_{\mathrm{2}}^{-}$) in samples to N₂O. The product N₂O was analyzed as described above (Weigand et al., 2016). For the shallow-water samples, where NO${}_{\mathrm{2}}^{-}$ constitutes over 10 % of the total NO${}_{\mathrm{2}}^{-}+$ NO${}_{\mathrm{3}}^{-}$ pool, mostly in the upper 200 m, sulfamic acid was added to remove NO${}_{\mathrm{2}}^{-}$ following the protocol described by Granger and Sigman (2009). The pooled standard deviation of all NO${}_{\mathrm{3}}^{-}$-only and NO${}_{\mathrm{2}}^{-}+$ NO${}_{\mathrm{3}}^{-}{\mathit{\delta }}^{\mathrm{15}}$N measurements, respectively, were 0.24 ‰ (for shallow samples, <200 m) and 0.08 ‰ (for deeper samples). Two international potassium nitrate (KNO₃) reference materials, IAEA-N3 (δ¹⁵N of 4.7‰, δ¹⁸O of 25.6‰) and USGS-34 (δ¹⁵N of −1.8 ‰, δ¹⁸O of −27.9‰), were used for standardization, and a lab N₂O standard in helium was run in parallel to monitor the consistency of mass spectrometry.

2.3 Stable oxygen and carbon isotope analysis

For the down-core records of δ¹⁸O and δ¹³C, we picked samples of all five species from the 250–400 µm size fraction throughout the core. For reference, the largest test sizes for T. sacculifer and G. siphonifera were observed in the 400–630 µm size fraction, whereas the largest test sizes for G. truncatulinoides were found in the >630 µm size fraction. To investigate the size-fraction-specific δ¹³C and δ¹⁸O, we chose specimens with sizes of 200–250, 250–315, 315–400, and >400 µm from two climatically distinct (end-member) settings: an interglacial (129 ka; Marine Isotope Stage (MIS) 5) and glacial (27 ka; Last Glacial Maximum (LGM)) time period. There were not enough G. bulloides tests within the two largest size fractions (315–400 and >400 µm); thus, size-specific δ¹³C and δ¹⁸O could not be evaluated for this species. For each sample measured, 30 specimens were gently cracked open with a watch glass, submerged in aliquots of ethanol (pure grade), and put in an ultrasonic bath for 20 s. Subsequently, the ethanol was decanted and the samples were dried at 60 °C in a clean oven. Clean sample fragments were placed in 4.5 mL Exetainer vials, flushed with helium, and digested by anhydrous orthophosphoric acid (H₃PO₄) at 70 °C. The liberated CO₂–He mixture was transported to the Thermo Finnigan GasBench II preparation device with He as the carrier gas and analyzed using a Thermo Finnigan DELTA V mass spectrometer (Breitenbach and Bernasconi, 2011). For each series of 60 samples, four different calcite laboratory standards were analyzed in quintuplicate. CaCO₃ standard weights were chosen so that they spanned the entire range of sample weights. VU Internal Carbonate Standards (VICS) were used to correct for sample-size-based fractionation effects. VICS, International Atomic Energy Agency (IAEA) standards, and two in-house carbonate standards (travertine and Carrara marble) were used to detect mass spectrometer drift during the run and to quantify the precision and accuracy of the corrected isotope data. After correction for these effects, the reproducibility of the standards was typically better than 0.1 ‰ (± 1 SD) for both δ¹⁸O and δ¹³C.

2.4 Age model

The age model for DSDP 516 was produced by graphically aligning benthic foraminifera oxygen isotopes (Auderset et al., 2024) with global benthic stack LR04 by Lisiecki and Raymo (2005) using the AnalySeries software (Paillard et al., 1996).

Figure 2 The 160 kyr time series of species-specific nitrogen, carbon, and oxygen isotopes at DSDP Site 516. (a) Benthic foraminifera δ¹⁸O stack (Lisiecki and Stern, 2016). MIS denotes Marine Isotope Stage. (b) Species-specific FB-δ¹⁵N measured at DSDP 516 for G. bulloides (turquoise), G. siphonifera (orange), G. truncatulinoides (green), T. sacculifer (blue), G. ruber ruber (pink), and G. ruber albus (gray). (c) N content for each species. (d) Species-specific calcite-δ¹⁸O and (d) calcite-δ¹³C.

Figure 3 Box plots of species-specific nitrogen, carbon, and oxygen isotopes at DSDP Site 516. (a) Species-specific FB-δ¹⁵N values averaged over the last 160 kyr for G. bulloides (turquoise), G. siphonifera (orange), G. truncatulinoides (green), T. sacculifer (blue), G. ruber ruber (pink), and G. ruber albus (gray). (b) N content for each species. Panels (c) and (d) present the species-specific δ¹⁸O and δ¹³C, respectively.

3 Results

Down-core changes in FB-δ¹⁵N and δ¹³C are small relative to interspecies differences. The weak changes in δ¹⁵N across the last 160 kyr are discussed elsewhere (Auderset et al., 2024). Here, we focus on the interspecies differences (Figs. 2, 3). FB-δ¹⁵N data show two distinct groups: T. sacculifer, G. ruber albus, and G. ruber ruber of ca. 4.2 ‰ (±0.42), 4.1 ‰ (±0.43), and 4.3 ‰ (±0.37), respectively, vs. G. truncatulinoides, G. bulloides, and G. siphonifera, which show more elevated δ¹⁵N values of 8.0 ‰ (±0.67), 9.3 ‰ (±0.45), and 8.8 ‰ (±0.46), respectively (Figs. 2a, 3a). The N contents range between 3.8 and 4.5 nmol mg⁻¹ (Figs. 2b, 3b). We observe similar N contents throughout the record and among the species. The studied foraminifera are well preserved, and the stability of the N content through the record implies little to no diagenetic loss of FB-N after burial through the sediment mixed layer (Fig. S1).

The δ¹⁸O data show a fairly even spread between the different species (Figs. 2c, 3c). Trilobatus sacculifer and G. ruber (albus and ruber) are more depleted in ¹⁸O (with δ¹⁸O values of 0.3 ‰–0.5 ‰) compared with G. truncatulinoides, G. bulloides, and G. siphonifera (1.1 ‰–1.9 ‰). Over the last 160 kyr, T. sacculifer, G. ruber (albus and ruber), and G. siphonifera generally follow the same trend. These species show a temporal variation of ∼0.7 ‰–1 ‰ with a maximum at 40 ka and minimum at 90 ka (Figs. 2c, 3c). Globorotalia truncatulinoides and G. bulloides show a smaller amplitude of variation of ∼0.5 ‰ and no clear δ¹⁸O maximum/minimum (Figs. 2c, 3c). Apart from G. bulloides with a δ¹³C range between −1.6 ‰ and −0.3 ‰, all species range between 0.6 ‰ and 1.7 ‰. Globigerina bulloides δ¹³C shows a substantial negative offset of ca. 1.5 ‰ from the other species and is more variable than all of the other species analyzed here (Figs. 2d, 3d).

All species investigated appear to show a positive relationship between test carbonate δ¹³C and test size, although some reversals occur in the largest (400 µm) size fraction (data for G. bulloides are for the two smallest size fractions only) (Fig. S2). We record different slopes of the linear fit in LGM and MIS 5 samples (the two end-member settings) for T. sacculifer and G. ruber (albus and ruber) (Table S1). The Pearson correlation coefficient (r value) for the LGM ranges between 0.60 and 0.98, whereas it is between 0.68 and 0.96 for MIS 5, with G. siphonifera and G. truncatulinoides having the lowest r values among the studied species. The slopes for δ¹³C vs. test size remain relatively constant for G. siphonifera (0.0013 vs. 0.0024) and G. truncatulinoides (0.0043 vs. 0.0038) between the LGM and MIS 5. However, the strength of the relationship differs between the two intervals, with r values for G. siphonifera increasing from 0.60 during the LGM to 0.86 in MIS 5, and values for G. truncatulinoides decreasing from 0.91 to 0.68 during the same period. T. sacculifer exhibits the highest average slope for both time slices (0.053), while G. siphonifera records the lowest (0.018). Notably, significant relationships (p<0.05) for δ¹³C are observed in G. ruber albus (p=0.02) and G. ruber ruber (p=0.03) during the LGM and in T. sacculifer (p=0.04) during MIS 5.

In contrast to δ¹³C, the test carbonate δ¹⁸O shows only a weak correlation with test size (Table S1), and the relationship varies down the core and between species (Fig. S3). The only significant relationship is observed for G. ruber albus in the combined time slice (p=0.02).

Seawater nitrate nitrogen isotopes and nitrate concentration profiles from CLIMODE and A13.5 stations are shown in Fig. S4 and discussed in the Supplement.

4 Discussion

4.1 FB-δ¹⁵N at DSDP Site 516

In the planktic foraminifera species T. sacculifer and G. ruber, dinoflagellate symbionts may recycle low-δ¹⁵N ammonium and thus keep FB-δ¹⁵N low, preventing the full ca. 3 ‰ elevation characteristic of a trophic level increase (Minagawa and Wada, 1984; Ren et al., 2012b; Smart et al., 2018). Therefore, the δ¹⁵N of T. sacculifer and G. ruber should more closely match the δ¹⁵N of their food source than in species without dinoflagellate symbionts. In oligotrophic areas, such as the subtropical South Atlantic gyre, surface nitrate is fully consumed by phytoplankton. With the role of euphotic zone N recycling described in Sect. 1 (Altabet, 1988; Fawcett et al., 2011), this leads to herbivorous zooplankton close to or slightly higher than the δ¹⁵N of this consumed nitrate, e.g., 3.6±1.0 ‰ (n=107) for zooplankton vs. 2.6 ‰ for thermocline nitrate in the Sargasso Sea (Montoya et al., 2002; Smart et al., 2018). With foraminifera feeding on a mixture of small and large phytoplankton and zooplankton, this yields a δ¹⁵N for the foraminifera diet that is close to the δ¹⁵N of the subsurface nitrate supply (Fawcett et al., 2011). At DSDP Site 516, the FB-δ¹⁵N of the dinoflagellate-bearing species T. sacculifer and G. ruber (Gastrich, 1987; Hemleben et al., 1989) ranges between 4 ‰ and 5 ‰. This value is lower than global mean pycnocline nitrate (of 6.25 ‰; Fripiat et al., 2021), suggesting that regional subsurface nitrate δ¹⁵N has been lowered by N₂ fixation, as has been observed in other subtropical gyres (Harms et al., 2019; Knapp et al., 2008; Marshall et al., 2022; Casciotti et al., 2008; Liu et al., 1996; Yoshikawa et al., 2015), especially the North Atlantic (Marconi et al., 2015, 2017). For comparison, in the Sargasso Sea – an ocean region with persistent N₂ fixation – shallow thermocline nitrate δ¹⁵N is 2.6 ‰ (Fawcett et al., 2015) and surface sediment FB-δ¹⁵N values are 2.4 ‰ (T. sacculifer) and 2.6 ‰ (G. ruber) (Smart et al., 2018).

In G. truncatulinoides, the bulk of the shell derives from calcification below 100 m, indicating that they spend much of their lives below the euphotic zone, where light conditions are unsuitable for photosymbionts (Reynolds et al., 2018; Schiebel and Hemleben, 2017). This is consistent with their high δ¹⁵N compared with dinoflagellate symbiont bearers. G. bulloides and G. siphonifera are more enriched in ¹⁵N than the other analyzed species at DSDP Site 516. This may indicate minimal nitrogen recycling associated with non-dinoflagellate symbionts, e.g., chrysophytes for G. siphonifera, such that this species groups with the symbiont-barren species (Faber et al., 1989, 1988; Ren et al., 2012b; Smart et al., 2018; Granger et al., 2024). As for the higher δ¹⁵N of G. bulloides and G. siphonifera relative to G. truncatulinoides, this might be explained by a higher δ¹⁵N for their food source. This possibility is pursued in Sect. 4.4.

Figure 4 Relationships between oxygen, carbon, and nitrogen isotopes. Interspecies comparison between G. ruber (albus and ruber), T. sacculifer, G. truncatulinoides, G. siphonifera, and G. bulloides at DSDP Site 516. Arrows indicate the distinction between shallow (warm) vs. deep (cold) dweller, remineralization vs. photosynthesis in the water column, and non-dinoflagellate-bearing/non-symbiotic vs. dinoflagellate-bearing foraminifera.

4.2 Comparison with carbonate oxygen and carbon isotopes at DSDP Site 516

To gain additional insight into the controls on FB-δ¹⁵N, we compare the data with our measurements of carbonate δ¹⁸O and δ¹³C of the foraminiferal shells. The δ¹⁸O differences among species have been used to reconstruct depth habitat (Fairbanks and Wiebe, 1980; Ravelo and Fairbanks, 1980). Minimum δ¹⁸O values in foraminiferal assemblages point to warmer waters and shallower depth habitats, which are likely to be required or at least best-suited for photosymbiont-bearing foraminifera (Fig. 4a) (see also the Supplement). By cross-plotting FB-δ¹⁵N and δ¹⁸O from Site 516, we see two distinct groups: (1) shallow dwellers with dinoflagellate symbionts with lighter FB-δ¹⁵N and δ¹⁸O and (2) deeper dwellers with non-dinoflagellate symbionts/symbiont-barren species recording heavier FB-δ¹⁵N and δ¹⁸O (Fig. 4b). Thus, unsurprisingly, the measurements are consistent with a shallow depth habitat for photosymbiosis.

The controls on the δ¹³C of planktic foraminiferal calcite are a matter of debate (Spero and Williams, 1989; Spero et al., 1991); however, size-specific δ¹³C measurements have been proposed as a proxy for photosymbiosis. Foraminiferal δ¹³C is likely sensitive to both environmental- and organism-scale δ¹³C gradients (induced by photosymbiont activity and foraminifera respiration), with a ¹³C enrichment for increased photosynthetic activity (Spero et al., 1997; Spero and DeNiro, 1987). However, the δ¹³C of the deepest dwelling, symbiont-barren species G. truncatulinoides at DSDP Site 516 does not distinguish it from even the most photosymbiotically active species (Figs. 2d, 3d). This is because depth habitat, variation in dissolved inorganic carbon, remineralization, and other environmental factors also influence carbonate δ¹³C, such that it alone is unlikely to be a reliable proxy for photosymbiosis. Surface mixed-layer DIC is typically enriched in ¹³C due to photosynthesis, whereas subsurface water DIC δ¹³C is lower due to bacterial respiration and remineralization of the low-δ¹³C organic matter arriving from the euphotic zone (Kroopnick, 1985). In a stratified ocean, such as the South Atlantic subtropical gyre, shallow-dwelling foraminifera should thus incorporate a higher δ¹³C in contrast to thermocline dwellers with lower δ¹³C. However, at DSDP Site 516, modern surface ocean δ¹³C_DIC seems to be only marginally higher (∼1.7 ‰) than the δ¹³C_DIC in the upper thermocline (∼1.4 ‰) (Fig. S7), which could explain why the calcite δ¹³C values of T. sacculifer and G. truncatulinoides at Site 516 are so similar. In our measurements, the shallow subsurface/thermocline dweller G. bulloides is the only species to be clearly lower in δ¹³C than the other foraminifera. It has been speculated that the presence of metabolically active cyanobacterial endobionts (Synechococcus) affects the carbon isotopes measured in the host G. bulloides through respiration, instead of photosynthesis from the endobiont, leading to ¹³C depletion instead of ¹³C enrichment in the calcifying microenvironment (Bird et al., 2017; Febvre-Chevalier, 1971; Spero and Lea, 1996). Alternatively, we suspect that this offset could be due to other effects associated with seasonality (Sect. 4.4.2) or a different environmental baseline for foraminifera originating from outside the South Atlantic gyre (Sect. 4.4.3).

In any case, the cross-plot between FB-δ¹⁵N and δ¹³C strongly discriminates between G. bulloides vs. the other foraminifera analyzed at Site 516 (Fig. 4c). Interestingly, G. siphonifera clusters in the same group as G. truncatulinoides, although both species are very different in terms of depth habitat and symbiotic relationship. Altogether, the δ¹⁵N–δ¹³C comparison might be interpreted as indicating that the non-dinoflagellate symbionts have a significant but variable effect on carbonate δ¹³C while having no appreciable effect on FB-δ¹⁵N (Fig. 4c). However, this view is complicated for δ¹³C by the observation that G. truncatulinoides, which is a symbiont-barren species, has a similar δ¹³C to species with a range of symbionts.

An alternative method proposed for the detection of photosymbiosis in fossil foraminifera is through the relationship of foraminiferal carbonate δ¹³C with test size (Spero and DeNiro, 1987). The argument is that the preferential uptake of ¹²CO₂ by the photosynthesizing symbionts raises the δ¹³C of the internal DIC pool from which the calcification proceeds (Spero et al., 1997). Larger individual foraminifera can host more symbionts per surface area of the individual, resulting in δ¹³C elevation compared with smaller foraminifera (Spero et al., 1991). Thus, a steeper rise in δ¹³C with increasing specimen size (steeper positive slope) indicates potentially higher photosymbiotic activity (Edgar et al., 2013; Norris, 1996; Shaw et al., 2021). According to this method, dinoflagellate-bearing G. ruber and T. sacculifer and potentially chrysophyte-hosting G. siphonifera would be predicted to have a steeper rise in δ¹³C with size than G. truncatulinoides or G. bulloides. However, for all investigated species except G. siphonifera, our measurements indicate similar slopes in δ¹³C vs. size, with dinoflagellate-bearing species T. sacculifer, and G. ruber (albus and ruber) showing no statistically significant difference in slopes compared to the non-symbiotic species G. truncatulinoides (Fig. S2, Table S1). The weak/near-zero slope for G. siphonifera δ¹³C (Fig. S2e) could be the result of the size fraction between 125 and 400 µm used in this study not capturing the larger, adult life stages of this species; we also observed a small number of G. siphonifera in the 400–630 µm fraction but did not measure them. Nevertheless, the anomalously weak/near-zero slope for G. siphonifera δ¹³C has been observed before. Bijma et al. (1998) showed a steeper slope for G. siphonifera type II than type I. They proposed that G. siphonifera type-I symbionts are less efficient at removing low-δ¹³C host-respired CO₂, allowing respiration to partially counteract the photosynthesis-induced δ¹³C rise in the microenvironment and shells during ontogeny. The significant contribution of host-respired CO₂ to symbionts in G. siphonifera type II is supported by the recent experimental study by Takagi et al. (2022).

That G. truncatulinoides falls into the same range of slopes (as well as absolute δ¹³C values) as T. sacculifer and G. ruber and has a steeper slope than G. siphonifera indicates that either the size fraction was too low and we observe the difference between juveniles and adults in G. truncatulinoides or that the previously proposed symbiosis metric of δ¹³C vs. size may not be sensitive enough to capture all photosymbiotic relationships.

However, using FB-δ¹⁵N as a symbiosis proxy may also have its complexities. In particular, the FB-δ¹⁵N differences between foraminifera with and without dinoflagellate symbionts is notably greater at DSDP Site 516 than reported at other sites across the global ocean (Costa et al., 2016; Martínez-García et al., 2014; Ren et al., 2012b; Schiebel et al., 2018; Smart et al., 2020, 2018). Interspecies FB-δ¹⁵N relationships might vary due to processes in the water column, such as seasonal changes in the supply of N affecting the δ¹⁵N of the food source (Smart et al., 2020, 2018) or lateral transport (Granger et al., 2024). In the next section, we compare DSDP Site 516 with core-top samples from geographically distinct regions to examine the variability in FB-δ¹⁵N differences amongst foraminifera species relative to the oceanographic environment.

Figure 5 Compilation of FB-δ¹⁵N values and offsets between dinoflagellate-bearing and non-dinoflagellate-bearing foraminifera from different locations. (a) Foraminifera from core tops in the Sargasso Sea (BATS) (Smart et al., 2018), Great Bahama Banks, Indonesia, Hawaii, South China Sea, Aotearoa / New Zealand (Ren et al., 2012b), and equatorial Atlantic 1–12 (Schiebel et al., 2018) as well as Holocene samples from the equatorial Pacific a–f (Costa et al., 2016), Ocean Drilling Program (ODP) Site 1090 (Martínez-García et al., 2014), ODP Site 662 (Auderset et al., 2024) and DSDP Site 516 (this study) (see Table S2 and Fig. S5 for more information about core sites). (b) Across all sites in panel (a), the average FB-δ¹⁵N offset between and dinoflagellate-barren and dinoflagellate-bearing foraminifera is categorized by symbiont relationship and proposed intensity of internal ammonium recycling. An asterisk (*) denotes suspected chrysophyte symbiont-bearing or symbiont-barren species, whereas two asterisks (**) denote an unknown symbiotic status.

4.3 FB-δ¹⁵N as a new proxy for photosymbiosis?

To further advance our understanding of FB-δ¹⁵N and its suitability as a metric of photosymbiosis, we now broaden our view beyond DSDP Site 516. We compile all available and published core-top FB-δ¹⁵N data that contain dinoflagellate-bearing and non-dinoflagellate-bearing foraminifera (Fig. S5, Table S2). The calculated offsets between the two categories across all core sites show that non-dinoflagellate-bearing foraminifera (Globorotalia menardii, Neogloboquadrina dutertrei, G. siphonifera, Globorotalia tumida, Globorotalia hirsuta, Globorotalia inflata, G. truncatulinoides, and G. bulloides) are consistently enriched in ¹⁵N compared with dinoflagellate-bearing foraminifera (G. ruber, T. sacculifer, and O. universa) (Fig. 5a). For the most part, FB-δ¹⁵N offsets are consistent across wide geographic ranges of the core sites over different latitudes and biogeochemical regimes. With this compilation, we first consider the average FB-δ¹⁵N values that arise for different species and their implications for FB-δ¹⁵N differences among different symbioses (Figs. 5b, 6).

Figure 6 Proposed interactions and processes influencing planktic foraminifera-bound nitrogen isotopes (FB-δ¹⁵N) in relation to symbiotic status and food sources. Symbiont-barren foraminifera or those hosting symbionts other than dinoflagellates record elevated FB-δ¹⁵N relative to their food source, as with typical zooplankton. In contrast, symbiont-bearing foraminifera exhibit lower FB-δ¹⁵N values due to their association with photosymbionts, which internally reassimilate the low-δ¹⁵N metabolic ammonium (NH${}_{\mathrm{4}}^{+}$) from the foraminiferal host in their photosynthetic growth, preventing or reducing the δ¹⁵N elevation that typically results from heterotrophy.

Figure 7 Chlorophyll a concentrations in living planktic foraminifera vs. FB-δ¹⁵N offsets between dinoflagellate-bearing and other non-dinoflagellate-bearing foraminifera. Chlorophyll a/biomass values were measured in foraminifera collected in the central and western Pacific Ocean and the tropical eastern Atlantic Ocean across different seasons (Takagi et al., 2019) and compared to the median of the FB-δ¹⁵N offsets in our core-top compilation (Figs. 5, S5; Table S2).

The calculated mean FB-δ¹⁵N offsets between dinoflagellate bearing and non-dinoflagellate/symbiont-barren foraminifera from the global compilation correlate with chlorophyll a (Chl a) measurements by Takagi et al. (2019). They analyzed living foraminifera from the central and western Pacific Ocean and the tropical eastern Atlantic Ocean across different seasons and estimated the intensity of photosymbiosis (Fig. 7). Based on Takagi et al. (2019), O. universa, T. sacculifer, G. ruber ruber, and G. ruber albus have a higher percentage of intracellular Chl a and thus photosymbiont activity. This is in contrast to G. truncatulinoides, G. bulloides, and G. inflata, with little/no Chl a inside of the tests, which record the highest FB-δ¹⁵N compared with dinoflagellate-bearing foraminifera (G. ruber, T. sacculifer, and O. universa) (Fig. 5b). Neogloboquadrina dutertrei and G. menardii are consistently higher in Chl a/biomass and lower with respect to their FB-δ¹⁵N offset than photosymbiont-barren G. bulloides and G. truncatulinoides. However, the strength of the Chl a/biomass vs. FB-δ¹⁵N relationship is not extraordinarily strong (r=0.72), suggesting that the Chl a content is not the only factor driving FB-δ¹⁵N (Fig. 7).

Globigerinella siphonifera stands out in the correlation of Chl a/biomass in the global compilation (Fig. 7), as it records similar levels of Chl a/biomass compared with some dinoflagellate symbiont-bearing foraminifera G. ruber (ruber and albus), despite their high FB-δ¹⁵N offset from dinoflagellate-hosting foraminifera reported at DSDP Site 516 (Fig. 3a) and in previous studies (Li et al., 2019; Ren et al., 2012b; Smart et al., 2018; Granger et al., 2024). Takagi et al. (2016) find that the chlorophyll content of G. siphonifera peaks before the final chamber formation, resulting in a minimum chlorophyll content during formation of the largest–second-largest chamber, which likely dominate calcite mass (and thus geochemistry). This may reflect digestion of its symbionts prior to gametogenesis (Faber et al., 1988), making G. siphonifera a symbiont-barren species in its later stages of calcification. In any case, this ontogenetic change could explain why FB-δ¹⁵N would identify G. siphonifera as having a weaker host–symbiont N cycle. As an alternative or additional explanation, the high FB-δ¹⁵N and relatively high Chl a of G. siphonifera may indicate a less active symbiotic relationship, whether due to less efficient internal ammonium recycling (Smart et al., 2018), low chrysophyte growth rates, and/or higher harvesting rates.

All such interpretations, however, are currently uncertain. Across species of foraminifera, symbiont physiology, photosynthetic rates, and host–symbiont interactions may vary, with unforeseen impacts on the N isotopes. Moreover, fixed N may enter foraminifera through more than just feeding. Depending on the physiology of the endosymbionts, their host, and environmental conditions, direct nitrate uptake by the foraminifera host and/or its symbionts is a possibility (Piña-Ochoa et al., 2010; Uhle et al., 1999).

Regardless, the uniquely low FB-δ¹⁵N of the dinoflagellate-bearing species supports higher photosynthetic rates (averaged over their lifetime) in dinoflagellate symbionts and, thus, a generally more important role for the symbiosis in the dinoflagellate-bearing species (e.g., as in O. universa; LeKieffre et al., 2020). The shallow depth (i.e., high-light) habitats of the dinoflagellate species are fully consistent with this interpretation, as is the dominance of dinoflagellate symbioses among modern symbiotic corals (Davy et al., 2012).

4.4 Deviations from the FB-δ¹⁵N photosymbiosis paradigm at Site 516

Despite the overall consistency observed across sites in the global FB-δ¹⁵N compilation, there are notable differences in interspecies offsets between the global compilation and DSDP Site 516 (Fig. 5a). In particular, among the foraminiferal species analyzed, the FB-δ¹⁵N differences between the three dinoflagellate-bearing species and the three other species (G. truncatulinoides, G. bulloides, and G. siphonifera) are unusually great at DSDP Site 516 (Fig. 8). These differences warrant investigation here, with a focus on their cause and thus their implications for interspecies FB-δ¹⁵N differences as a paleoproxy for foraminiferal photosymbiosis.

Figure 8 FB-δ¹⁵N offsets between non-dinoflagellate-bearing foraminifera vs. dinoflagellate-bearing foraminifera (the average of G. ruber albus, T. sacculifer, and O. universa), with DSDP Site 516 having uniquely high FB-δ¹⁵N offsets. DSDP Site 516 exhibits stronger ¹⁵N enrichment for non-dinoflagellate-bearing foraminifera than other sites from the global core-top compilation (see Table S1 and Fig. S5).

4.4.1 Depth habitat

The isotopic composition of suspended PN varies systematically with depth in the water column (Altabet, 1988; Altabet et al., 1991). Suspended PN has its lowest δ¹⁵N in the euphotic layer, increasing with depth below the euphotic zone as a consequence of the preferential release of ¹⁴N ammonium during microbial decomposition. The PN δ¹⁵N increase from the surface to ∼500 m depth can be up to 6 ‰ (Altabet, 1988; Altabet et al., 1991; Hannides et al., 2013). Therefore, planktic foraminifera living below the euphotic zone might be expected to have a higher δ¹⁵N for their feeding source than surface dwellers.

At Site 516, our N isotope results show clear differences between two groups of species. The group of G. truncatulinoides, G. bulloides, and G. siphonifera have a δ¹⁵N composition that is 3 ‰ to 5 ‰ higher than the group of T. sacculifer and G. ruber. These observations are in good agreement with previous studies (Ren et al., 2012b; Smart et al., 2018), including samples from upper-ocean net tows (surface 200 m), moored sediment traps, core tops, and down-core sediments. Planktic foraminifera living below the euphotic zone, when feeding on suspended PN, such as G. truncatulinoides, could incorporate food that is enriched in ¹⁵N (Mintenbeck et al., 2007). To date, there are not yet any published depth profile data for PN or nitrate isotopes close to core Site 516. Based on data from the North Atlantic (Altabet, 1988; Altabet et al., 1991) and South Atlantic surface particles (Mino et al., 2002), PN δ¹⁵N may be up to 9 ‰ higher at 1000 m water depth.

Figure 9 Estimation of the N isotopes of the South Atlantic gyre based on North Atlantic gyre nitrate isotopic measurements. (a) Seawater nitrate δ¹⁵N in the North Atlantic gyre system, with distinct isotopic signatures outside (black diamonds, CLIMODE; this study) and inside (blue diamonds, BATS; Fawcett et al., 2015) the gyre system. (b) Seawater nitrate δ¹⁵N from outside the South Atlantic gyre system (black diamonds, profile A13.5; this study). The hatched area with blue lines indicates the inferred possible seawater nitrate δ¹⁵N for a location within the South Atlantic gyre, based on the difference between inside and outside the North Atlantic gyre and the nitrate isotopic data from the South Atlantic. Colored circles indicate Holocene FB-δ¹⁵N from Site 516. Locations for seawater nitrate δ¹⁵N profiles are shown in Fig. 1.

However, several lines of evidence suggest that depth habitat is not the main cause of the difference in δ¹⁵N between the two groups. The two morpho- and genotypes of G. siphonifera (types I and II) are surface-to-thermocline dwellers (Bijma et al., 1990; Schiebel and Hemleben, 2017), but they have a higher δ¹⁵N than the deeper-dwelling G. truncatulinoides (Hemleben et al., 1985) (Fig. 2a). Other differences in depth habitat inferred from foraminiferal δ¹⁸O and δ¹³C (Figs. S6 and S7 in the Supplement) are also not consistent with the interspecies δ¹⁵N offsets that would be expected if depth were the main driver of FB-δ¹⁵N. For example, of the species that we measured at Site 516, G. truncatulinoides is the deepest dweller according to its δ¹⁸O values, but it has a δ¹⁵N that is lower than that of G. siphonifera and G. bulloides. In addition, at DSDP Site 516, G. siphonifera δ¹⁵N is more similar to G. truncatulinoides during interglacials but more similar to G. bulloides during glacials. However, the δ¹⁸O of G. siphonifera maintains its position in δ¹⁸O space relative to T. sacculifer and G. ruber, which would indicate no major change in the relative depth habitat of G. siphonifera through the glacial cycle. These observations suggest that depth habitat per se is not the main driver of the grouping that we see in the FB-δ¹⁵N data. Depth also cannot explain the larger-than-average FB-δ¹⁵N offset at Site 516, given the higher FB-δ¹⁵N values for G. siphonifera and G. bulloides than for G. truncatulinoides.

4.4.2 Seasonality

Seasonality can also play a role in determining which nitrogen isotope signals are incorporated by foraminifera. While total foraminifera production often follows the seasonality of phytoplankton productivity in a given environment, there are major differences among species (Schiebel et al., 2001). In temperate latitudes and parts of the subtropical gyres, most nitrate supply from below occurs during the winter, with a nitrate-fueled phytoplankton bloom in the spring giving way to intensive ammonium recycling between phytoplankton and zooplankton in the summer–autumn (Dugdale and Goering, 1967; Eppley and Peterson, 1979). Depending on the most productive season of a given species of planktic foraminifera, it may feed within a nitrate-based or ammonium-based ecosystem, with the latter being associated with a lower δ¹⁵N for suspended PN, zooplankton, and sinking PN (Fawcett et al., 2011). This distinction is apparent in the seasonality of foraminifera δ¹⁵N (Smart et al., 2020, 2018). Thus, if we assume the seasonality of production is similar in the South Atlantic compared with the North Atlantic gyre, we would expect Site 516 summer bloomers T. sacculifer and G. ruber to be lower in δ¹⁵N than G. truncatulinoides, which subsist in deep waters year-round but exhibit peak fluxes to the seafloor in winter–spring (Deuser, 1987; Hemleben et al., 1985; Salmon et al., 2015) when they reproduce in surface waters around the time of peak phytoplankton production (Schiebel et al., 2002; Reynolds et al., 2018). However, the offset that we observe between T. sacculifer and G. ruber vs. G. truncatulinoides is 4 ‰–4.5 ‰, which is substantially higher than typical seasonal offsets between summer and winter PN δ¹⁵N in oligotrophic regions (e.g., Smart et al., 2018). Thus, if seasonality is the explanation for the particularly large FB-δ¹⁵N interspecies offsets at DSDP Site 516, it would appear that some special source of seasonality is required for the South Atlantic subtropical gyre.

4.4.3 Lateral transport

In austral winter–spring, between July and September, the oligotrophic South Atlantic gyre is affected by wind-driven mixing and an Ekman-transport-driven incursion of nutrient-rich waters with high-δ¹⁵N nitrate from the Southern Ocean (Figs. 1, S8). Globigerina bulloides, perhaps the most opportunistic among the species discussed here, is abundant at the southern margin of the Atlantic subtropical gyre around 35° S (Boltovskoy, 1962), thus recording the higher δ¹⁵N associated with high degrees of consumption of a nitrate source with a δ¹⁵N that is similar to that of Subantarctic Mode Water (Smart et al., 2015). Indeed, it has been shown that G. bulloides reaches highest abundances when food supply is increased (Kretschmer et al., 2018; Mortyn and Charles, 2003; Schiebel et al., 1995). Globigerinella siphonifera, although it is a subtropical species, similarly yields the highest shell fluxes at times of highest productivity (Jonkers and Kučera, 2015), which would be the winter–spring season in the South Atlantic. For both G. bulloides and G. siphonifera, we observed highly variable specimen counts across the last 160 kyr (Fig. S9), in strong contrast to the uniformly high abundances of T. sacculifer, G. ruber albus, and G. truncatulinoides throughout the record. Therefore, the large FB-δ¹⁵N offset between summer-blooming species (T. sacculifer and G. ruber) and G. bulloides and G. siphonifera at Site 516 could reflect increased nutrient influx from the Southern Ocean with an elevated δ¹⁵N signature in austral winter–spring during northward movement of the subtropical front. As previously documented, not only nutrients but also planktic foraminifera and/or their particulate food sources (PN) can be transported from other oceanic regions to the South Atlantic (Peeters et al., 2004; Smart et al., 2020; Granger et al., 2024).

Here, we compare the nitrate profiles for “inside-gyre” (BATS; Fawcett et al., 2015) vs. “outside-gyre” (CLIMODE; this study) stations in the North Atlantic (Fig. 9a) to evaluate baseline (nitrate δ¹⁵N) differences between the two regions, providing a reference point for the baseline (nitrate δ¹⁵N) differences that we could expect in the South Atlantic. Subsurface nitrate at 200 m inside the North Atlantic gyre is ca. 3 ‰ lower than subsurface nitrate outside the gyre, which is due to gyre-related N₂-fixation processes keeping thermocline nitrate δ¹⁵N low. A similar upper-ocean nitrate δ¹⁵N gradient is observed in other subtropical gyres in the Indian and Pacific oceans, with ∼2.5 ‰ lower δ¹⁵N inside the gyre vs. outside (Harms et al., 2019; Yoshikawa et al., 2018, 2015; Marconi et al., 2024; Marshall et al., 2023). While no similar data exist from inside the South Atlantic gyre, a hypothetical inside-gyre δ¹⁵N profile based on the southern outside gyre would put the subsurface nitrate δ¹⁵N at 4 ‰ (inside gyre) vs. 7 ‰ (outside gyre) (Fig. 9b), closely matching the FB-δ¹⁵N of T. sacculifer and G. ruber at Site 516. Symbiont-barren G. truncatulinoides has the expected ∼3 ‰ δ¹⁵N offset above inside-gyre nitrate, i.e., the full trophic enrichment typically observed for marine zooplankton (Minagawa and Wada, 1984). G. bulloides and G. siphonifera, in contrast, appear to be elevated by an additional 1 ‰–2 ‰. This additional offset could be due to a Southern Ocean (i.e., outside-gyre) influence, where subsurface nitrate derived from Subantarctic Mode Water is enriched in ¹⁵N due to incomplete nutrient consumption (Sigman et al., 2000; Smart et al., 2015). Indeed, the distinct δ¹³C of G. bulloides has been interpreted as reflecting its Southern Ocean origin (Fig. S7 in the Supplement; Ninnemann and Charles, 1997). In our interpretation, both the relatively high δ¹⁵N of subantarctic-sourced nitrate and the greater potential for the incursion of extra-subtropical water into the small South Atlantic gyre contribute to the abnormally large FB-δ¹⁵N difference at DSDP Site 516 between G. bulloides/G. siphonifera and the dinoflagellate-bearing species. The South Atlantic may be more prone to these effects than the North Atlantic gyre or other sites from the core-top compilation due to the smaller geographic extent of the gyre and/or its proximity to the Southern Ocean.

Figure 10 Proposed different processes affecting FB-δ¹⁵N at DSDP Site 516. Shallow-dwelling, summer-blooming, dinoflagellate-bearing foraminifera located within the oligotrophic gyre record generally lower FB-δ¹⁵N values than symbiont-barren foraminifera or foraminifera hosting symbionts other than dinoflagellates that generally live in a deeper (depth) habitat, bloom during the winter/spring season, and/or live outside oligotrophic gyres. PON stands for particulate organic nitrogen.

5 Conclusions

FB-δ¹⁵N is a promising new geochemical tool to identify fossil foraminiferal photosymbiosis, particularly the strong symbiosis with dinoflagellates, and also has the potential to identify symbiosis with non-dinoflagellate symbionts (e.g., chrysophytes and pelagophytes) (Figs. 5, 6, 10). The FB-δ¹⁵N range of the non-dinoflagellate symbiont group may indicate that non-dinoflagellate symbioses in foraminifera are relatively weak in terms of the internal N demand that they generate, which could indicate that they generate substantially less photosynthetic energy for the host compared with dinoflagellate endosymbionts. This possibility warrants testing with other approaches.

While FB-δ¹⁵N offsets are remarkably consistent in our global compilation, DSDP Site 516 stands out for its larger-than-expected FB-δ¹⁵N offsets between the dinoflagellate-bearing foraminifera (G. ruber and T. sacculifer) and the non-dinoflagellate-bearing G. siphonifera and G. bulloides. Site 516 sits near the core of the uniquely small South Atlantic subtropical gyre. As a subtropical gyre, it hosts convergence, fed by inflow from the surrounding surface ocean. Accordingly, we propose that DSDP Site 516 is influenced by the input to the region of G. bulloides and G. siphonifera specimens and/or their N sources that were sourced from the higher-δ¹⁵N nitrate of the subantarctic zone to the south and/or the more proximal temperate ocean waters surrounding the gyre. With this caveat regarding regions with strong environmental gradients, our findings support the use of interspecies FB-δ¹⁵N offsets to investigate the emergence and evolution of photosymbiosis in foraminifera and potentially other animals.