Re-examining extreme carbon isotope fractionation in the coccolithophore Ochrosphaera neapolitana

arising from Y.-W. Liu et al. Nature Communications https://doi.org/10.1038/s41467-018-04463-7 (2018)

In coccolithophores, stable isotopes recorded in both the calcite exoskeleton (coccoliths), and organic carbon (C_org), can reflect their physiological response to environment, and thereby have a wide usage in paleoclimate and biogeochemistry studies. Recently, Liu et al.¹ reported that coccolithophore Ochrosphaera neapolitana has much more positive carbon isotope fractionations relative to dissolved inorganic carbon (DIC) in both coccolith and C_org compared with those published previously for other species^2,3,4 and attributed such unexpected positive carbon isotope fractionations to a unique carbon pathway in this species. However, we find that these extreme isotopic fractionations should be attributed to the poor constraints in DIC carbon isotope ratios instead of the coccolithophores’ physiological response to pCO₂. More careful measurements of DIC carbon isotope would benefit data interpretations and comparisons in future laboratory culture works focusing on phytoplankton’s response to ocean acidification.

In order to study this unusual isotopic fractionation, we carried out another independent culture of O. neapolitana under a low CO₂ environment (pCO₂ ≈ 254ppm and CO₂(aq) ≈ 8.7 μM). The results of carbon isotope fractionation show no exaggerated positive values in neither coccolith nor C_org (Fig. 1). The carbon isotope fractionation between coccolith and DIC (Δ¹³C_{coccolith-DIC} = δ¹³C_coccolith − δ¹³C_DIC) is −1.86 ± 0.43‰ (standard deviation of three biological replicates) and that of C_org (Δ¹³C_Corg-DIC = δ¹³C_Corg − δ¹³C_DIC) is −23.14 ± 0.57‰, which are in the same range as published carbon isotope fractionations of other species.

Fig. 1: Carbon isotope fractionations (Δ¹³C) of coccolithophores in different laboratory culture studies.

a Carbon isotope fractionation between PIC (particulate inorganic carbon from harvested cells) and DIC (dissolved inorganic carbon from seawater). b Carbon isotope fractionation between POC (particulate organic carbon from harvested cells) and DIC. Markers plotted in red and blue are Ochrosphaera neapolitana in this study and in Liu et al.¹, respectively. Markers in other colors are results from other publications^{2,3,4, 12,13,14,15} using different species of coccolithophores including Coccolithus pelagicus, Gephyrocapsa oceanica, Calcidiscus leptoporus, and Emiliania huxleyi.

We propose that the extreme positive carbon isotope fractionations and fractionation trend with pCO₂, originally reported by Liu et al.¹ might be caused by inaccurate estimation of the DIC carbon isotope ratio (δ¹³C_DIC). Instead of directly measuring δ¹³C_DIC during culture, the authors assumed that δ¹³C_DIC reached equilibrium with CO₂ sources through the whole experiment. However, in reality, the δ¹³C_DIC could be positively shifted by two processes, (1) the isotopic disequilibrium between CO₂(g) and DIC during bubbling and (2) the selective uptake of light carbon by algae photosynthesis during culture.

As previously suggested, seawater culture medium should be bubbled at least overnight to achieve a stable carbonate system⁵. Yet, the isotopic equilibrium is much slower than the chemical equilibrium⁶, because the carbon atoms need to be fully exchanged between gas phase and liquid phase before reaching an isotopic equilibrium. Moreover, all culture experiments in Liu et al.¹ were carried out using 38 L glass aquaria, thus increasing the time for such a large DIC pool to reach isotopic equilibrium with the gas phase. In Liu et al.¹, the details in aeration process, such as aeration time and initial δ¹³C_DIC, were not described. Based on the final pCO₂ in all treatments being lower than the target pCO₂ by at least 20%, we infer that the seawater media were pre-bubbled before the incubation of coccolithophores and that there was no further CO₂ aeration during the culture. To evaluate the extent of isotopic disequilibrium in the pre-bubbling process, here we carry out isotopic simulations to trace the carbon atom exchanging process between DIC and CO₂ source (more details in Methods and Supplementary Note 1).

One key parameter in the simulations is the carbon atom exchanging rate constant between CO₂(g) and CO₂(aq) (k_E). In our previous work, we measured this constant in two photobioreactor systems featuring aeration with specified CO₂ concentration; this constant ranged from 3.4 × 10⁻⁵ to 8.7 × 10⁻⁵ mol s⁻¹ atm^{−1 7}. We conduct simulations using k_E ranging from 10⁻⁴ to 10⁻³ mol s⁻¹ atm⁻¹. For reproducing the DIC carbon isotope evolution in Liu et al.¹, a k_E of ~3.6 × 10⁻⁴ is more realistic considering their gas flux was 1.5 L min⁻¹, 6 times as our bubbling system in ETH Zurich. Another important parameter is the initial carbon isotope ratio difference between DIC and CO₂ source. Based on equilibrium δ¹³C_DIC described by Liu et al.¹, we back-calculated the isotopic signatures of the CO₂ sources employed among treatments in Liu et al.¹. These range from −14‰ to −38‰ (Supplementary Table S1). A larger initial difference can potentially cause a more significant disequilibrium between DIC and CO₂(g) leading to an underestimation of δ¹³C_DIC (as simulated in Fig. 2) and thereby more positive coccolith and C_org carbon isotope fractionations. For example, the carbon isotope disequilibrium in ‘280 ppm’ treatment could be larger than the disequilibrium in other two treatments by up to 8‰ even after two weeks aeration (green line in Fig. 2a). Hence, we infer that the isotopic disequilibrium between gas phase and liquid phase during the pre-bubbling process is the main reason why the calculated carbon isotope fractionations in ‘280 ppm’ treatment were much more positive than the results in the other two treatments, instead of the lower pCO₂.

Fig. 2: Simulated DIC carbon isotope ratios evolution during bubbling.

Panel a, b, c are simulations for different pCO2 and initial δ¹³C_DIC. Blue curves: k_E = 10⁻⁴ mol s⁻¹ atm⁻¹ (a slower equilibrium). Red curves: k_E = 10⁻³ mol s⁻¹ atm⁻¹ (a faster equilibrium). Green curves: k_E = 3.6 × 10⁻⁴ mol s⁻¹ atm⁻¹ (potential exchanging rate for Liu et al.¹). The black dashed lines represent the isotopic equilibrium δ¹³C_DIC. The disequilibrium between DIC and gas in 280 ppm treatment after two weeks of bubbling is marked by the vertical green arrows.

Photosynthesis preferentially takes up the light carbon leading to a positive shift of remaining DIC. Ignoring the carbon isotope effect of calcification (much smaller effect compared with photosynthesis), this positive shift in DIC carbon isotope ratio can be estimated by mass balance (Supplementary Note 3). Given a cell density of 10⁵ cell mL⁻¹ as in Liu et al.¹, the carbon isotope fractionation could be as large as +3‰. We suggest that this DIC carbon isotope shift during culture can explain ~20% of the abnormal carbon isotope fractionation in Liu et al.¹, especially the large carbon isotope differences among replicates within the same treatment.

Our culture data and simulations show evidence that the carbon isotope fractionation results in Liu et al.¹ could have a significant bias due to the lack of direct δ¹³C_DIC measurements. Despite acknowledging their values might have an offset (the absolute values on fractionation could differ from real values), they still claimed that the trend of carbon isotope fractionation on pCO₂ should be robust. Nevertheless, such trend in Fig. 3 of Liu et al.¹ would already not be significant when we account for the different disequilibrium offsets among the three treatments. Indeed, simulations of δ¹³C_DIC evolution suggest that the offset in ‘280 ppm’ treatment could be as large as 10‰ more positive than the other two treatments. Thus, their measurements do not support their conclusion that the calcification and photosynthesis of O. neapolitana share the same carbon pool.

Carbon isotope techniques in laboratory cultures have been widely used to trace the ocean acidification effects on phytoplankton, and thereby calibrate a robust pCO₂ proxy in the paleo-climate field. Considering the importance of culture studies with isotopic measurement, cautions should be exercised in experiment design. First, a larger volume of culture medium entails a longer isotopic equilibrium time during the CO₂ aeration process, which should be noted for future work especially using bubbling methods. More importantly, the δ¹³C_DIC should be measured directly to assess fractionation correctly in future isotopic studies. Only measuring the DIC carbon isotope at the beginning of culture is not enough for the batch cultures, because the positive shifts of DIC carbon isotope due to photosynthesis could be as large as 3‰. By following the recommendations described here, we can make the isotopic results comparable and help to better understand the phytoplankton’s response to ocean acidification.

Methods

Laboratory culture

Ochrosphaera neapolitana (RCC1357) was precultured in K/2 medium without Tris buffer⁸ using artificial seawater (ASW) supplemented with NaHCO₃ and HCl to yield an initial DIC of 2050 µM. In triplicate, 1-L bottles were filled with 150 mL of seawater medium with air in the bottle headspace and inoculated with a mid-log phase preculture at an initial cell concentration of 10⁴ cells mL⁻¹. Cultures were grown at 18 °C under a warm white LED light at 100 ± 20 µE on a 16h-light/8h-dark cycle. Bottles were orbitally shaken at 60 rpm to keep cells in suspension. Cell growth was monitored with a Multisizer 4e particle counter and sizer (Beckman Coulter). At ~1.4 × 10⁵ cells mL⁻¹, cells were diluted up to 300 mL to 2–3 × 10⁴ cells mL⁻¹ and harvested after 2 days of more exponential growth up to 7.9 ± 0.6 × 10⁴ cells mL⁻¹. More detailed culture results are listed in the Supplementary Note 1.

Immediately after harvesting, pH was measured using a pH probe calibrated with Mettler Toledo NBS standards (it should be noted here that high ionic strength calibration standards would be optimal for pH measurement of liquids like seawater). There was a carbonate system shift during the batch culture and more details are shown in Supplementary Fig. S1. Cells in 50 mL were pelleted by centrifuging at ~1650 × g for 5 min. Seawater supernatant was analyzed for DIC and δ¹³C_DIC by injecting 3.5 mL into an Apollo analyzer and injecting 1 mL into He-flushed glass vials containing H₃PO₄ for the Gas Bench.

For seawater DIC, an Apollo SciTech DIC-C13 Analyzer coupled to a Picarro CO₂ analyzer was calibrated with in-house NaHCO₃ standards dissolved in deionized water at different known concentrations and δ¹³C values from −4.66 to −7.94‰. δ¹³C_DIC in media were measured with a Gas Bench II with an autosampler (CTC Analytics AG, Switzerland) coupled to ConFlow IV Interface and a Delta V Plus mass spectrometer (Thermo Fischer Scientific). Pelleted cells were snap-frozen with N₂ (l) and stored at −80 °C. For PIC analysis, pellet was resuspended in 1 mL methanol and vortexed. After centrifugation, the methanol phase with extracted organics was removed and the pellet containing the coccoliths was dried at 60 °C overnight. About 300 mg of dried coccolith powder were placed in air-tight glass vials, flushed with He and reacted with five drops of phosphoric acid at 70 °C. PIC δ¹³C and δ¹⁸O were measured by the same Gas Bench system. The system and abovementioned in-house standards were calibrated using international standards NBS 18 (δ¹³C = −5.01‰, δ¹⁸O = +23.00‰) and NBS 19 (δ¹³C = +1.95‰, δ¹⁸O = +2.2‰). The analytical error for DIC concentration and δ¹³C is <10 μM and 0.1‰, respectively.

POC and PON were determined from cells harvested on pre-combusted QFF filters and deep-frozen until analysis. Inorganic carbon from cells on filters was removed by fuming sulfurous acid during 24 h. Filters were placed inside a desiccator on a porous tray and 50 mL sulfurous acid below was fumed with a vacuum pump. Gases were evacuated and filters were further dried at 60 °C overnight. Right before Elemental Analysis (EA), filters were compacted and wrapped into tin cups with the help of tweezers and a press. Samples loaded on a 96-well plate were combusted in the oxidation column at 1020 °C of a Thermo Fisher Flash-EA 1112 coupled with a Conflo IV interface to a Thermo Fisher Delta V-IRMS (isotope ratio mass spectrometer). Combustion gas passed through a reduction column at 650 °C producing N₂ and CO₂ which were separated by chromatography and into a split to the IRMS for an on-line isotope measurement.

Simulations of carbon isotope evolution during aeration

The DIC carbon isotope evolution model is simplified from the model in Zhang et al.⁷ The exchanging rate (with a unit of mol s⁻¹) between CO₂(g) and CO₂(aq) depends on the CO₂ gradient and exchanging rate constant (k_E, with a unit of ppm s⁻¹): ER =  k_E(CO_2(g)–k_HCO_2(aq)), where the k_H is Henry’s law constant with a unit of ppm µM⁻¹. The evolutions of DIC and DIC carbon isotope ratios during CO₂ aeration can be calculated by four differential equations:

$$\frac{{d{{{{{\mathrm{C}}}}}}}}{{dt}}=\,\frac{{k}_{{{{{{\mathrm{E}}}}}}}}{V}\left({{{{{\mathrm{G}}}}}}-{{{{{\mathrm{C}}}}}}\,{k}_{{{{{\mathrm{{H}}}}}}}\right)+\left({k}_{-1}{{{{{{\mathrm{H}}}}}}}^{+}+{k}_{-4}\right){{{{{\mathrm{C}}}}}}-\left({k}_{+1}+{k}_{+4}\,{{{{{{{\mathrm{OH}}}}}}}}^{-}\right)\,{{{{{\mathrm{B}}}}}} \; \; {{{{{{\mathrm{XB}}}}}}}1$$

(1)

$$\frac{{d}^{13}{{{{{\mathrm{C}}}}}}}{dt}=\frac{{k}_{{{{{{\mathrm{E}}}}}}}}{V}\left({\,}^{13}{{{{{\mathrm{G}}}}}}\,{\alpha }_{g2aq}-{\,}^{13}{{{{{\mathrm{C}}}}}}\,{k}_{{{{{{\mathrm{H}}}}}}}{\alpha }_{aq2g}\right)+\left({k}_{+1}^{13}+{k}_{+4}^{13}{{{{{\mathrm{O}}}}}}{{{{{{\mathrm{H}}}}}}}^{-}\right){\,}^{13}{{{{{\mathrm{B}}}}}}\,{{{{{{\mathrm{X}}}}}}}^{13}{{{{{\mathrm{B1}}}}}}-\left({k}_{-1}^{13}\,{{{{{{\mathrm{H}}}}}}}^{+}+{k}_{-4}^{13}\right){\,}^{13}{{{{{\mathrm{C}}}}}}$$

(2)

$$\frac{{d{{{{{\mathrm{B}}}}}}}}{{dt}}=-\,\left({k}_{-1}{{{{{{\mathrm{H}}}}}}}^{+}+{k}_{-4}\,\right){{{{{\mathrm{C}}}}}}+\left({k}_{+1}+\,{k}_{+4}\,{{{{{{{\mathrm{OH}}}}}}}}^{-}\right)\,{{{{{\mathrm{B}}}}}} \; \; {{{{{{\mathrm{XB}}}}}}}1$$

(3)

$$\frac{{d}^{13}{{{{{\mathrm{B}}}}}}}{dt}=-\left({k}_{-1}^{13}{{{{{{\mathrm{H}}}}}}}^{+}+{k}_{-4}^{13}\right){\,\!}^{13}{{{{{\mathrm{C}}}}}}+\left({k}_{+1}^{13}+{k}_{+4}^{13}\,{{{{{\mathrm{O}}}}}}{{{{{{\mathrm{H}}}}}}}^{-}\right){\,\!}^{13}{{{{{\mathrm{B}}}}}} \; \; {{{{{{\mathrm{X}}}}}}}^{13}{{{{{\mathrm{B}}}}}}1$$

(4)

where capital letters G, C, B, H, and OH represent CO₂(g), CO₂(aq), HCO₃⁻ + CO₃²⁻, H⁺, and OH⁻, respectively. The V stands for volume. The α is the isotopic fractionation, e.g. α_g2aq represents the carbon isotope fractionation of CO₂ gas diffusion into liquid phase. The XB1 and X¹³B1 are the fraction of HCO₃⁻ in (HCO₃⁻ + CO₃²⁻) and the fraction of H¹³CO₃⁻ in (H¹³CO₃⁻ + ¹³CO₃²⁻). The XB1 can be calculated by ${{{{{\rm{XB}}}}}}1=\frac{1}{1+\frac{K2}{\left[{{{{{{\mathrm{H}}}}}}}^{+}\right]}}$ and the X¹³B1 can be calculated by ${{{{{{\rm{X}}}}}}}^{13}{{{{{\mathrm{B}}}}}}1=\frac{1}{1+\frac{K2}{\left[{{{{{{\mathrm{H}}}}}}}^{+}\right]}{\alpha }_{{{{{{{\mathrm{CO}}}}}}_{3}}-{{{{{{\mathrm{HCO}}}}}}_{3}}}}$, where the ${\alpha }_{{{{{{{\mathrm{CO}}}}}}}3-{{{{{{\mathrm{HCO}}}}}}}3}$ is the carbon isotope fractionation between CO₃²⁻ and HCO₃⁻⁹. The k₊₁ is the reaction rate constant of CO₂ hydration, which can be calculated by ${{{{{\rm{ln}}}}}}{k}_{\!+1}=1246.98-\,\frac{61900}{{T}_{k}}-183{{{{{\rm{ln}}}}}}{T}_{k}$¹⁰. The k₋₁, the reaction rate constant of HCO₃⁻ dehydration, can be calculated from k₊₁, ${k}_{-1}=\,\frac{{k}_{+1}}{K1}$ . The k₋₄ and k₊₄ are the reaction rate constants of CO₂ hydroxylation and HCO₃⁻ dihydroxylation. The k₊₄ is calculated by$\,{{{{{\rm{ln}}}}}}{k}_{+4}=17.67- \,\frac{2790.47}{{T}_{k}}$¹⁰ and ${k}_{-4}={k}_{+4}\frac{{K}_{{{{{{\mathrm{w}}}}}}}}{K1}$, where K_w is the stoichiometric ion product of water. The K1 and K2 are the first and second dissociation constants of carbonic acid and in this study, we employed equations from¹¹, in which the K1 and K2 were calculated for pH in NBS scale. The reaction rate constants for ¹³C (${k}_{-1}^{13}$, ${k}_{+1}^{13}$, ${k}_{-4}^{13}$ and ${k}_{+4}^{13}$). The initial values for these differential equations are described in the Supplementary Note 2.