Stable Carbon Isotopes of Phytoplankton as a Tool to Monitor Anthropogenic CO₂ Submarine Leakages

Abstract

This study aims to validate the stable carbon isotopic composition (δ¹³C) of phytoplankton as a tool for detecting submarine leakages of anthropogenic CO_2(g), since it is characterised by δ¹³C values significantly lower than the natural CO₂ dissolved in oceans. Three culture experiments were carried out to investigate the changes in δ¹³C of the diatom Thalassiosira rotula during growth in an artificially modified medium (ASW). Three different dissolved inorganic carbon (DIC) concentrations were tested to verify if carbon availability affects phytoplankton δ¹³C. Simultaneously, at each experiment, T. rotula was cultured under natural DIC isotopic composition (δ¹³C_DIC) and carbonate system conditions. The available DIC pool for diatoms grown in ASW was characterised by δ¹³C_DIC values (−44.2 ± 0.9‰) significantly lower than the typical marine range. Through photosynthetic DIC uptake, microalgae δ¹³C rapidly changed, reaching significantly low values (until −43.4‰). Moreover, the different DIC concentrations did not affect the diatom δ¹³C, exhibiting the same trend in δ¹³C values in the three ASW experiments. The experiments prove that phytoplankton isotopic composition quickly responds to changes in the δ¹³C of the medium, making this approach a promising and low-impact tool for detecting CO_2(g) submarine leakages from CO_2(g) deposits.

1. Introduction

The 2015 Paris UNFCCC (United Nations Framework Convention on Climate Change) agreement has set the maximum global temperature increase to 2 °C to significantly reduce the risks and impacts of climate change [1]. To achieve this challenging goal, global emissions need to be reduced at about 3% year⁻¹, corresponding to 37 Gt CO_2(g) year⁻¹ [2]. Several strategies need to be adopted to decrease atmospheric CO_2(g) concentration, such as containing the rate of emissions, developing zero-carbon energy technologies, fuel-switching, improving efficiency, implementing negative emissions technologies and developing and improving carbon capture and storage (CCS) [2,3,4]. CCS has emerged as one of the most promising options among the mitigation technologies by storing carbon dioxide in geological reservoirs, such as depleted oil and gas fields and deep, saline formations [5], either on-shore or off-shore [4]. The International Energy Agency has assessed that, to cope with the purpose of 2 °C warming limitation, CCS should capture about 6000 Mt CO₂ year⁻¹ by 2050, contributing 15–20% to global CO_2(g) emissions reduction [6]. Thus far, nineteen CCS sites are already operational [7], and, among these, six off-shore sites are currently active [8]. Although a well-engineered storage site is not expected to leak, the success and future of CCS depend on the demonstrable effectiveness of storage integrity and, therefore, on the implementation of reliable monitoring strategies for the detection of leakages and their quantification to an acceptable degree of confidence [3,9]. Geological discontinuities, abandoned exploratory boreholes and operational malfunction have been identified among the mechanisms that could potentially facilitate CO_2(g) leakage [10]. The dissolution of the released CO_2(g) in seawater leads to lower pH, and thus, it could critically affect the surrounding marine environment. Among the main adverse effects, shell formation in calcifying organisms could be inhibited, the dissolution and desorption of metals could be enhanced and, once mobilised, they can accumulate in marine organisms, interfering with physiological processes, inhibiting growth or even causing death [11,12,13,14]. Concerning primary producers, except calcareous organisms that are detrimentally affected by the dissolution of CO_2(g), the contradictory effects of high dissolved CO₂ concentration have been reported [14,15,16,17,18]. Phytoplankton and seaweed species seem to benefit from high CO_2(aq) levels by enhancing the carbon and nitrogen fixation and stimulating photosynthesis [11,19,20]. However, pH decrease could affect the physiological processes of some phytoplankton species, including photosynthesis, uptake rates of trace metals and toxin production of some harmful algal species. Moreover, natural phytoplankton assemblages and community composition could be modified by high dissolved CO₂ concentrations [14]. Therefore, effective management of storage sites is required to early detect leakages and prevent risks for the surrounding environment.

Monitoring strategies have been established to detect leakages from storage sites using surface and downhole seismic methods, direct downhole measurements and seabed and water column monitoring [8,21]. Concerning the environmental monitoring, water column and sediment sampling is crucial for the chemical and geochemical characterisation of spatial and temporal dynamics of the natural baseline, thus allowing the identification of anomalies and avoiding false positive and/or negative results [21,22]. However, the temporal sampling frequency may not permit the detection of small leakages depending on plume dispersion and carbonate buffering or may not be able to cover the entire possible affected area [8]. The use of a multidisciplinary approach is encouraged; however, biological monitoring is still underutilised, though it is a low-cost effective tool [12].

Stable isotopes have been largely applied to estimate the effects of anthropogenic pressures on natural ecosystems [23]. Carbon stable isotope analysis is considered a suitable tool for tracing the fate of CO_2(g) in the reservoir and for detecting leakages from storage sites, provided that the isotopic composition (δ¹³C, sensu Coplen [24]) of injected CO_2(g) and its dissolution products is different from that of the surrounding environment [25,26]. The isotopic composition of CO_2(g) (δ¹³C_CO2(g)) stored in CCS is predominantly controlled by the δ¹³C of the feedstock (e.g., fossil fuels, biomass and limestone) used in the CO_2(g)-generating process (e.g., combustion, gasification, cement and steel manufacture and fermentation) [5,27], and the δ¹³C of most of captured CO_2(g) is predicted to range between −60‰ and −20‰. Despite some overlap, this range is rather distinct from the expected natural baselines; thus, δ¹³C represents a valid geochemical tracer [27]. In particular, CO_2(g) derived from fossil fuels is generally characterised by δ¹³C values similar to the isotopic composition of bulk coal (−28.46 to −23.86‰ [28]) or of methane (from −60 to −20‰ [5]) and is thus easier to be distinguished from baseline conditions.

The δ¹³C of atmospheric CO_2(g) has changed significantly since the 19th century, from a value of about −6.5‰ to a present-day value below −8.0‰ as a consequence of the addition of ¹³C-depleted carbon from fossil fuel and land-use changes [29] (the so-called Suess effect [30]). δ¹³C of the total dissolved inorganic carbon (DIC) in the ocean ranges, on average, from 0 to −2‰ [31]; however, substantial spatial-temporal variations have been detected [32,33]. Physical and biological processes affect the δ¹³C of DIC (δ¹³C_DIC), resulting in relatively low values in areas of CO_2(g) invasion and in ¹³C-enriched DIC in areas of outgassing [33]; thus, δ¹³C_DIC could range from −6.56‰ to +3.10‰ [34]. At the present marine pH, bicarbonate (HCO₃⁻) is the dominant species composing DIC; therefore, its isotopic composition (δ¹³C_HCO3) is generally very close to δ¹³C_DIC, whereas CO_2(aq) is ¹³C-depleted by about ~10‰ [35], depending on the temperature isotopic fractionation and the concentrations of the carbonate species dissolved in seawater that are driven mainly by temperature, pH and salinity. Thus, the seasonal and spatial variability described for δ¹³C_DIC also applies to both δ¹³C_HCO3 and δ¹³C_CO2(aq).

DIC uptake by primary producers represents one of the biological processes that mainly contribute to the modifications of δ¹³C_DIC. Indeed, during carbon uptake for photosynthesis, isotopic fractionation (ε, sensu Coplen [24]) occurs because of the isotopic discrimination operated by the enzyme involved in carboxylation (ribulose-1,5-bisphosphate carboxylase/oxygenase—RuBisCO). The lighter isotope (¹²CO₂) is fixed at a rate 1.03 times faster than the heavier (¹³CO₂), leading to ¹³C-depleted values in phytoplankton in respect to the inorganic carbon source [36] and, consequently, enriching the residual DIC. The enzyme RuBisCO discriminates carbon by about 22‰ to 31‰, depending on the different enzyme forms that express a variable affinity for CO_2(aq) [37]. These maximum ε values, however, are rarely observed because several factors affect the CO_2(aq) concentration at the active site of the enzyme. These factors include the dissolved CO₂ concentration [38], phytoplankton growth rates [39,40], cell size or geometry [40], temperature and nutrient availability [41], light intensity [39,42,43], day length [44], temperature effects on enzyme kinetics [37], utilisation of HCO₃⁻ through CO₂-concentrating mechanisms (CCMs) [45,46] and species variability [44,47]. Therefore, phytoplankton isotopic composition directly depends on δ¹³C_DIC and its spatial and temporal variations. Phytoplankton δ¹³C values globally range between −31.0‰ and −16.5‰, displaying higher values in the equatorial upwelling regions and lower values at latitudes >60° in the northern hemisphere, and exhibit a high variability in the temperate regions due to seasonal dynamics in climatology and phytoplankton life cycle [32].

The difference between the isotopic composition of captured CO_2(g) and the natural dissolved CO₂ baseline makes δ¹³C the most widely analysed tracer, as it allows leakage monitoring with minimal economic and environmental impacts compared to other tracers [5]. Most of the applications of the carbon stable isotope analysis are related to monitoring of the movement of injected CO_2(g) inside the reservoir [25,48,49,50]. The direct monitoring of the changes in δ¹³C_DIC could result not suitable to detect transient leaks because of the rapid mixing in the surrounding environment of CO_2(g) leaking from a storage site. During an experimental CO_2(g) leak into the sediments of North Sea [10,51], the analysis of pore-water δ¹³C_DIC allowed the identification of the leak earlier than the detection of significant changes in bicarbonate concentration [10], confirming the reliability of carbon stable isotope analysis as an inherent tracer of CO_2(g) leakage. However, it has also been highlighted that low levels of leakage may be hardly detected and quantified because of carbonate buffering and CO_2(g) transport. Phytoplankton, instead, could represent a record of the changes induced by CO_2(g) leakage, because their isotopic composition directly depends on the δ¹³C of the available DIC pool. A considerable fraction of phytoplankton can vertically migrate along the water column to acquire multiple resources and moves towards the bottom to access the higher availability of nutrients regenerated at the sediment–water interface [52]. In a storage site, this active movement towards depth allows therefore the uptake of the CO_2(aq) that eventually leaks out. If stored CO_2(g) is characterised by δ¹³C different than natural CO_2(aq), the phytoplankton isotopic composition is expected to significantly change from the baseline of the area that needs to be defined to take into account for the natural seasonal variability. Since stored CO_2(g) is generally characterised by δ¹³C lower than natural δ¹³C_DIC, phytoplankton δ¹³C is thus expected to change towards more negative values. An efficient CCS monitoring should therefore include, besides the geochemical analysis, the evaluation of the biological parameters that could give an integrated contribution to the detection of CO_2(g) leakages.

This study aims to explore the feasibility of using the phytoplankton stable carbon isotopic composition as a tool for detecting anthropogenic CO_2(g) leakage from the seabed. Through isotopic fractionation during photosynthetic DIC uptake, phytoplankton δ¹³C is expected to change, thus representing an important contribution to the identification of sporadic leakages that could not be detected by the direct analysis of the geochemical properties of the water column, provided that the isotopic composition of injected CO_2(g) and its dissolution products is different from that of the surrounding environment. Three pairs of culture experiments were designed in a model case approach, reproducing the isotopic composition of the carbonate system modified by an anthropogenic CO_2(g) submarine leakage but maintaining pH levels in the typical marine range to allow phytoplankton optimal growth. A single diatom species was grown in an artificially modified medium characterised by strongly ¹³C-depleted dissolved CO₂, and three different DIC concentrations were tested to verify if inorganic carbon availability would affect the phytoplankton algal metabolism. Simultaneously, at each experiment, the same diatom species was cultured under natural carbonate system and δ¹³C_DIC conditions to better interpret the isotopic results considering this condition as the reference baseline.

2. Materials and Methods

2.1. Experimental Setup

Three culture experiments (EXP-1, EXP-2 and EXP-3) with the diatom Thalassiosira rotula were conducted within two 2L-photobioreactors (Kbiotech^®, Lugano, Switzerland) under controlled conditions of light (light:dark cycle: 14:10 h, intensity 100 µE m⁻² s⁻¹) and temperature (20 ± 0.02 °C) (Figure 1 and

Table 1. Description of the experiments. Initial cell abundance (cells mL⁻¹) and dissolved inorganic carbon (DIC) concentration (µmol L⁻¹); initial medium pH (pH_NBS) and experiment duration (h) for the three experiments (EXP-1, EXP-2 and EXP-3) carried out in two bioreactors, one filled with natural seawater (NAT) and the other with artificial seawater medium (ASW).

Experiment	Medium Tested	Experiment Code	Initial Cell Abundance (Cells mL−1)	Initial DIC Concentration (µmol L−1)	Initial pH (pHNBS)	Experiment Duration (h)
EXP-1	ASW	1-ASW	250	1952	8.14	96
	NAT	1-NAT	250	2222	8.32	96
EXP-2	ASW	2-ASW	1000	893	8.19	140
	NAT	2-NAT	1000	2243	8.37	140
EXP-3	ASW	3-ASW	1000	5098	8.19	120
	NAT	3-NAT	1000	2350	8.32	96

). During each experiment, diatoms were simultaneously grown in the two bioreactors filled respectively with different media (Table 1). In one of the two bioreactors, microalgae were grown in a natural seawater medium (NAT), whereas in the second one an artificial seawater medium (ASW) was used.

The ASW medium was prepared to reach a modified isotopic composition of DIC. Since CO_2(g) leakages from storage sites induce variations in DIC concentration, for each experiment, ASW medium was prepared differently to test whether altered DIC levels would affect the diatoms δ¹³C in the isotopically modified environment. The different DIC concentrations tested were not coupled with pH modification; indeed, typical marine pH values (8.0–8.2) were preserved to avoid alterations in phytoplankton physiology (Table 1). DIC concentration in the ASW medium for EXP-1 (1-ASW) was comparable to natural marine levels (~2000 µmol L⁻¹), and during EXP-2 (2-ASW), about half the concentration was tested (~900 µmol L⁻¹), whereas, in EXP-3 (3-ASW), more than twice the natural DIC concentration was reached (~5000 µmol L⁻¹). On the other hand, the DIC concentration was maintained in the natural range (~2000 µmol L⁻¹) for the three NAT experiments.

For NAT medium, a sterile-filtered (0.2-µm pore size) Adriatic Sea water was used, whereas the artificial seawater for ASW was prepared based on the artificial seawater medium ESAW (enrichment solution with artificial water) designed for coastal and open ocean phytoplankton [53], modified as follows. The solutions for the ASW medium were prepared with ultrapure Milli-Q water degassed by bubbling pure nitrogen (N_2(g)) to remove the natural dissolved CO₂. The required inorganic carbon concentrations were not achieved by the addition of sodium bicarbonate (NaHCO₃) as described in Andersen [53] recipe, because a ¹³C-depleted DIC source needed to be reached. NaHCO₃ was replaced by a stoichiometric quantity of NaOH, and the final concentration of DIC was achieved by bubbling CO_2(g) (as a mixture of CO₂-free air and compressed CO_2(g)) into medium solution after the addition of NaOH, until the typical marine pH value was reached (8.0–8.3). Bubbling was performed with an industrial compressed CO_2(g) to achieve a modified DIC source after verifying the different isotopic composition of the gas from the current atmospheric δ¹³C_CO2(g) (~−8‰). Since no changes in δ¹³C_CO2(g) due to Rayleigh fractionation occurred during cylinder utilisation, the same CO_2(g) cylinder was used for the preparation of the three ASW media experiments (δ¹³C of CO_2(g) used was −48.2‰ ± 0.2‰ as the average of the measurements performed during the preparation of the three different ASW media using the same gas cylinder).

Both media were supplied with nutrients and microelements, as proposed by Andersen [53] for the ESAW medium and were introduced into the photobioreactors by filtration (hydrophilic PTFE sterile filter, nominal pore size 0.2 µm, Merck KGaA, Darmstadt, Germany), with the purpose of minimising bacterial interference and to prevent gas exchange with the atmosphere.

Before the beginning of the experiments, an equal quantity of diatoms was inoculated into the two media (Table 1). A diatom species was selected, since diatoms represent the most abundant microalgae taxa worldwide due to their ability to grow in a wide range of environments and account for about 40% of marine primary production [54,55]. Moreover, carbon isotopic fractionation in diatoms is not affected by the precipitation of calcite that instead occurs in calcifying organisms [56]. The aliquot of microalgae for the inoculation was collected from clonal cultures of Thalassiosira rotula (species identity was confirmed by genetic analysis, Appendix A) established in September 2012 from a sample collected in the Gulf of Trieste (North Adriatic Sea) (45°42.06’ N, 13°42.60’ E). The strain was maintained in 100-mL glass Erlenmeyer flasks filled with modified L1 medium obtained from sterile-filtered (0.2-µm pore size) Adriatic Sea water and supplemented with nutrients, metals and vitamins. The culture was grown at 15 ± 2 °C and at a light intensity of about 53 μE m⁻² sec⁻¹ under a controlled light:dark regime (14:10 h). Cells were transferred within sterilised 2-L glass Erlenmeyer flasks to allow growth until the exponential phase was reached for the inoculation in the ASW and NAT media.

Periodic samplings were performed during each culture experiment through a peristaltic pump to minimise the atmospheric gas exchange. During the whole duration of the experiments, the variation of pH, temperature and dissolved oxygen was continuously monitored and recorded by online probes. Intermittent CO₂-free air flow was maintained (on:off cycle: 10:50 min), and a continuous gentle stirring prevented stratification and favoured medium mixing (Figure 1).

The duration of each experiment depended on the beginning and persistence of the stationary phase of microalgae (Table 1).

2.2. Sampling and Analysis for Phytoplankton Abundance and Media Characterisation

For phytoplankton abundance (cells mL⁻¹), an aliquot of 10 mL of culture was sampled every day and immediately fixed with prefiltered and neutralised formaldehyde (1.6% final concentration) [57]. Cell counts were performed in triplicates in a Sedgewick Rafter counting chamber at a light microscope (Leica DM2500, Leica Microsystems Srl, Wetzlar, Germany) at a magnification of 100–200×, counting a minimum of 200 cells in random fields or transects. Growth rate of the diatoms was calculated in the exponential phase of the culture [53] as

$$\mu =(\ln {\mathrm{cell}}_{\mathrm{tf}}-\ln {\mathrm{cell}}_{\mathrm{ti}})/\left({\mathrm{t}}_{\mathrm{f}}-{\mathrm{t}}_{\mathrm{i}}\right)$$

(1)

where µ (d⁻¹) is the specific growth rate, ${\mathrm{cell}}_{\mathrm{ti}}$ represents the abundance of phytoplankton (cells mL⁻¹) at the beginning of the time interval (${\mathrm{t}}_{\mathrm{i}}$; days) in which the culture exhibited exponential growth and ${\mathrm{cell}}_{\mathrm{tf}}$ the abundance (cells mL⁻¹) at the end of the time interval analysed (${\mathrm{t}}_{\mathrm{f}}$; days).

The pH (National Bureau of Standards, NBS, scale) was continuously recorded by online probes (EasyFerm BIO Arc 225, Hamilton Bonaduz AG, Bonaduz, Switzerland) previously calibrated on the National Bureau of Standards (NBS) scale with certified buffer solutions of pH 7 ± 0.02 and 9 ± 0.03 (at 25 °C, Merck KGaA, Darmstadt, Germany).

To measure total alkalinity (A_T, µmol kg_SW⁻¹), samples of 100-mL culture suspension were filtered over precombusted (450 °C for 4 h) Whatman (Cytiva, Global Life Sciences Solutions USA LLC, Marlborough, MA, USA) GF/F filters (nominal porosity 0.7 µm) to remove phytoplankton cells [58]. Each bottle was poisoned with saturated mercuric chloride (HgCl₂) to halt biological activity and stored at 4 °C in the dark until analysis. Total alkalinity was determined by potentiometric titration in an open cell (SOP 3b [59]), as described by Ingrosso et al. [60].

Samples for dissolved inorganic nitrogen (DIN, as the sum of nitrite—NO₂, nitrate—NO₃ and ammonium—NH₄); phosphate (PO₄) and silicate (Si(OH)₄) were filtered over precombusted (450 °C for 4 h) Whatman (Cytiva, Global Life Sciences Solutions USA LLC, Marlborough, MA, USA) GF/F filters (nominal porosity 0.7 µm) in acid-washed polyethylene vials and kept frozen (−20 °C) until laboratory analysis. Inorganic nutrient concentrations were determined calorimetrically with a QuAAtro (Seal Analytical Inc., Mequon, WI, USA) autoanalyser according to Hansen and Koroleff [61]. Detection limits reported by the analytical methods were 0.01 μmol L⁻¹, 0.02 μmol L⁻¹, 0.02 μmol L⁻¹, 0.01 μmol L⁻¹ and 0.01 μmol L⁻¹ for NO₂, NO₃, NH₄, PO₄ and Si(OH)₄, respectively. The accuracy and precision of the analytical procedures are annually checked through the quality assurance program (AQ1) QUASIMEME Laboratory Performance Studies (Wageningen, The Netherlands), and the relative coefficient of variation for five replicates was less than 5%. Internal quality control samples were used during each analysis.

Concentrations of total dissolved inorganic carbon (DIC) and the other carbonate system parameters, including dissolved CO₂ (CO_2(aq)) and bicarbonate (HCO₃⁻), were calculated using the CO2SYS program [62] through in situ A_T, medium pH (pH_NBS at 20 °C), temperature, salinity, phosphate and silicate data. Carbonic acid dissociation constants (i.e., pK1 and pK2) of Mehrbach et al. [63] refitted by Dickson and Millero [64] were used for the computation, as well as the Dickson constant for the ion HSO₄⁻ [65] and borate dissociation constant of Uppström et al. [66].

The abundance of prokaryotes (cells mL⁻¹) was estimated by a FACSCanto II (Becton Dickinson, Franklin Lakes, NJ, USA) flow cytometer, equipped with an air-cooled laser at 488 nm and standard filter setup. Samples were treated according to Marie et al. [67], as described by Celussi et al. [68].

2.3. Sampling and Analysis of Medium and Phytoplankton Carbon Stable Isotopic Composition and Calculation of Isotopic Fractionation

Samples for the determination of the stable carbon isotope ratio of CO_2(g) of the cylinder used for ASW media preparation and the stable carbon isotope ratio of dissolved inorganic carbon (DIC) were collected in 12-mL Exetainer^® (Labco Limited, Ceredigion, UK) evacuated glass tubes, preventing gas exchange with the atmosphere.

Stable carbon isotopic composition of CO_2(g) (δ¹³C_CO2(g)) was determined with an isotope ratio mass spectrometer (IRMS) Europa Scientific 20-20 (Crowe, UK) with trace gas preparation module ANCA-TG for trace gas samples.

Samples for stable isotopic composition of dissolved inorganic carbon (δ¹³C_DIC) were filtered over precombusted (450 °C for 4 h) Whatman GF/F filters (nominal porosity 0.7 µm) to remove phytoplankton cells, and the analysis was performed by injecting aliquots of the sample into evacuated septum tubes with phosphoric acid. Released CO_2(g) was then analysed with continuous-flow IRMS Europa 20-20 (Crowe, UK) with an ANCA-TG trace gas separation module. In order to determine the optimal extraction procedure for water samples, two standard Na₂CO₃ solutions were prepared with a known δ¹³C value of −10.8‰ ± 0.2‰ and −4.12‰ ± 0.2‰, respectively.

The carbon isotopic composition of the phytoplankton was determined as bulk particulate organic carbon (δ¹³C_POC) by filtering aliquots of culture suspension (~20 mL) over precombusted (450 °C for 4 h) Whatman (Cytiva, Global Life Sciences Solutions USA LLC, Marlborough, MA, USA) GF/F filters (nominal porosity 0.7 µm, ø 25 mm). To assess the potential influence of prokaryotes on the bulk δ¹³C_POC, other aliquots of culture suspension (~20 mL) were filtered over precombusted Whatman (Cytiva, Global Life Sciences Solutions USA LLC, Marlborough, MA, USA) GF/F filters after a prefiltration on a 10-µm nylon mesh to remove phytoplankton cells (δ¹³C_POC<10µm). The dimension of the mesh for prefiltration was chosen according to the average Thalassiosira rotula cell dimension of 20 µm [69]. Samples for δ¹³C_POC and δ¹³C_POC<10µm analysis were collected in duplicate, rinsed with deionised water, oven-dried at 60 °C and stored in desiccators until analysis. Before the analysis, filters for δ¹³C_POC and δ¹³C_POC<10µm were treated with 1-M HCl to remove carbonates and inserted into tin capsules for further analysis. Measurements were performed on a Europa Scientific 20-20 IRMS (Crowe, UK) with an ANCA-SL preparation module for solid and liquid samples.

All stable carbon isotope values are given conventionally in δ-notation in per mil deviation (‰) from the Vienna Peed Dee Belemnite (VPDB) standard. IRMS was calibrated against reference materials: IAEA-CH-6 and IAEA-CH-7 (International Atomic Energy Agency, Vienna, Austria) for δ¹³C_POC analysis and “CO₂ ISO-TOP, High” (Messer Group GmbH, Bad Soden, Germany) with a δ¹³C value of −4.3 ± 0.2‰ for δ¹³C_CO2(g) and δ¹³C_DIC. The precision based on replicate analyses was ±0.2‰ for all stable isotope analyses (δ¹³C_CO2(g), δ¹³C_DIC, δ¹³C_POC and δ¹³C_POC<10µm).

A mass balance relation was used to calculate the isotopic composition of dissolved CO_2(aq) (δ¹³C_CO2(aq)) and HCO₃⁻ (δ¹³C_HCO3⁻) from δ¹³C_DIC by the temperature-fractionationation relationships from Rau et al. [70], based on Mook et al. [35]:

$${\delta }^{13}{\mathrm{C}}_{{\mathrm{CO}}_{2\left(\mathrm{aq}\right)}}={\delta }^{13}{\mathrm{C}}_{\mathrm{DIC}}+23.644─\frac{9701.5}{{\mathrm{T}}_{\mathrm{K}}},$$

(2)

$${\delta }^{13}{\mathrm{C}}_{{\mathrm{HCO}}_3^{-}}={\delta }^{13}{\mathrm{C}}_{{\mathrm{CO}}_2}-24.12-\frac{9866}{{\mathrm{T}}_{\mathrm{K}}},$$

(3)

where T_K is the absolute temperature in Kelvin.

Diatom isotopic fractionation associated with photosynthetic CO_2(aq) fixation was reported as the isotopic enrichment factor (ε_CO2-POC) [71] and was calculated relative to the isotopic composition of dissolved CO₂ in the bulk medium according to Farquhar et al. [72] and Freeman and Hayes [73]:

$${\mathsf{\epsilon }}_{\mathrm{CO}2-\mathrm{POC}}=1000\frac{\left({\delta }^{13}{\mathrm{C}}_{{\mathrm{CO}}_{2\left(\mathrm{aq}\right)}}-{\delta }^{13}{\mathrm{C}}_{\mathrm{POC}}\right)}{\left(1000+{\delta }^{13}{\mathrm{C}}_{\mathrm{POC}}\right)}.$$

(4)

Despite the presence of CO₂-concentrating mechanisms that allow diatoms to use both HCO₃⁻ and CO_2(aq), the isotopic fractionation was calculated based on δ¹³C_CO2(aq) since all inorganic carbon is converted to CO_2(aq) before being fixed and to allow for comparison with literature data.

Considering the experimental set-up, the closed-system Rayleigh-type fractionation relationships [74] were investigated to discuss the data from the natural seawater experiments by applying the following equation [71]:

$${\delta }^{13}{\mathrm{C}}_{{\mathrm{DIC}}_{\mathrm{r}}}={\delta }^{13}{\mathrm{C}}_{{\mathrm{DIC}}_{\mathrm{r}0}}\times f^{\left(\mathsf{\alpha }-1\right)},$$

(5)

where ${\delta }^{13}{\mathrm{C}}_{{\mathrm{DIC}}_{\mathrm{r}}}$ is the isotopic composition of the residual fraction ($f$) of unutilised substrate remaining (0 ≤ f ≤ 1), ${\delta }^{13}{\mathrm{C}}_{{\mathrm{DIC}}_{\mathrm{r}0}}$ is the isotopic composition of DIC at the beginning of the three NAT experiments (when f = 1 and corresponds to −2.3‰) and α is the fractionation factor and was calculated as (ε/1000) + 1 using the theoretical value of RuBisCO isotopic fractionation (ε) of 20‰ [75].

To verify the possible application of the methodological approach described to CCS sites where stored CO_2(g) is characterised by different δ¹³C values, the equation by Zhang et al. [76] for carbon isotopic fractionation during CO_2(g) dissolution in seawater was applied:

$${\mathsf{\epsilon }}_{\mathrm{DIC}-\mathrm{CO}2}=0.0144\times \mathrm{T}\times f{\mathrm{CO}}_3^{2-}-0.107\times \mathrm{T}+10.53,$$

(6)

where T is the temperature in °C and $f{\mathrm{CO}}_3^{2-}$ is the carbonate fraction of total DIC measured at the beginning of the experiments. For this calculation, the carbonate fraction measured at the beginning of the ASW experiments was used.

2.4. Statistical Analysis

The Spearman rank correlation test was used to investigate the time variation of the variables and the relationships between phytoplankton abundance, abiotic parameters and isotopic composition.

In order to verify the differences between natural and artificial seawater media and among the three experiments, the Kruskal-Wallis ANOVA test was applied to the variables. Kruskal-Wallis ANOVA and Spearman rank tests were conducted using STATISTICA 7 (StatSoft, Inc., Tulsa, OK, USA), and only significant data (p < 0.05 and p < 0.01) are presented. A multivariate analysis was performed to verify the overall difference between the two media applying a non-metric multidimensional scaling analysis (nMDS) ordination model. Phytoplankton and prokaryotes abundance data were previously square root-transformed to avoid a skewness of data that could have affected the analysis. An nMDS ordination was performed based on the triangular resemblance matrix constructed on z-normalised data using the Euclidean distance method. The analysis of similarity (one-way ANOSIM) between the ASW and NAT was also performed, and a similarity profile (SIMPROF) analysis was applied on the cluster analysis. The nMDS, ANOSIM and SIMPROF analyses were performed using PRIMER 7 (PRIMER-E Ltd., Plymouth, UK) software [77].

3. Results

3.1. Phytoplankton Growth under Different Conditions and Media Characterisation

The Thalassiosira rotula abundance significantly increased (p < 0.05) during the three experiments and nearly followed a typical exponential growth in both natural (NAT) and artificial (ASW) medium (Figure 2). The highest abundance was measured at the end of the experiments, and the maximum value was reached in 2-ASW at 144 h (4.17 × 10⁴ ± 5.42 × 10² cells mL⁻¹). Despite the equal diatom inoculation in the two media at each experiment (250 cells mL⁻¹ in EXP-1, 1000 cells mL⁻¹ in EXP-2 and EXP-3; Table 1), in the ASW, the diatom growth was generally slower than in the NAT. Consequently, the exponential phase was not coincident in the two conditions, and the growth rates (µ) were not calculated in the same time interval. The stationary phase of the culture was generally achieved in the NAT (Figure 2b) within the first 48 h of the culture, and the average growth rate for the three experiments resulted in 1.1 ± 0.3 d⁻¹. On the other hand, stationary conditions in the ASW (Figure 2a) were generally reached after 72 h from the inoculation, and the average growth rate for the three experiments was 1.0 ± 0.2 d⁻¹. Despite the slight dissimilarities in growth rates, the diatom abundances between the two conditions were not significantly different. At the end of the 1-NAT experiment (96 h), the diatom abundance decreased to 1.96 ± 0.22 × 10² cells mL⁻¹, indicating the beginning of the senescent phase. The data from this last sampling, however, were included in the discussion, since the δ¹³C_POC values detected a fall in the range of the 2-NAT and 3-NAT experiments and still derived from fresh phytoplanktonic organic matter.

At the beginning of the experiments, the pH levels (Figure 3a,b and Table 1) in the ASW (8.17 ± 0.03) fell within the natural marine pH range, as in the NAT (8.34 ± 0.03), suggesting that the modification in the DIC concentration induced for the ASW medium preparation did not alter pH from the natural seawater levels. A significant increase (p < 0.05) was observed in pH values in both media (up to 9.50; the maximum measured at 96 h in 2-NAT; Figure 3a,b). A different pattern was observed in 2-ASW (Figure 3a). In this experiment, the pH values only increased after 72 h, but the high values observed in the other two ASW experiments were not reached (2-ASW maximum pH: 8.32 at 140 h, 1-ASW maximum pH: 8.83 at 96 h and 3-ASW maximum pH: 9.12 at 96 h). Considering the whole experiments duration and excluding 2-ASW, no significant differences were detected in pH values between the NAT and ASW.

Dissolved inorganic carbon concentration in the NAT after diatom inoculation (2272 ± 68 µmol L⁻¹; Figure 3d and Table 1) fell within the typical range for the Adriatic Sea [60,78,79]. On the other hand, the DIC concentration measured in the ASW (Figure 3c and Table 1) at the beginning of the experiments confirmed the effectiveness of the modification imposed on carbonate system during the ASW medium preparation. Indeed, in 1-ASW, a typical marine DIC concentration was measured (1952 µmol L⁻¹), whereas lower (893 µmol L⁻¹) and higher (5098 µmol L⁻¹) concentrations were detected at the beginning of 2-ASW and 3-ASW, respectively. The DIC concentration significantly decreased (p < 0.01) in the NAT during the three experiments, reaching the minimum value at 96 h in 2-NAT (1059 µmol L⁻¹; Figure 3d). A decrease in DIC concentration was also detected in the three ASW experiments 48 h after the inoculation, reaching the lowest DIC value at 72 h in 2-ASW (566 µmol L⁻¹).

The dissolved CO₂ concentration at the beginning of the three ASW experiments depended on DIC concentration (Figure 3e). Indeed, the lowest initial concentration was detected in 2-ASW (5.81 µmol L⁻¹), the highest in 3-ASW (29.71 µmol L⁻¹) and a natural marine CO_2(aq) concentration (13.94 µmol L⁻¹) was measured in 1-ASW. On the other hand, for the three NAT experiments, the initial CO_2(aq) concentration (Figure 3f) fell within the typical marine range (9.09 ± 0.76 µmol L⁻¹). The dissolved CO₂ concentration significantly decreased (p < 0.01) in both the ASW and NAT during the three experiments. However, considering only 2-ASW, CO_2(aq) concentration remained almost constant and decreased only after 72 h. In the ASW, the minimum concentration was detected at the end of EXP-1 (1.34 µmol L⁻¹), whereas in the NAT, concentrations lower than 0.20 µmol L⁻¹ were detected after 96 h in EXP-2 and EXP-3. Despite the different initial CO_2(aq) concentrations, considering the whole duration of the experiments, no significant differences were detected between the ASW and NAT.

3.2. Isotopic Characterisation of the Media and Phytoplankton

The initial isotopic composition of dissolved inorganic carbon (δ¹³C_DIC) was significantly different (Kruskal-Wallis ANOVA, p < 0.01) between the two conditions tested. Indeed, the initial ASW δ¹³C_DIC exhibited very low values in the three experiments (−44.2 ± 0.9‰; Figure 4a), whereas, in the NAT, the DIC isotopic composition (−2.3‰ ± 1.4‰; Figure 5b) fell within the typical range for the Adriatic Sea [79]. During diatom growth, δ¹³C_DIC in the ASW (Figure 4a) significantly increased (p < 0.01). The minimum values were recorded at the beginning of each experiment and the maximum on the last day of the culture, reaching the highest value (−30.1‰) at 96 h in 1-ASW. No significant differences were detected among the three ASW experiments, though, in 3-ASW, δ¹³C_DIC exhibited a slight increase, reaching the maximum value of −41.3‰ at the end of diatom culture (120 h). On the contrary, no significant time variations during diatom growth were detected in δ¹³C_DIC in the NAT experiment (Figure 4b), with values ranging from −3.8‰ (1-NAT at 7 h) to 1.6‰ (2-NAT at 96 h).

At the beginning of the culture experiments, the phytoplankton isotopic composition (δ¹³C_POC) did not show significant differences between the two conditions tested (−25.2‰ ± 1.3‰; Figure 4c,d); however, microalgae δ¹³C exhibited an opposite trend in the ASW than NAT during the diatom growth. In the ASW (Figure 4c), the δ¹³C_POC values significantly decreased (p < 0.01), reaching the lowest values at the end of culture experiments, until the minimum detected in 2-ASW (−43.4‰ at 140 h). On the other hand, in the NAT (Figure 4d), δ¹³C_POC values slightly increased after the inoculation (p < 0.01), reaching the highest values at the end of the diatom culture (maximum −17.7‰ at 72 h in 3-NAT).

In the artificial seawater condition, the highest isotopic enrichment factor (ε_CO2-POC, Figure 5a) associated with CO_2(aq) uptake by T. rotula was detected at the beginning of the experiments (−29.2‰ ± 1.0‰). The enrichment values in the ASW significantly changed during phytoplankton growth (p < 0.01), reaching 0.9‰ at 140 h in 2-ASW. Slighter variations were detected in carbon isotopic enrichment in the NAT (Figure 5b), with values ranging from 13.8‰ ± 2.7‰, calculated at the beginning of the diatom cultivation, to 6.7‰, detected on the last day of the longest experiment (2-NAT at 140 h).

4. Discussion

4.1. Validation of the Experimental Set-Up and Diatom Growth in the Natural and Modified Media

The isotopic composition of DIC measured in ASW confirms that the experimental set-up succeeded in modifying the isotopic composition of the carbonate species dissolved in water. Indeed, δ¹³C_DIC values at the beginning of ASW experiments (−44.2‰ ± 0.9‰) approached the isotopic composition of the strongly depleted anthropogenic CO_2(g) used for media preparation (δ¹³C_CO2(g): −48.2‰ ± 0.2‰). Such low δ¹³C_DIC values are hardly detected in natural systems, even in lakes where a broad range of δ¹³C_DIC values is expected to be found (down to −31.1‰ [80]). On the other hand, δ¹³C_DIC in the NAT (−2.3‰ ± 1.4‰) fell within the typical marine range [79] and resulted significantly higher (Kruskal-Wallis ANOVA, p < 0.01) than values detected in the ASW.

Both the ASW and NAT were characterised by initial pH values typical for the marine environment. pH may play a significant role in limiting the rate of primary production, growth and total abundance of phytoplankton [81]. Moreover, the medium pH also affects the relative concentration of the dissolved carbonate species, thus finally conditioning the isotopic composition of CO_2(aq). It was therefore crucial to maintain comparable pH values in both the ASW and NAT in order not to affect microalgae growth and isotopic response during diatom cultivation.

Thalassiosira rotula abundances significantly increased (p < 0.01) in both the ASW and NAT during the three experiments (Figure 2). A general delay in microalgae growth was observed in the ASW, probably due to the acclimation time needed after the inoculation in the modified medium. However, no significant differences in terms of cell abundance were detected between the ASW and NAT.

For the ASW experiments, three different initial DIC concentrations were tested to verify the effects of lower (2-ASW; 893 µmol L⁻¹) and higher (3-ASW; 5098 µmol L⁻¹) DIC availability with respect to the typical marine concentration (1-ASW; 1952 µmol L⁻¹). The diatom abundance was not affected by the different DIC concentrations, as no differences were detected among the three ASW experiments, and, in addition, the highest T. rotula abundance was detected at the end of 2-ASW, suggesting that the low DIC concentration tested during this experiment should not be limiting. These results are consistent with the experimental findings of Clark and Flynn [82] and Clark et al. [83], which showed that, at low DIC concentrations and maintaining initial pH levels in the natural range, the phytoplankton growth was not carbon limited. On the other hand, the higher availability of DIC in 3-ASW neither stimulated microalgae growth, therefore confirming that diatoms are capable of growth over a wide range of DIC availability [83]. Similar results were obtained from the mesocosm experiments performed by Maugendre et al. [84] and Esposito et al. [85], who observed no increase in phytoplankton abundance at high CO_2(aq) concentrations. Several studies indicated contrasting effects of high CO_2(aq) concentrations on the phytoplankton; indeed, the growth rates of the diatoms resulted in being stimulated, not affected or inhibited [17]. Phytoplankton growth, however, should be considered as the result of the interaction of several factors, such as pH variation, temperature, nutrients and/or light limitation, acting in synergy with the changes in the carbonate system [86]. The experiments performed by Li et al. [87] also proved that elevated CO_2(aq) has no detectable effects on the diatom T. weissflogi but showed limited growth and photosynthesis at lowered CO_2(aq) levels (0.8~3.4 µmol L⁻¹). However, the experiments performed by Li et al. [87] were conducted at CO_2(aq) concentrations lower than 2-ASW (5.81 µmol L⁻¹ at the beginning of the cultivation, Figure 3e) and at an initial higher pH (Li et al. [87]: 8.63~9.02, 2-ASW: 8.19). Li et al. [87] addressed the suppression of diatom growth to the shift in carbonate system composition towards higher concentrations of HCO₃⁻ because of the high pH, resulting in higher energy demand for inorganic carbon uptake via concentrating mechanisms. Indeed, diatoms, as many other phytoplankton species, have developed CO₂-concentrating mechanisms (CCMs) for actively taking up inorganic carbon and for increasing CO_2(aq) concentration around RuBisCO [88]. These mechanisms have evolved to prevent carbon limitations. The high pH of the modern sea and the slow CO_2(aq) diffusion rate in water are limiting factors for the dissolved CO₂ availability, usually making CO_2(aq) concentrations in water not sufficient to saturate RuBisCO fixation rates [88,89], representing less than 1% of total DIC [58]. CCMs increase the available inorganic carbon by active CO_2(aq) and HCO₃⁻ transport and by the reversible dehydration of bicarbonate to CO_2(aq) mediated by intra- and extracellular carbonic anhydrase (CA) enzymes, being CO_2(aq) the carbon species ultimately used for photosynthesis. Experimental and genetic data confirmed that both HCO₃⁻ and CO_2(aq) are possible substrates for diatom photosynthesis [43,88,89,90] and that CCMs grant full-carbon saturation of RuBisCO catalytic sites, supplementing inorganic carbon by direct HCO₃⁻ uptake even at high CO_2(aq) concentrations [91]. Diatoms are thus able to maintain high photosynthetic rates in response to large variations in the marine carbonate system [91]. The strain of T. rotula used for the experiments was cultivated from a sample collected in the Gulf of Trieste. This semi-enclosed basin is characterised by an elevated variability in its oceanographic properties due to fluctuations in riverine inputs and climatological forcing factors that strongly affect inorganic nutrient concentrations, salinity, pH and the carbonate system [92]. Thus, the microalgae used for the experiments are adapted to a wide range of pH and variations in the carbonate system, as already reported for other coastal phytoplankton communities [17,93,94].

The inorganic carbon uptake for photosynthesis increased pH in both the ASW and NAT, and the experiments were stopped before reaching high pH levels (>9.50) that could have inhibited microalgae growth [81]. Only the 2-ASW experiment differed in pH variation, since the values detected were always lower than 8.50, and pH values increased only after the third day of diatom cultivation. This difference in pH variation could be addressed to the variation in CO_2(aq) concentration. In 2-ASW, indeed, the CO_2(aq) concentration remained almost constant for the first 72 h of the T. rotula cultivation. Since the diatom growth was comparable to the other experiments, it results that the carbon supply was supported by CCMs by the uptake of HCO₃⁻, thus not significantly affecting the medium pH.

Diatoms were cultivated in nutrient-replete conditions (Figure S1) to avoid stress due to nutrient limitations. The DIN and PO₄ initial concentrations were comparable between the ASW and NAT, and, during diatom growth, their concentrations significantly decreased (p < 0.05) but were not completely consumed.

Based on the above considerations, the differences observed in phytoplankton isotopic composition between the ASW and NAT should not depend on dissimilarities in the growth conditions. Indeed, T. rotula should not have suffered any stress for carbon or nutrient availability during the three experiments performed. Similarly, the diatoms were not influenced by the two conditions tested, as no significant differences were detected in terms of cell abundance and growth between the NAT and ASW. Thus, the ASW and NAT δ¹³C_POC values should be considered as driven mainly by the different δ¹³C_DIC values.

4.2. Time Variation of Phytoplankton Isotopic Composition as a Result of Different DIC Sources and the Inorganic Carbon Uptake

The isotopic composition of phytoplankton (δ¹³C_POC) at the beginning of the experiments did not differ between the ASW and NAT, resulting, on average, in −25.2‰ ± 1.3‰. The aliquots of T. rotula for inoculation in the two media were sampled from the same strain of microalgae culture—thus, the similar δ¹³C_POC in the two conditions at the beginning of the three experiments. The initial δ¹³C_POC values in both the ASW and NAT are comparable to previous measurements performed on samples collected in the Gulf of Trieste [95,96], since the culture was prepared from a sample collected in this area and are typical for marine phytoplankton in estuarine environments.

After the inoculation in the two media, the diatoms began to uptake the DIC available, with a consequent gradual modification of the carbon isotopic composition. In the ASW, the δ¹³C_POC values significantly decreased (p < 0.01) and reached strongly ¹³C-depleted values at the end of the cultivation (−43.4‰ at 140 h in 2-ASW). On the other hand, in the NAT, the isotopic composition of the POC reached at the end of the experiments (−17.7‰, the maximum measured at 72 h in 3-NAT) remained in the natural marine range and of the typical phytoplankton values measured in the open Northern Adriatic Sea (−20.6‰ [97]).

The variations of the δ¹³C_POC values observed in both media depend on the isotopic fractionation due to the preferential uptake of ¹²C for photosynthesis. If no other DIC source is available or supplied, as in the case of the experiments here described, the residual DIC pool gradually enriches in ¹³C during the photosynthetic uptake. As also observed in mesocosm experiments [85], this typical residual source enrichment (i.e., increasing the δ¹³C_DIC values) naturally occurs during intense bloom events as a consequence of the slow CO_2(g) equilibration compared to the fast biological uptake [81,86]. Both the ASW and NAT experiments reproduce a blooming condition. The diatom growth, indeed, increased the medium pH by the DIC consumption and decreased the nutrient concentration. Moreover, in the ASW, the δ¹³C_DIC values increased (Figure 6a), whereas the isotopic composition of the DIC in the NAT (Figure 6b) was not significantly modified by the diatoms carbon uptake, though the δ¹³C_DIC values slightly increased (Figure 4b) during diatom cultivation. On the other hand, a clear different pattern was observed in the δ¹³C_POC values between the ASW and NAT (Figure 6a,b).

With the proceeding of diatom growth, in both the ASW and NAT, the fraction of available ¹²C in the DIC pool gradually decreased because of the preferential uptake for photosynthesis. This process in the NAT did not change significantly δ¹³C_DIC values, since the microalgae were in isotopic equilibrium with the medium. However, considering the fast growth of microalgae, the gradual reduction of the ¹²C in the available DIC resulted in an evident ¹³C-enrichment of the phytoplankton with increasing δ¹³C_POC values during diatom growth (Figure 6a). In the ASW, despite the significant increase (p < 0.01) in δ¹³C_DIC values, as the result of the preferential uptake of the ¹²C by the diatoms, the δ¹³C_POC values showed an opposite behaviour than in the NAT (Figure 6a). This misleading isotopic anomaly depended on the source of DIC provided in the ASW and on the experimental set-up. The experiments aimed to verify how the isotopic composition of the diatoms changed during the growth in a modified medium; therefore no acclimation time followed the inoculation. The Thalassiosira rotula strain was grown under natural marine δ¹³C_DIC until inoculation, and once transferred in the ASW medium, microalgae started to take up the modified DIC available. As discussed in the previous section (Section 4.1), the ASW was characterised by low δ¹³C_DIC values; therefore, the lightest isotope (¹²C) was much more abundant than in the NAT. The diatoms had therefore a higher availability of ¹²C for photosynthesis, and this resulted in the significant (p < 0.01) and quick decrease in δ¹³C_POC values (Figure 6a) and the following increase in δ¹³C_DIC. Notwithstanding that the variation of δ¹³C_DIC during diatom growth was similar in all ASW experiments, in 3-ASW, a slighter increase was observed. The culture medium for 3-ASW was prepared to obtain high DIC concentration (Figure 3c); consequently, ¹²C was even more abundant than in 2-ASW and 1-ASW. Since the diatom growth and δ¹³C_POC followed the same trend as 1-ASW and 2-ASW, the residual DIC in 3-ASW maintained a relatively higher amount of ¹²C than the other ASW experiment during the T. rotula culture; therefore, a slighter variation was observed in the δ¹³C_DIC.

Esposito et al. [85] found a similar decrease in δ¹³C_POC values during mesocosm experiments after the addition of seawater bubbled with ¹³C-depleted CO_2(g). The POC isotopic composition reflected the trend of δ¹³C_DIC, and the minimum δ¹³C_POC was reached at the end of the experiments, although higher values than those measured in our experiments were reached. To our knowledge, extremely low δ¹³C_POC values, similar to those detected at the end of our experiments, were only measured by Poulain et al. [98] (−58‰ ± 4‰) in a mixed-species phytoplankton culture (used as a zooplankton food source) grown in a medium where ¹³C-depleted industrial CO_2(g) was bubbled, even though additional medium information (e.g., pH, DIC concentration and CO_2(g) bubbling duration) was not provided to allow additional consideration.

The isotopic composition of T. rotula in the ASW resulted significantly different (Kruskal-Wallis ANOVA, p < 0.01) from that of diatoms grown in the NAT. The inverse correlation (p < 0.01) of δ¹³C_POC with both δ¹³C_DIC and diatom abundance in the ASW supports the hypothesis that the isotopic composition of T. rotula depends mostly on the isotopic composition of the available DIC source. The different DIC concentrations in the ASW experiments seem not to affect the final phytoplankton δ¹³C, as no significant differences were detected in δ¹³C_POC among the three ASW experiments, confirming that the DIC concentration is not becoming a limiting factor and does not affect the δ¹³C_POC [99,100].

The presence of prokaryotes was monitored to consider its potential effect on the bulk POC isotopic composition, since the prokaryotes abundance increased (p < 0.05, Figure S2) in both media throughout the duration of the experiments. The isotopic composition of the fraction of POC smaller than 10 µm (δ¹³C_POC<10µm), assumed to depend mostly on the prokaryotes biomass, followed the same trend as δ¹³C_POC (p < 0.01) in both the ASW and NAT. Previous studies [101,102] revealed that the heterotrophic prokaryotes isotopic composition is generally close to or similar to that of the organic substrate provided. This is consistent with the results here discussed, because, during algal growth in bioreactors, the only substrate for prokaryotes was the phytoplankton-derived organic matter [103]. Slighter variations, however, were observed in δ¹³C_POC<10µm compared to δ¹³C_POC during phytoplankton growth as the result of a delay in the incorporation of the freshly produced phytoplankton material, as already discussed by De Kluijver et al. [104] in labelling experiments. Considering the strong positive correlation between δ¹³C_POC<10µm and δ¹³C_POC in both the ASW and NAT, it could be therefore assumed that the increase in prokaryotes biomass might have been not sufficient to modify the bulk POC isotopic composition.

A multivariate analysis (nMDS; Figure 7) was performed to investigate the global difference between the two conditions tested. The isotopic composition of POC, DIC and δ¹³C_POC<10µm, phytoplankton and prokaryotes abundances, pH and inorganic nutrients were the parameters included in the analysis. Considering the time variation of δ¹³C_POC and the comparability of the diatom abundance in the three ASW experiments, it was concluded that the different DIC concentrations tested did not affect the phytoplankton carbon uptake; therefore, neither DIC nor CO_2(aq) concentrations were included in the nMDS analysis to avoid the strong effect of the different concentrations tested in the analysis. Two groups can be clearly distinguished (Figure 7), and the separation corresponds to the two different media in which the diatoms were grown (ASW and NAT). The strongest separation of the groups was evident at the final sampling times, when the highest difference in δ¹³C_POC values was detected between the ASW and NAT. The analysis of similarity (one-way ANOSIM) supported the significant global differences between ASW and NAT (p < 0.05), even if with a certain degree of overlapping of the data (R = 0.358) probably related to the common starting condition of the microalgae culture. Indeed, the SIMPROF analysis highlighted as a unique group the data collected within the first day of cultivation, whereas the separation of the NAT and ASW became gradually wider following the phytoplankton growth in the two culture media.

The multivariate analysis confirmed that carbon stable isotopes could be considered as valid tracers of the origin of dissolved inorganic carbon. Indeed, both the δ¹³C_DIC and δ¹³C_POC values significantly depend on the different origin of dissolved CO₂, and phytoplankton δ¹³C quickly responded to the change of growth medium. The time variation of phytoplankton δ¹³C in the ASW experiments resulted in a decrease of ~3.1‰ (calculated from the regression line, Figure 6a) in only one day of cultivation, corresponding to an average ¹³C-depletion of ~18‰ in less than one week. Conversely, the derived increase in δ¹³C_DIC corresponded to ~1.7‰ d⁻¹ (Figure 6a), which is comparable to the increasing trend identified for δ¹³C_POC in the NAT (~1.4‰ d⁻¹, Figure 6b). The modification of the phytoplankton isotopic composition in the diatoms grown in the ASW is, therefore, faster and broader than the variation of medium δ¹³C_DIC. These results are consistent with the mesocosm experiments performed by Esposito et al. [85] that highlighted a significant drop of δ¹³C_POC values in the ¹³C-depleted mesocosm about five days after the bloom phase. The δ¹³C_POC decrease observed by Esposito et al. [85] is comparable to the POC ¹³C-depletion in the ASW after ~1.5 days, confirming that our experimental approach is able to fast detect changes in the phytoplankton isotopic composition.

This result should be considered when dealing with CO_2(g) leakages from storage sites. In a real CCS site, the chemical composition of the water column quickly changes after a leakage, and δ¹³C_DIC is modified depending on the CO_2(g) concentration released and on the environmental conditions. However, the δ¹³C_DIC values return to background levels due to mixing currents that rapidly disperse CO_2(g) both horizontally and vertically [11]. Therefore, a CO_2(g) leakage could be detected by δ¹³C_DIC analysis only if the sampling is close enough in time. On the other hand, considering the rapid variation of diatom δ¹³C, the assessment of CO_2(g) leakages through the analysis of the isotopic composition of phytoplankton gives the opportunity to gather time-integrated information, and thus, even low and sporadic emissions, that otherwise would be hardly detectable from the seawater analysis, could be identified.

4.3. Phytoplankton Isotopic Fractionation

The isotopic composition of phytoplankton depends on the fractionation occurring during inorganic carbon uptake [73] and was described as isotopic enrichment factor (ε_CO2-POC) [71]. Isotopic fractionation is mainly the result of the activity of the enzyme RuBisCO during photosynthetic CO_2(aq) fixation, which discriminates carbon by about 22‰ to 31‰, depending on the CO_2(aq) affinity of the different RuBisCO forms [37], thus resulting in ¹³C-depleted values in the phytoplankton. However, these maximum ε_CO2-POC values are rarely observed, because several physiological and environmental factors affect CO_2(aq) concentration at the active site of the enzyme [37,38,39,40,41,42,43,44,45,46,47].

For the experiments here discussed, a single diatom species—Thalassiosira rotula—was grown in two different media (ASW and NAT) under equal conditions of light availability, temperature and nutrient concentration. Since T. rotula growth was similar in the ASW and NAT (Section 4.1), the significant differences (Kruskal-Wallis ANOVA, p < 0.01) of the ε_CO2-POC values detected between the two conditions tested should be addressed to the different isotopic composition of the phytoplankton and DIC and, eventually, to the availability of CO_2(aq), even if the DIC concentration seemed not to affect either the diatom growth or its isotopic composition (Section 4.1 and Section 4.2).

Extremely negative ε_CO2-POC values were calculated in the artificially modified medium (−29.2‰ ± 1.0‰) at the beginning of the experiments, and an increasing trend (p < 0.01) was observed during diatom growth. These low values depend on the experimental setup, since aliquots of T. rotula were inoculated into a strongly ¹³C-depleted medium without an acclimation period; thus, after inoculation, the diatom δ¹³C values were higher than δ¹³C_DIC. The largest availability of ¹²C and the quick utilisation of DIC in the ASW experiments resulted in negative ε_CO2-POC values that progressively increased during algal growth as a consequence of the gradual reduction in the difference between the source (DIC) and product (POC) isotopic compositions. A switch to positive ε_CO2-POC values was observed after 96 h of cultivation due to the modifications induced in δ¹³C_DIC and δ¹³C_POC because of the DIC uptake (Figure 5a).

The isotopic enrichment in the natural medium, on the contrary, was characterised by decreasing values (p < 0.05), following the gradual ¹³C-enrichment of the DIC pool because of the uptake of ¹²C for photosynthesis [105]. The ε_CO2-POC values in the NAT (13.8~6.7‰) were lower than the in vitro RuBisCO ε-values calculated for the marine diatom Skeletonema costatum (18.5‰ [106]) but fell in the range of the measurements performed in a mesocosm experiment (9.6~16.5‰ [85]) where diatom was the predominant phytoplankton taxon.

As previously discussed, DIC concentration does not affect the phytoplankton isotopic composition; however, the observed decrease in ε_CO2-POC values in the NAT are consistent with previous studies demonstrating that phytoplankton ¹³C-discrimination is lower the less DIC is available [105,107]. On the other hand, a higher carbon supply is expected to induce more pronounced isotopic fractionation (i.e., an increase in ε_CO2-POC values) [85,107,108]. The results from the 3-ASW experiment, in which a higher DIC concentration was tested, are in contrast with this assumption. Indeed, no differences were detected in the ε_CO2-POC values in the ASW experiments due to the different DIC availability, showing that the only change in the DIC concentration did not have a large effect on the δ¹³C_POC, as also recently reported by Tuerena et al. [99]. In both the ASW and NAT, the ε_CO2-POC values were strongly correlated with δ¹³C_POC (p < 0.01) and diatom abundance (p < 0.05); thus, the microalgae growth and the carbon isotopic fractionation occurring during photosynthetic DIC uptake should be considered the main drivers for the variations observed in the phytoplankton δ¹³C.

The theoretical relationship between the isotopic composition of the phytoplankton and DIC in the natural seawater experiments was investigated by applying the closed-system Rayleigh-type fractionation model [109]. The isotopic fractionation associated with the DIC uptake was assumed to correspond to the theoretical value of RuBisCO isotopic fractionation (ε) of 20‰ [75]. The resulting relationships between the δ¹³C of the cumulative POC (green curve), source carbon (DIC, blue curve) and incremental POC (red curve) are shown in Figure 8, together with the experimental δ¹³C_POC (green circles) and δ¹³C_DIC (blue circles) data from the three NAT experiments. According to the literature data, the enzymatic isotopic fractionation can vary between 22‰ and 31‰ [37]; however, by applying any other isotopic fractionation value included in this range the shape of the lines does not significantly change.

The experimental results of the δ¹³C_POC (green circles) plot in-line with the cumulative product (green line), while the δ¹³C_DIC data (blue circles) plot out of the source carbon curve (blue line); thus, the processes occurring in the system cannot be described only by the closed-system Rayleigh-type fractionation model. Indeed, the literature data indicate that the photosynthetic ¹³C fractionation can be described as a passive-diffusion phytoplankton photosynthesis, where the isotopic fractionation is associated with diffusive transport of CO_2(aq) and enzymatic, intracellular carbon fixation [45,70]. Furthermore, the phytoplankton δ¹³C values are not solely controlled by the DIC/CO_2(aq) concentrations, but several factors both physiological (e.g., phytoplankton growth rates and cell dimension) and environmental (e.g., temperature, nutrient concentration and light intensity) [39,42,43,44,70] affect the isotopic fractionation. Indeed, during the experiments, the ε_CO2-POC values were not constant either in the ASW or in the NAT. Moreover, the presence of CCMs in diatoms allows the uptake of both CO_2(aq) and HCO₃⁻ [88,89]. Intra- and extracellular carbonic anhydrase (CA) enzymes have been recognised to operate in the reversible dehydration of bicarbonate to CO_2(aq), since it is the species ultimately used for photosynthesis. Bicarbonate is relatively ¹³C-enriched in respect to CO_2(aq) by about ~10‰ [35], but a comparable kinetic fractionation occurs during the dehydration of bicarbonate by CA, making CA-mediated HCO₃⁻ uptake indistinguishable from direct CO_2(aq) uptake [111] from an isotopic point of view. However, if no HCO₃⁻ leaks out of the cell and all the bicarbonate taken up is converted into CO_2(aq), no fractionation occurs during the process of intracellular dehydration, making the substrate for photosynthesis 9–10‰ enriched than CO_2(aq), thus affecting the resulting phytoplankton isotopic ε_CO2-POC values [112]. In this process, pH was identified as an important controlling factor, since it regulates the level of CA activity, the induction of CCMs and affects carbon leakage from the cell in relation to the CO_2(aq) and HCO₃⁻ uptake and conversion [112]. The increase in pH detected in both conditions during diatom growth has clearly moved the carbonate system equilibrium towards higher concentrations of bicarbonate; thus, both in the ASW and NAT, HCO₃⁻ should have played an important role as a photosynthetic substrate. No differences were encountered in the pH trend between the ASW and NAT; hence, the bicarbonate contribution should have been comparable and, thus, should not cover a significant role in the interpretation of the differences detected in δ¹³C_POC. Moreover, as discussed in Section 4.1, in 2-ASW, the pH variation was slighter, and a lower DIC concentration was tested, implying that the contribution of HCO₃⁻ was reduced compared to the other experiments performed. However, the δ¹³C_POC, δ¹³C_DIC and, thus, ε_CO2-POC values followed the same trend as 1-ASW and 3-ASW, confirming that the different origin of CO_2(aq) (i.e., the different isotopic composition) was the main factor affecting the phytoplankton isotopic composition and that the variation in DIC concentration and of the contribution of the different carbonate species did not significantly affect the diatom isotopic composition.

4.4. Application of the Method to CCS Sites: Limitations and Future Perspectives

The available DIC pool for diatoms grown in the artificially modified medium was characterised by δ¹³C_DIC values significantly different from the typical marine range. Through photosynthetic isotopic fractionation, the microalgae isotopic composition rapidly changed, reaching significantly ¹³C-depleted values (decrease of ~3.1‰) in just one day. Moreover, the modification of diatom δ¹³C resulted not affected by the different concentrations of dissolved inorganic carbon, exhibiting the same trend in δ¹³C_POC values for all the DIC concentrations tested.

The results of the experiments described are encouraging for the future application of this method as a monitoring approach at CCS sites. However, the condition tested represents an extreme case, and further experiments should be performed. The δ¹³C_CO2(g) used for the ASW experiments simulates the leakage of stored CO_2(g) derived from fossil fuels combustion. However, the different feedstocks and CO_2(g)-generating processes are expected to expand the range of stored δ¹³C_CO2(g) values [5,27]. The first requirement for the application of the approach described is that δ¹³C_DIC values after CO_2(g) leakage are significantly different from the natural baseline. To test whether different δ¹³C_CO2(g) leaking would result in detectable differences, the equation for carbon isotope fractionation during CO_2(g) dissolution in seawater by Zhang at al. [76] (Equation (6), Section 2.3) was applied using different δ¹³C_CO2(g) values and assuming the atmospheric δ¹³C_CO2(g) corresponding to −8.0‰. The different δ¹³C of leaking CO_2(g) tested were −15.0‰, −20.0‰, −25.0‰ and −35.0‰. The equation was also applied to the δ¹³C value of CO_2(g) used for the three ASW media preparations (−48.2‰) (Figure 9).

To assess more confidently whether this approach would be useful on specific CCS sites, the natural baseline must be defined in detail to verify if the variations in δ¹³C of the DIC and phytoplankton would overlap with the natural seasonal variations. The detection of leakages would be more difficult if δ¹³C_DIC values resulting from leaking CO_2(g) dissolution were similar to the natural baseline. Among the different δ¹³C_CO2(g) values tested, leaking CO_2(g) characterised by a δ¹³C of −15‰ represents the worst scenario to be distinctly detected by isotopic analysis. In that case, if the natural δ¹³C_DIC in the CCS area can reach the lowest value of −4‰, the leakage would be detectable if its concentration would finally correspond to at least 70% of the total DIC (Figure 9). On the other hand, if δ¹³C_DIC naturally never reaches values lower than −1‰, the leakage might be detectable if it corresponds to 40% of the total DIC (Figure 9). The resulting differences in phytoplankton isotopic composition are much more complicated to be modelled, since several factors affect fractionation during photosynthesis [37,38,39,40,41,42,43,44,45,46,47,70]. Obviously, if the leaking CO_2(g) dissolution results in significant differences in δ¹³C_DIC values in respect to the natural baseline, also significant differences in δ¹³C_POC are expected accordingly. Therefore, for the correct implementation of this method, it is important to know the natural variations of δ¹³C_DIC and δ¹³C_POC at the storage site and the isotopic composition of stored CO_2(g) at the CCS.

The variation of δ¹³C_DIC calculated might slightly overestimate the final δ¹³C_DIC in case the whole DIC derives from the leaked CO_2(g). Indeed, assuming that the DIC completely derives from the dissolution of CO_2(g) used for the experiments (−48.2‰), the calculated δ¹³C_DIC result is −39.8‰ (Figure 9), which is higher than the measured δ¹³C_DIC at the beginning of the three ASW experiments (−44.2‰ ± 0.9‰), but it still falls within the limits of the expected range. The equation by Zhang et al. [76], however, assumes that pH is constant and represents the results of the complete mixing of the water mass after leaking CO_2(g) dissolves and reaches equilibrium; thus, it does not reproduce the leakage conditions.

As highlighted by the release experiment by Blackford et al. [10], the porewater δ¹³C_DIC values are modified earlier than the detection of significant changes in the bicarbonate concentration and more consistently than the water column δ¹³C_DIC by CO_2(g) leakages. The direct monitoring of the changes in the δ¹³C_DIC could therefore result not sufficient in detecting transient leaks because of carbonate buffering and of the rapid mixing in the surrounding environment of CO_2(g) leaking from a storage site. The primary producers’ isotopic composition, instead, could represent a record of the changes induced by CO_2(g) leakage, because their isotopic composition directly depends on the δ¹³C of the available DIC pool. Despite that phytoplankton could vertically migrate along the water column [52] and, thus, represents an important record of the modifications occurring in δ¹³C_DIC after CO_2(g) leakages, these organisms are not able to contrast the strong currents that eventually can transport them horizontally far from the leakage site. Moreover, our experiments were conducted in a model case approach, where the whole carbonate system isotopic composition was modified and a single diatom species was tested, maintaining the pH levels within the natural range to allow phytoplankton optimal growth. Since storage sites will be predominantly hosted in coastal and shelf seas [11], and simulation results showed that the highest variations due to CO_2(g) leakages are predicted to affect the seabed [113], further experiments and investigations should be performed focusing on tycopelagic and benthic primary producers (microalgae and/or seagrass) that are less prone to water currents. Moreover, benthic primary producers represent reliable biomonitoring because of the low spatial movement ability and would allow easier detection of transient and small leakages, since they live attached to the seafloor through which CO_2(g) is leaking.

For the future applications of this method, transient leakages characterised by different δ¹³C of injected CO_2(g) should be explored, both by culture and release field experiments, and a multispecies approach should also be tested to consider the interspecific variability of carbon isotopic fractionation. As already highlighted in other studies (e.g., [5,10,21,22]), it is critical to extensively define the natural baseline at the storage site from both a geochemical and biological perspective. The seasonal changes that naturally characterise phytoplankton and DIC isotopic composition [32,33] and seabed community cycles [10] should be deeply understood to allow the correct identification of leakages and not misinterpret the observed deviations. Furthermore, for the coastal storage sites, the definition of the natural baseline would be even more challenging and important, since the anthropogenic and riverine inputs and the higher productivity of these areas represent additional factors that strongly affect the phytoplankton and DIC δ¹³C [10,33,114]. Moreover, the isotopic baseline should be defined, including the recent monitoring data, to take into account the decrease induced on the δ¹³C_DIC and δ¹³C_POC by the Suess effect [99]. It has been demonstrated that the decrease in δ¹³C_DIC because of the invasion of anthropogenic CO_2(g) from the atmosphere is already responsible for the decrease in δ¹³C_POC by an extra 0.5–0.8‰ consistently for every cell size [99].

5. Conclusions

The experiments presented here reveal that the primary producer’s isotopic composition quickly responds to changes in the seawater δ¹³C_DIC; thus, this approach represents a valuable contribution for ocean management in the evaluation of the impacts of CO_2(g) released from carbon storage sites. These results highlight the relevance of the application of a multidisciplinary approach that includes biological monitoring for the identification of CO_2(g) leakages, since the phytoplankton isotopic composition provides time-integrated information on the dissolved inorganic carbon pool, allowing the identification of sporadic emissions that could be hardly detected by seawater sampling. The definition of a detailed baseline is a primary requirement in this field to allow the identification of the seasonal changes that could interfere with the observed variations in the isotopic composition of both the seawater and primary producers.

Considering the importance and the urge to reduce climate change effects, research on CCS monitoring strategies represents a primary field of study in which the biological approach using inherent tracers is still underused and under-investigated. The results of the experiments described are encouraging for the future application of the analysis of the primary producer’s carbon stable isotopic composition for monitoring CCS sites, and more effort deserves to be dedicated in this direction.