Semi-continuous cultures of the two diatom species Fragilariopsis curta (isolated from the Weddell Sea by Thomas Mock during ANT XVI/3 in 1999) and Odontella weisflogii (isolated in the Atlantic sector of the Southern Ocean by Bank Beszteri during ANT-XXIX/5, 2013) were grown at 4°C in sterile-filtered (0.2 μm) unbuffered natural Antarctic seawater (30.2) enriched with silicate, trace metals and vitamins according to F/2 medium (Guillard and Ryther, 1962). Nitrate and phosphate were added in concentrations reflecting the Redfield N:P ratio of 16:1, 100 and 6.25 μmol L⁻¹ respectively (Redfield, 1958).

Based on photosynthesis-irradiance-curves (PE-curves) with both species acclimated to ambient pCO₂ and 150 μmol photons m⁻² s⁻¹ light treatments were chosen. Both species were grown in triplicates under low, medium and high light (LL = 20, ML = 200, HL = 500 μmol photons m⁻² s⁻¹) conditions at a 16:8 h light:dark light cycle using light-emitting diodes (LED) lamps (SolarStinger LED SunStrip Marine Daylight, Econlux). The three light treatments were further continuously bubbled through a frit with humidified air of the three CO₂ partial pressures (pCO₂) of 180, 380, and 1000 μatm (low, ambient and high pCO₂). A gas flow controller (CGM 2000, MCZ Umwelttechnik) was used to generate the CO₂ gas mixtures from CO₂-free air (<1 ppmv CO₂, Nitrox CO2 RP280, Domnick Hunter ltd.) and pure CO₂ (Air Liquide Deutschland ltd., Germany). Triplicates of both species were exposed to all light and pCO₂ treatment combinations. All replicates of one experimental treatments were run in parallel.

For each light treatment chlorophyll a (Chl a) fluorescence was calibrated against cell number and pH, allowing us to monitor the status of the culture using fluorescence alone. Cultures were grown in 1 L glass bottles (custom made), kept in exponential phase and harvested at cell densities of 224,000 ± 97,000 and 732 ± 170 cells per mL for the small Fragilariopsis (~5 μm) and the much larger Odontella (>50 μm) respectively to prevent drift of carbonate chemistry.

Only during the pre-acclimation phase to all experimental conditions of at least 10 days, cultures were diluted two to three times using pre-equilibrated medium. During the main experiment, cells grew for 3.8 ± 1.7 days until all parameters were sampled.

The pH was measured every other day and at the final sampling day using a pH-ion meter (pH-Meter 827, Metrohm), calibrated (3 point calibration) with National Institute of Standards and Technology-certified buffer systems. The pH remained constant at 8.39 ± 0.03, 8.02 ± 0.03, and 7.76 ± 0.02 for the low, ambient and high pCO₂ treatments respectively (Table 1). Total Alkalinity (TA) was measured by duplicate potentiometric titrations (TW alpha plus, SI Analytics, Brewer et al., 1986) of filtrated samples (Whatman GF/F glass fiber filters, ~ 0.6 mm), having been stored at 4°C in 150 mL borosilicate bottles until analysis. Certified reference material (CRMs provided by Prof. A. Dickson, Scripps, USA; batch no. 111; reproducibility ±13 μmol kg⁻¹) was used to correct TA for systematic errors. Dissolved inorganic carbon (DIC) samples were sterile-filtered (0.2 μm) and stored at 4°C in 5 mL gas-tight borosilicate bottles without headspace until analysis. DIC was measured colourimetrically in duplicates with a QuAAtro autoanalyzer (Seal Analytical, Stoll et al., 2001). The analyzer was calibrated with NaHCO₃ solutions (salinity of 35, achieved by NaCl addition) with concentrations ranging between 1800 and 2300 mmol DIC kg⁻¹. Certified reference materials (reproducibility ± 8 μmol kg⁻¹) were used to correct for errors in instrument performance such as baseline drifts. The carbonate system was calculated based on TA, pH, silicate, phosphate, temperature and salinity using the CO2Sys program (results shown in Table 1, Pierrot et al., 2006) choosing the equilibrium constant of Mehrbach et al. (1973) refitted by Dickson and Millero (1987).

Samples for cell counts of every treatment were fixed with acid Lugol's solution (10% final concentration) and stored at 3°C in the dark until analysis. Cells were counted on an inverted light microscope (Axio Observer.D1; Zeiss) after sedimentation for 24 h in 10 mL Utermöhl chambers (Hydro-Bios). Growth rates were determined from samples taken right after at the beginning of the experiment (N₀) and at the day of sampling (N_fin) and for each sample >400 cells were counted. Cell specific growth rate (μ, day⁻¹) was calculated as

where Δt is the incubation duration in days.

Particulate organic carbon (POC) and nitrogen (PON) were determined from gently filtered (<20 mmHg) subsamples of each treatment using pre-combusted GF/F filters (15h, 200°C; Whatman). Samples were stored at −20°C and defrosted prior to analysis (>12 h, 60°C), acidified with 0.1 mol HCl L⁻¹ and dried over night at 60°C. Samples were analyzed using an elemental analyzer (EURO EA Elemental Analyzer, Euro Vector). Contents of POC and PON were corrected using blank measurements and normalized to filtered volume and cell densities to yield cell quotas. Production rates of POC and PON, were calculated by multiplication of the cellular quota with the specific growth rate of the respective treatment.

Pigment concentrations of chlorophyll a, chlorophyll c2, fucoxanthin, diadino- and diatoxanthin were determined using High-Performance-Liquid-Chromatography (HPLC). Samples were gently filtered onto GF/F (Whatman) filters, immediately frozen and stored at −80°C for later analysis. Pigments were extracted in 90:10 acetone:water for 24 h at 4°C in the dark and filtered (4 mm nylon syringe filters, 0.45 μm pore size, NalgeneC; Labware) prior to analysis. Analyses were performed on a LaChromElite® system consisting of a chilled autosampler L-2200, a DAD detector L-2450 (VWR-Hitachi International GmbH) and a Spherisorb ODS-2 column (25 cm × 4.6 mm, 5 μm particle size; Waters). The system used a LiChrospher® 100 RP-18 guard cartridge for pigment separation applying a gradient following Wright et al. (1991) detecting peaks at 440 nm which were identified and quantified via co-chromatography of pigment standards obtained from DHI Lab Products (ORT, Denmark) using the software EZChrom Elite ver. 3.1.3.

From concentrations of diadinoxanthin (DD) and its de-epoxidised equivalent diatoxanthin (DT) the de-epoxidation state (DES) was calculated according to

Photophysiological parameters were measured using a Fast Repetition Rate fluorometer (FRRf, FastOcean PTX; Chelsea Technologies) in combination with a FastAct Laboratory system (Chelsea Technologies). Measurements were conducted at growth temperature (4°C). Samples were dark-acclimated for 10 min prior to measurement. The duration of the dark acclimation phase was chosen after pre-testing different time intervals (5, 10, 20, 30 min) to ensure that all photosystem II (PS II) reaction centers were fully oxidized and non-photochemical quenching was relaxed. Excitation wavelength of the fluorometer's LED was 450 nm with an automated adjustment of the light intensity (between 0.66 and 1.2 × 10²² photons m⁻² s⁻¹). The single turnover mode was used with 100 flashlets saturation phase on a 2 μs pitch and 40 flashlets relaxation phase on a 40 μs pitch in order to cumulatively saturate PS II. Estimation of minimum (F₀) and maximum chlorophyll a (Chl a) fluorescence (F_m) was based on iterative algorithms for induction (Kolber et al., 1998) and relaxation phase (Oxborough et al., 2012). Minimum and maximum Chl a fluorescence were used to calculate the apparent maximum quantum yield of photochemistry in PS II (F_v/F_m) according to the following equation:

Additional Chl a fluorescence measurements of every treatment were performed in response to increasing incident irradiances (E) generating photosynthesis-irradiance-curves (PE-curves; irradiances ranged between 0 and 1800 μmol photons m⁻² s⁻¹) using 15 steps with an acclimation duration of 5 min per light step and six subsequent Chl a fluorescence measurements. From these fluorescence measurements, the light-adapted minimum (F′) and maximum (Fm′) fluorescence of the single turnover acquisition was estimated. The effective PSII quantum yield under ambient light (Fq′/Fm′) was derived according to the equation (Fm′/F′)/Fm′ (Genty et al., 1989). Absolute electron transport rates (absETR, e⁻ PSII⁻¹ s⁻¹) at the different light steps of the PE-curve were calculated as (Suggett et al., 2004, 2009; Huot and Babin, 2010):

where σ_PSII is the functional absorption cross section of PSII photochemistry (in nm² quanta⁻¹) and E denotes the instantaneous irradiance (μmol photons m⁻² s⁻¹). Light-use characteristics were analyzed by fitting irradiance-dependent ETRs according to Ralph and Gademann (2005), including maximum absolute ETR (ETR_m), minimum saturating irradiance (I_K) and maximum light utilization efficiency (α). The PE-Curve was followed by another 10 min of dark acclimation with single turnover flashlets in order to assess PSII recovery (yield recovery). Yield recovery was calculated from the F_v/F_m measured before and after the PE-curve and given as % of the initial F_v/F_m (before the PE-curve). Non-photochemical quenching (NPQ, Equation 5) was calculated following the Stern-Volmer equation.

From the single turnover measurement of dark-adapted cells, also the time constant for electron transport at the acceptor side of PSII (τ_Qa, μs), the connectivity factor of adjacent PSII light-harvesting pigment matrices (p, dimensionless) and the concentration of functional PSII reaction centers ([RCII], nmol m⁻³), were derived according to Oxborough et al. (2012), using FastPro8 software (Version 1.0.50, Kevin Oxborough, CTG Ltd.).

All data are given as replicate means (n = 3) ± SE. To test for significant differences between treatments, two-way analyses of variance (ANOVA) with additional normality (Shapiro-Wilk) and post hoc (Holm–Sidak method) tests were performed (α = 0.05). In addition to this, to test for direct effects between two particular treatments standard t-tests were used. All statistical analyses were carried out with SigmaPlot 12.3 (SysStat Software Inc.). Different letters in figures and tables indicate statistical differences between treatments based on post-hoc tests.

Despite the large difference in cell size both investigated species showed comparable growth rates (Figures 1A,B). Several studies, investigating polar diatom species, of very different cell sizes, reported growth rates that are comparable to those found in this study (Gilstad and Sakshaug, 1990; Kropuenske et al., 2009; Arrigo et al., 2010; Boelen et al., 2011; Petrou et al., 2014). Hence, within polar diatoms growth rates generally seem not to be greatly affected by cell size, but much more by abiotic factors such as temperature and light availability. Growth rates of Fragilariopsis curta (Fragilariopsis) generally increased from low (LL) to medium light (ML) conditions and slightly decreased toward high light (HL, Figure 1A). At ambient pCO₂ (380 μatm), Fragilariopsis increased growth by 20% from LL to ML, whereas it decreased growth by 58% between ML and HL conditions. Under high pCO₂ (1000 μatm), no significant light effects were detectable in growth. At low pCO₂ (180 μatm), however, a strong increase in growth (86%) occurred between LL and ML, with no detectable changes from ML to HL. There were no overall trends with increasing pCO₂ discernible within all light treatments. At LL, growth of Fragilariopsis was suppressed in low pCO₂ treatments compared to the other two pCO₂ treatments, whereas at ML both low and high pCO₂ treatments grew less than the ambient treatment, yet only the former was statistically significant. The reverse pattern occurred at HL, exhibiting decreased growth at ambient pCO₂, 22% compared to low and 39% toward high pCO₂ treatments.

In Odontella weisflogii, (Odontella), in the ambient and high pCO₂ treatments growth increased from LL to ML by 58% and 79%, respectively, yet both pCO₂ treatments showed no changes between ML and HL (Figure 1B). Within the low pCO₂ treatments, no light effects on growth were found from ML to HL, yet at LL despite several attempts Odontella could not grow. At LL and ML, growth of Odontella decreased from ambient to high pCO₂. No differences between pCO₂ treatments were detectable at HL. At ML and HL, low pCO₂ acclimated cells grew slower than the ones acclimated to ambient pCO₂ while they reached similar growth rates as cells grown at high pCO₂.

The production rate of particulate organic carbon (POC) revealed species-specific trends (Figures 1C,D). In Fragilariopsis, at ambient pCO₂ POC production significantly decreased (84%) from LL to HL. Under high pCO₂, POC production decreased by 51% from LL to ML and showed no further changes from ML to HL. Under low pCO₂, POC production increased about 86% from LL to ML and decreased about 67% from ML to HL. At LL, POC production rates increased about 91% from low to ambient pCO₂ and decreased by 39% in response to high pCO₂. At ML, POC production decreased about 53% from low to high pCO₂, whereas at HL no differences between pCO₂ treatments were found.

In Odontella, POC production rates increased from LL to ML by 86 and 93% at ambient and high pCO₂, respectively. At ambient pCO₂, POC production rates decreased from ML to HL. Within high and low pCO₂ treatments, no changes were detected between ML and HL. At LL and ML, the ambient pCO₂ treatments had higher POC production rates than high pCO₂ treatments. Low pCO₂ treatments revealed lower POC production rates at ML, when compared to the two other pCO₂ treatments. No differences between pCO₂ treatments were observed at HL.

In Fragilariopsis, the ratio of cellular carbon to nitrogen (C:N, mol mol⁻¹) showed effects of both light and pCO₂. Under all three pCO₂ treatments C:N ratios decreased from LL to ML, but increased from ML to HL under high pCO₂ (Figure 1E). Under LL, effects of pCO₂ became evident as C:N ratios were reduced under high compared to low and ambient pCO₂, whereas the opposite was true under HL with increased C:N ratios under high pCO₂. Odontella, however, showed no effects of light on the C:N ratio (Figure 1F). Only under ML, the ratio of C:N increased from low to ambient pCO₂ with no further increase toward high pCO₂.

Photosynthesis-irradiance curves (PE-curves; Figure 2) displayed a strong light effect as maximum electron transport rates (ETRm), light saturation point (I_K) and light use efficiency (α, Table 2) increased with increasing irradiance in both Fragilariopsis and Odontella, but there were no differences in I_k and α between LL and ML in both species. The pCO₂ did not affect absolute ETRs (absETR) within LL treatments in both species (Figures 2A,B). Under ML, ETRm and α increased from low to ambient pCO₂ in both species, but remained unaltered between ambient and high pCO₂. Only at HL, an effect of pCO₂ on absETR (Figures 2E,F) and I_K (Table 2) was evident in Fragilariopsis and Odontella, with decreased ETRm and I_K values in the low compared to ambient and high pCO₂ treatments, respectively.

The maximum quantum yield (F_v/F_m) is an indicator for overall fitness and revealed species-specific patterns under the different experimental treatments (Figures 3A,B). In Fragilariopsis, F_v/F_m decreased from LL to ML and remained unchanged between ML and HL in the ambient pCO₂ treatments. At LL, the low pCO₂ treatments exhibited the lowest yield. Besides this, no pCO₂ effects were found. In Odontella, F_v/F_m did not change between LL and ML in the ambient pCO₂ treatments, whereas it decreased from LL to ML in the high pCO₂ treatments. Between ML and HL, F_v/F_m decreased in all pCO₂ treatments.

To test whether both species experienced short-term light stress or any damage in PSII, potentially induced during the PE-curve, a second dark acclimation phase right after the PE-curve was carried out followed by another F_v/F_m measurement, named yield recovery (given as % of initial F_v/F_m, Figures 3C,D). The two species showed diverging patterns of yield recovery. In Fragilariopsis, the recovery potential increased with increasing light intensity. In Odontella, a reversed bell shape was observed with a decrease in recovery from LL to ML and an increase from ML to HL. Yield recovery was not affected by pCO₂ under all light conditions in Odontella, whereas in Fragilariopsis the recovery increased from low to high pCO₂ conditions at LL and ML, but remained unaffected by pCO₂ at HL.

Cellular pigment content generally decreased with increasing light intensity (Table 3). In both species and for all light-harvesting pigments (Chl a, Chl c2, fucoxanthin, β-carotene, Table 3), a significant decrease between LL and ML was observed in all pCO₂ treatments and only minor or no decreases were observed between ML and HL conditions. Besides the light effect, Fragilariopsis showed a pCO₂ effect on all light-harvesting pigments (Chl a, Chl c2, fucoxanthin and β-carotene, Table 3). To prevent repetition only the CO₂-dependent change in Chl a will be described exemplarily for all light-harvesting pigments. At LL, cellular Chl a concentrations increased from low to ambient pCO₂ and decreased toward high pCO₂. At ML and HL, no pCO₂ effects were detected. In contrast to Fragilariopsis, in Odontella cellular Chl a concentrations remained unaltered in response to changes in pCO₂(Table 3).

The functional absorption cross section of PSII (σ_PSII, Table 4), is a measure of the size of the functional “target area” of light-harvesting antenna. In Fragilariopsis, σ_PSII decreased with increasing light intensity, whereas there was no change detectable in Odontella. However, effects of pCO₂ on σ_PSII were found in both species. Under low compared to ambient and high pCO₂, σ_PSII was significantly lower at ML and HL in both diatoms.

Light had an influence on re-oxidation times of the primary electron acceptor Q_a (τ_Qa, Table 4), but with different effects for the two investigated species. In Fragilariopsis, at LL, low pCO₂ treatments showed the highest re-oxidation times with no differences between ambient and high pCO₂, whereas at ML the re-oxidation time increased from low to high pCO₂, and at HL no differences between pCO₂ treatments were detectable. In Odontella, at ambient pCO₂, τ_Qa decreased from ML to HL. In the high pCO₂ treatments of Odontella τ_Qa increased with light. There was no light effect present in low pCO₂ treatments. When light was limiting (LL), τ_Qa was lower at high than at ambient pCO₂. At ML and HL, in Odontella τ_Qa was less at low compared to ambient and high pCO₂. At HL, the high pCO₂ treatments showed a higher τ_Qa than ambient treatments (results are insignificant due to a lack of statistical power).

The connectivity between PSIIs (p, Table 4) was influenced by pCO₂ in Fragilariopsis. Low pCO₂ treatments showed decreased connectivity compared to ambient and high pCO₂ treatments at LL and ML. In Odontella, in all pCO₂ treatments p decreased from ML to HL. At ML, p increased from low to ambient pCO₂ and remained constant between ambient and high pCO₂. At HL the high pCO₂ treatments showed the lowest p. Yet, statistical results for Odontella were not significant.

In Fragilariopsis, the concentration of PSII reaction centers ([RCII]) varied with light, decreasing significantly between LL and ML in the ambient and high pCO₂ treatments, respectively and increasing between ML and HL at ambient, but not at high pCO₂ (Table 4). Only in the low pCO₂ treatments, [RCII] was not altered in response to increasing light intensities. At LL, [RCII] increased with increasing pCO₂ in Fragilariopsis, whereas at ML the ambient pCO₂ treatments showed higher concentrations than the low pCO₂ treatments and no difference from the high pCO₂ treatments. Compared to Fragilariopsis, in Odontella [RCII] exhibited a different pattern with light. Concentrations of RCII increased from LL to ML in ambient and high pCO₂ treatments, but decreased from ML to HL in all pCO₂ treatments. At LL, [RCII] did not change in response to pCO₂ in Odontella. At ML, [RCII] increased from low to ambient pCO₂, but not from ambient to high pCO₂. Only at HL, the low and high pCO₂ treatments displayed lower concentrations of RCII than the ambient pCO₂ treatments.

The two diatom species showed different light-dependent trends in the de-epoxidation state (DES), an indicator for the dissipation of excess light energy (Table 3). In Odontella DES increased with increasing light intensity in all pCO₂ treatments, whereas Fragilariopsis showed a bell shaped pattern. DES increased from LL to ML in Fragilariopsis, and decreased toward HL in all pCO₂ treatments. Besides the low pCO₂ treatments, no de-epoxidation was observed in Fragilariopsis at LL. Furthermore, in Fragilariopsis at ML and HL DES was lower in the low compared to ambient pCO₂ treatments. Odontella exhibited elevated DES only in the low pCO₂ treatments at ML and HL.

The NPQ in Fragilariopsis showed an increase with increasing light intensity, while Odontella displayed a bell shaped pattern with highest quenching rates at ML (Figure 4). For the low pCO₂ treatments, NPQ reached higher values at LL in Fragilariopsis and at HL in Odontella.

Under ambient pCO₂, an increase in light intensity from low light (LL) to medium light (ML) stimulated growth by 42% in Odontella, with no further stimulation between ML and high light (HL, Figure 1B), indicating that growth was saturated at ML in this species. Other studies also reported that growth was stimulated by light when increased from lower to higher irradiances in various diatoms and prymnesiophytes (Bartual and Gálvez, 2002; Arrigo et al., 2010; Boelen et al., 2011). Congruently, growth of Fragilariopsis was also stimulated by 20% from LL to ML, but declined by 58% from ML to HL (Figure 1A). Such a decline at high light intensities was already found in the diatom Phaeodactylum tricornutum and the prymnesiophyte Phaeocystis antarctica (Arrigo et al., 2010; Li et al., 2014). In Fragilariopsis, particulate organic carbon (POC) production rates and quantum yield of photosynthesis (F_v/F_m), a measure of a cell's photosynthetic performance, were highest under LL and decreased from LL to ML, indicating saturation of photosynthesis already at LL conditions and an onset of light stress under ML (Figures 1C, 3A). In contrast, POC production increased from LL to ML in Odontella. This finding suggests that even under LL conditions Fragilariopsis was able to fix POC at a maximum rate while for Odontella this was only the case at ML, thus indicating species-specific light optima for photosynthesis. Despite the decrease in POC production from LL to ML in Fragilariopsis, growth increased suggesting that cell volumes might have changed. However, cell size measurements did not show any changes in both species and in all treatments (data not shown). Considering further that cell volume of Odontella was on average 2000 times larger than of Fragilariopsis (cell volumes estimated based on Hillebrand et al. (1999), data not shown) this potentially implied a much higher carbon demand in the former. For both species from ML to HL, we observed a significant decline in POC production (Figures 1C,D) and in Odontella a decrease in F_v/F_m (Figure 3B). In contrast to this observation, ETRm increased in both species from LL to HL, indicating no light saturation (Table 2). Due to this, the observed reduced POC production rates, between LL and HL in Fragilariopsis and between ML and HL in Odontella, might have resulted from a saturation of the Calvin-Benson-Cycle. The Calvin-Benson-Cycle is considered to be the rate-limiting step of photosynthesis under excessive light conditions, thus creating the demand of alternative electron pathways such as Mehler reaction, midstream oxidase pathways and cyclic electron transport around PSI to dissipate excess electrons (Behrenfeld and Milligan, 2012). Cyclic electron transport would rise the transthylakoid pH gradient and consequently lead to a higher production of ATP at the expense of NADPH and thereby would result in lower POC fixation rates (Falk and Palmqvist, 1992). Congruently with the changes in POC production the photosynthetic performance (F_v/F_m) declined from LL to ML in Fragilariopsis and from ML to HL in Odontella (Figures 1C,D, 3A,B). With increasing irradiance, F_v/F_m was commonly found to decrease in various temperate and Antarctic phytoplankton species (Boelen et al., 2011; Hoogstraten et al., 2012; Li et al., 2014) implying a lower photosynthetic capacity through photoinhibition and eventually photodamage of PSII. The observed species-specific changes in F_v/F_m may have resulted from a decreased ratio of photons absorbed per electrons generated in PSII. The amount of electrons generated depends on the cell's ability to capture light energy via light-harvesting pigments and the amount of functional PSII reaction centers. Less active PSII can result from closure of reaction centers (Schreiber, 2004) through photodamage (e.g., damage of D1 protein of PSII; Alderkamp et al., 2010), which can be an initial stage of photoacclimation process yielding reduced concentrations of PSII (Sakshaug et al., 1997).

For both diatoms, cellular concentrations of light-harvesting pigments (Ch a, Chl c2, fucoxanthin) significantly decreased similarly between LL and ML conditions, suggesting an acclimation to increased light by reducing light-harvesting pigment concentration for both species (Table 3). Yet, between ML and HL concentrations of light-harvesting pigments remained similar in both species, indicating a lower limit of pigment content for light absorption. We, however, found that concentrations of functional reaction centers ([RCII]) and F_v/F_m decreased significantly between ML and HL in Odontella while the absorption cross section of PSII (σ_PSII) remained unchanged (Figure 3B, Table 4). This shows that Odontella experienced stress at HL compared to ML. In comparison, F_v/F_m, [RCII] and σ_PSII already declined from LL to ML in Fragilariopsis (Figure 3A, Table 4). This finding suggests a higher susceptibility of this species to light stress already at moderate light intensities, yet these did not affect growth rates until HL conditions were applied.

To counteract photodamage, phytoplankton cells possess various strategies such as non-photochemical quenching (NPQ) including the operation of the xanthophyll cycle (XC), therefore dissipation of excess light energy in Odontella and Fragilariopsis was investigated via the concentrations of cellular diadinoxanthin (DD), its de-epoxidised form diatoxanthin (DT) and the de-epoxidation state (DES). An increase in DES (Table 3) with incident light is a response commonly found in phytoplankton (Casper-Lindley and Björkman, 1998; Arrigo et al., 2010; Petrou et al., 2011). NPQ increased from LL to HL in Fragilariopsis while in Odontella it increased mainly from LL to ML, but decreased from ML to HL (Figure 4). In Odontella, DES rose from LL to HL whereas there was no XC activity detectable in Fragilariopsis at LL, but there was a decrease from ML to HL (Table 3). The discrepancy in the trends of NPQ and DES from ML to HL in both species could result from smaller and much larger cellular DD+DT pools in Odontella and Fragilariopsis (−60 and +300%; Table 3), respectively under these conditions changing the general capacity for energy dissipation.

Overall, the two diatom species showed diverging responses to increasing light at ambient pCO₂ during acclimation. Fragilariopsis was characterized by highest POC production rates and photosynthetic performance (F_v/F_m) under LL, but experienced light stress with increasing light intensity, as POC production rates decreased and NPQ increased from LL to HL. Hence, it showed good ability for energy dissipation even under HL conditions and thus seemed better able to cope with light stress. In comparison, Odontella showed an increase in POC production rates from LL to ML, indicating no saturation at LL. With increasing light intensities, Odontella was susceptible to light stress as POC production, F_v/F_m and NPQ declined indicating a limited potential to tolerate HL conditions.

Phytoplankton blooms in Antarctic coastal waters occur annually and have been found to sometimes coincide with a severe drawdown of inorganic carbon through high photosynthetic carbon fixation rates in surface waters (Sweeney, 2003; Arrigo et al., 2008). In order to sustain photosynthesis under these low CO₂ conditions, phytoplankton cells need to up-regulate their carbon concentrating mechanisms (CCMs) leading to increased energy demands and potentially impeding carbon fixation (Raven et al., 2014).

Low pCO₂ in combination with LL was found to have the strongest impact on growth and carbon fixation in comparison to all other treatments. Under these conditions, both factors light and CO₂ were limiting, causing lowest growth and POC production rates in Fragilariopsis while Odontella did not grow (Figures 1A–D). Commonly, the CCM is up-regulated under low pCO₂ in order to supply CO₂ to RubisCO, which is highly energy demanding (Hopkinson et al., 2011). Yet, in conjunction with LL conditions, the production of energy equivalents during photosynthesis might not be sufficient to maintain carbon fixation at a maximum, potentially causing the here observed minimal growth and POC production in Fragilariopsis. Additionally, light-harvesting pigment concentrations (Table 3), [RCII] and p (Table 4) were lowest under these conditions, indicating a small efficiency of individual PSII reactions centers with a low probability of excitation distribution among them. Furthermore, an excess of excitation energy was clearly visible as NPQ during PE-curves (Figure 4) was highest and yield recovery thereafter was lowest (Figure 3C). Overall, Fragilariopsis revealed great difficulties to grow under LL in conjunction with low pCO₂. Nonetheless, it grew whereas Odontella could not cope with these highly stressful conditions. With increasing light availability the energetic constraints under LL were alleviated in both species, as potentially a larger share of photosynthetically generated energy (ATP and NADPH) was available to fuel their CCMs. Hence, at ML in conjunction with low pCO₂ Odontella was able to grow and fix POC. This stimulative light effect was also found in Fragilariopsis, strongly increasing growth and POC production from LL to ML (Figures 1A,C).

Within the ML treatments, pCO₂ was also found to stimulate growth rates of Odontella and Fragilariopsis by 8 and 42% from low to ambient pCO₂, respectively (Figures 1A,B), suggesting optimal growth conditions for both species at ambient pCO₂. A similar CO₂-dependent increase in growth was previously reported for the two polar diatoms Rhizosolenia cf. alata and C. debilis (Riebesell et al., 1993; Trimborn et al., 2013). POC production was enhanced from low to ambient pCO₂ levels in Odontella (Figure 1D) potentially resulting from a higher diffusive CO₂ supply and therewith a lower energy demand. However, for Fragilariopsis we observed a CO₂-dependent decline in POC production (Figure 1C). This decline was accompanied by a reduced capability for light absorption as cellular light-harvesting pigment concentrations were decreased at ambient compared to low pCO₂ (Chl a, Chl c2, fucoxanthin, Table 3). Please note that Fragilariopsis experienced light stress already at ML when grown at ambient pCO₂, hinting toward a higher susceptibility to light stress in this species. Through the down-regulation of the CCM from low to ambient pCO₂, it may serve less as an energy sink and thereby increasing light stress exerted by excessive light (Gao et al., 2012). This was, however, not reflected in any change in F_v/F_m. In Odontella, under ML absETRs, ETRm, I_K and α decreased from low to ambient pCO₂ while POC production increased indicating that carbon uptake might not have been sufficient to saturate RubisCO and the Calvin-Benson-Cycle. The reason for this might be its higher cell volume compared to Fragilariopsis and thereby higher amount of carbon necessary to saturate the Calvin-Benson-Cycle and to sustain growth. Increased absETR under these conditions could further indicate cycling of electrons around PSI and thus generation of energy equivalents that could be funneled into carbon uptake. At HL, we observed an increase of the photosynthetic efficiency from low to ambient pCO₂treatments of both species, displaying higher absETR (Figures 2E,F), ETRm, I_K, α (Table 2), σ_PSII, τ_Qa (Odontella only), p and [RCII] (Fragilariopsis only, Table 4) while NPQ and DD+DT pools decreased in both species, indicating lowered energy dissipation with increasing pCO₂. Hence, in the low pCO₂ treatments the HL was not utilized, but cells rather dissipated the excess light energy absorbed. As the CCM of both species was potentially down-regulated from low to ambient pCO₂, the higher photosynthetic efficiencies as well as lowered energy dissipation at HL suggest that the CCM does not represent an important sink for excess energy under high light conditions (Gao et al., 2012).

Overall, this study revealed that low pCO₂ concentrations were stressful for both Antarctic diatoms, but especially in conjunction with light limiting conditions. According to our results, the hypothesis that the regulation of the CCM aids in the dissipation of excess energy at high light intensities cannot be conclusively answered for the two species tested here.

Under OA, Antarctic diatoms might benefit from diffusive CO₂ uptake as has been shown for temperate species (Burkhardt et al., 2001; Rost et al., 2003; Trimborn et al., 2009), decreasing the demand for CCM operation. Especially under limiting light conditions, the lowered energy demand through down-regulation of the CCM was found to stimulate growth and carbon fixation in temperate phytoplankton species (Wu et al., 2010; Li et al., 2014). Similarly, the temperate coccolithophorid Emiliania huxleyi showed increased POC production rates under OA when irradiances are low, yet under higher light intensities this effect was amended (Rokitta and Rost, 2012). In our investigated species, however, growth and POC production remained either unaffected or declined under OA and LL (Figures 1A–D) contradicting the aforementioned assumption of a stimulative effect of OA, but going in line with other findings for Antarctic phytoplankton species (Hoogstraten et al., 2012). We even observed for Fragilariopsis negative effects of OA under LL on light-harvesting pigment concentrations (Chl a, Chl c2, and fucoxanthin; Table 3) potentially compensated by an increased number of active RCII (Table 4). Interestingly, in Odontella light harvesting pigments (Chl c2, fucoxanthin, Table 3) were increased under elevated pCO₂, indicating a maximization of light absorption under LL. In agreement with this, an OA-dependent induction of chlorophyll-fucoxanthin protein genes were reported for the temperate diatom Phaeodactylum tricornutum (Li et al., 2015). In addition to this, enhanced rates of mitochondrial respiration were previously reported under OA in combination with relatively LL levels in the temperate diatom Thalassiosira pseudonana (Yang and Gao, 2012). Together with the species-specific photoacclimation responses under OA, one could hypothesize that the two investigated species may have experienced higher metabolic costs, potentially causing the observed decline in POC production.

For both species, the negative OA effects were amplified under ML conditions. Next to negative responses in growth and POC production (Figures 1A–D), the two species displayed OA-dependent photoacclimation. While Odontella revealed lowered photochemical efficiencies (Figure 3B), Fragilariopsis had less active RCIIs with reduced connectivity between PSIIs (Table 4) potentially counteracted by increased light harvesting pigmentation (Chl a and fucoxanthin, Table 3). For both species, acclimation to high light conditions was indicated by a larger cellular DD+DT pool size (Table 3). This is in line with previous suggestions that down-regulated CCMs serve less as a sink for excess light energy under OA in conjunction with saturating light intensities (Gao et al., 2012). However, this positive effect of CCM operation under excessive irradiances was not found when comparing low and ambient pCO₂ treatments under HL (see discussion above). Unexpectedly, this negative OA effect was alleviated under HL conditions in our two tested species. In this case, growth rates and POC production remained unaffected (Figures 1A–C). Only for Fragilariopsis, POC production was stimulated by OA and HL. For the latter, C:N ratios were also found to raise (Figure 1E). For the two coccolithophores Gephyrocapsa oceanica and Coccolithus pelagicus ssp. braarudii, OA also increased POC production rates, but C:N ratios were either stimulated (200 μmol photons m⁻² s⁻¹, Rickaby et al., 2010) or reduced (100 μmol photons m⁻² s⁻¹, Jin et al., 2013) in G. oceanica. In our study, the CO₂-dependent increase in C:N ratios under HL was mainly due to lowered cellular PON content (data not shown), the underlying reason for this, however, remains unclear.

From our results, we can conclude that OA in conjunction with all tested irradiances did not lead to stimulation in growth or POC production in any of the tested species. It was, however, evident that increasing light intensities during acclimation caused OA-dependent photoacclimation responses and negatively impacted POC production. Only under HL, this OA effect was alleviated in Odontella, while in Fragilariopsis POC production was even increased, potentially resulting from a shift in carbon allocation. Congruently we cannot support the finding that a down-regulation of the CCM and consequent decrease of energy dissipation therein increases light stress under OA in conjunction with HL in these Antarctic diatoms.

Antarctic coastal waters were found to have high iron concentrations and form highly productive and extensive blooms over the season (Holm-Hansen et al., 1989; Martin et al., 1990; Prézelin et al., 2000; Garibotti et al., 2003; de Jong et al., 2012). Annual patterns of phytoplankton species succession and their presence in either spring or summer blooms can be related to species-specific abilities to acclimate, utilize and tolerate different light intensities. The physiological characteristics of both species we observed match well their seasonal occurrence. In line with its presence in spring (Garibotti et al., 2005; Annett et al., 2010), Fragilariopsis showed greater ability than Odontella to grow under LL conditions. Accordingly, it was characterized by highest POC production rates under limiting light as well as a good ability to endure high light stress when mixed to the water surface. Odontella was still light limited under LL, requiring higher irradiances coinciding with its occurrence in high abundances in summer when the upper mixed layer is shallow and stable (Mitchell and Holm-Hansen, 1991; Nelson and Smith, 1991). Congruently, it was able to endure longer periods of high irradiances. Yet, at the peak of a bloom when cell density is high, light is reduced and CO₂ gets drawn down by high photosynthetic activity, this creates very unfavorable growth conditions for Odontella.

Strong species-specific physiological responses were apparent in response to different future climate scenarios, mimicking either OA in conjunction with increased irradiances due to shallower UML depth or OA associated with decreased daily irradiances due to deepening of the UML through increased winds. Under ML, both species experienced light stress, which was further amplified by OA. Yet, in the HL OA-scenario Fragilariopsis was able to tolerate the HL conditions. Also under the LL OA-scenario, neither of the investigated species showed an OA-dependent stimulation in growth or POC production. Hence, OA mainly showed negative effects on growth and carbon fixation in both diatom species, implying that OA could potentially reduce the strength of the biological carbon pump under the tested light scenarios.

The results of this study indicate that physiological traits can help to explain the spatial distribution of diatom species in the current Southern Ocean. We further demonstrate that the here tested future climatic scenarios could negatively affect growth and carbon fixation of both diatom species with potential implications for a future SO. Furthermore, the results of this study demonstrated that the effect of OA was strongly modulated by the irradiance regime, pointing out the need to conduct multiple stressor experiments to better understand the impact of climate change on Southern Ocean key phytoplankton species.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer PJ and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review