Strong shift from HCO3− to CO2 uptake in Emiliania huxleyi with acidification: new approach unravels acclimation versus short-term pH effects

Effects of ocean acidification on Emiliania huxleyi strain RCC 1216 (calcifying, diploid life-cycle stage) and RCC 1217 (non-calcifying, haploid life-cycle stage) were investigated by measuring growth, elemental composition, and production rates under different pCO2 levels (380 and 950 μatm). In these differently acclimated cells, the photosynthetic carbon source was assessed by a 14C disequilibrium assay, conducted over a range of ecologically relevant pH values (7.9–8.7). In agreement with previous studies, we observed decreased calcification and stimulated biomass production in diploid cells under high pCO2, but no CO2-dependent changes in biomass production for haploid cells. In both life-cycle stages, the relative contributions of CO2 and HCO3 − uptake depended strongly on the assay pH. At pH values ≤ 8.1, cells preferentially used CO2 (≥ 90 % CO2), whereas at pH values ≥ 8.3, cells progressively increased the fraction of HCO3 − uptake (~45 % CO2 at pH 8.7 in diploid cells; ~55 % CO2 at pH 8.5 in haploid cells). In contrast to the short-term effect of the assay pH, the pCO2 acclimation history had no significant effect on the carbon uptake behavior. A numerical sensitivity study confirmed that the pH-modification in the 14C disequilibrium method yields reliable results, provided that model parameters (e.g., pH, temperature) are kept within typical measurement uncertainties. Our results demonstrate a high plasticity of E. huxleyi to rapidly adjust carbon acquisition to the external carbon supply and/or pH, and provide an explanation for the paradoxical observation of high CO2 sensitivity despite the apparently high HCO3 − usage seen in previous studies. Electronic supplementary material The online version of this article (doi:10.1007/s11120-014-9984-9) contains supplementary material, which is available to authorized users.


Introduction
Marine phytoplankton account for *50 % of global primary production and are the main drivers of the marine ''particulate organic carbon'' (POC) pump Field et al. 1998). Calcifying phytoplankton species also contribute to the ''particulate inorganic carbon'' (PIC) pump and thereby play a dual role in regulating marine biogeochemical cycling of carbon through their effects on surface ocean alkalinity (Broecker and Peng 1982;Zeebe and Wolf-Gladrow 2007). One key species of calcifying phytoplankton is the cosmopolitan and bloom-forming coccolithophore Emiliania huxleyi, which has been established as a model organism over the recent decades (Paasche 2002; Raven and Crawfurd 2012;Read et al. 2013;Westbroek et al. 1993). While the calcifying diploid lifecycle stage of this species has been intensively studied in field and laboratory experiments, the non-calcifying haploid stage has only recently gained attention due to its important ecological role. In blooms of diploid E. huxleyi, which are usually terminated by viruses, the haploid life-cycle stage functions as a virus-resistant backup population (Frada et al. 2012). Furthermore, the presence and absence of calcification in the differing life-cycle stages of E. huxleyi make them ideal candidates to investigate the cellular mechanisms of calcification and their interaction with photosynthesis under increasing oceanic CO 2 concentrations (Mackinder et al. 2010;. Increasing pCO 2 in oceanic surface water directly affects carbonate chemistry by elevating the concentration of dissolved inorganic carbon (DIC) and shifting the carbon speciation toward higher CO 2 and H ? concentrations, a phenomenon often referred to as ocean acidification (OA; Caldeira and Wickett 2003;Wolf-Gladrow et al. 1999). Compared to preindustrial values, pH is expected to drop by 0.4-0.5 units until the end of this century. In several studies testing the effects of OA on E. huxleyi, diploid strains were found to exhibit strong, yet opposing responses in terms of biomass and calcite production. While biomass production was either unaffected or stimulated by increased pCO 2 , calcification typically decreased and malformations of coccoliths increased (e.g., Hoppe et al. 2011;Langer et al. 2009;Riebesell et al. 2000). Bach et al. (2011) suggested that biomass production is stimulated by increasing CO 2 concentration at sub-saturating conditions, whereas calcification is specifically responsive to the associated decrease in pH. Such differential CO 2 and pH effects on biomass and calcite production are supported by the observation that OA distorts ion homeostasis and shifts the metabolism from oxidative to reductive pathways Taylor et al. 2011). In a number of studies, the sensitivity of E. huxleyi toward OA has been attributed to its mode of inorganic carbon (C i ) acquisition, which is intrinsically responsive to changes in carbonate chemistry. Thus, for understanding the differential responses to OA, one needs to look at this crucial process of C i assimilation.
Like most phytoplankton, E. huxleyi operates a CO 2 concentrating mechanism (CCM), which utilizes CO 2 and/ or HCO 3 uptake systems to accumulate CO 2 in the vicinity of RubisCO, and employs the enzyme carbonic anhydrase (CA) to accelerate the inter-conversion between these C i species (see Reinfelder 2011 for review). For a long time, the CCM in E. huxleyi was assumed to rely on the CO 2 delivery by calcification (Anning et al. 1996;Sikes et al. 1980). More recently, however, studies have demonstrated that C i fluxes for photosynthesis and calcification are independent (Herfort et al. 2004;Rost et al. 2002;Trimborn et al. 2007), and that these two processes may even compete for C i substrates . Most studies performed on the CCM of E. huxleyi to date yielded moderately high substrate affinities for C i , which decreased slightly under OA scenarios (e.g., Rost et al. 2003, Stojkovic et al. 2013).
Moreover, low activity for extracellular CA and high contribution of HCO 3 uptake for photosynthesis have been reported (e.g., Herfort et al. 2002;Stojkovic et al. 2013;Trimborn et al. 2007). This high apparent HCO 3 usage is puzzling, however, as it suggests biomass production to be rather insensitive to OArelated changes in CO 2 supply, which is in contrast to what studies usually have observed.
Most physiological methods characterizing the CCM and its functional elements are performed under standardized assay conditions, including a fixed pH value, and thus differing from treatment conditions. The pH and the concominant C i speciation can, however, influence the cell's physiology, in particular its C i acquisition. When identifying the cause-effect relationship in OA responses, it is difficult to separate the effects of changes in C i speciation from concomitant changes in H ? concentrations. Changes in external pH have been shown to directly drive changes in cytosolic pH in E. huxleyi, which, in turn, affected H ? gradients and membrane potentials (Suffrian et al. 2011;Taylor et al. 2011). This effect could indirectly impact secondary active transporters, e.g., the Cl -/HCO 3 antiporter (Herfort et al. 2002;Rokitta et al. 2011). Moreover, the protonation of amino acid side chains can affect activity, specificity, and kinetics of enzymes and transporters involved in cellular processes (Badger 2003;Raven 2006). Hence, aside from altered concentrations of C i species, pH itself could directly impact the mode of CCM (Raven 1990). These possible effects of the assay pH on C i acquisition should be accounted for when performing experiments to characterize the CCM.
One common approach to determine the C i source for photosynthesis is the application of the 14 C disequilibrium method (Espie and Colman 1986), which has proven suitable for the study of marine phytoplankton in laboratory cultures (e.g., Elzenga et al. 2000;Rost et al. 2006a) and in natural field assemblages (e.g., Cassar et al. 2004;Martin and Tortell 2006;Tortell and Morel 2002;Tortell et al. 2008). The method makes use of the relatively slow chemical conversion between the CO 2 and HCO 3 in the absence of CA (Johnson 1982), allowing for a differential labeling of these C i species with 14 C. This method is typically performed at pH of 8.5 (''assay pH''), deviating strongly from most natural in situ values and even more from the pH values applied in OA-experiments (''acclimation pH''). In this study, we aimed to disentangle the short-term effect of assay pH from the long-term effect of acclimation history on the photosynthetic C i source of E. huxleyi. To this end, we grew haploid and diploid lifecycle stages at present-day (380 latm) and elevated pCO 2 (950 latm), and measured the responses in growth, elemental composition, and production rates. These low and high pCO 2 -acclimated cells were then tested for their preferred C i source by applying the 14 C disequilibrium method, with assay conditions set to a range of ecologically relevant pH values (pH 7.9-8.7). The reliability of this new approach was tested by performing sensitivity studies.
Prior to experiments, cells were acclimated to the respective pCO 2 and light conditions for at least 7 days (i.e., more than 10 generations). Prior to initiating cultures, medium was pre-aerated for at least 36 h with humidified, 0.2 lm-filtered air comprising pCO 2 values of 380 or 950 latm (equivalent to 38.5 and 96.3 Pa, or *15 and *35 lmol kg -1 , respectively). Gas mixtures were created by a gas flow controller (CGM 2000 MCZ Umwelttechnik, Bad Nauheim, Germany) using pure CO 2 (Air Liquide Deutschland, Düsseldorf, Germany) and CO 2 -free air (CO2RP280, Dominick Hunter, Willich, Germany). Sampling and measurements were done 4-8 h after the beginning of the light period (i.e., at midday) in exponential growth at densities of 40,000-60,000 cells mL -1 . Cultures showing a pH drift of [ 0.05 were excluded from further analyses.
The carbonate system (Table 1) during the acclimations was assessed based on measurements of pH and total alkalinity (TA). The pH NBS of the cultures was measured potentiometrically and corrected for temperature (pHmeter 3110; WTW, Weilheim, Germany). The electrode (A157, Schott Instruments, Mainz, Germany) was three-point calibrated with NBS certified standard buffers and the measurement uncertainty was 0.03 pH units. TA was determined by potentiometric titration (Dickson 1981; TitroLine alpha plus, Schott Instruments). Measurements were accuracy-corrected with certified reference materials (CRMs) supplied by A. Dickson (Scripps Institution of Oceanography, USA). Calculation of the carbonate system was performed using CO2sys (Pierrot et al. 2006). Input parameters were pH NBS and TA, as well as temperature (15°C), salinity (32.4), and pressure (1 dbar, according to 1 m depth; Hoppe et al. 2012). For all calculations, phosphate and silicate concentrations were assumed to be 7 and 17 lmol kg -1 , respectively, based on assessments of the media. Equilibrium constants for carbonic acid, K 1 and K 2 given by Mehrbach et al. (1973) and refit by Dickson and Millero (1987) were used. For the dissociation of sulfuric acid, the constants reported by Dickson (1990) were employed.
Cell growth was assessed by daily cell counting with a Multisizer III hemocytometer (Beckman-Coulter, Fullerton, CA, USA) and the specific growth rates (l) were calculated from daily increments (cf., . For the determination of total particulate carbon (TPC), POC and particulate organic nitrogen (PON), cell suspensions were vacuum-filtered (-200 mbar relative to atmosphere) onto pre-combusted (12 h, 500°C) GF/F filters (1.2 lm; Whatman, Maidstone, UK), which were dried at 65°C and analyzed with a EuroVector CHNS-O elemental analyzer (EuroEA, Milano, Italy). Before quantification of POC, filters were HCl-soaked (200 lL, 0.2 M) and dried to remove calcite. PIC was assessed as the difference between TPC and POC. By multiplying the POC and PIC cell quotas with l, the respective production rates were derived (cf., . For Chl a measurements, cells were filtered onto cellulose nitrate filters (0.45 lm; Sartorius, Göttingen, Germany) and instantly frozen in liquid nitrogen. Chl a was extracted in 90 % acetone (v/v, Sigma, Munich, Germany) and determined fluorometrically (TD-700 fluorometer, Turner Designs, Sunnyvale, USA) following the protocol by Holm-Hansen and Riemann (1978). The calibration of the fluorometer was carried out with a commercially available Chl a standard (Anacystis nidulans, Sigma, Steinheim, Germany).

C disequilibrium method
The C i source for photosynthesis was determined by applying the 14 C disequilibrium method (Elzenga et al. 2000;Espie and Colman 1986;Tortell and Morel 2002). In this method, a transient isotopic disequilibrium is induced by adding a small volume of a 14 C i ''spike'' solution with a relatively low pH (typically 7.0) into larger volume of buffered cell suspension with a relatively high pH (typically 8.5). The cell suspension contains dextran-bound sulfonamide (DBS) to eliminate possible external CA activity. Due to the pH-dependent speciation of DIC, the relative CO 2 concentration of the spike is high (*19 % of DIC at pH 7.0), compared to the cell suspension (*0.3 % of DIC at pH 8.5). When adding the spike to the cell suspension, the majority of the CO 2 added with the spike Photosynth Res (2014) 121:265-275 267 converts into HCO 3 until equilibrium is achieved (Johnson 1982;Millero and Roy 1997). Consequently, the specific activity of CO 2 (SA CO 2 , dpm (mol CO 2 ) -1 ) is initially high and exponentially decays over time (Fig. 1). The slope of the 14 C incorporation curve of a ''CO 2 user'' is, therefore, initially much steeper than during final linear 14 C uptake, when isotopic equilibrium is achieved. In contrast, the slope of 14 C incorporation for ''HCO 3 users'' changes only marginally over time because SA HCO À 3 stays more or less constant during the assay.
Quantification of the relative proportion of CO 2 or HCO 3 usage was done by fitting data with the integral function of the 14 C fixation rate (Elzenga et al. 2000;Espie and Colman 1986;Martin and Tortell 2006). The function includes terms representing the instantaneous fixation rate of DI 14 C, the fractional contribution of CO 2 f CO 2 ð Þ or HCO 3 usage 1 À f CO 2 ð Þto the overall C i fixation and the specific activity (SA, dpm mol -1 ) of these substrates at any given time (Eq. 1; Espie and Colman 1986;Elzenga et al. 2000;Tortell and Morel 2002). Strictly speaking, as HCO 3 and CO 3 2cannot be differentially labeled, 1 À f CO 2 also comprises the potential fraction of CO 3 2used. dpm In this equation, V DI14C is the total rate of 14 C uptake; f CO 2 is the fraction of uptake attributable to CO 2 ; a 1 and a 2 are the temperature-, salinity-, and pH-dependent firstorder rate constants for CO 2 and HCO 3 hydration and dehydration, respectively; t is the time (s); DSA CO 2 and DSA HCO 3 are the differences between the initial and equilibrium values of the specific activities of CO 2 and HCO 3 -, respectively; and SA DIC is the specific activity of DIC. During steady-state photosynthesis, V DI14C and f CO 2 are assumed to be constant so that changes in the instantaneous 14 C uptake rate reflect only changes in the specific activity of CO 2 and HCO 3 -. In the present study, the 14 C disequilibrium method was modified to enable measurements over a range of ecologically relevant pH values (7.90-8.70). In order to maintain a suitably large initial isotopic disequilibrium DSA CO 2 =SA DIC ð Þ , the pH of the 14 C spike solutions needs to be adjusted in conjunction with the pH of the assay buffer. We, thus, used either MES or HEPES buffers to set the pH of spike solutions over the range of 5.75-7.30 (see Table 2 for exact pH values of assay and spike buffers). For the assays, 10-30 9 10 6 cells were concentrated via gentle filtration over a polycarbonate filter (2 lm; Millipore, Billerica, MA, USA) to a final volume of 15 mL. During this filtration procedure, cells were kept in suspension, while the medium was gradually exchanged with buffered assay medium of the appropriate pH value. Assay media and spike buffers were prepared at least 1 day prior  to the assay and stored in closed containers to avoid CO 2 exchange and pH drift. The pH value and temperatures of all buffers were measured immediately prior to assay runs. DIC concentration of the assay buffers was determined colorimetrically according to Stoll et al. (2001) using a TRAACS CS800 autoanalyzer (Seal Analytical, Norderstedt, Germany), and measurements were accuracy-corrected with CRMs supplied by A. Dickson (Scripps Institution of Oceanography, USA).
To initiate the assays, a volume of 4 mL buffered concentrated cell suspension was transferred into a temperature-controlled, illuminated glass cuvette (15°C; 300 lmol photons m -2 s -1 ) to which 50 lM DBS was added (Ramidus, Lund, Sweden). Cells were continuously stirred in the light for at least 5 min prior to spike addition to reach steady-state photosynthesis. Spike solutions were prepared by adding NaH 14 CO 3 solution (1.88 GBq (mmol DIC) -1 ; GE Healthcare, Amersham, UK) into a final volume of 200 lL of pH-buffered MilliQ water (various buffers at 20 mM; Table 2), yielding activities of *370 kBq (10 lCi). Following the spike addition, 200 lL subsamples of the cell suspension were transferred into 2 mL HCl (6 M) at time points between 5 s and 12 min. Addition of these aliquots to the strong acid caused instant cell death and converted all DIC and PIC to CO 2 . DI 14 C background was degassed in a custom-built desiccator for several days until samples were dry. Deionized water (1 mL) was then added to re-suspend samples prior to addition of 10 mL of scintillation cocktail (Ultima Gold AB, GMI, Ramsey, MN, USA), and the sample was vortexed thoroughly.
Acid-stable (i.e., organic) 14 C activity in samples was counted with a Packard Tri-Carb Liquid Scintillation Counter (GMI). Blank samples, consisting of cell-free medium, were treated alongside the other samples. In the few cases where no blanks were available, time zero values were approximated by extrapolating the y-axis intercept from linear fitting of the first three data points of the 14 C incorporation curves. Total radioactivity of the NaH 14 CO 3 stock solution was regularly quantified and compared to expected values to estimate loss of radioactivity or changes in counting efficiency. In all spike solutions, measured radioactivity ranged between 80 and 100 % of the theoretical values, and the actual radioactivity levels were used in the calculation of the specific activities. Blank-corrected data were fitted (Eq. 1), using a least-squares-fitting procedure. Applied fit parameters are given in Table 2. Furthermore, a detailed Excel spread sheet for calculating the fit parameters in dependence of the applied conditions (e.g., pH, temperature and DIC concentrations) is provided as Supplementary Material. Please note that in the calculation of initial and final specific activities, we accounted not only for changes in concentrations of 14 C i species but also for  changes in concentrations of DI 12 C, 12 CO 2 , and H 12 CO 3 upon spike addition. If these changes are neglected, DSA CO 2 =SA DIC will be significantly overestimated, leading to an underestimation of f CO 2 (Eq. 1, Table 2, Supplementary material). We used a numerical sensitivity study to examine how offsets in parameters such as pH, DIC concentrations, radioactivity, temperature, or blank values influence the derived estimates of f CO 2 . First, theoretical 14 C incorporation curves for ''HCO 3 users'' f CO 2 ¼ 0:25 ð Þand ''CO 2 users'' f CO 2 ¼ 0:80 ð Þwere generated for two assay pH values (7.90 and 8.50) and used as a reference, assuming fixed values of DIC concentrations of 2,300 lmol kg -1 , assay temperature of 15°C, spike solution temperature of 23°C and spike radioactivity of 370 kBq. In a second step, model fits were obtained using slight offsets in these parameters (e.g., pH 7.95 and 7.85 instead of 7.90) to obtain the effect of parameter variability on f CO 2 estimates. Sensitivity toward over-and underestimation of pH, temperature, DIC concentration, and radioactivity was tested. We further assessed the effects of blank values (±100 dpm) on f CO 2 estimates as a function of different final 14 C incorporation rates.

Statistics
All experiments were performed using at least biological triplicates (i.e., three independent, but equally treated cultures). When data were normally distributed (Shapiro-Wilk test) and showed equal variance (Equal-Variance Test), significance in difference between pCO 2 treatments was tested by performing student 0 s t-tests. When samples were not normally distributed or did not show equal variance, a rank sum test was performed instead. Null hypotheses were rejected when p B 0.05, unless otherwise indicated.

Results
In diploid cells of E. huxleyi, the specific growth rate l and PIC quotas did not change significantly in response to elevated pCO 2 (Table 3). While there was a small decrease in PIC production rates (-11 %), POC quotas and production rates increased strongly under elevated pCO 2 (?77 and ?55 %, respectively). In conjunction with these changes, the quotas and production rates of TPC also increased (?28 and ?23 %, respectively). The PIC:POC ratios of diploid cells decreased from 1.4 to 0.7 under elevated pCO 2 , while the POC:PON ratios increased from 6.3 to 8.8. Chl a quotas were largely unaffected by the pCO 2 treatments, although Chl a:POC ratios decreased significantly from 0.022 to 0.012 pg pg -1 under elevated pCO 2 , owing to the change in POC quotas. In haploid cells, neither l, elemental quotas or the respective production rates showed any significant response to elevated pCO 2 Table 3 Growth rates, elemental quotas and production rates, elemental ratios, as well as pigment composition of haploid (1N) and diploid (2N) cells of E. huxleyi, cultured at low (380 latm) and elevated pCO 2 (950 latm): l (day -1 ), POC quota (pg cell -1 ), POC production (pg cell -1 day -1 ), PIC quota (pg cell -1 ), PIC production (pg cell -1 day -1 ), TPC quota (pg cell -1 ), TPC production (pg cell -1 day -1 ), PON quota (pg cell -1 ), PON production (pg cell -1 day -1 ), PIC:POC ratio (mol:mol), POC:PON ratio (mol:mol), Chl a quotas (pg cell -1 ), and Chl a:POC ratios (pg:pg)  (Table 3). Similarly, Chl a quotas, Chl a:POC, and POC:PON ratios were all unaffected by the experimental CO 2 manipulations in the haploid strain. Under both pCO 2 acclimations, diploid cells were shown to be predominant ''CO 2 users'' under low assay pH (f CO 2 * 1.0 at pH 7.9; Fig. 2a). With increasing assay pH, however, we observed a significant increase in relative HCO 3 utilization. HCO 3 uptake was induced at assay pH C 8.3 (equivalent to CO 2 concentrations B 9 lmol L -1 ), reaching considerable contribution at high assay pH (f CO 2 * 0.44 at pH 8.7). In contrast to the strong effect of the assay pH, the tested pCO 2 acclimations had no effect on the pH-dependent C i uptake behavior (Fig. 2a). In other words, both low and high pCO 2acclimated cells showed the same short-term response of f CO 2 to assay pH. Like the diploid stage, haploid cells progressively changed from high CO 2 usage at low assay pH (f CO 2 * 0.96 at pH 7.9) to substantial HCO 3 contributions when assays were conducted in high pH assay buffers (f CO 2 * 0.55 at pH 8.5; Fig. 2b). HCO 3 uptake became relevant at pH C 8.1 (equivalent to CO 2 concentrations B 14 lmol L -1 ), particularly in low pCO 2 -acclimated cells. Except for haploid cells measured at pH 8.1, no significant differences in f CO 2 were observed between the low and high pCO 2 acclimations (Fig. 2b).
The sensitivity analysis showed that an offset in the input pH of the buffered assay cell suspension (± 0.05 pH units) led to deviations in f CO 2 of B 0.09 (i.e., 9 percentage points) in ''CO 2 users'' and B 0.02 in ''HCO 3 users'' (Fig. 3a). An offset in the input temperature of the assay buffer (± 2°C) led to a deviation in f CO 2 of B 0.09 in ''CO 2 users'' and B 0.03 in ''HCO 3 users'' (Fig. 3a). An offset in the input pH of the spike (± 0.05 pH units) changed the f CO 2 estimates by B 0.08 in ''CO 2 users'' and B 0.03 in ''HCO 3 users'' (Fig. 3a). Applying an offset in the input temperature of the spike (± 2°C) caused a deviation in f CO 2 by B 0.06 in ''CO 2 users'' and had practically no effect on f CO 2 in ''HCO 3 users'' (B 0.01; Fig. 3a). An offset in the input DIC concentration of the buffer (± 100 lmol kg -1 ) affected f CO 2 by B 0.08 in ''CO 2 users'' and B 0.03 in ''HCO 3 users''. Regarding the radioactivity of the spike (± 37 kBq), deviations in f CO 2 were B 0.12 in ''CO 2 users'' and B 0.04 in ''HCO 3 users.'' Irrespective of CO 2 or HCO 3 usage, offsets in blank estimations (± 100 dpm) led to deviating f CO 2 by B 0.27, but only when equilibrium 14 C fixation rates were B 1 dpm s -1 (Fig. 3b). When steady-state 14 C incorporation rates were C 2 dpm s -1 (i.e., average rate in diploid cells) and C 4 dpm s -1 (i.e., average rate in haploid cells), the deviations in f CO 2 due to offsets in the blanks were B 0.17 and B 0.11, respectively.

Acclimation responses
This study corroborates previous findings on the general sensitivity of the diploid life-cycle stage of E. huxleyi toward OA (e.g., Feng et al. 2008;Langer et al. 2009;Riebesell et al. 2000). While growth rate was unaffected, OA reduced PIC production and stimulated POC production (Table 3). Consequently, the PIC:POC ratio was strongly decreased under OA, indicating a redirection of C i fluxes between these two processes. Transcriptomics have previously attributed this redirection to an inhibition of calcification in response to impaired signal-transduction and ion-transport, as well as to stimulation in the production of glycoconjugates and lipids ). In our study, also the TPC production increased significantly under OA (Table 3), indicating that not only C i is allocated differently, but also the overall C i uptake increases with the increasing pCO 2 . Our data further suggest that less energy is required for the C i acquisition under OA as more POC and TPC could be produced even though the Chl a quota  (Table 3). Improved energy-use efficiencies under OA have previously been proposed for the diploid life-cycle stage of E. huxleyi .
In strong contrast to the diploid strain, the haploid lifecycle stage of E. huxleyi was insensitive toward OA with respect to growth rate and elemental composition ( Table 3). The ability of the haploid cells to maintain homeostasis under OA has also been observed by . Even though the haploid cells appeared non-responsive toward OA on the phenomenological level (i.e., growth, elemental composition), transcriptomics have revealed significant changes at the subcellular level, such as an upregulation of catabolic pathways under OA . Based on the comparison of the life-cycle stages, Rokitta and co-workers concluded that the OA sensitivity in diploid cells originates from calcification, differences in C i acquisition or both.
A number of studies have shown that E. huxleyi has moderately high C i affinities and uses HCO 3 as the primary C i source (e.g., Herfort et al. 2002;Rost et al. 2006b;Stojkovic et al. 2013), irrespective of the degree of calcification (Trimborn et al. 2007;. These characteristics would suggest E. huxleyi to be rather insensitive toward OA and the associated rise in CO 2 concentration, contrary to most results obtained for the diplont. As discussed below, this apparent discrepancy could originate from differences in conditions applied during short-term physiological measurements and those conditions cells experience in the long-term acclimation.

Modes of C i acquisition
Our results demonstrate that the C i source of both life-cycle stages of E. huxleyi is significantly influenced by the pH of the assay medium and the resulting carbonate chemistry (Fig. 2). With increasing pH in assay buffers, cells progressively changed from predominant CO 2 usage at lower pH values (B 8.1) to significant HCO 3 contribution at higher pH (C 8.3). Surprisingly, this change occurred irrespectively of the pCO 2 conditions in the acclimation. To our knowledge, such a strong short-term pH-dependence in C i acquisition has not been previously reported, which is most likely due to the fact that assays are typically performed under standardized pH values. Measuring physiological responses under one reference condition have the advantage that consequences of different acclimations can readily be compared in terms of altered capacities of certain processes, e.g., enzyme activities or transport rates. However, determination of the C i source at one standard pH appears to impose a methodological bias, and our results, therefore, bear direct relevance to the interpretation of previous laboratory observations.
In view of the short-term pH effect on C i acquisition, the contribution of HCO 3 as a photosynthetic C i source in E. huxleyi may have possibly been overestimated in previous studies. This overestimation is likely to be the most significant in those studies when 14 C disequilibrium assays were conducted at pH 8.5 (e.g., Rost et al. 2007). By looking at the C i source determined at an assay pH mimicking the acclimation condition, we can now re-evaluate and in fact explain the responses of 3 Sensitivity in f CO2 estimates for ''CO 2 users'' (f CO2 ¼ 0:80) and ''HCO 3 users'' (f CO2 ¼ 0:25) at low pH (7.9, in gray) and high pH (8.5, in white) A toward negative (inverted filled triangle) and positive (filled triangle) offsets in the pH, temperature, and DIC concentration of the assay buffer (pH Assay , T Assay , and [DIC]), as well as toward offsets pH, temperature, and radioactivity of the spike (pH Spike , T Spike , and RA), and B toward negative (inverted filled triangle) and positive (filled triangle) offsets in blank measurements (±100 dpm) in dependence of the final 14 C incorporation rates. Sensitivity was assessed based on theoretical curves with constraints of a [DIC] Assay = 2,300 lM, T Assay = 15°C, T Spike = 23°C, and RA Spike = 37 kBq. Dashed lines indicate f CO2 values as expected for optimal experimental conditions E. huxleyi toward elevated pCO 2 . When assessing f CO 2 using assay buffers of pH 7.9 and 8.1 (equivalent to the acclimation pH of high and low pCO 2 treatments), we observed predominant CO 2 uptake under both conditions (Fig. 2). Being ''CO 2 user'', cells were thus able to directly benefit from changes in the CO 2 concentrations in our acclimations (*15 lmol kg -1 at 380 latm and *38 lmol kg -1 at 950 latm). For a ''HCO 3 user'', however, it would be difficult to argue for a beneficial OAeffect as HCO 3 concentrations do not differ much between treatments (*1,930 lmol kg -1 at 380 latm and *2,130 lmol kg -1 at 950 latm). Our results thus suggest that biomass production in diploid cells not only profits from the declined calcification at high pCO 2 , as suggested by  but also from the higher CO 2 supply under OA. As CO 2 usage is considered to be less costly than HCO 3 uptake (Raven 1990), this could also explain the higher energy-use efficiency observed for E. huxleyi .
Although the haploid life-cycle stage of E. huxleyi exhibited a pH-dependent C i uptake behavior that was similar to the diploid (Fig. 2), the haploid cells did not show any CO 2 -dependent stimulation in biomass production (Table 3). This could partly be related to the fact that the biomass production cannot profit from a down-scaling of calcification, simply because this process is absent in the haploid life-cycle stage. The lack of significantly stimulated biomass buildup under OA could also be attributed to the concomitant upregulation of catabolic pathways, such as higher lipid consumption, which is a specific feature of the haploid cells . After all, the similar C i uptake behavior of both life-cycle stages confirms that photosynthetic HCO 3 usage is not tied to calcification (Herfort et al. 2004;Trimborn et al. 2007;Bach et al. 2013) and that the preference for CO 2 or HCO 3 is predominantly controlled by carbonate chemistry.
Our findings clearly demonstrate that the acclimation history, in both life-cycle stages, has little or no effect on the C i usage of the cells (Fig. 2). In other words, the instantaneous effect of the assay conditions dominates over acclimation effects. We cannot preclude, however, that cells acclimated to higher pH values, where CO 2 supply becomes limiting, may increase their capacity for HCO 3 uptake and acclimations effects would then be evident. Notwithstanding the potential for some acclimation effects, the extent to which short-term pH and/or CO 2 levels in the assay medium directly control cellular C i usage is striking. This implies that even though E. huxleyi did not use significant amounts of HCO 3 for photosynthesis, it must constitutively express a HCO 3 transporter in all acclimations. Without the presence of a functional HCO 3 transport system we could otherwise not explain the capacity for significant HCO 3 uptake under short-term exposure to high pH (even in high pCO 2 -acclimated cells).
In the diploid life-cycle stage, HCO 3 transporter may be constitutively expressed to fuel calcification, as HCO 3 was identified as the main C i source for this process (Paasche 1964;Rost et al. 2002;Sikes et al. 1980). If CO 2 supply for photosynthesis becomes limiting, HCO 3 transport could then also fuel photosynthesis. In the haploid cells, which do not calcify, we nonetheless observed the same capacity for HCO 3 uptake, which suggests that HCO 3 uptake capacity represents a fundamental component of the CCM of both life-cycle stages of E. huxleyi. Whether levels of protons or CO 2 concentrations are the main trigger for the shift between CO 2 and HCO 3 uptake remains unclear, even though there is strong evidence that CO 2 supply is the main driver for the responses in photosynthesis (Bach et al. 2011).

Sensitivity analyses
In our sensitivity study, the applied offsets in pH (± 0.05 pH units), temperature (± 2°C), DIC of the assay buffer (± 100 lM), and spike radioactivity (± 37 kBq) were larger than typical measurement errors to represent ''worst-case scenarios''. None of these offsets caused f CO 2 estimates to deviate by more 0.12 in any of the pH treatments (Fig. 3a). When adequate efforts are taken to control these parameters (e.g., using reference buffers, thermostats), methodological uncertainties are thus negligible. DIC concentrations and radioactivity, however, are often not measured and in view of the potential drift over time, offsets can easily exceed typical measurement errors and lead to severe deviations in f CO 2 . For instance, 14 CO 2 out-gassing causes the spike solution to progressively lose radioactivity. This loss of 14 C can easily be [ 20 % over the course of weeks or months, despite the high pH values of the stock solution and small headspace in the storage vial (Gattuso et al. 2010).
The average final 14 C fixation rates, which depend on the biomass and radioactivity used, were 2.1 ± 0.8 dpm s -1 in the runs with diploid and 6.6 ± 2.2 dpm s -1 in those with haploid cells (Fig. 3b). In these ranges, offsets in blank values (± 100 dpm) can lead to biases in the estimated f CO 2 by up to 0.20 (Fig. 3b). This strong sensitivity highlights the need to thoroughly determine blank values, but also to work with sufficiently high biomass and/or radioactivity to maximize 14 C incorporation rates. When working with dense cell suspensions, however, self-shading or significant drawdown of DIC during the assay might bias results. Higher label addition generally increases the resolution of the assay and lowers the consequences of offsets in the blank value. It should be noted, however, that high concentrations of 14 C in Photosynth Res (2014) 121:265-275 273 spike solutions can affect not only the isotopic but also the chemical conditions in the cuvette (e.g., pH and DIC). Overall, our sensitivity study revealed that the 14 C disequilibrium method is a straightforward and robust assay, which is very useful for resolving the C i source of phytoplankton over a range of different pH values. It is important to realize, however, the pH of assay buffers has the potential to significantly affect the C i uptake behavior of cells.

Conclusions
Our data clearly demonstrate that both life-cycle stages of E. huxleyi predominantly use CO 2 as C i source for photosynthesis under typical present-day and future CO 2 levels, but constitutively express HCO 3 transporters allowing them to directly use HCO 3 when CO 2 becomes limiting. Under bloom conditions, where pH values can easily increase to 8.5 or higher, cells might, therefore, be able to maintain efficient C i acquisition. Future research needs to investigate whether and how the assay pH governs the mode of C i acquisition also in other coccolithophores species or phytoplankton taxa and how this may alter the energy budget of cells. Results from previous studies may need re-consideration in the light of our data showing strong short-term pH effects on C i uptake of phytoplankton.