Emiliania huxleyi strain RCC1266, originally isolated from shelf water around Ireland (49^o30^ʹ N, 10^o30^ʹ W), was obtained from the Roscoff algal culture collection. Cells were maintained in semi-continuous cultures in artificial seawater (ASW) media prepared according to Berges et al. (2001) without the addition of NaHCO₃, with a salinity of 33psu, under 200μmol photons m⁻² s⁻¹ of photosynthetically active radiation (PAR; measured using a LI-190SA quantum sensor, Beijing Ligaotai Technology Co. Ltd., China) and a 12:12h light/dark cycle (light period: 8:00 to 20:00) and 20°C in a thermo-controlling incubator (GXZ, Dongnan Instrument Company). The DIC-free ASW media was enriched with 64μmolL⁻¹ NO₃⁻, 4μmolL⁻¹ PO₄³⁻, f/8 concentrations for trace metals and vitamins (Guillard and Ryther, 1962). To disentangle the effects of carbonate system parameters on elemental contents and biomacromolecules of E. huxleyi, calculated amounts of Na₂CO₃ (1.20molkg⁻¹, filtered by PTFE filter, 0.22μm pore size, Filter-Bio, Nantong) and hydrochloric acid (HCl, 2.00molkg⁻¹) were added stepwise into the media to achieve four different carbonate chemistry conditions (Table 1). Initial CO₂ concentration and pH_Total (Total scale) value were set to 400μatm and 8.04 under low CO₂ and high pH treatment (1980μmolkg⁻¹ DIC, LCHpH, present DIC and pH treatment), 400μatm and 7.70 under low CO₂ and low pH treatment (910μmolkg⁻¹ DIC, LCLpH, reduced pH), 1,000μatm and 8.04 under high CO₂ and high pH treatment (4,930μmolkg⁻¹ DIC, HCHpH, ocean carbonation), and 1,000μatm and 7.70 under high CO₂ and low pH treatment (2,160μmolkg⁻¹ DIC, HCLpH, ocean acidification), respectively. In each condition, the ASW was put into the incubator at 20°C for 24h and then filtered (0.22μm pore size, Polycap 75 AS, Whatman) and carefully pumped into autoclaved 100ml, 1,120ml, and 2,350ml polycarbonate bottles with no headspace to minimize gas exchange. 100ml seawater was used to determine initial total alkalinity (TA) and pH_Total of seawater, 1,120ml bottles were used to acclimate cells to experimental conditions (one replicate), and the main experiment culture was conducted in 2350ml bottles (four replicates). The cells were inoculated to achieve an initial concentration of about 3,000 cell ml⁻¹ and cultured in each experimental condition for 2days and then diluted to the initial cell density again (Supplementary Figure S1). This process was performed four times under each treatments, and cells were maintained in exponential growth phase for a minimum of 13 generations (Supplementary Table S1) and then transferred from 1,120ml to 2,350ml bottles at the same time. The volume of culture inoculum was calculated to match the volume of media taken out from the bottles before inoculation. Culture bottles were rotated 10 times until cells were mixed at 9:00, 14:00 and 18:00 (Beijing Time). In the second days of the main experiments, cell densities were lower than 70,000 cells ml⁻¹ under all treatments, and sub-samples in each incubation bottle were harvested with different volume for measurements of TA, pH_Total, cell density, total particulate carbon (TPC), POC, and nitrogen (PON), protein, carbohydrate, and pigment. It should be mentioned that Barcelos e Ramos et al. (2010) reported that E. huxleyi can rapidly alter the rates of essential metabolical processes in response to changing seawater carbonate chemistry within 26h. In this study, E. huxleyi cells were incubated under each treatment with 4 biological replicates for 2days, thus the results found here were comparable with other studies (see discussion section; Feng et al., 2017; Bi et al., 2018).

In the second days of the main experiments, to reduce impact of respiration of E. huxleyi on seawater pH value and total alkalinity (TA), 40ml and 50ml samples were, respectively, filtered (PTFE filter, 0.22μm pore size, Nantong) 5h after the onset of the light period (at 13:00) and used to measure the pH_Total and TA. The bottles were filled from bottom to top with overflow, and closed immediately without a headspace. The pH_Total was measured immediately at 20°C using a pH meter which was calibrated with buffers (Tris•HCl, Hanna) at pH 4.01, 7.00 and 10.00. The pH value was not corrected with a standard buffer of defined pH in seawater (Dickson et al., 2007). TA samples were stored at 4°C for a maximum of 7days (Zhang et al., 2021), and TA was measured at 20°C by potentiometric titration (AS-ALK1+, Apollo SciTech) according to Dickson et al. (2003). Phosphate concentration was not measured at the beginning and end of the incubations. In this study, the carbonate system was estimated from TA, pH_Total, temperature, salinity, and phosphate (4μmolL⁻¹) using the CO2SYS program (Pierrot et al., 2006) with carbonic acid constants, K₁ and K₂, taken from Roy et al. (1993).

Twenty fiveml samples were taken daily 6h after the onset of the light period (at 14:00) and fresh ASW with the same carbonate chemistry as in the initial treatment conditions were added as top-up. Cell density was measured using a Z2 Coulter Particle Count and Size Analyzer (Beckman Coulter). The diameter of detected particles was set to 3–7μm in the instrument, which excluded detached coccoliths (Müller et al., 2012). Cell concentration was also measured by microscopy (ZEISS), and variation in measured cell concentrations was±8% between the two methods. Growth rate (μ) was calculated for each replicate according to the equation: μ=(ln N_t−ln N₀)/d, where N_t and N₀ refer to the cell concentrations in the second day and beginning of the main experiment, respectively, and d was the growth period in days.

The photosynthetic fluorescence parameters were determined using a pulse amplitude modulated fluorometer (WATER–PAM, Walz, Effeltrich, Germany). 2ml samples were taken from the incubation bottles and kept in the dark for 15min at 20°C. The assay light levels (A-PAR) between 0 and 1,120μmol photons m⁻² s⁻¹ were applied in nine steps and 45s each in fast light response curve measurements. The instant minimal (F₀ʹ) and maximal fluorescence (F_mʹ) were, respectively, determined at the end of each A-PAR and saturating light pulse (800ms, 3,000μmol photons m⁻² s⁻¹). The effective photochemical quantum yield (Yield) was calculated as: Yield=(F_mʹ−F₀ʹ)/F_mʹ (Baker, 2008). The relative electron transport rate (rETR) was calculated as: rETR=Yield×A-PAR (Ralph and Gademann, 2005). The parameters of the photosynthesis vs. irradiance curves (P–I curves) were analyzed as follows: rETR=rETR_max×tanh (a×A-PAR/rETR_max; Jasby and Platt, 1976). The maximal relative electron transport rate (rETR_max) represents the light-saturating level of rETR, and light use efficiency (a) was derived from the slope of each electron transport rate (ETR) vs. light curve. Saturating light intensity, I_k, is calculated from the expression rETR_max/a and is characteristic for the onset of light saturation.

One hundredml samples were filtered onto pre-combusted GF/F filters (at 450°C for 6h) with low vacuum pressure (<0.02MPa) at 15:00 and soaked in 5ml 90% acetone overnight at 4°C in the dark. The extracts were centrifuged at 4300×g for 10min to remove glass fibers. Absorbance (A) of the supernatant on 750, 664, 647, and 480nm was measured using a spectrophotometer (SP-722, Shanghai Spectrum Instruments, China). The chlorophyll a (Chl a) concentration was calculated as: Chl a (μgml⁻¹)=−1.93×(A₆₄₇–A₇₅₀)+11.93×(A₆₆₄–A₇₅₀). The carotenoid concentration was calculated as: Carotenoid (μgml⁻¹)=4×(A₄₈₀–A₇₅₀; Jeffrey and Humphrey, 1975; Davies, 1976).

Two hundredml samples for determinations of TPC and POC (PON) were, respectively, gently filtered onto GF/F filters (pre-combusted at 450°C for 6h) at the same time (15:30) in each treatment, rinsed three times with DIC–free ASW media, and then stored in the dark at −20°C. For determination of POC (PON), samples were fumed with HCl for 12h to remove inorganic carbon. TPC and POC (PON) samples were dried at 60°C for 12h and analyzed using an Elementar CHNS analyzer (Vario EL cube, GmbH, Germany). PIC quota was calculated as the difference between TPC and POC quota (Fabry and Balch, 2010). PIC, POC, and PON production rates were calculated by multiplying cellular contents with μ (d⁻¹), respectively, (Supplementary Figure S2).

After mixing, 700ml and 800ml samples for determinations of protein and carbohydrate were, respectively, filtered onto polycarbonate filters (25mm diameter, 0.6μm pore size, Nuclepore, Whatman) and onto pre-combusted GF/F filters at 15:30 and then stored in the dark at −80°C. Protein samples were put into 2ml MP Biomedical tubes (Lysing, Matrix D) containing large ceramic beads. After being freeze-dried, protein samples were extracted by a mixture of 0.7ml 1X protein extraction buffer (0.10mmolL⁻¹ 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (protease inhibitor), 125.00mmolL⁻¹ ethylene diamine tetraacetic acid, 28.07mmolL⁻¹ Tris base, 21.10mmolL⁻¹ Tris-HCl, 1085.89mmolL⁻¹glycerol and 73.44mmolL⁻¹ lithium dodecyl sulfate). Cells were lysed using a FastPrep–24 machine (MP Biomedicals, United States) at 6.5ms⁻¹ with 4cycles and 1min each, and samples were chilled in an ice bath for 2min between 2cycles. Then, the samples were centrifuged at 10,000×g for 5min (Centrifuge 5,418 R, Eppendorf, Germany), and extracted protein in the supernatant was determined at 562nm using the BCA Assay with Bovine gamma globulin as a standard, using a spectrophotometer (Ni et al., 2016).

Carbohydrate samples were firstly treated with 12.00molL⁻¹ sulfuric acid (H₂SO₄) in the dark for 1h and then diluted by milliQ water for a final H₂SO₄ concentration of 1.20molL⁻¹. Samples were then sonicated for 5min and vortexed for 30s, and boiled at 90.0°C for 3h in a water bath (Pakulski and Benner, 1992). The concentration of monosaccharide was analyzed at 490nm by phenol–sulfuric reaction using a spectrophotometer with D-glucose as standard (Masuko et al., 2005).

The elemental content of macromolecular pools was calculated based on a mean elemental stoichiometry for protein, carbohydrates, Chl a and carotenoid (Geider and LaRoche, 2002). Two-way ANOVA were used to determine the main effects of CO₂ concentration and pH value and their interactions on each variable. Tukey post hoc tests were conducted to identify significant differences between two CO₂ concentrations or two pH values. Normality of residuals was tested using a Shapiro–Wilk test, and a Levene test was used to analyze for homogeneity of variances. All statistical calculations were performed using R Core Team (2018) with the packages carData, lattice, and nlme.

Initial CO₂ concentrations ranged from 400 to 420μatm under low DIC (LC) treatments and were 1,010μatm under high DIC (HC) treatments (Table 1). Initial pH values were 8.04 and 7.70 in high and low pH conditions, respectively. During the experiment, DIC concentrations decreased by, on average, 4.57 to 14.42% under various carbonate chemistry treatments. Correspondingly, CO₂ concentrations decreased by, on average, 10.44 to 20.61%, and pH_Total values increased by less than 0.07 under different experimental treatments (Table 1). Under low pH treatment, low DIC concentration (LCLpH) represents low CO₂, bicarbonate (HCO₃⁻), carbonate (CO₃²⁻) and TA concentrations, and under high pH treatment, high DIC concentration (HCHpH) represents high CO₂, HCO₃⁻, CO₃²⁻, and TA concentrations (Table 1).

Compared to present DIC concentration and pH value (LCHpH), ocean acidification (high DIC concentration and low pH value, HCLpH) did not significantly affect growth rate, rETR_max, light use efficiency (a), and saturating light intensity (I_k; all p>0.10) and increased chlorophyll a (Chl a) and carotenoid content by 13.56 and 7.17%, respectively (both p<0.01; Figures 1, 2, gray). Under the same pH treatments, the effects of elevated DIC concentration on growth rate, Chl a and carotenoid contents, rETR_max, a, and I_k were positive, which can be seen by comparing these variables at HC concentrations with their paired LC concentrations (Figures 1, 2). Compared to low DIC concentration, high DIC concentration increased growth rate by 18.96% at high pH and by 10.34% at low pH, increased Chl a content by 33.83% at high pH and by 65.02% at low pH, increased carotenoid content by 53.55% at high pH and by 45.51% at low pH (Figures 1A–C), increased rETR_max by 22.50% at high pH and by 21.99% at low pH, increased a by 20.88% at high pH and by 12.40% at low pH (all p<0.05), respectively (Figures 2B,C; Tables 2, 3) and did not significantly increase I_k under both high and low pH treatments (both p>0.10; Figure 2D).

Under the same CO₂ concentrations, the effects of reduced pH value on growth rate, Chl a and carotenoid contents, rETR_max, a and I_k were negative, which can be seen by comparing these variables in low pH condition with their paired high pH condition (Figures 1, 2). Compared to high pH value, low pH value reduced growth rate by 12.82% at LC and by 19.14% at HC, reduced Chl a content by 31.18% at LC and by 15.14% at HC, reduced carotenoid content by 26.35% at LC and by 30.21% at HC (Figures 1A–C), reduced rETR_max by 21.59% at LC and by 21.91% at HC (all p<0.01), reduced a by 8.16% at LC and by 14.60% at HC (p=0.10 at LC; p<0.01 at HC) and reduced I_k by 14.68% at LC and by 8.52% at HC (p<0.01 at LC; p=0.14 at HC), respectively (Figures 2B–D). These results showed that elevated DIC concentration compensated for the negative effects of reduced pH on growth rate, Chl a and carotenoid contents, rETR_max, a and I_k.

Compared to present DIC concentration and pH value (LCHpH), ocean acidification (HCLpH) reduced PIC content by 40.93% and PIC/POC ratio by 44.05%, and increased POC content by 13.75% and did not significantly affect PON content (p=0.24) and POC/PON ratio (p=0.62; Figure 3, gray). The effects of elevated DIC concentration on cellular PIC, POC, and PON contents, and on PIC/POC ratio and POC/PON ratio were pH dependent (Figure 3). In high pH condition, elevated DIC concentration increased PIC content by 37.93%, POC content by 44.43%, PON content by 34.09% (all p<0.01; Tables 2, 3) and did not significantly affect PIC/POC ratio and POC/PON ratio. In low pH condition, elevated DIC concentration increased PIC content (p<0.01), PIC/POC ratio (p=0.01) and POC/PON ratio (p=0.07) did not significantly affect POC content and significantly reduced PON content.

Under both low and high DIC concentrations, reduced pH value reduced PIC contents by 53.74–57.17% and PIC/POC ratios by 39.62–63.53% (Figures 3A,D). The effects of reduced pH value on POC and PON contents were DIC dependent (Figures 3B,C). In low DIC concentration, reduced pH value increased POC content by 18.03% and PON content by 27.09% (both p<0.01); in high DIC concentration, it reduced POC content by 21.24% and PON content by 19.78% (both p<0.01; Figures 3B,C). These results showed that reduced pH value mainly reduced PIC content and PIC/POC ratio and reduced pH value and elevated DIC concentration acted antagonistically to affect cellular POC and PON contents.

Compared to present DIC concentration and pH value (LCHpH), ocean acidification (HCLpH) did not significantly affect protein content and increased carbohydrate content by 38.20% (Figure 4, gray). The effect of elevated DIC concentration on cellular protein content was pH dependent (Figure 4A). In high pH condition, elevated DIC concentration increased protein content by 60.42%, and in low pH condition, it did not significantly affect protein content (p=0.06). In low CO₂ concentration, reduced pH value did not significantly affect protein content (p=0.05), and in high CO₂ concentration, it significantly reduced protein content (p<0.01).

The effect of elevated DIC concentration on carbohydrate content was positive, which can be seen by comparing the carbohydrate contents in high DIC concentrations with their paired low DIC concentrations (Figure 4B). Elevated DIC concentration increased carbohydrate content by 55.43% in high pH condition and by 30.02% in low pH condition. The effect of reduced pH value on carbohydrate content was DIC concentration dependent. Reduced pH value did not significantly affect carbohydrate content in low DIC concentration and reduced carbohydrate content by 11.08% in high DIC concentration (Figure 4B, p<0.01). These results showed that elevated DIC concentration increased protein content in high pH condition but not in low pH condition and mainly increased carbohydrate content under both low and high pH treatments (Figures 4A,B).

Compared to present DIC concentration and pH value (LCHpH), ocean acidification (HCLpH) did not significantly reduce the percentage of POC allocated to protein (p=0.08; Figure 5A, gray), significantly increased the percentage of POC allocated to carbohydrate by 21.52% (Figure 5B, gray) and did not significantly affect the percentage of PON allocated to protein (p=0.82; Figure 5E, gray). Under high or low pH treatments, elevated DIC concentrations did not significantly affect the percentages of POC allocated to protein (Protein–C: POC; p=0.07 at high pH; p=0.88 at low pH; Figure 5A; Tables 2, 3). Reduced pH did not significantly affect the percentages of POC allocated to protein in low DIC concentration (p=0.27) and significantly reduced this percentage in high DIC concentration.

Elevated DIC concentration did not show significant influence on the percentage of POC allocated to carbohydrate (Carbohydrate–C: POC) under high pH treatment (p=0.40) and significantly increased this percentage by 34.98% under low pH treatment (Figure 5B). Under low or high DIC concentrations, reduced pH value did not significantly affect the percentages of POC allocated to carbohydrate (p=0.21 at LC; p=0.06 at HC). The percentages of POC allocated to Chl a (Chl a–C: POC) and carotenoid (Carotenoid–C: POC) were 0.45–0.78% and 0.38–0.65%, respectively, under different carbonate chemistry treatments (Figures 5C,D).

Under high or low pH treatments, the effect of elevated DIC concentration on the percentage of PON allocated to protein (Protein–N: PON) was positive, which can be seen by comparing this percentage in high DIC concentrations with their paired low DIC concentrations (Figure 5E). Elevated DIC concentration increased the percentage of PON allocated to protein by 19.73% in high pH condition and by 9.86% in low pH condition (p=0.11). Under low or high DIC concentrations, the effect of reduced pH value on percentage of PON allocated to protein was negative, which can be seen by comparing this percentage in low pH conditions with their paired high pH conditions (Figure 5E). Reduced pH value reduced the percentage of PON allocated to protein by 13.95% in low DIC concentration and by 21.04% in high DIC concentration (both p<0.05). These results showed that elevated DIC concentration and reduced pH value acted synergistically to increase the percentage of POC allocated to carbohydrate (Figure 5B), and antagonistically to affect the percentage of PON allocated to protein (Figure 5E).

Under four carbonate chemistry treatments, growth rate or PIC production rate significantly and positively correlated with rETR_max (R²=0.93 and 0.88, both p<0.01; Figures 6A,B). In addition, the percentage of POC allocated to protein (Protein–C: POC) or the percentage of PON allocated to protein (Protein–N: PON) significantly and positively correlated with growth rate (R²=0.61 and 0.83, both p<0.01; Figures 6C,D).