Two species of Alexandrium isolated from the coastal waters of the Korea Strait were used. Alexandrium affine (Aa-MEDHO0710) and A. pacificum (Ap-MEDHO0210) were isolated from Mijo Harbor (34°42′ N, 128°03′ E) in September 2007 and Jangmok Bay (34°59′ N, 128°40′ E) in February 2010, respectively. Cultures of these strains were maintained in f/2 medium without silicate at 20–23°C and 32–34 psu under a 12 : 12 h light-dark cycle with ~100 μmol photons m⁻² s⁻¹, typically containing a mixture of single cells and two- and more than three-cell chains. Alexandrium species-specific primer sets targeting the internal transcribed spacer region (5′-TGTGCTTGACTTTTACATGA-3′ and 5′-GAGCAGCACAGATACA-3′ for A. affine and 5′-TTCAATGCAAAACATTGACCT-3′, and 5′-CTGTTTGCATTTCTCTAGTT-3′ for A. pacificum) were applied to verify the identification of species by conventional PCRs (Supplementary Fig. S1).

Cells of A. affine and A. pacificum were acclimated in batch mode to the experimental temperatures and pCO₂ levels for 14 days before the start of the experiment. Subsequently, A. affine and A. pacificum were cultured semi-continuously with the abundance of 100 cells mL⁻¹ and 200 cells mL⁻¹, respectively. Compared with target pH values, these relatively low cell densities allowed us to maintain the pH with minor variations (<0.10 units). Cells incubated in 250-mL Erlenmeyer flasks at 33–34 psu under an incident irradiance of 150 μmol photons m⁻² s⁻¹ provided by four 36W daylight fluorescent lamps (Dulux L 36W/86; Osram, Münich, Germany), with a light-dark cycle of 14 : 10 h.

The five target temperatures were controlled using a thermostatically controlled water bath, which allowed control of the temperature within ±0.5°C (Supplementary Fig. S2). The temperature was logged every 30 min using data loggers (Onset Computer Corporation, Pocasset, MA, USA). A given target pCO₂ level was manipulated by adding an appropriate amount of CO₂-saturated seawater (CSS), no more than 10 μL. CSS was obtained by bubbling reagent-grade CO₂ through seawater in gas-tight containers with the pH dropping below 5. Samples were collected in triplicates from each semi-continuous culture for pH_T measurements to reset the target pCO₂ level. The photosynthetic efficiency and cell counts of the samples were also measured.

All inoculums were transferred weekly to keep them in the exponential growth phase until the start of the experiment. During the semi-continuous cultivation in Erlenmeyer flasks (100 mL capacity), cultures were maintained at the lower section of the exponential phase (Supplementary Fig. S3), and were optically thin to minimize self-shading, CO₂ drift, and nutrient limitation.

The inorganic carbonate chemistry was subsequently calculated with the help of CO2SYS program (Lewis and Wallace 1998) using the total pH scale (pH_T), total alkalinity (A_T), temperature, and salinity of culture seawater (Lim et al. 2020). The pH_T on the total hydrogen scale was monitored for each culture every day using a high-resolution UV-VIS spectrophotometer (Agilent 8453; Agilent Technologies, Pal Alto, CA, USA) fitted with 20-mm path length special optical glass spectrophotometric cells (18B-SOG-20, Starna). Absorbances of 30 mL seawater samples kept at 25°C were measured after adding 80 μL aliquots of m-cresol purple indicator dye (Sigma-Aldrich Chemical Co., St. Louis, MO, USA) at wavelengths 434, 578, and 730 nm.

Total alkalinity (A_T) measurements of each culture were obtained at the beginning and the end of the 10-day incubation using potentiometric titration with acid in a semi-closed cell system, consisting of a Metrohm 765 Dosimat titrator (Metrohm, Zofingen, Switzerland) connected to a pH meter (Orion 920A; Thermo Fisher Scientific, Waltham, MA, USA). The electromotive force during titration with acid (0.13 N HCl) was recorded using Q-basic software (Millero et al. 1993). A_T values were calibrated against certified seawater reference material (Andrew Dickson, Scripps Institution of Oceanography).

Live cells were enumerated every day in a 1 mL Sedgewick-Rafter counting chamber on an inverted microscope (IX73; Olympus, Tokyo, Japan). Specific growth rates (μ) were calculated for each replicate of the applied treatments by fitting an exponential function through all cell counts over time as follows:

, where N_t denotes the cell count at time t, and N₀ denotes the cell count at the start of each experiment.

Lugol-fixed samples from the harvest day were used for the measurement of chain-forming cells using an inverted microscope (IX73; Olympus). For each replicate, at least 100 single or chain cells were randomly assigned for each treatment.

The Cardinal Temperature Model with Inflection (CTMI) in which the maximum specific growth rate (μ_max) is described as a function of temperature (T, °C) alone was used (Rosso et al. 1993). The cardinal temperature parameters were obtained by fitting the CTMI defined as follows:

In CT_min ≤ T ≤ CT_max, where CT_min (°C) and CT_max (°C) are the hypothetical lower (minimum) and upper (maximum) critical temperatures, respectively, through which the growth rate is zero. However, we represented CT_min and CT_max as the temperatures at which the growth rate was 5% of the maximum growth rate due to their ambiguity, that determined the temperature tolerance range (T_tol). The thermal performance breadth (TB₈₀) was also defined as the 80th percentile of the CTMI fits where the performance was close to optimal. T_opt (°C) is the temperature at μ_max, and μ_opt is the specific growth rate at T_opt.

To determine photosynthetic activity, the maximum efficiency of photosystem II (PSII) photochemistry (i.e., the maximum quantum yield of PSII) was determined using a pulse amplitude-modulated (PAM) chlorophyll-a fluorometry (Phyto-PAM fluorometer; Walz GmbH, Effeltrich, Germany). The maximum quantum yield (F_v/F_m) for each culture was measured after being acclimatized to darkness for 15 min. Where F_v (variable fluorescence) is calculated as the difference between F_o and F_m fluorescence, where F_o is the minimum level of fluorescence in the absence of photosynthetic light at open reaction centers, and F_m is the maximum fluorescence yield induced by a saturation light pulse at closed reaction centers of PSII (Genty et al. 1989).

For experimental measures (carbonate chemistry, maximum efficiency of PSII photochemistry, growth rate, and chain formation), the main and interaction effects of temperature (five levels ranging from 15 to 31°C for A. affine or 30°C for A. pacificum) and pCO₂ (ambient and elevated levels) to assess significant responses were evaluated by one- or two-way analyses of variance (ANOVA). Data were initially checked for conformity to normality using the Shapiro-Wilk test, and for homogeneity of variances using Levene’s test. If a significant ANOVA was found, differences among the group means were evaluated using Turkey’s post hoc tests for multiple comparisons. All statistical analyses were performed in SPSS Statistics version 25 (IBM Corp., Armonk, NY, USA).

The values of F_v/F_m (maximum quantum efficiency of PSII) for A. affine gradually increased as temperature increased from 15 to 23°C and remained high at 27 and 31°C. For A. pacificum, the F_v/F_m was highest at 23°C, and showed a clear decreasing trend beyond this temperature (Fig. 2).

The shape of TPCs and cardinal temperature parameters differed between the species. For instance, the maximum growth rate (μ_max) of A. affine observed at an optimal temperature (T_opt) of 26°C was 0.54 d⁻¹, which is faster than A. pacificum that had an average growth rate of 0.51 d⁻¹ at a T_opt of 25.3°C (Fig. 3). Critical thermal minimum (CT_min) and critical thermal maximum (CT_max) were 15.6 and 35.3°C, respectively, for A. affine, and 7.0 and 30.7°C, respectively, for A. pacificum. Thermal breadth (TB₈₀) also differed between the species, with the values 8.4 and 7.8°C for A. affine and A. pacificum, respectively (Table 3).

Percentage of chain-forming cells varied from 28 to 84% at 15 and 27°C for A. affine, and from 8 to 57% at 30 and 23°C for A. pacificum (Fig. 4). A. pacificum showed a lower proportion of cells with chains than A. affine. There were also differences in chain-length fractionation (CF) between the species (Fig. 5). A. affine showed more and various CF than A. pacificum, with a range of 5–40% of short (2–3 cells), middle (4–8 cells), and long (>8 cells) chains against 2–35% of primarily short CF across the experimental temperature range.

There was no significant interaction between temperature and pCO₂ on the F_v/F_m for both species. However, A. affine was significantly affected by temperature only (p < 0.001), and A. pacificum was significantly affected by both temperature and pCO₂ (p < 0.001) (Fig. 2, Supplementary Table S1). Exposure to elevated pCO₂ lead to only minor increases in the F_v/F_m of A. affine within a range of 0.4 to 0.8, with the most prominent negative effect being observed at the lowest temperature (15°C) (Fig. 2). Compared to A. affine, the values of F_v/F_m of A. pacificum were somewhat lower at temperatures beyond the optimal temperatures under both ambient and elevated pCO₂ conditions.

A. affine showed a significant increase in growth rate as a response to elevated pCO₂ compared to ambient conditions (p < 0.001) (Supplementary Table S2). In the same way, its maximum growth rate (μ_max) increased by 11% (from 0.54 to 0.60 d⁻¹), CT_max increased from 35.3 to 37.1°C, and CT_min decreased from 15.6 to 14.9°C, while T_opt remained unchanged at 26.0°C for both ambient and elevated pCO₂ conditions. Thermal breadth (TB₈₀) increased from 8.4 to 10.2°C only under elevated pCO₂ conditions. For A. pacificum there was no significant effect of elevated pCO₂ on the growth rate (p = 0.489) (Supplementary Table S2), even its maximum growth rate (μ_max) increased from 0.51 d⁻¹ under ambient condition to 0.58 d⁻¹ under elevated pCO₂ conditions. T_opt decreased from 25.3 to 24.6°C, CT_min increased from 7.0 to 9.8°C, and CT_max decreased from 30.7 to 30.4°C under elevated pCO₂ when compared with ambient conditions. Thermal breadth (TB₈₀) also decreased from 7.8 to 7.3°C when such comparison was made. The TB₈₀ indicates that A. affine was not stressed by the experimental treatments, pointing to its broad temperature tolerance resulting in a wide fundamental niche in respect to the future climate.

The chain-forming capacity did not vary significantly between ambient and elevated pCO₂ conditions (p > 0.01), but significant differences between temperatures occurred (p < 0.001) (Supplementary Table S3). The percentage of chain-forming cells was higher in A. affine than in A. pacificum, that found to markedly increase at their optimal growth temperatures (Fig. 4). A. affine exposed to elevated pCO₂ responded by building up chains of various lengths and had higher (but not significantly higher) proportion of chain-forming cells relative to that of ambient pCO₂ cultures. In contrast, A. pacificum had only a few cell chains with a slightly higher proportion in ambient than in elevated pCO₂ conditions, showing that chain formation was not related to CO₂ treatment.

We also tested whether species responded differently to elevated pCO₂ for chain-forming trait, and found that only A. affine exhibited a more substantial increase in the proportion of middle chain-length (4–8 cells chain⁻¹) fractionation, while no change in short chain-length (2–3 cells chain⁻¹) and long chain-length (>8 cells chain⁻¹) fractionation was found under elevated pCO₂ conditions (Fig. 5, Supplementary Table S4). In A. pacificum, elevated pCO₂ did not cause a change in CF (Fig. 5, Supplementary Table S5).

The pH drift and CO₂ exchange in the culture flask were dependent on the growth rate and concentration of cells. In nature, pH and pCO₂ may vary considerably algal blooming, with drastic increases in pH and reductions in CO₂ levels at the bloom peak (Brandenburg et al. 2021). It is unclear whether cells become inhibited by high pH or limited by low CO₂ during the experiment; nevertheless, by allowing pH drift to occur, the experiment represents changes in carbonate chemistry due to biomass build-up due to some treatments. During the semi-continuous culture, flasks were kept at the lower section of the exponential growth phase (Supplementary Fig. S3) and were maintained at, or below, the starting cell concentration of ca. 100 cells mL⁻¹ for A. affine and ca. 200 cells mL⁻¹ for A. pacificum to avoid self-shading, nutrient limitation, and minimize CO₂ exchange. Such experimental cell concentrations are far away from those found in bloom episodes in nature.

The maximum quantum yield (F_v/F_m) has been used as a stress indicator in phototrophic organisms for characterizing the physiological status (Kim et al. 2019). Two Alexandrium tended to increase their F_v/F_m in response to elevated pCO₂, but elevated pCO₂ had a significant effect only on the F_v/F_m of A. pacificum (Supplementary Table S1). The F_v/F_m of A. affine showed neither a significant difference between pCO₂ conditions nor considerable deterioration across temperature gradients during the experiment. It indicates that A. affine is not stressed by the combined treatments, pointing to tolerate broad temperature resulting in a wide fundamental niche in respect to those environmental factors. Various studies support this and generally describe Alexandrium species as eurythermal marine species (Lee et al. 2019a, Lim et al. 2019). Also, an overall increase of the F_v/F_m as considered a marker of its photosynthetic ability reflects its current health status. However, Chl-a fluorescence analysis is not always a reliable method when considering an identifiable F_v/F_m to measure the potential for photosynthetic activity. Note that these measurements represent only a snapshot of bio-optical characteristics of PSII photochemistry that occur within the reaction center complex of PSII (van de Poll et al. 2020).

There was a clear difference between the species in their growth performance response to elevated pCO₂ (Fig. 3). Overall, the growth rates of A. affine showed an apparent increase under elevated pCO₂, whereas A. pacificum showed a slight, insignificant increase (Supplementary Table S2). The optimal temperature (T_opt) of A. pacificum decreased from 25.3°C under ambient to 24.6°C under elevated pCO₂ conditions and was lower than that of A. affine (estimated at 26.0°C under both conditions) (Table 3). The sharp decline in growth rate above the T_opt of A. pacificum indicates its high sensitivity to heat stress once T_opt is surpassed during summer days in the shallow waters of Korea Strait. In contrast, A. affine had a higher critical thermal maximum (CT_max) compared to A. pacificum, which might indicate a more extraordinary competitive ability of A. pacificum to withstand heatwaves prevalent during warm season in the Korea Strait. This could be why the abundance of A. pacificum in the Korea Strait does not increase in summer (Lim et al. 2019, Lee et al. 2020a). Although A. affine is a seasonal species that blooms during the spring-summer months, it is surprising a low abundance of species remains in deep-water of the Korea Strait during the winter months (Lee et al. 2020a), which could have shaped its wide thermal tolerance range (T_tol) and thermal performance breadth (TB₈₀).

In this laboratory experiment, thermal breadth (TB₈₀) of A. affine widened by 21%, while that of A. pacificum narrowed by 6% with elevation of pCO₂ levels (Table 3, Fig. 3). Thus, A. affine can be considered as a thermal generalist species, which can tolerate and grow under a wide range of temperatures, with a wide geographical distribution (Farrell 2009). Since both species share habitat and locality in the Korea Strait, the considerable variation in cardinal temperatures may reflect thermal niche partitioning between species, likely arising from enhanced resource use efficiency and interaction in the temperate pelagic system. Two Alexandrium species increased their maximum growth rate (μ_max) in response to elevated pCO₂ (Table 3). Based on the observed changes in cardinal temperatures, we anticipate the abundance of A. affine to eventually increase in the Korea Strait, given the warming of surface waters and its synergistic effect with increasing pCO₂.

Elevating CO₂ concentration would favor RuBisCO’s carboxylation reaction over oxygenation activity, and further gains in photosynthesis would be granted. The role of CCM is to suppress photorespiration and support adequate carbon fixation by enhancing the chemical conditions in the vicinity of the primary CO₂ assimilating enzyme RuBisCO (Raven et al. 2017). Although we did not measure CCM activity, dinoflagellates with a high CCM activity also revealed a high CCM plasticity (van de Waal et al. 2019). Therefore, Alexandrium also may possess effective and flexible CCMs. Especially A. affine with high growth rate may have down-regulated their CCMs in the high CO₂ condition to reallocate energy towards other cellular processes such as carbon metabolism, cell growth, and chain formation.

Most phytoplankton species maintain a relatively uniform morphology over time, but others actively change their size by forming chains. Chains form when daughter cells do not separate at the end of cell division, generating highly elongated shapes. Cell size and shape of phytoplankton can be mediated by many essential processes, including light capture, nutrient acquisition, predation, and motility (Pahlow et al. 1997, Smayda 2010, Selander et al. 2011, Bergkvist et al. 2012, Bjærke et al. 2015, Lovecchio et al. 2019). The swimming speed of phytoplankton increases with chain length because chain formation increases propulsion force rather than hydrodynamic drag (Fraga et al. 1989). Chain-forming dinoflagellates, such as Alexandrium species, were observed to swim much faster than single cells (Fraga et al. 1989, Selander et al. 2011, Lovecchio et al. 2019) and to have the ability to migrate vertically between surface waters, during the daytime to harvest light, and deep waters, during the night to consume nutrients (Ryan et al. 2010). Chain length plasticity has been suggested to provide adequate refuge to phytoplankton to reduce encounters with grazers and regulate the balance between motility and grazer avoidance (Selander et al. 2011, Bergkvist et al. 2012, Kang et al. 2019a, Jang and Jeong 2020, You et al. 2020). However, our results from laboratory cultures suggest that maximum capacity of chain formation by Alexandrium species is attained in the absence of grazing pressure, which occurs in the absence of chemical cues released by the grazers. Although the chain formation extent of A. affine and A. pacificum showed varied responses with temperature, no significant effects of pCO₂ on the chain formation of both species were observed (Fig. 4, Supplementary Table S3). Moreover, chain length correlates to the growth rate in both species, suggesting that chain length might also depend on growth conditions. In addition, an enhanced response of chain length was observed only in A. affine, showing that elevated pCO₂ significantly enhances the middle CF (4–8 cells chain⁻¹) (Fig. 5). Differential chain formation responses of A. affine vs. A. pacificum indicate species-specific responses in physiological adjustment and carbon allocation strategies for their growth to cope with OA.

In conclusion, the Alexandrium species are cryptic, sharing similar morphological characteristics, and show phenotypic plasticity within traits in response to elevated pCO₂. Two Alexandrium can acclimate their physiology to local environmental conditions within their plastic response range. Although the strains were isolated from the same region, the two species differed strongly in their chain-forming capacity, cardinal temperatures, and response directions under elevated pCO₂.