Emiliania huxleyi BOF 92 was established in 1993 from strain 5/90/25j, which was originally isolated in 1990 from the North Atlantic by John C. Green (Paasche et al., 1996; Paasche, 1999). In the present study, this strain was obtained from Jimei University (Xiamen, China). The E. huxleyi strain was initially able to calcify but lost the capacity for calcification after having been cultured in the laboratory (Nanninga and Tyrrell, 1996), and did not resume the capability to calcify as shown in a recent study (Fu and Gao, 2022). A modified IMR medium (Eppley et al., 1967), with N O 3 − = 24.7 μM and P O 4 3 − = 2.5 μM, was used to culture the E. huxleyi cells. The photon flux was set at 100 μmol photons m^-2 s^-1 (measured by a US-SQS/WB spherical micro quantum sensor, Walz, Germany) with a 12:12 light:dark period. For the short-term experiment, the culture temperature was set at five different levels, 14°C, 17°C, 21°C, 24°C, and 28°C. During the short-term experiment, sterilized seawater with the addition of IMR nutrients was renewed every 3 days to achieve stable carbonate chemistry (Gao, 2021), and the cell concentrations were 1.5-2.0×10⁴ cells ml^-1and and 8.0-15.0×10⁴ cells ml^-1 after and before the medium renewal, respectively. All vessels were sterilized by autoclave to avoid contaminations. Periodic microscopic examination verified that the cultures remained mono-algal, but we cannot guarantee the axenic nature of the cultures. Three independent replicates for each treatment were grown for about ten generations (≥10), and various physiological parameters were collected at the end of this period.

The stock culture of Emiliania huxleyi BOF 92 has been maintained for many years at 17°C, which represents the annual mean temperature of its isolation site. For the long-term warming experiments, 17°C and 21°C were selected as control and seawater warming treatments, respectively. Three independent replicate cultures were used for each treatment. The long-term semi-continuous culture lasted for 656 days from March 6, 2019 to December 21, 2020. The light conditions and nutrients used in the long-term culture were the same as those in the short-term experiment. The sterilized seawater with the addition of IMR nutrients was renewed every 5-10 days, and the cell concentration before dilution ranged 5.0-10×10⁴ cell ml^-1 and was set at ~500 cells ml^-1 after each dilution. The growth was recorded and cells were regularly collected before dilution throughout the experiment except from Day 348 (February 17, 2020) to Day 437 (May 16, 2020). In this period, the cultures were diluted regularly but not sampled for measurements of physiological parameters due to the COVID-19 pandemic.

The specific growth rate and cell size (diameter) were measured using a Z-2™ Coulter particle Count and Size Analyzer (Beckman Coulter, USA), and the particle size range was set as 3-8 μm. The specific growth rate (μ) was calculated as: μ = ( 1 n C 2 - 1 n C 1 ) / ( T 2 - T 1 ) , where C₂ and C₁ were the cell concentrations at time T₂ and T₁, respectively. The number of generations (G) was estimated as: G = ( 1 n C 2 - 1 n C 1 ) / 1 n 2

To determine the cellular contents of chlorophyll a (Chl a), cells were collected on the GF/F membrane (Whatman, 25mm, USA) by suction filtration within a pressure of 0.01MPa to prevent cell damage from excessive pressure. The collected samples were extracted overnight in 10 ml of absolute methanol at 4°C in darkness. After centrifugation (5000g for 10 minutes), the absorption values of the supernatant were measured with a UV-VIS spectrophotometer (DU800, Beckman Coulter, USA) across the wavelength range 400-800nm. The concentrations of Chl a were calculated according to Porra (2002),

where A_x is the absorbance under the x-wavelength, V_E is the volume of the methanol extraction, V_C is the volume of the algae solution, and C is the concentration of algae cells (cells ml^-1).

In short-term experiments, the measured specific growth rate of E. huxleyi cultured at 14°C, 17°C, 21°C, 24°C, and 28°C was fitted to temperature according to Schoolfield et al. (1981) and Padfield et al. (2016):

where μ(T) represents the specific growth rate at temperature T (kelvin - K), E_a is the activation energy (eV) for growth, E_h is the deactivation energy (eV) characterizing high-temperature-induced inactivation above the deactivation temperature T_h (K), k is Boltzmann’s constant with a value of 8.62×10^-5 eV K^-1, and T_c represents the reference temperature (17°C, 290.15K). The optimum growth temperature (T_opt) was calculated from T_c, T_h, E_a, and E_h as (Yi et al., 2020):

In the short-term experiment, net photosynthesis and dark respiration rates of E.huxleyi cells were determined using a Clark-type oxygen electrode (Hansatech, Norfolk, UK) under the growth temperatures and irradiance (100 µmol photons m⁻² s⁻¹). Cells were gently collected on polycarbonate membranes (1.2 μm, Millipore, Ireland) and resuspended with sterilized seawater. The resuspended cells were placed in an oxygen electrode chamber equipped with a magnetic stirrer. Both net photosynthesis rate and dark respiration rate were measured for 10 minutes and calculated from the slope of oxygen concentration vs time. After measurement, the resuspended cells were counted by the Z-2™ Coulter particle Count and Size Analyzer, and the calculated photosynthesis and respiration rates are presented as oxygen concentration change per cell per hour (fmol O₂ cell^-1 h^-1).

Cells were collected onto pre-combusted (450°C for 6h) Whatman GF/F membranes (25 mm) and frozen at -20°C. Before the measurement, the filter membranes were placed in a closed container filled with HCl for 24 hours to completely remove any inorganic carbon and then dried in an oven at 80°C for 24 hours. Subsequently, all the filters were analyzed by a Perkin Elmer Series II CHNS/O Analyzer 2400 (Perkin Elmer Waltham, MA). Production rates (P) of particulate organic nitrogen (PON) and particulate organic carbon (POC) were calculated as P (pg cell^-1 d^-1) = specific growth rate (d^-1) × cellular PON or POC content (pg cell^-1), respectively.

A Multi-color PAM fluorescence monitoring system (Walz, Effeltrich, Germany) was employed to measure the photosynthetic performance of E. huxleyi cells in the short-term experiment. The minimal fluorescence (F_o) for 15 minutes dark-adapted cells was induced by a modulated red measuring light of low irradiance (~5 μmol photons m^-2 s^-1), maximum fluorescence (F_m) was measured during a 0.8 s saturating flash (~10,000 μmol photons m^-2 s^-1), steady-state fluorescence (F) was monitored periodically under a ~100 μmol photons m^-2 s^-1 actinic light, and the 0.8 s saturating flash was applied to induce the maximum fluorescence under actinic light (F_m’). The maximum quantum yield of photosystem II (PSII) (F_v/F_m), the effective quantum yield of PSII (YII), and the non-photochemical quenching (NPQ) were calculated as (F_m-F_o)/F_m, (F_m’-F)/F_m’, and (F_m – F_m’)/F_m’, respectively. For the measurements of rapid light curves, cells were illuminated for 20 s with 10 incremental steps of increasing actinic irradiance (PAR). A saturation flash was applied following each irradiance to generate corresponding F and F_m’ values. Relative electron transport rates (rETR) were calculated as (F_m’-F)/F_m’ × PAR. Photosynthetic parameters α (photosynthetic light harvesting efficiency), β (a photoinhibition constant), and rETR_s (potential rETR maximum without photoinhibition) were obtained by fitting the rETR data to the function (Platt et al., 1980): rETR = rETRs × [1 - e ^{-α×PAR/rETRs}] × e ^{-β×PAR/rETRs}. The minimum saturating light intensity, I_k, was calculated as rETR_max/α, and the maximal rETR, rETR_max, was calculated as rETR s × ( α α + β ) × ( β α + β ) β / α .

After having been grown in their respective temperature conditions for 656 days, corresponding to ~790 generations for 17°C and ~830 generations for 21°C, cells were transferred to the reciprocal temperature to acclimate for another 7 days (~9 generations). Such shift experiments were used to evaluate whether evolutionary changes stemming from the selection environment occurred (Tong et al., 2017 and references therein). The specific growth rates, the cellular POC, PON contents, as well as the ratio of POC/PON were measured in the shift experiment.

Data analyses were performed with SPSS19.0 (SPSS Inc., Chicago, USA) or R language (version 4.1.1). One-way ANOVA with post-hoc Tukey HSD test was used to test the differences among treatments in the short-term experiment. The homogeneity of variance was examined using Levene’s test. Differences were considered to be significant at p< 0.05. For data collected in the long-term experiment, generalized additive models (GAMs) were built (R language and package “mgcv”) to test the effects of warming on the time series of specific growth rate, diameter, Chl a content, and cellular contents of POC and PON (van Rij et al., 2019). The GAMs for the time series of Chl a, POC, and PON contents indicated that these parameters changed linearly with time, so linear mixed effects modes (LMMs) were used instead (R language and package “nlme”; Pinheiro and Bates, 2006). The pairwise comparisons between control and warming treatments on each day were achieved using the R packages “itsadug” and “emmeans” for GAMs and LMMs, respectively. The building, selection, and validation of the statistical models followed the principles and guidelines provided in Pinheiro and Bates (2006); Zuur et al. (2009) and Wood (2017). The tested statistical models and the corresponding values of the Akaike information criterion (AIC) are presented in the Supplementary Material (Table S1-8) , and AIC values were calculated using the R language function “AIC”.

As shown in Figure 1A , the specific growth rate (SGR) of E. huxleyi in short-term cultures displayed a significant difference among the five temperature treatments (One-way ANOVA, p< 0.05), with an increase from 14°C to 23°C and a sharp decrease from 23°C to 28°C (post-hoc Tukey HSD test, p<0.05 for all pairwise comparisons). The cells survived at 28°C but cannot replicate, so the calculated SGR was about 0. Diameters measured by the Z-2™ Coulter ( Figure 1B ) indicated that the size of cultured cells was significantly larger under the highest temperature conditions (One-way ANOVA, p<0.05). Compared with 14°C, the average cell diameter increased by 29.1% at 28°C (post-hoc Tukey HSD test, p<0.05). Since cells cultured at 28°C could not be harvested in sufficient numbers, only 4 temperature treatments, i.e. 14°C, 17°C, 21°C, and 24°C, were analyzed for photosynthetic and respiratory performance. As with SGR, both photosynthetic O₂ evolution and respiratory rates increased gradually with temperature and peaked at 24°C ( Figure 1C ). In these four various temperature treatments (at 14°C, 17°C, 21°C and 24°C), the O₂ evolution rates were 5.1, 8.3, 12.7, and 26.8 fmol O₂ cell^-1 h^-1, and the respiratory rates were 6.3, 7.4, 10.1 and 15.1 fmol O₂ cell^-1 h^-1, respectively. Due to the low-density of E. huxleyi cells cultured in our long-term experiment, insufficient cells were available to measure both photosynthetic O₂ evolution and respiratory rates. Results from the short-term thermal response curve for growth ( Figure 1A ) showed that the activation energy E_a, the deactivation energy E_h, and the deactivation temperature T_h were 0.34 eV, 17.80 eV, and 24.62°C (297.77 K), respectively. The calculated optimal growth temperature for E. huxleyi BOF 92 was around 22.9°C (296.1 K).

With the increasing temperature, cellular Chl a content increased gradually, displaying the maximum value of ~0.25 pg cell^-1 and a significant difference at 24°C (One-way ANOVA, p<0.05, Figure 2A ), while the cellular contents of POC and PON showed no significant change between 17°C and 21°C (post-hoc Tukey HSD test, p=0.372 and 0.418, respectively, Figure 2B ). Compared with 17°C or 21°C, the POC and PON contents of cells grown at 14°C or 24°C were significantly higher by ~ 5% (post-hoc Tukey HSD test, p<0.05, Figure 2B ). The ratios of POC to PON of cells grown under different temperatures remained within the range of 7.5 to 8.2. In terms of POC and PON production rate, values mainly depended on SGR and were positively correlated with increasing temperature ( Figure 2C ).

The chlorophyll fluorescence parameters determined using the Multi-color PAM are shown in Table 1 . The maximum photochemical efficiency (F_v/F_m) and effective photochemical efficiency (Y(II)) of coccolithophores photosystem II showed no significant difference among the five different culture temperature conditions (One-way ANOVA, p = 0.75 for F_v/F_m, p = 0.66 for Y(II)), with an average value of F_v/F_m from 0.64 to 0.66 and an average value of Y(II) from 0.53 to 0.59. The value of rETR_max at 17°C was significantly higher than that of cells grown under the other temperature treatments, which were 20.6% (post-hoc Tukey HSD test, p<0.05), 25.3% (post-hoc Tukey HSD test, p<0.05), and 59.8% (post-hoc Tukey HSD test, p<0.01) higher than at 14°C, 21°C, and 24°C, respectively. The light utilization efficiency, i.e. values of α, exhibited the highest value under 24°C, which was significantly higher than that of cultures grown at 14°C, 17°C and 21°C (post-hoc Tukey HSD test, p<0.05 for 14°C and 17°C, p<0.01 for 21°C). In terms of β, the degree of photoinhibition, no significant difference was observed between cells grown at 14°C and 17°C (post-hoc Tukey HSD test, p=0.797), but its value was significantly higher than that of cells from the two elevated temperatures (post-hoc Tukey HSD test, p<0.05 for both 21°C and 24°C). Similarly, variation of the minimum saturation light intensity (I_k) also showed the highest value at 17°C, which was significantly higher than those of two elevated temperatures (post-hoc Tukey HSD test, p<0.05 for both 21°C and 24°C).

During the long-term culture (656 days) at 17°C and 21°C, cells had been growing for ~790 and ~830 generations, respectively. The average SGRs under 17°C was 0.84 d^-1 which was slightly lower than that (0.88 d^-1) at 21°C; the elevated seawater temperature thus increased the growth of E. huxleyi by 5.28% ( Figure 3A , Supplementary Table S2 ). The GAM analysis demonstrated that throughout the whole process, the elevated seawater temperature significantly improved SGR in the first 34 days and last 162 days, and no significant difference in SGR was detected between the control and warming treatments from Day 35 to Day 493 ( Figure 3B ). The average cell diameter in control and warming treatments were 4.59 μm and 4.64 μm, respectively ( Figure 3C ; Supplementary Table S4 ). Although the GAM analysis found that the elevated temperature did enlarge the cell size (Wald tests, p<0.001, Table S4 ), the pairwise comparisons between control and warming treatments at each day showed no significant difference on most days ( Figure 3D ). Similar results were also observed in the cellular Chl a content (Wald tests, p=0.0205 for temperature, Figure 4 , Table S6 ).

During the long-term culture at 17°C and 21°C, no significant differences ( Supplementary Table S8 ) were observed in both cellular contents of POC ( Figure 5A ) and PON ( Figure 5B ), as well as in the ratio of POC to PON ( Figure 5C ). However, there was a trend that the cellular contents of POC and PON increased gradually with the time. This was supported by the analysis based on the linear mixed-effects model, which indicated that both the cellular contents of POC and PON were significantly affected by the time (Wald tests, p<0.001, Supplementary Table S8 ).

After the 17°C-selection cells were acclimated to 21°C for 7 days, their SGR significantly increased by 22% compared to the cells continuously grown at 17°C (post-hoc Tukey HSD test, p<0.05). In contrast, SGR was observed to decrease by 14%, as cells were transferred from 21°C to 17°C (post-hoc Tukey HSD test, p<0.05, Figure 6A ). Changes of temperatures showed no significant effects on cellular POC contents, PON contents, as well as the ratio of POC to PON ( Figure 6B for POC contents, 6C for PON contents, 6D for POC : PON). These results suggested that no adaptative changes occurred during the long-term seawater warming treatments.

Acceleration of metabolic activity at elevated temperatures is a common response among phytoplankton, which could further prompt their growth (Lund, 1949; Feng et al., 2008; Schaum et al., 2017). However, the temperature rising from 21°C to 24°C further increased photosynthesis and respiration but reduced specific growth rate in the present work. The temperature elevation from 21°C to 24°C stimulated the ability to obtain carbon but also increased the cellular demand for carbon due to increased respiration, leading to reduced rate of growth. Since cellular POC production rates play an important role in global carbon cycles (Poulton et al., 2007), warming from 17°C to 21°C over 800 generations promoted the role of E. huxleyi BOF 92 in primary productivity.

While the annual mean temperature of its isolation site was about 17°C, the optimal growth temperature and the upper thermal limit of E. huxleyi BOF 92 were found in this work to be ~23°C and ~28°C, respectively. This supports the general understanding that the optimal temperatures of phytoplankton strains from high latitude oceans are considerably higher than their mean annual temperatures (Thomas et al., 2012; Brandenburg et al., 2019). Therefore, Emiliania huxleyi living in high latitude oceans might benefit from ocean warming irrespective of other environmental changes that would synchronously occur with it. That the cell size increased at thermal stress (28°C) in this study can be attributed to retardation of nutrient utilization by the cells, leading to reduced cell division (Smith and Kalff, 1982; Savage et al., 2004; Litchman et al., 2007; Marañón, 2015).

In the long-term selection experiment, varied response of phytoplankton to warming has been recognized to be related to evolutionary adaptation. For example, in a green alga Chlorella vulgaris, population growth was initially (~10 generations) limited at a high temperature because its respiration was more sensitive to temperature than photosynthesis, leading to less carbon utilization for growth (Padfield et al., 2016). However, after acclimating for 100 generations under the elevated temperature, the down-regulation of respiration would drive phytoplankton to overcome the metabolic constraint and allocate more fixed carbon to growth (Padfield et al., 2016). In Chlamydomonas reinhardtii, the increase in photosynthetic rate, but not a trade-off between respiration and growth, was suggested to be the main adaptive response to warming (Schaum et al., 2017). For E. huxleyi BOF 92 examined in this work, warming at 21°C increased its cellular photosynthesis and respiration increased by 53% and 37%, respectively, compared to 17°C ( Figure 1C ). Such stimulations to both photosynthesis and respiration may be the main reason for the enhancement of growth observed in the early stage of the culture period, but was inconsistent with the lack of effect in the middle stage of the culture period. On the other hand, the growth enhancement after ~600 generations and the shift experiment indicated that selection cultures at 21°C did not bring about any adaptation due to the warming, though both cell quotas of POC and PON gradually increased with duration time ( Figure 5 ). This implies that the non-stressful warming only resulted in plastic acclimatory changes over about 800 generations. In a recently published work on the diatom Skeletonema dohrnii (Cheng et al., 2022), warming (24°C vs. 20°C; 28°C vs. 20°C) significantly decreased the cellular POC and PON contents after 300 generations of selection, but the changes were completely recovered after 700 generations of selection. This variation of POC and PON contents was accompanied by significant proteome plasticity (Cheng et al., 2022). The mechanism behind the inconsistent warming effects on growth in our long-term experiment is unclear. Since we did not obtain proteomic or transcriptomic data, future work is needed to link the phenotypic responses to the molecular responses, which may help interpret similar phenomena (Schaum et al., 2018; Jin et al., 2022).

The result that warming-selected cells over about 800 generations shifted back to their ancestral traits ( Figure 6 ) suggest that E. huxleyi cells maintained their phenotypic plasticity. While warming could increase the fitness of phytoplankton (Schlüter et al., 2014; Listmann et al., 2016), evolutionary changes can be dependent on thermal stress levels or latitudes. In warm seas, some diatoms adapted to warming with trade-offs in photosynthetic efficiency and growth rate (Jin and Agustí, 2018). By contrast, the warming treatment (21 °C) for E. huxleyi BOF 92 from high latitude in our study was below its optimal growth temperature (~23°C), resulting in no adaptation over 800 generations. This implies that mild warming treatment that exerts no selective pressure can hardly give rise to evolutionary change (Xu et al., 2023).

In terms of the long-term adaptation of phytoplankton to environmental changes, its evolutionary process seems to be strain-specific (Supplementary material, Table S9 ). In coccolithophorids, the existence of coccoliths, i.e. the calcification process, could affect its adaptability to environmental changes. For example, in non-calcified Gephyrocapsa oceanica NIES-1318, elevated pCO₂ initially prompted its growth and carbon, nitrogen assimilation rates from generations ~670 to ~1550 (Jin et al., 2013). However, in the same strain with coccoliths, high CO₂-selected populations exhibited reduced growth and calcification rates and enhanced POC and PON production (Tong et al., 2018). E. huxleyi BOF 92 cultured in the present study used to be a calcified strain (Paasche, 1999), but lost its calcification capability during the laboratory cultures which lasted for about 30 years. Low light, high nutrients and absence of solar ultraviolet radiations might be responsible for the loss of ability to calcify (Paasche, 2002 and references therein; Guan and Gao, 2010). Nevertheless, the naked cells exhibit the same other morphological features as those of calcified cells (Paasche, 2002), and the measured growth rate and cell size were also comparable with previous reports (Paasche, 1999). Our data in the present study may partially imply the physiological responses of E. huxleyi BOF 92 to future warming, but may not be representative of all E. huxleyi generally, bearing in mind that this strain might differ genetically from its ancestral form.

Increasing temperature is a key factor that directly determines phytoplankton distribution, production, and metabolic function (Morán et al., 2010; Thomas et al., 2012; Toseland et al., 2013). Its interaction with other environmental drivers, e.g., light irradiance, has also been reported to affect the acclimation of phytoplankton to temperature change (Feng et al., 2008; Edwards et al., 2016). The growth irradiance used in this study (100 μmol photons m^-2 s^-1) is lower than the saturating irradiancies for both growth and photosynthesis (> 200 μmol photons m^-2 s^-1) of E. huxleyi BOF 92 (Nanninga and Tyrrell, 1996; Paasche, 2002). Higher irradiance can stimulate the growth and reproduction of E. huxleyi BOF 92, resulting in shorter generation times. Theory predicts that populations with shorter generation times evolve faster because of the accumulation of more DNA replication errors per unit time (Weller and Wu, 2015). Therefore, growing E. huxleyi BOF 92 population under higher and more favorable irradiance might promote its adaptation to warming. In addition, the increasing seawater temperature induced by additional atmospheric CO₂ accumulation is accompanied by other environmental changes, such as ocean acidification. Studies exploring the adaptative responses of phytoplankton to the combination of ocean warming and acidification suggest that ocean warming might exert a greater impact than ocean acidification and the latter might serve as a modulating factor (Schlüter et al., 2014; Zhong et al., 2021; Jin et al., 2022; Xu et al., 2023; Supplementary Material Table S9 ). For phytoplankton living in the natural dynamic environment, either an acclimation or an adaptation strategy is essential in determining “winners” and “losers” under the influences of climate changes (Somero, 2010). Different timescales of exposure to a perturbation can result in similar or different physiological responses of phytoplankton species from different regions; when mild warming treatment is applied to phytoplankton strains isolated from cold seas, as in the present work, phenotypic plasticity prevails.