Adaptive phenotypic evolution of Skeletonema costatum to ocean acidification and warming with trade-offs from a multi-year outdoor experiment

Abstract

Human-induced climate change is increasing variability in marine environments, significantly affecting marine organisms and ecosystems. While marine diatoms can adapt to ocean acidification and warming in stable laboratory settings, their responses to long-term environmental changes under natural variability remain unclear. To investigate this, we cultivated Skeletonema costatum in outdoor semi-continuous cultures for over 3 years, exposing them to fluctuating natural light and temperature that tracked the in situ sea surface temperatures. We simulated current and future ocean conditions through four treatments: ambient CO₂ and temperature (LTLC), elevated CO₂ (LTHC), elevated temperature (+4°C, HTLC) and combined increases (HTHC). After 1396 days, we assessed populations in two assay environments (20°C, 400 ppm CO₂ and 24°C, 1000 ppm CO₂) for adaptations in growth rate, pigment composition and photosynthesis. The HTLC-selected group showed the highest growth rates in the HTHC assay environment, while the LTLC-selected group grew fastest in the LTLC assay environment, indicating adaptive evolution. Furthermore, populations selected under elevated conditions exhibited lower fitness in LTLC environments, highlighting a trade-off and underscoring the complexity of evolutionary adaptation in marine diatoms. Understanding these mechanisms is crucial for predicting phytoplankton dynamics and their role in marine ecosystems, especially in response to climate change.

1. Introduction

The variability of the marine environment shapes the ecological distribution and evolution of living organisms. Since the industrial revolution, increased human activities, such as fossil fuel combustion and tropical deforestation, have led to a rise in atmospheric CO₂ levels from 280 ppm to around 420 ppm [1]. The ocean absorbs about 30% of anthropogenic CO₂ emissions, causing a reduction in seawater pH [2,3], which is the so-called ocean acidification. At the same time, oceans have absorbed over 93% of the increased energy in Earth’s climate system due to global warming, leading to a steady rise in sea surface temperatures [4]. According to the RCP 8.5 model, CO₂ levels are projected to surpass 900 ppm by the end of this century, with global average sea surface temperatures expected to increase by approximately 1 to 3°C by 2100 [5]. These two global environmental changes, ocean acidification and ocean warming, are key drivers of population dynamics of marine phytoplankton and, thus of primary production in the oceans [6].

Marine diatoms can contribute around 20% of global primary production and play a key role in the biogeochemical cycles of biogenic elements, such as nitrogen, phosphrus and silicon [7,8]. The dynamics of diatom biomass and evolution are driven by environmental variations through bottom-up control [9,10]. Changes in seawater CO₂ concentrations are known to affect various physiological aspects of diatoms, including growth rate, cellular metabolism, enzyme activity and lipid accumulation [11–13]. Long-term ocean acidification is expected to affect the growth and carbon fixation of phytoplankton, with both direct and indirect impacts on the marine ecosystem [14]. At the same time, marine diatoms have a strong capacity for rapid evolutionary adaptation to elevated CO₂ levels, relying on their short turnover times, high population densities and substantial standing genetic variation [15,16]. Therefore, the long-term responses of marine diatoms to ocean acidification may differ compared with short-term acclimation [17]. In fact, it has been demonstrated that the physiological responses of phytoplankton to ocean acidification vary across different timescales, with numerous experimental studies highlighting the adaptability of these microorganisms to prolonged exposure to ocean acidification [18–20].

Temperature plays a critical role in shaping the physiological processes of diatoms, influencing their growth, metabolism and ecological dynamics [18,21,22]. In the short term, temperature fluctuations can directly affect photosynthesis rates and respiration, often resulting in immediate changes in growth rate and morphology [23]. Diatoms exhibit a range of thermal tolerances, with certain species thriving within specific temperature windows [24]. Over the long term, repeated or sustained temperature stress could trigger evolutionary adaptations in these microalgae, including altered thermal tolerance, metabolic efficiency changes and community composition shifts [25,26].

From the perspective of evolutionary biology, adaptive evolution refers to the gradual development of traits better suited to a population’s current environmental conditions over long-term environmental selection through phenotypic adaptation and genetic variation [10]. Experimental evolution in the laboratory is commonly applied to investigate the potential adaptive evolution of microalgae under a controlled environment [27,28]. In such experiments, the algal populations are exposed to specific environmental pressures (such as high CO₂ concentrations, high temperature or nutrient limitations), allowing them to adapt over multiple generations [29]. After long-term selection, the phenotypic changes of populations selected under specific environments are tested under assay conditions by establishing reaction norms or reciprocal transplant experiments [17,27,30]. Additionally, whole-genome sequencing and transcriptomic analysis can examine from a molecular perspective whether specific genes and genetic variations associated with adaptive evolution have emerged [26,31]. At the same time, trait trade-offs are frequently detected after long-term selections owing to constraints in energy and resource allocations [31,32]. For example, high-temperature-evolved populations usually exhibit elevated performance under higher temperatures while underperforming when transferred back to lower temperatures, relative to non-selected populations [33]. Both the adaptive evolution and trade-offs revealed from the experimental evolution provide insights into the mechanistic understanding of the evolutionary adaptation of organisms to changing environmental conditions, and possible ecological consequences for marine ecosystems. However, most of the existing experimental evolution studies were carried out under stable environmental conditions in laboratory settings, and often did not encapsulate the variability of the natural environment. For example, temperature fluctuation has been shown to significantly influence the adaptation of thermal tolerance in the diatom Thalassiosira pseudonana [23]. Therefore, there is a need for studies conducted under more realistic environments, such as outdoor experiments, which can introduce fluctuations of the environmental drivers.

Skeletonema costatum, a chain-forming diatom and a major species in the Yellow Sea, plays a significant role in carbon sequestration through its unique carbon-concentrating mechanisms (CCMs), which support high carbon utilization efficiency under fluctuating environments [34]. Despite its ecological importance, this species is less studied than other model organisms. Most existing research on S. costatum has focused on short-term experiments [13,35,36]. However, its evolutionary adaptability to long-term ocean acidification and warming remains poorly understood. To address this, we conducted a long-term outdoor experiment (>3 years) simulating future scenarios projected for the late twenty-first century [37], incorporating natural sunlight and seasonal temperature fluctuations. Populations were maintained under four conditions: current ocean (LTLC, 400 ppm CO₂, baseline environmental temperature variation), ocean acidification (LTHC, 1000 ppm CO₂, baseline environmental temperature variation), ocean warming (HTLC, 400 ppm CO₂, +4℃ to baseline temperature) and future ocean (HTHC, 1000 ppm CO₂, +4℃ to baseline temperature). Following selection, short-term assays were performed under LTLC and HTHC conditions to evaluate potential adaptive evolution and trade-offs by comparing phenotypic traits such as growth, photosynthesis, respiration and chlorophyll content (figure 1).

Figure 1.

Experiment design. Four selected environmental conditions were established: ambient CO₂ and temperature (LTLC), elevated CO₂ (LTHC), elevated temperature (HTLC) and elevated CO₂ combined with elevated temperature (HTHC). After 1396 days of growth under the selection conditions, the cells were transferred to two assay regimes: ambient CO₂ and temperature (LTLC) and elevated CO₂ and temperature (HTHC), for 8 days. Dilutions were performed every 5 to 21 days for the selection experiment and every 2 days for the assay environment. CO₂ levels are shown in ppm.

2. Material and methods

(a). Strain and experimental setup of the selection experiment

Diatom Skeletonema costatum was isolated from coastal waters near Gaogong Island (119.5° E, 34.7° N) in Jiangsu Province on 15 December 2021. The stocks were precultured at 20°C under a light intensity of 120 µmol photons m⁻² s⁻¹, with a 12 : 12 h light : dark cycle and maintained in f/2 + Si culture medium [38]. During the experiment, sub-populations were maintained semi-continuously in water baths on a rooftop for 1396 days under controlled conditions. Four selected environmental conditions were established (figure 1): ambient CO₂ and temperature (LTLC), elevated CO₂ (LTHC), elevated temperature (HTLC), and elevated CO₂ and temperature (HTHC). The ambient CO₂ level was approximately 400 ppm (outdoor air), while the elevated level was set at 1000 ppm, following the RCP8.5 scenario from the IPCC 2022 report [39]. To simulate natural sea temperature variations, the ambient temperature (LT) was adjusted daily (except during periods of extreme heat or cold in summer and winter, when populations could not sustain significant growth according to our pre-test), following sea surface temperature at 12.00 from the central Yellow Sea (123.6° E, 35.6° N) obtained from the China Oceanic Information Network (https://www.oceanguide.org.cn/IndexHome). The elevated temperature (HT) was set 4°C higher than ambient to simulate late twenty-first-century ocean warming [37]. The natural temperature variation ranged from 4.4 to 29.4°C for the ambient temperature (LT) and from 8.4 to 33.4°C for the elevated scenario (HT), and temperatures were recorded (electronic supplementary material, figure S1), with critical minimum (7°C) and maximum (30°C) temperatures marked to avoid the extinction of the populations during the selection experiment.

Algal strains were cultivated in quartz tubes, centrally positioned in a rooftop tank to ensure unobstructed sunlight exposure. The culture medium consisted of sterilized natural seawater (autoclaved at 105°C for 15 min) collected from the Yellow Sea and enriched following the f/2 + Si culture medium formulation [38], with major nutrient concentrations of 882 μM N, 36.2 μM P and 102 μM Si. CO₂ concentrations were maintained by bubbling filtered gas (0.22 μm, Millipore Express) into the culture medium overnight before use. The elevated CO₂ concentration (1000 ppm) was achieved by mixing outdoor air and pure CO₂ gas using a CO₂ enricher (HP 1000G-D, Wuhan Ruihua, China), while the ambient CO₂ level (approx. 400 ppm) was maintained by directly bubbling outdoor air. To ensure equilibration, all media were bubbled with target CO₂ gas overnight. Temperature was controlled using a water bath system in a tank, regulated by a circulating thermostat (HYH-2DR-B, Zhejiang SenSen, China) paired with a submersible pump (YQB-7500, Zhejiang SenSen, China) to ensure consistent water circulation. To simulate light conditions in the mixing layer of the central Yellow Sea, cultures were covered with a grey neutral mesh (55% transmittance). The light intensities of photosynthetic active radiation (PAR), ultraviolet A (UVA) and ultraviolet B (UVB) were recorded and calculated based on transmittance during the outdoor experiment (electronic supplementary material, figure S2). The daily averaged intensities ranged from 2.19 to 207.13 W m⁻² for PAR, 0.07 to 22.34 W m⁻² for UVA, and 1.56 to 104.98 W m⁻² for UVB*100, excluding extreme days (54, 55 and 57, when there were snowfalls), where values of 0 were recorded for each light type. Cultures were maintained semi-continuously with four replicates and were diluted every 5 to 21 days depending on the growth rate. The initial cell concentration for the starting cycle was 500 cells ml⁻¹. Before each dilution, the final cell concentration in each cycle was maintained below 5 × 10⁴ cells ml⁻¹ to ensure carbonate system stability across all treatments. Based on our experience, this concentration range does not cause significant changes in the pH of the culture medium. During periods when the populations exhibited faster growth rates and could reach final concentrations above 5 × 10⁴ cells ml⁻¹, we estimated the initial cell concentration required to achieve a final cell concentration of 5 × 10⁴ cells ml⁻¹ after 7 days of culture. This estimation was based on the growth rate observed in the previous cycle, and the predicted initial concentration was used as the target for dilution. Although the pH values were not recorded in detail here, they were regularly checked to ensure the stability of the carbonate system during the experiment, with pH values of 8.15 ± 0.1 for the LC treatments and 7.78 ± 0.1 for the HC treatments. Algal cultures were gently mixed three times per day to keep cells in suspension. The selection experiment began on 25 December 2020 and concluded on 21 October 2024.

(b). Experimental setup of the assay experiment

After 1396 days of growth under their respective CO₂ and temperature conditions (corresponding to 1172 generations in the LTLC treatment, 1204 generations in the LTHC treatment, 1303 generations in the HTLC treatment and 1457 generations in the HTHC treatment) the first three of the four replicates from each treatment (owing to resource constraints) were transferred to two assay regimes: a stable temperature of 20°C with a CO₂ level of 400 ppm (LTLC) and a stable temperature of 24°C with a CO₂ level of 1000 ppm (HTHC), both with the same 12 : 12 h light–dark cycle (with a light intensity of 200 µmol photons m⁻² s⁻¹, ca 43.5 W m⁻², to allow comparison with the unit used for outdoor light intensity), in their respective incubators (GXZ, Ningbo Jiangnan); see figure 1. Cultures were maintained under the two assay environments for 8 days (more than 10 generations each) before measurements were made. Dilutions were performed every 2 days during the assay experiment, with an initial concentration of 5000 cells ml⁻¹ and final concentration maintained below 5 × 10⁴ cells ml⁻¹ to ensure carbonate system stability. The growth rate and other traits were assessed during the final dilution cycle. Cultures were shaken three times per day to ensure the suspension of algal cells.

(c). Determination of growth rate and generation

Sub-samples were collected during mid-day on dilution days and transferred to a 0.1 ml phytoplankton counting chamber (DSJ-01, Xiamen Dengxun) for cell counting under a microscope (DM500, Leica). The relative growth rate was calculated using the formula:

$${\displaystyle \mu =\frac{\text{ln}{N}_1-\text{ln}{N}_0}{t_1-t_0},}$$

where N₁ and N₀ represent the cell concentration on the final day (day t₁) before dilution and the initial cell concentration after dilution in the previous culture period (day t₀), respectively.

The number of generations produced during each dilution cycle in the selection experiment was calculated using the formula:

$${\displaystyle \text{no. generations}=\frac{\mu }{\text{ln }2}\times t,}$$

where μ represents the relative growth rate during one dilution cycle, t is the time duration of the cycle (in days), and ln 2 represents the predicted growth rate for one generation (when the cell number doubles). The number of accumulated generations was thus calculated by summing the generations from all 172 dilution cycles across the experiment.

(d). Photosynthesis and dark respiration

The net photosynthesis and dark respiration rates were determined using a Clark-type oxygen electrode (Oxygraph+, Hansatech). At the end of the transplant experiment, cells were harvested during the middle of the light period. They were concentrated by gentle filtration onto a polycarbonate membrane (0.8 μm pore size, Millipore) and resuspended in fresh medium. Specifically, cells from 100 ml of culture medium were concentrated to 2 ml, with final cell concentrations ranging from 1 × 10⁶ to 2 × 10⁶ cells ml⁻¹. Net oxygen evolution was assessed by measuring oxygen production under light conditions (200 µmol photons m⁻² s⁻¹, consistent with the growth condition). Dark respiration was determined by measuring the decrease in oxygen concentration in the dark over a 10 min period. The rates of net oxygen evolution and dark respiration at each assay temperature were normalized to cell abundance in each sample aliquot, expressed as pmol O₂ cell⁻¹ h⁻¹.

In this study, carbon use efficiency (CUE) was calculated using gross photosynthesis rate (GP) and dark respiration rate (R), as described by the following formula [40]:

$${\displaystyle \mathrm{C}\mathrm{U}\mathrm{E}=1–R/\mathrm{G}\mathrm{P}.}$$

(e). Pigment determination

At the end of the transplant experiment, 50 ml of cells were collected by filtering through a GF/F filter (25 mm, Whatman, UK) using a vacuum filtration pump, ensuring that the pressure remained below 0.02 MPa throughout the filtration process. Filters were then placed in 5 ml of methanol and incubated in darkness at 4°C for 24 h. After incubation, the samples were centrifuged at 4°C and 5000g for 5 min using a centrifuge (Biofuge Primo R, Thermo, Germany). The absorbance of the supernatants was then measured at wavelengths of 480, 510, 632, 665 and 750 nm using a spectrophotometer. The Car/Chl a ratio was calculated based on the equations below and then converted according to the dilution factor of 10:

$${\displaystyle \mathrm{C}\mathrm{h}\mathrm{l}a(\text{μ}{\mathrm{g}\mathrm{m}\mathrm{l}}^{-1})=13.2654(A_{665}–A_{750})-2.6839(A_{632}–A_{750}),}$$

$${\displaystyle \mathrm{C}\mathrm{a}\mathrm{r}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{o}\mathrm{i}\mathrm{d}(\text{μ}{\mathrm{g}\mathrm{m}\mathrm{l}}^{-1})=7.6((A_{480}-A_{750})-1.49(A_{510}–A_{750})).}$$

(f). Photochemical efficiency

To assess the maximum photochemical efficiency of photosystem II (F_v/F_m) and the rapid light response curves (RLCs) of the algal cells after the assay regimes, the AquaPen-C (AP-C100, PSI, Czech Republic) was used. Before measuring F_v/F_m, the algal suspension was concentrated to an appropriate volume and dark-adapted for 15 min. For the rapid light response curve, the photosynthetically active radiation intensities were set at 10, 20, 50, 100, 300, 500 and 1000 μmol photons m⁻² s⁻¹. The relative electron transport rate (rETR) was calculated using the following formula:

$${\displaystyle {\mathrm{r}}_{\mathrm{E}\mathrm{T}\mathrm{R}}=\mathrm{P}\mathrm{A}\mathrm{R}\times {\mathit{Y}}_{\mathrm{I}\mathrm{I}}\times 0.5,}$$

where PAR represents the light intensity, and Y_II denotes the effective quantum yield of photosystem II. The following equation was used for curve fitting [41]:

$${\displaystyle P=P_{\mathrm{m}}\times \mathrm{t}\mathrm{a}\mathrm{n}\mathrm{h}(\alpha \times \mathrm{P}\mathrm{A}\mathrm{R}/P_{\mathrm{m}}),}$$

where α represents the efficiency of the photosynthetic apparatus in utilizing light energy, P represents the rETR and P_m represents the maximum rETR (rETR_max). The light saturation point (I_k) is calculated using the following formula:

$${\displaystyle {\mathrm{I}}_{\mathrm{k}}={\mathrm{r}\mathrm{E}\mathrm{T}\mathrm{R}}_{\mathrm{m}\mathrm{a}\mathrm{x}}/\alpha .}$$

(g). Statistical analysis

Generalized additive mixed effects models (GAMMs) were used to test the effects of elevated CO₂ (HC) and temperature (HT) on the growth rate of S. costatum after long-term selection experiment. Model fitting and selection started with the full model: ${\displaystyle f({\mathrm{C}\mathrm{O}}_{2\ast }\text{temperature}+s(\text{time}),\text{random}=\text{replicate})}$, and compared with simplified models. The full model has the lowest corrected Akaike information criterion (AICc) value and was used for computing statistical results (electronic supplementary material, tables S1 and S2). For the assay experiment, one-way ANOVA followed by post hoc Tukey test was used to test the differences in traits across the selected populations in responses to each assay condition (electronic supplementary material, table S3 and §S2.2). Students’ t‐test was used to test the differences in traits between two assay responses for populations from each selected treatment (electronic supplementary material, §S2.3). Significance level was set at p < 0.05. Data visualization and statistical analysis were conducted using R (v. 4.4.3) and Origin (v. 2024b).

3. Results

(a). Growth dynamics during the long-term selection experiment

The growth rate of S. costatum monitored during the experimental period varied over time, and was influenced by fluctuations in temperature and solar irradiance. Results from the GAMM indicated that both CO₂ and temperature levels had significant impacts on the growth rate of S. costatum, with significant interactions between these two factors (p < 0.05, electronic supplementary material, tables S1 and S2). The growth rate of S. costatum fluctuated in accordance with seasonal temperature variations (figure 2a), and distinct patterns were observed between the LT- and HT-selected populations. Generally, the LTHC group exhibited higher growth rates than the LTLC group throughout most of the cultivation period, except for specific periods such as days 6−114, 327−399, 804−818 and 1256−1318. The HT groups tended to show higher growth rates during autumn and winter when ambient temperatures were lower (e.g. days 0−200), but sometimes displayed reduced growth during summer (days 200−280), when ambient temperatures were already high. Specifically, the highest growth rate was observed in the HTHT treatment on day 649 in autumn (2.12 ± 0.01 day⁻¹), while the lowest was −0.11 ± 0.04 day⁻¹ on day 580 in summer. In the LTLC treatment, growth ranged from a maximum of 1.82 ± 0.01 day⁻¹ (day 649, autumn) to a minimum of −0.19 ± 0.02 day⁻¹ (day 811, winter). The LTHC group showed a similar seasonal trend, with growth rates ranging from 1.87 ± 0.07 day⁻¹ (day 635, autumn) to −0.12 ± 0.07 day⁻¹ (day 1165, winter). For the HTLC treatment, the maximum and minimum growth rates were 2.00 ± 0.01 and −0.18 ± 0.01 day⁻¹, respectively, observed on day 174 (spring) and day 818 (winter). The HTHC-selected populations underwent the most generations (1456) during the selection experiment, followed by the HTLC-selected populations (1300). Populations selected under LTLC and LTHC conditions underwent a similar number of generations (1182 and 1216, respectively), both lower than those under the other two conditions (figure 2b).

Figure 2.

The growth dynamics of Skeletonema costatum in the long-term selection experiment. (a) Growth rates as a function of time under different selection environments: ambient CO₂ and temperature (black), elevated CO₂ (red), elevated temperature (blue) and elevated CO₂ and temperature (green) in the long-term selection experiment. The data points are means ± s.d. of quadruplicate cultures (n = 4). The pale-coloured points represent ambient temperature during the selection experiment (b): the light red represents elevated temperature and light blue is for the natural variation.

(b). Growth rates in the short-term assay experiment

The growth rate of S. costatum differed significantly among selection treatments the LTLC assay environment (one-way ANOVA, p = 0.031, figure 3), with the LTLC‐selected group exhibiting the highest rate (1.40 ± 0.01 day^⁻¹) and the HTHC‐selected group the lowest (1.19 ± 0.09 day^⁻¹), a 14.7% difference, and relative fitness showed the same ranking (figure 3). Under the HTHC assay environment, growth rates again varied significantly among selection treatments (one-way ANOVA, p = 0.011, figure 3), with the HTLC‐selected group performing best (1.62 ± 0.03 day⁻¹; Tukey test, p < 0.05) and the LTLC‐selected group lowest (1.38 ± 0.04 day^−¹; Tukey test, p < 0.05), representing a 14.7% increase (figure 3). Across assay environments, populations selected under LTHC, HTLC and HTHC conditions all exhibited significant differences in both growth rate when measured under LTLC versus HTHC assays (Student’s t‐test, p < 0.05, figure 4b).

Figure 3.

The growth rate of Skeletonema costatum in the assay experiment. The x-axis represents the selection environments, while the different colours of the boxes represent the two assay environments. The right y-axis shows the relative fitness calculated from growth rate. Boxplots represent the distribution of values from triplicate cultures (n = 3). The bold horizontal line within each box indicates the median; the lower and upper edges of the box respectively represent the 25th and 75th percentiles (interquartile range, IQR); whiskers extend to 1.5 times the interquartile range; data points beyond the whiskers are plotted as individual outliers. Lowercase letters indicate significant differences between groups under the LTLC assay environment, and uppercase letters indicate differences under the HTHC assay environment (Tukey’s test, p < 0.05). Asterisks below lines denote significant differences between assay environments for populations selected under the corresponding environments (Student’s t‐test: *p < 0.05, **p < 0.01).

Figure 4.

The dark respiration (a), gross photosynthesis rate (b) and carbon use effiiency (CUE) (c) of Skeletonema costatum in different CO₂ and temperature regimes in the short-term assay experiment. The x-axis represents the selection environment. Boxplot elements and statistical annotations are described in figure 3.

(c). Photosynthesis and respiration in the short-term assay experiment

After acclimation to the LTLC environment for 8 days, there were no significant differences in dark respiration rates (p = 0.099, figure 4a) and net photosynthetic rates (p = 0.157) among different selected groups. However, after acclimation to the HTHC environment, significant differences were observed in dark respiration rates (p = 0.017, figure 4a), GP (p = 0.037, figure 4b) and CUE (p = 0.04, figure 4c). Specifically, the dark respiration rate of the LTLC-selected group was the highest (0.11 ± 0.00 pmol O₂ cell⁻¹ h⁻¹, Tukey test, p < 0.05), which was 41.8% higher than that of the HTHC-selected group (0.07 ± 0.02 pmol O₂ cell⁻¹ h⁻¹, Tukey test, p < 0.05). Additionally, under HTHC assay conditions, the dark respiration rate of the HTHC-selected group (0.05 ± 0.01 pmol O₂ cell⁻¹ h⁻¹, Tukey test, p < 0.05) was significantly lower than that of the LTLC group. In terms of GP, no significant differences were observed among groups acclimated to the LTLC environment (p = 0.240, figure 4b). However, the HTLC-selected group exhibited the highest GP values under HTHC (0.58 ± 0.03 pmol O₂ cell⁻¹ h⁻¹, Tukey test, p < 0.05) assay conditions.

Notably, under the HTHC assay condition, CUE significantly differed among the selected populations (p = 0.004, figure 4c). Specifically, the HTHC-selected populations exhibited the highest values (0.89 ± 0.02, Tukey test, p < 0.05), followed by the HTLC- and LTHC-selected groups, while the LTLC-selected populations having the lowest CUE (0.79 ± 0.02, Tukey test, p < 0.05).

(d). Chl a and carotenoid contents in the short-term assay experiment

After acclimation to the LTLC assay environment for 8 days, there were no significant differences in chlorophyll a (Chl a) among groups (one-way ANOVA, p = 0.235, figure 5a), among which the LTHC-selected group had the highest mean value (0.32 ± 0.07, Tukey test, p < 0.05). Carotenoids (Car) showed significant differences among groups (one-way ANOVA, p = 0.049, figure 5b), with the LTLC-selected group having a significantly higher carotenoid content (0.13 ± 0.02) than the HTHC-selected group (0.06 ± 0.01, Tukey test, p < 0.05). The Car/Chl a ratio was not statistically significant (p = 0.058, figure 5c), but it showed a decreasing trend from the LTLC- to the HTLC-selected group (from 0.43 to 0.26). The Chl a content of HTLC-selected populations was higher under HTHC assay environment compared with the LTLC (Students’ t‐test, p < 0.05). The HTHC-selected populations also exhibited significant differences in carotenoid content and the Car/Chl a ratio between assay environments, with higher values observed under HTHC conditions (Student's t‐test, p < 0.05).

Figure 5.

The Chl a (a), carotenoid (b) and Car/Chl a (c) of Skeletonema costatum in the short-term assay experiment. The x-axis represents the selection environment. Boxplot elements and statistical annotations are described in figure 3.

For the HTHC assay condition, Chl a showed no significant differences among groups (p = 0.138, figure 5a), even though the mean Chl a content of the HTLC-selected group (0.38 ± 0.01) was higher than that of the LTLC-selected group (0.26 ± 0.08, Tukey test, p < 0.05). Car did not show significant differences among groups (p = 0.749, figure 5b). The Car/Chl a ratio showed significant differences among groups (p = 0.046, figure 5c), with the LTLC-selected group having a significantly higher ratio (0.49 ± 0.09, Tukey test, p < 0.05) than the HTLC-selected group (0.32 ± 0.04, Tukey test, p < 0.05).

(e). Photochemical performance in the short-term assay experiment

Across the selected populations acclimated to each assay environment, LTLC and HTHC assay environments, there were no significant differences in F_v/F_m (0.61–0.64 versus 0.59–0.64) and light energy utilization efficiency α (LTLC: 0.30–0.32; HTHC: 0.30–0.32) among groups (one-way ANOVA, p > 0.05, table 1). rETR_max exhibited significant differences among groups in both assay environments (LTLC: p < 0.001; HTHC: p = 0.002, figure 6f). Specifically, the HTLC-selected populations achieved the highest rETR_max under the HTHC assay condition, while the lowest rETR_max value was found in the HTHC-selected populations under the LTLC assay environment. For most of the selected populations, rETR_max values differed between the two assay environments (p < 0.05), except for the LTHC-selected population. The light saturation point (I_k) showed significant differences under LTLC assay conditions (p < 0.001, table 1). Specifically, the I_k values of the LTLC- and LTHC-selected populations were significantly higher than those of the HTLC- and HTHC-selected populations (Tukey test, p < 0.05, figure 6d). Under HTHC assay conditions, there were no significant differences in I_k values among the different selected populations. No significant difference was detected for α, both between selection treatments under the same assay treatment and between assay treatments for the same selection treatment.

Table 1.

Parameters of the photochemical performance of Skeletonema costatum of different selected treatments in the assay experiment. The F_v/F_m represents the maximum photochemical efficiency of photosystem II, rETR_max indicates relative maximal electron transport rate, α is the light use efficiency, I_k (μmol photons m⁻² s⁻¹) is the light intensity for the saturated electron transport rate. The values are means ± s.d. of triplicate cultures (n = 3). Significant differences among selection treatment populations in response to each assay are indicated by different letter superscripts (lowercase letters represent LTLC assay environment, while uppercase letters represent HTHC, Tukey test, p < 0.05). Significant differences between the assay environments of populations selected under the same selection environments are indicated by the superscript asterisks (Student's t‐test, p < 0.05).

assay	selection	Fv/Fm	rETRmax	α	I k
LTLC	LTLC	0.63 ± 0.00a	170.88 ± 5.72a*	0.31 ± 0.00a	554.38 ± 12.85a*
	LTHC	0.64 ± 0.01a*	157.58 ± 2.65a	0.32 ± 0.01a	497.86 ± 11.47a
	HTLC	0.61 ± 0.02a	117.05 ± 11.70b*	0.30 ± 0.02a	394.22 ± 31.82b*
	HTHC	0.61 ± 0.01a	108.10 ± 9.47b*	0.30 ± 0.02a	365.32 ± 26.13b
HTHC	LTLC	0.59 ± 0.03A	127.85 ± 10.26B*	0.30 ± 0.00A	420.39 ± 39.23A*
	LTHC	0.61 ± 0.01A*	155.22 ± 4.14AB	0.32 ± 0.03A	489.74 ± 66.18A
	HTLC	0.64 ± 0.06A	181.25 ± 10.99A*	0.32 ± 0.01A	568.79 ± 42.73A*
	HTHC	0.64 ± 0.01A	164.47 ± 9.17A*	0.30 ± 0.04A	567.89 ± 120.04A

Figure 6.

Photochemical performance of Skeletonema costatum selected under different environments in two assay experiment. Rapid light response curves of rETR under LTLC (a) and HTHC (b) assay environments. The fitted curves represent data from different selected populations: Ambient CO₂ and temperature (black), elevated CO₂ (red), elevated temperature (blue), and elevated CO₂ and temperature (green). Values represent means ± SD of triplicate cultures (n = 3). Parameters derived from the fitted curves include Fv/Fm (c), Ik (d), α (e), and rETR_max (f). The x-axis represents the selection environment, while different colors of the boxes represent the two assay environments. Boxplot elements and statistical annotations are described in figure 3.

4. Discussion

It has been demonstrated that phytoplankton have the potential to evolve under selective environmental changes over hundreds of generations in laboratory experiments [42]. However, investigating the potential for adaptive evolution under conditions more representative of natural environments remains crucial [43]. This study demonstrates the adaptive evolution and physiological trait trade-offs of the diatom S. costatum after more than 3 years of growth in monocultures mimicking ocean acidification and warming. From the long-term growth tracking, the results indicated that elevated CO₂ and temperature interactively impact the growth dynamics of S. costatum, leading to differences in the number of accumulated generations among populations selected under different regimes. Specifically, S. costatum selected under HTHC conditions produced the most generations, followed by those selected under HTLC condition. In the short-term assay experiments, we evaluated the performance of populations under a common environment to assess potential adaptive evolution and trade-offs. Populations selected under high-temperature (HT) conditions showed higher fitness in the HT assay but lower fitness in the low-temperature (LT) assay compared with LT-selected populations. These findings support the concept of local advantage [44], where populations perform better in their native environments. Further evidence and implications of this adaptation are discussed in the following sections.

(a). The evolutionary adaptation in physiological traits

In this study, we transitioned the selected populations to LTLC (current conditions) and HTHC (future scenarios) after more than 3 years (1396 days) of selection across four different treatments. We aimed to investigate potential evolutionary adaptations to elevated CO₂ and temperature. The experimental results revealed significant physiological trait differences among the various selection groups, indicating adaptations driven by these elevated environmental factors. Specifically, the HTLC-selected group exhibited a pronounced advantage in growth rate under HTHC conditions, while the LTLC-selected group achieved the highest growth rates in its native LTLC environment. This phenomenon exemplifies the local adaptation characteristic of ‘local advantage’ [44]. The adaptations observed in physiological traits, as evidenced through our long-term experiments, will be further discussed in the subsequent sections. For instance, the HTLC-selected population exhibited higher relative fitness under conditions of elevated temperature and CO₂, demonstrating a pronounced adaptability to the stresses associated with these environments. In contrast, the LTLC-selected population was better suited to the relatively low CO₂ and temperature conditions, likely owing to the fixation of advantageous alleles through directional selection in response to their specific environmental context [45]. CUE did not differ among populations selected under the LTLC assay environment, while populations selected under HTLC and HTHC conditions showed higher CUE compared with LTLC populations in the HTHC assay environment, indicating adaptive evolution under elevated temperatures despite the CO₂ level. This change may be attributed to reduced respiration rates in the HTHC-selected populations and increased gross photosynthesis in HTLC-selected populations compared with LTLC-selected populations. Previous studies on Phaeodactylum tricornutum have demonstrated reduced metabolism (lower respiration rate) in response to high CO₂ [31], which supports the balancing mechanisms of resource acquisition and maintenance as described by metabolic theory [46]. Notably, this adaptation was observed only in populations exposed to both elevated temperature and elevated CO₂ (HTHC) in our experiment, highlighting the interactive effects of temperature and CO₂ on the adaptation process. Furthermore, the HTLC-selected population showed higher gross photosynthesis rates under the HTHC assay environment, which contrasts with findings from a study on Thalassiosira pseudonana, where selection under fluctuating high temperatures alone led to reduced gross photosynthesis after 300 generations [26]. However, since our experiment also incorporated fluctuations in natural light, direct comparison with other studies is challenging. Ultimately, our study suggests that experimental evolution under fluctuating environmental conditions closer to natural variability might involve more complex underlying mechanisms.

Changes in pigment content and photochemical parameters provide compelling evidence for the occurrence of adaptive differentiation among phytoplankton populations. The HTHC-selected population demonstrated significantly higher levels of carotenoid contents and carotenoid: chlorophyll a (Car/Chl a) ratio in the HTHC assay environment compared with LTLC assay, indicative of an effective strategy to mitigate photoinhibition through an increased allocation of photoprotective pigments. Conversely, the light saturation point (I_k) and maximum rETR_max of the LTLC-selected population exhibited a notable decline in the HTHC environment. This decrease may reflect altered energy conversion efficiencies as a consequence of prolonged adaptation to lower temperature conditions [47]. The opposite trend was observed for the I_k of the HTLC-selected population and the rETR_max of both the HTLC- and HTHC-selected populations, where these traits show higher performance under the HTHC assay environment compared with the LTLC environment. Adaptation to elevated temperature (and high CO₂) appears to shift the light‐harvesting strategy of diatom populations: HTLC- and HTHC-selected cells show a higher I_k when assayed under HTHC conditions, maximizing performance under their selection environment [31]. Conversely, these same populations exhibit lower I_k values under LTLC conditions, relative to the other selection treatments, reflecting a trade-off that optimizes photosynthetic efficiency in home environments. These results contribute to the growing body of literature on the evolutionary responses of the phytoplankton physiological traits to environmental changes under fluctuating conditions.

(b). Evolutionary trade-offs in physiological traits

The observed growth–performance trade-offs across different assay environments strongly indicate that evolutionary adaptations to specific temperature–CO₂ combinations entail metabolic costs that constrain phenotypic plasticity in novel conditions. Although the HTLC-selected population exhibited the highest growth rate under the HTHC assay environment, its growth rate decreased significantly under the LTLC assay, with a relative fitness below 1.0 compared with the LTLC-selected population. This trade-off suggests the presence of antagonistic pleiotropy, where genetic variants that are beneficial in one environment can be detrimental in another [48]. Notably, the LTLC-selected populations exhibited substantially lower CUE in the HTHC assay environment compared with populations selected under high CO₂, elevated temperature or both. In contrast, no significant differences in CUE were observed among the selected populations under the LTLC assay environment. The CUE of HTLC- and HTHC-selected populations under the LTLC assay environment was comparable to that of the LTLC-selected population, while their CUE under elevated temperature conditions was significantly higher. This suggests that selection under elevated temperature and CO₂ enhanced the plasticity of CUE without compromising performance under lower temperature conditions. Research has shown that warming serves as the primary driver of evolutionary adaptation in the marine diatom Thalassiosira weissflogii, rather than ocean acidification alone [18]. This highlights the constraints and trade-offs encountered by diatoms in adapting to both ocean acidification and warming, particularly concerning temperature response traits such as optimal temperatures for photosynthesis, maximum net photosynthetic oxygen evolution rates and activation energy [49].

Moreover, the lack of significant differences (p > 0.05) between the HTLC and HTHC groups across multiple parameters, for example, the growth rate, CUE and F_v/F_m, may suggest convergent evolutionary strategies under high-temperature regimes. The differences in photosynthetic performance, the rETR_max, between the two assay environments in populations selected under LTLC, HTLC and HTHC conditions suggest a trade-off in electron transport capacity under elevated temperature or under both elevated temperature and high CO₂ conditions [31]. Specifically, the rETR_max of populations selected under HTLC and HTHC conditions was higher under the HTHC assay environment compared with LTLC, whereas the LTLC-selected population showed the opposite trend. These findings are consistent with evolutionary theory, which suggests that adaptation to specific environmental conditions can lead to reduced phenotypic plasticity, as evidenced by the limited cross-environmental performance of LTLC-adapted populations [50]. Additionally, the observed variations in the carotenoid-to-chlorophyll a (Car/Chl a) ratio further underscore the metabolic costs associated with photoprotective pigment synthesis under conditions of environmental mismatch.

5. Conclusion and future perspective

This study demonstrates that the evolutionary adaptation of phytoplankton to elevated CO₂ and temperature conditions involves complex metabolic trade-offs and environment-specific adaptive characteristics. The experiments showed that different populations, after long-term selection under simulated ocean acidification and warming, exhibited both adaptive evolution and trade-offs in their relative fitness when assessed in the assay environment, indicating evolved phenotypic plasticity in this diatom. The underlying mechanisms involve a reduction in respiration, increase in gross photosynthesis and consequent increase in CUE of populations adapting to high-temperature environments; however, this comes with trade-offs, such as decreased production of photoprotective pigments and reduced photochemical efficiency when populations are exposed to environments that differ from their selection conditions i.e. at lower temperature and CO₂. Although further exploration is needed to address adaptation strategies in a more mechanistic way, such as through omics analysis, this study provides evidence indicating the adaptability of the coastal diatom S. costatum to elevated CO₂ and temperature under fluctuating natural sunlight and ambient temperature.

The adaptability of diatoms to changing environments may play a key role in their survival and persistence in marine ecosystems, particularly against the backdrop of global climate change. However, fluctuations in the marine environment extend beyond the CO₂ levels, temperature and light conditions simulated in our experimental design. Other factors, such as nutrient variation, pollutant contamination and predator–prey relationships, also contribute as bottom-up and top-down controls on marine phytoplankton. Therefore, significant research efforts are still needed to investigate and understand the evolutionary potential and trade-offs of phytoplankton under more complex experimental systems. Such knowledge will enhance our ability to predict the future dynamics of marine primary producers and their roles in global primary production and biogeochemical cycles.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

All data and codes can be found in the repository for this paper [51].

Supplementary material is available online [52].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

Y.W.: data curation, formal analysis, investigation, methodology, visualization, writing—original draft; S.W.: data curation, investigation, methodology, validation; M.X.: data curation, investigation, methodology, validation; Z.L.: conceptualization, formal analysis, supervision, validation, visualization, writing—review and editing; J.X.: conceptualization, funding acquisition, project administration, resources, supervision, validation, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed herein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

Funding was received from the National Natural Science Foundation of China (No. 42361144840).