Decreased photosynthesis and growth with reduced respiration in the model diatom Phaeodactylum tricornutum grown under elevated CO2 over 1800 generations

Studies on the long‐term responses of marine phytoplankton to ongoing ocean acidification (OA) are appearing rapidly in the literature. However, only a few of these have investigated diatoms, which is disproportionate to their contribution to global primary production. Here we show that a population of the model diatom Phaeodactylum tricornutum, after growing under elevated CO2 (1000 μatm, HCL, pHT: 7.70) for 1860 generations, showed significant differences in photosynthesis and growth from a population maintained in ambient CO2 and then transferred to elevated CO2 for 20 generations (HC). The HCL population had lower mitochondrial respiration, than did the control population maintained in ambient CO2 (400 μatm, LCL, pHT: 8.02) for 1860 generations. Although the cells had higher respiratory carbon loss within 20 generations under the elevated CO2, being consistent to previous findings, they downregulated their respiration to sustain their growth in longer duration under the OA condition. Responses of phytoplankton to OA may depend on the timescale for which they are exposed due to fluctuations in physiological traits over time. This study provides the first evidence that populations of the model species, P. tricornutum, differ phenotypically from each other after having been grown for differing spans of time under OA conditions, suggesting that long‐term changes should be measured to understand responses of primary producers to OA, especially in waters with diatom‐dominated phytoplankton assemblages.

Diatoms play a critical role in the carbon cycle as they are responsible for about 20% 2 of global primary production (Field et al., 1998). They contribute about 35% to the 3 primary productivity of the oligotrophic oceans and nearly 75% to that of the coastal 4 zone and other nutrient-rich systems (Nelson et al., 1995). In addition, due to the   Polycarbonate bottles used for culturing cells were placed in one incubator. They 19 were manually shaken once a day and their positions relative to the light source were 20 randomly changed, so that cells were not systematically exposed to different levels 21 of illumination. The pH was measured before and after dilution by a pH meter 22 (Orion 2 STAR, Thermo Scientific) calibrated with standard National Bureau of 1 Standards (NBS) buffers. CO2 partial pressure was measured by a CO2 detector 2 (M170, Vaisala Oyj). Carbonate chemistry parameters were calculated by CO2SYS 3 software using pH and pCO2. Values of pH and pCO2 were measured periodically 4 (but not at every dilution due to logistical difficulties) during the long-term exposure.

5
Cell density was maintained below 1.2 × 10 5 per ml (exponential growth phase) and 6 the carbonate chemistry remained stable over the culture period, with pH variations 7 ˂ 0.08 units. Carbonate chemistry of media at the end of experiment are shown in 8   Table S1. Experimental set-up 11 For the long-term experiment, Aquil medium was pre-aerated either with ambient 12 outdoor air (400 μatm pCO2) for the long-term ambient air treatment (hereafter 13 termed LCL) or with CO2 enriched air (1000 μatm pCO2) for long-term CO2 14 enriched air treatment (hereafter termed HCL) before dilution. The latter condition 15 was achieved within a CO2 plant growth chamber (HP1000G-D, Ruihua), in which 16 CO2 partial pressure was continuously monitored and maintained at 1000 ± 50 μatm. 17 The cultures were diluted every 5 to 7 days to maintain a stable carbonate system 18 within the LCL and HCL treatments. Three separate bottles were maintained during 19 each growth cycle (5 ~ 7 days), and then the triplicate cultures of LCL or HCL 20 treatment were pooled by treatment at the end of the growth cycle, and then diluted 21 to separate bottles with fresh medium equilibrated with the CO2 levels. This means 22 that there is a single population for each treatment, and triplicate bottles were used to 1 measure growth rate and cell size of each population in the growth cycle. Cells were 2 grown under LCL and HCL conditions for 1035 days (about 1860 generations) 3 before having their traits measured.

4
To investigate short-term OA effects, LCL cells were inoculated into the medium 5 pre-aerated with 1000 μatm pCO2 air at the end of the long-term experiment above 6 and grown for another 20 generations (HC treatment) before sampling. LCL cells 7 rather than ancestral cells were used so that the entire experiment could be measured 8 at once to avoid time-block effects, so the LCL population growing in air is the 9 "control" phenotype to which the short-and long-term phenotypes (HC and HCL)   Chlorophyll and carotenoid contents 21 Cells for determination of pigment contents were filtered onto GF/F filters (25 mm, Chl c (µg ml -1 ) = -6.0138 × (A665 -A750) + 28.8191 × (A632 -A750).   re-suspended into 20 mmol L -1 Tris buffered medium whose pH was pre-adjusted by 6 freshly prepared hydrochloric acid and sodium hydroxide to their corresponding 7 culture medium values. Use of Tris-buffer was necessary to maintain the appropriate 8 pH at the higher cell densities used for oxygen evolution experiments. Samples with 9 a known concentration around 2×10 6 cells per milliliter were injected into the 10 oxygen electrode chamber, and were magnetically stirred. Samples for dark 11 respiration determination were filtered and re-suspended in the same way as 12 mentioned above, and oxygen consumption rates were monitored in the dark. Rates 13 of oxygen evolution and consumption were recorded when they became constant.
14 Additionally, photosynthetic rates of LCL and HCL populations were determined by 15 the 14 C method (see supporting information for detail) at day 558 and 588.

17
The population responses to elevated CO2

18
Responses to elevated CO2 of individual parameters for the HC or HCL population 19 were calculated as the fold difference compared to the LCL control. For both 20 population responses, a value of 1 indicates no response, a value < 1 indicates a 21 negative response and a value > 1 indicates a positive response to elevated CO2. The

22
value was set to 1 when no significant difference was detected between HC or HCL 1 and LCL populations. with 95% confidence intervals. Significant differences in growth rates and cell sizes 9 between LCL and HCL populations for every measured growth cycle from days 503 10 to 1035 were analyzed using independent samples t tests with 95% confidence 11 intervals. Specific growth rates and cell sizes of both populations for every measured 12 growth cycle and p values of independent t tests are shown in Tables S2 and S3.

15
Growth rate and mean cell size 16 The specific growth rates ranged between 0.96 ± 0.11 ~ 1.50 ± 0.01 d -1 in both LCL 17 and HCL populations over the entire experiment. From days 503 to 960, 33 of the 54 18 growth cycles showed no significant differences in growth rate between LCL and 19 HCL populations ( Fig.1a  Mean cell sizes (width) ranged from 4.6 ± 0.02 ~ 5.6 ± 0.02 μm during the entire 10 experiment, and the differences between the two populations varied over time.  Table S3, independent t test, p ˃   respectively. There were no significant differences in α found among populations. Chlorophyll-normalized dark respiration under the corresponding growth conditions 9 showed differing patterns: the HC population had a 179% enhancement of dark 10 respiration compared to the LCL population ( Fig. 3a, one-way ANOVA, p < 0.001).

11
However, dark respiration in the HCL population was statistically the same as that of

17
When normalized to per cell, the HC population had a 129% increase of dark 18 respiration per cell compared to cells of LCL population (Fig. 3b, one-way ANOVA, 19 p < 0.001), and the rate was reduced by 41.9% in the HCL population relative to the 20 LCL population (one-way ANOVA, p = 0.048). There were no statistically 21 significant differences for net oxygen evolution per cell among populations (Fig. 3d). The population responses to elevated CO2 6 Differences between the HC and HCL populations are summarized in showed a positive response to elevated CO2. In contrast, the HCL population showed 10 a differing pattern: values for growth rate, cell size, dark respiration, and 11 chlorophyll-normalized photosynthesis were all below 1, and the value for 12 chlorophyll content was greater than 1.  short-term responses for this species.

22
The long-term consequences of OA on the main marine phytoplankton groups 1 (cyanobacteria, diatoms, dinoflagelates, and coccolithophores) are beginning to be 2 studied, and differences in the responses (direction and rate) of the main 3 phytoplankton groups to OA on a long timescale will lead to shifts in marine 4 community composition and thus the frequencies of some groups in ecosystems.  However, it is still worth noting that the present study did not measure 10 evolutionary response to OA (which would have been revealed by switching HC 11 selection line cells to the LC condition). Thus we cannot confirm whether the 12 long-term response shown in the present study is a plastic or an adaptive response.

13
Regardless of the attributes of the long-term response though, the present study 14 provides evidence that the response of phytoplankton to OA may depend on the 15 timescale for which they are exposed. The response of long-term OA exposure 16 population in the present study may provide some clues and information for future 17 evolutionary studies on diatom species. Only one population per treatment was used 18 in the present study, which limits our ability to attribute the results to long-term OA phytoplankton to OA may depend on the timescale for which they are exposed.
2 Differences in the responses (direction and rate) of the main phytoplankton groups to 3 OA and other factors on a long timescale will influence ecosystem composition and 4 primary production.    the incubation, cells were gently filtered onto GF/F filters (25mm, Whatman) and the 7 filters were placed into scintillation vials. Filters were exposed to hydrochloric acid 8 fumes overnight, and dried at 50 o C for 6 h. Scintillation cocktail was added to the 9 vials before assimilated radiocarbon was counted by a liquid scintillation counter (Tri-10 Carb 2800TR, PerkinElmer).