Culturing and all experiments were performed in a climate control “cold” laboratory where temperature was set at 0°C and humidity (<50%, dew point −10°C) was permanently controlled. Culturing and experiments were performed in axenic conditions inside the temperature-controlled laboratory. Axenic F. cylindrus (CCMP3323) was grown in natural seawater (sampled in Baffin Bay, Canadian Arctic, 67.48 N; 63.79 W) enriched with f/2 (Guillard, 1975). The sea water was prefiltered through a polypropylene 1 μm filter (Polypropylene felt filter bag 18–1/2L, 1 μm, Cole Parmer, Canada) and sterilized with a PolyCap 0.2 μm (Whatman™, United Kingdom). Cultures were pre-acclimated to each experimental light conditions for at least 3 weeks before the start of the experiment; they were gently stirred with a magnetic stirrer and aerated with air bubbling filtered through an activated carbon filter and a 0.2 μm HEPA filter (Carbon CAP, HEPA-Vent, Whatman™, United Kingdom). Cells were grown in triplicate in 3 L jacketed cylinder reactors at the temperature of 0°C. The temperature was controlled by the circulation of thermostated ethylene glycol through the jacket. Light was supplied uniformly by a custom illumination system with an array of LEDs (LXML-PR01, 445 nm; LXML-PB01, 470 nm; LXML-PM01, 505 nm; LXML-PM01, 530 nm; LXM2-PD01, 630 nm; LXM3-PD01, 660 nm; LXML-PD01, 4100 K; LXML-PL01, amber; LUXEON REBEL, LUMILEDS, Germany) that allowed the modulation of light intensity, the photoperiod and the recreation of a “white” light similar to under sea-ice solar spectrum (Supplementary Figure S1). PAR (Photosynthetically Available Radiations) light intensity was measured continuously in the centre of each reactor using a 4π PAR sensor (QSL 2101, Biospherical Instruments Inc., CA, United States) and PAR intensity was automatically adjusted according to the continuous reading of the PAR sensor to maintain PAR intensity inside the reactor vessel to the targeted value. To better highlight the responses to photoperiod changes, the optimal light intensity for the growth of F. cylindrus was used, i.e. 30 μmol photons m^-2 s^-1 of PAR (Morin et al., 2020; Croteau et al., 2022). Importantly, cultures were maintained optically thin, in order to control the light field inside the reactors, by a daily dilution semi-continuous growing to maintain a cell concertation of 10⁶ cells mL^-1.

The cells were acclimated to five photoperiods: 0 h Light: 24 h Dark (0L:24D), 6L:18D, 12L:12D, 18L:6D and 24L:0D (Figure 2). Each experiment lasted 52 h, and the time laps between samplings was at minimum 2 h and at maximum 6 h to perform cell counts, samplings for pigment analyses and measurements of photosynthetic performance (i.e., Rapid Light Curves, see below). The time laps between sampling were adjusted depending on the daylength and the moment of the day, for a tighter coverage during the light compared to the dark period. At the last sampling point, samples were taken for elemental composition analysis and photosynthetic carbon fixation estimation (Production Light Curves, see below). All measured parameters, their definition and unities are found in Table 1. All samplings and measurements were performed in the temperature-controlled laboratory under a green light of low intensity to prevent light alteration of cell photosynthetic activity and pigments. Cultures acclimated at 0:24 were not diluted, the cultures acclimated 6L:18D, 12L:12D, 18L:6D were diluted during the first hour of the dark period of the photoperiods and the cultures acclimated at 24:0 were diluted at the 10^th hour of the 24 h cycle.

Cell concentration was measured using a particle sizing and counting analyser (Multisizer 4 Coulter Counter, Beckman Coulter, United States). We considered counts between 2 and 15 μm cell size. A dilution of the culture sample was performed before analysis to remain within the cell concentration measurement range of the analyser. Dilution was carried out with the electrolyte of the analyzer, i.e., salted Milli-Q™ water (35 g L^-1. NaCl). The growth rate (μ, in day^-1) was calculated between two dilutions according to Andersen (2005), and the hourly growth rate (μ^H, in h^-1) between two sampling points.

The total particulate carbon (TPC), total particulate nitrogen (TPN), and the dry algal biomass were determined by filtering culture samples on pre-burned (450°C for 4 h) 25 mm GF/F glass-fibre filters (Whatman™, United Kingdom). The CHN analyses were performed with a PerkinElmer 2,400 Series II CHNS/O elemental analyzer (PerkinElmer, Waltham, United States). Acetanilide (99.999%, Perkin Elmer) was used as a standard.

Pigments were quantified by High Performance Liquid Chromatography (HPLC, Agilent Technologies 1,200 Series, Agilent, United States). Culture samples were filtered onto 25 mm GF/F glass fiber filters (Whatman™, United Kingdom) in the cold laboratory, immediately frozen in liquid nitrogen and stored at −80°C until further analysis. Before extraction, 50 μL of internal reference apo-carotene (trans-b-Apo-80-carotenal, 1.98 mg L^-1) was added to the 3 mL extraction buffer (100% methanol HPLC grade) for each sample and to the blanks. Filters were ground by sonication (Sonicator Ultrasonic Processor XL 2010) for 12 s on ice and centrifuged for 5 min at 3,700×g. The extracts were filtered through a 0.22 μm polytetrafluoroethylene syringe filter and placed in an automatic sampling vials filled with argon to limit oxidation. A volume of 100 μL was injected into a C8 Symmetry^® column (150 × 4.6 mm; 3.5 μm, Waters, France) of the HPLC system. Elution by the solvent gradient was monitored by a P4000 Thermo Separation (TSP) pump as indicated in Zapata et al. (2000). The pigments were detected with a TSP UV 6000 LP absorbance detector (400–700 nm). Pigment quantification was carried out using pigment standards provided by D.H.I. Water & Environment (Horsholm, Denmark) and the internal reference (apo-carotene). The de-epoxidation state of diadinoxanthin in diatoxanthin (DES) was calculated as in Equation 1: D E S = D t x D d x + D t x × 100 (1)

A blue light source (λ = 450 ± 20 nm) Pulse Amplitude Modulated (PAM) fluorometer (WATER-ED/B, Heinz Walz GmbH, Germany) was used to measure the photosynthetic performance of cultures on dark acclimated (30 min) samples. Rapid Light Curves (RLCs) were performed with eight 30 s steps of increasing light intensity from 0 to 420 µmol photons m^-2 s^-1. The dark-acclimated photochemical efficiency of photosystem II (PSII, F_V/F_M) was calculated as in Equation 2: F V / F M = F M − F 0 F M (2) were F₀ and F_M are, respectively, the minimum and maximum levels of dark acclimated chlorophyll fluorescence. The relative electron transport rate (rETR) was calculated as in Equation 3: r E T R = E × F M ′ − F ′ F M ′ (3) where E is the light intensity, and F’ and F_M’ are, respectively, the steady-state and maximum fluorescence levels of light acclimated cells. The determination of rETR for each of the eight intensities of the RLCs allowed to build rETR vs. E curves that were fitted according to Eilers and Peeters (1988) in order to extract photosynthetic parameters (see Barnett et al., 2015): rETR_max, α, E_k, and E_opt; see Supplementary Table S1 for definitions.

The ouput of the RLCs also allowed to calculate the non-photochemical quenching (NPQ) as in Equation 4: N P Q = F M − F M ′ F M ′ (4)

The determination of NPQ for each RLCs step allowed to build a NPQ vs. E curve used to estimate E50_NPQ by fitting it with the model of Serodio and Lavaud (2011). Additionally, the maximal NPQ (NPQ_max) was the maximal value obtained for the highest RLC step (420 µmol photons m^-2 s^-1).

The partitioning of absorbed excitation energy in PSII was determined by the complementary PSII quantum yields method (Hendrickson et al., 2004; Klughammer and Schreiber, 2008; Xu et al., 2019), and computed as: Φ_PSII = ( F M ′ − F ′ F M ′ ), Φ_NPQ = F ′ F M ′ − F ′ F M and Φ_NO = ( F ′ F M ); (see Table 1 for definitions).

Under our growth conditions, F. cylindrus shows an absorption coefficient of 0.86 (unitless) for the Water-PAM blue light (450 nm) and of 0.39 (unitless) for the growing “white” light spectrum. F’ and F_M’ were measured during RLC at 16 µmol photons m^-2 s^-1 providing 13.8 µmol photons m^-2 s^-1 of photosynthetic useable radiation, PUR, equivalent to 34.5 µmol photons m^-2 s^-1 of “white” light PAR as used for the photoperiod treatments (i.e., the RLC light intensity the closest to our experimental PAR of 30 µmol photons m^-2 s^-1). Absorption coefficient and PUR were calculated as described by Guérin et al. (2022).

The production vs. light curve (P vs. E, Steeman-Nielsen, 1975) were measured by adding 20 μL of a solution of 74 MBq mL^-1 of H¹⁴CO₃ (GE Healthcare, United States) to 40 mL of a culture sample. Then, 28 subsamples of 1 mL were illuminated for 20 min at increasing light intensities from 0 to 442 μmol m^-2 s^-1 photons at 0°C. Furthermore, six supplementary sub samples were incubated for 20 min and 24 h, under the culture respective growth light intensity and photoperiods (for 24 h incubation), to quantify the net and gross primary production respectively (NPP and GPP). Immediately after the extinction of the light, 50 μL of buffered formalin was added to each incubation vial to stop the reaction. Samples were acidified with 250 μL of 50% HCl (v./v.) for at least 3 h to allow the venting of inorganic ¹⁴C (Poulain et al., 2014) before liquid scintillation counting in a Tri-Carb2910 TF (Perkin Elmer, United States). The P vs. E curves were fitted to the equation of Eilers and Peeters (1988) to compute the parameters P_max, P-α, P-E_k and P-E_opt (see Table 1 for definitions).

A one-way ANOVA followed by Tukey’s HSD post hoc test were used to tests differences in the measured parameters means between treatments (Supplementary Table S1). Normality of residuals and homogeneity of variances were tested using Shapiro-Wilk and Bartlett’s test respectively. Analysis were performed using R version 4.2.2 (2022-10-10).

The growth rate (µ, Figure 3A) increased with daylength and reached its maximum (0.25 ± 0.04 d^-1) for the cells acclimated to 18 h Light: 6 h Dark (18L:6D). P_max (Figure 3; Table 1 for the definition of all parameters) decreased with increasing daylength and was close to 0 under 0L:24D. The gross primary production (Figure 3B) increased with the daylength, while the net primary production (Figure 3B) followed the same trend as the growth rate. The intracellular C (Table 2) was maximal for 6L:18D cells (19.2 ± 2.7 pg cell^-1) and decreased with the increase in daylength until reaching a minimum for 18L:6D (3.10 ± 0.5 pg cell^-1). The C/N (Table 2) was relatively similar between photoperiods (around 5.1 ± 0.5 g g^-1) as well as the Chlorophyll a (Chl a)/C (20.2 ± 5.0 mg g^-1), although the lowest and highest were observed under 6L:18D and 24L:0D, respectively.

Intracellular Chl a (Figure 3) increased with daylength, and the highest values were observed at both extreme photoperiods (0L:24D and 24D:0L, Figure 3). The photosynthetic pigments (Chl c and β-Carotene, fucoxanthin, Figure 3; Table 2) remained stable relative to Chl a among photoperiods, at the exception of fucoxanthin which decreased with daylength. Photoprotective Ddx and Dtx (Table 2) contents nearly doubled from 6 h to 18 h daylength and reached a maximum under 0L:24D (20.7 ± 0.6 mol 100 mol Chl a ⁻¹). The de-epoxidation state of Ddx into Dtx (DES, Figure 3) was low and similar (average 7.3% ± 0.7%) under 6L:18D, 12L:12D and 18L:6D and reached a significant maximum under 24L:0D (34.1% ± 4.6%).

F_V/F_M (Figure 3) was the highest under 0L:24D (0.660 ± 0.005) and 12L:12D (0.65 ± 0.01). No significant differences have been found between the three other treatments with a value of 0.62 ± 0.01. Φ_PSII (Figure 3) was maximal for 12L:12D and 18L:6D growing cells and divided by a factor of three under 0L:24D. rETR_max (Figure 3) was similar for 6L:18D and 12L:12D growing cells (average 21.1 ± 1.0) and the lowest under 0L:24D. NPQ_max (Figure 3) was the highest under 0L:24D and 18L:6D (1.5) and halved under 24L:0D. Φ_NPQ (Figure 3) was maximal under 0L:24D and halved under the other light conditions (average 0.10 ± 0.03) while Φ_NO was similar under all photoperiods (average 0.61 ± 0.04). E50_NPQ (Table 2) was the highest under 12L:12D (152 ± 39 µmol photons m^-2 s^-1) and the lowest under 24L:0D and 0L:24D.

Beyond the above described photoacclimation status based on average ± SD of three independent cultures monitored during 52 h, many of the examined parameters showed daily variations under light:darkness alternation, although the light intensity was kept stable (30 µmol photons m^-2 s^-1) during the light period.

Cultures grown under 6L:18D, 12L:12D and 18L:6D regimes showed a steady increase of the hourly growth rate (μ^H, h^-1, Figure 4A) from the beginning of the light period reaching a maximum after 4–6 h, followed by a plateau (12L:12D and 18L:6D treatments), and a decrease to a minimum during darkness (Figure 4A). In parallel, F_V/F_M (Figure 4B) rapidly increased during the first hours of the light period, reaching a maximum before the middle of the light period. F_V/F_M decrease systematically started before the onset of the darkness period, and continuing until light onset (Figure 2B b, 6L:18D, 12L:12D and 18L:6D). This pattern was modelled onto the length of the light period (i.g. the longer the light period, the later the value reach is maximum) and repeated itself over the next 24 h of monitoring. In contrast, F_V/F_M values were stable under treatments with no light:darkness alternation (Figure 4B, 0L:24D and 24L:0D). Other photosynthetic and photoprotective parameters (rETR_max, NPQ_max, Figures 5A, B), as well as the photoprotective pigments (Ddx + Dtx, DES, Figure 5C; Supplementary Figure S8) showed a similar pattern of daily variations with some modulations, while all of them showed the same stable pattern as F_V/F_M under 0L:24D and 24L:0D (Supplementary Figures S2–S12).

However, different types of patterns were observed during the light period: i) decrease after reaching their daily maximum (i.e., rETR_max NPQ_max and Ddx + Dtx Figure 5 also E_k, E_opt, ϕ_PSII Supplementary Figures S2, S3, S5), ii) a steady increase during the light period for reaching a maximum at the light-darkness transition (i.g. E50_NPQ and ϕ_PSII, Supplementary Figures S5 and S7) at the exception of the 18L:6D which fall in to the first group, iii) decrease during the first hours of illumination before increasing again (ϕ_NO Supplementary Figure S6). Also, for some parameters, 6L:18D cells showed a different pattern than 12L:12D and 18L:6D cells: i) E_k 6L:18D (Supplementary Figure S2) showed an inversed pattern compared to 12L:12D and 18L:6D cells, i.e., it decreased at light and increased under darkness; ii) NPQ_max 6L:18D (Figure 5B) was stable under darkness when it increase for 12L:12D and 18L:6D cells; iii) Ddx + Dtx 6L:18D (Figure 5C) reached its maximum by the end of the light period instead of the middle for 12L:12D and 18L:6D cells; iv) 6L:18D β-Carotene synthesis (Supplementary Figure S12) showed daily variations when it was steady for 12L:12D and 18L:6D cells.

When plotted as a function of the Zeitgeber Time (ZT), a standardized 24-hour cycle where 0 indicates the beginning of the illumination and 12 the end of the illumination, and as a function of the cumulative light dose (LD) received over the daylight period (Figure 6), some synchronisation appeared either with ZT and/or LD, depending on parameters. Most photosynthetic parameters daily oscillations were synchronized with the light period, reaching their maximum around the sixth hour of ZT (F_V/F_M and rETR_max, Figures 6A, B) but also earlier at the fourth hour (α and E_opt, Supplementary Figures S14A, C). The same holds true for the photochemistry energy usage with maximum at the fourth, 6 and 10^thh, for Φ_PSII, Φ_NO and Φ_NPQ respectively (Supplementary Figure S13) and the non-photochemical quenching parameters (Figures 6C, D). However, NPQmax (Figures 6C, G), ΦNPQ (Supplementary Figures S13B, E) which while having a maximum between the fourth to sixth hours of the ZT the maximum also coincides with 0.5 LD at the exception of Φ_NPQ 6L:18D where a maximum was reach during the dark period. Regarding pigments, only the β-carotene and Ddx + Dtx synthesis daily oscillations coincide with ZT (Figure 6D and supplementary 16. D), while fucoxanthin and the chlorophylls were not (Supplementary Figure S16).

Fragilariopsis cylindrus grows efficiently under close to natural conditions, i.e., low temperature (0°C) and low light intensity (30 µmol photons m^-2 s^-1), and it shows a maximum growth rate (0.25 ± 0.05 d^-1) in part with previous reports (Arrigo et al., 2010; Kropuenske et al., 2010; Morin et al., 2020; Croteau et al., 2021; Croteau et al., 2022). High F_V/F_M and close C/N and Chl a/C values, as well as similar E_k and E_opt values, between photoperiods witnesses of healthy, non-nutrient limited and steady-state acclimated cultures (Frigstad et al., 2014; Niemi et al., 2015; Kvernvik et al., 2021). Using a light intensity which matches the growth optimum for F. cylindrus (Morin et al., 2020; Croteau et al., 2022), we observed a proportional growth rate increase with daylength until it reached a maximum with 18 h of light, i.e., further increase in daylength showed to have no further benefits (Guérin et al., 2022). This pattern is explained by a precise balance between the pigment content, photochemistry and photoprotection, which ultimately defines a certain level of production and growth at 0°C. F. cylindrus adjusted its light-harvesting and production capabilities with the light dose, i.e., with more light energy available it decreased its Fx content (but not Chl c and β-carotene), P-α as well as P_max, likely by adjusting the Rubisco content (Lacour et al., 2022), in order to maintain its investment in light energy harvesting versus C fixation (Chl a/C), as well as its optimal intensity for C fixation (P-E_k). In parallel, the maximal potential for NPQ (NPQ_max) increased proportionally with the total pool of xanthophylls Ddx + Dtx and Dtx content with constant DES (≈7%). This general pattern of response was achieved by precisely adjusting the PSII photochemical light energy usage versus energy losses as illustrated by similar values for effective (Φ_PSII, Φ_NPQ, Φ_NO) and maximal (F_V/F_M, rETR_max) parameters with 6 h, 12 h and 18 h daylength. This PSII activity adjustment was paralleled with the central maintenance of the light intensity range between the maximal versus the optimal photochemistry (E_opt/E_k).

This general picture however needs to be modulated with two main observations. First, the 6 h additional light from 6 L to 12 L and from 12 L to 18 L did not show the same features. With 12 h light, the light dose acclimation was close to reaching a plateau (reached with 18 h) while with 6 h, F. cylindrus did not reach its maximal acclimation capacity. This was illustrated by a lower Chl a/C and growth rate, when compared to 12L:12D conditions. Our second observation is the fact that when the photoperiod increases from 18 h to 24 h light, the response of cells was not directly related to the increase in light dose (from 2 to 2.67 mol photons m^-2 d^-1), i.e., in addition the cells had to face the absence of regular darkness periods under 24 h day length. The combination of a higher light dose with the absence of regular darkness periods is known to respectively generate a stronger excitation pressure on PSII (Lepetit et al., 2013; Lepetit et al., 2017) and a disruption of photodamaged PSII repair in darkness (Xu et al., 2017), driving imbalance between functional and non-functional PSII, and a decrease in photosynthetic efficiency (Lavaud et al., 2016). Here, it was illustrated by a slightly but significantly lower F_V/F_M, Φ_PSII and rETR_max which was nevertheless not reflected in P_max nor in the growth rate. The somewhat maintenance of photochemistry was ensured by a more effective dissipation of the excess light energy (Φ_NPQ) supported in part by a higher DES and content in Dtx. However, NPQ_max was stable, indicating that additional Dtx molecules were not all involved in NPQ (as illustrated by a lower NPQ_max/Dtx + Dtx), but instead likely participated in direct ROS scavenging (Lepetit et al., 2010; Lavaud and Lepetit, 2013; Schuurmans et al., 2015); and/or the number of Dtx binding sites, provided by LHCx proteins, was stable (Lepetit et al., 2013; Lepetit et al., 2017; Buck et al., 2019). It is noteworthy that the synthesis of additional Dtx molecules was “de novo”, i.e., with no change in Ddx content (Supplementary Figure S17), and it amounted to 1/3 of total Dtx by the end of the 24L:0D treatment monitoring. This feature was reported before in several temperate diatom species (Lavaud et al., 2004; Goss et al., 2006; Blommaert et al., 2017), and was hypothesised to bring additional photoprotection under strong and prolonged light stress (Lavaud and Lepetit, 2013; Blommaert et al., 2017). In parallel, the irradiance range where photochemistry is optimal/maximal was stretched (higher E_opt/E_k and P-E_opt/P-E_k) (Falkowski and Raven, 2013) and the induction of NPQ got closer to E_k (E50_NPQ/E_k closer to 1), and the opposite to E_opt. This overall strategy allowed the cells to make the most of the excess in light energy without modifying their ability to harvest light (no major change in pigment content, in α and P-α), as proposed before (Lacour et al., 2020), driving a smooth transient acclimation towards 24L:0D conditions. However, mitigating constant illumination is more costly for the cells, as highlighted by the greater difference between NPP and GPP, compared to cells growing under light:dark cycle, which resulted in the observed lower growth rate.

Finally, it is worth noting the apparent discrepancy between photochemistry performances and production parameters: while the former are mainly similar among photoperiods (except for 24L:0D), the later showed a more contrasted response with a significantly higher P_max for the shortest daylength (6 h) and decreasing values with increasing daylength. Such pattern suggests that for 6L:18D, the rate of C fixation is likely maximized to compensate for the low light dose in order to provide the necessary energy to meet the needs for growth and nighttime processes (such as PSII repair; Xu et al., 2017). It therefore means that under 6L:18D most of the photosynthetic electrons are likely used for C fixation, while when daylength is longer (12 h–24 h) a larger proportion is diverted to other processes among which NPQ is not the major one as it does not tremendously increase, even under 24 h daylength. Such decoupling between the photochemical electron transport rate and C fixation is not uncommon in cold adapted diatoms (Goldman et al., 2015; Hancke et al., 2015; Lacour et al., 2017; Schuback et al., 2017; Lacour et al., 2022). These authors proposed that part of the photosynthetic electrons is used through alternative pathways to produce ATP and NAPDH before entering the Calvin cycle machinery, thus avoiding the need to reoxidize the newly fixed C, thus allowing a more efficient use of light energy. The evidence of such cyclic electron flow(s) has been specifically reported in F. cylindrus (Goldman et al., 2015; Lacour et al., 2017). This process may play a major role in the acclimation of F. cylindrus, and other polar diatoms, to changes in daylength and light dose, and it supports the above described fine tuning between light harvesting, photochemistry, photoprotection and production (Lacour et al., 2022).

Acclimation of polar diatoms and F. cylindrus to prolonged darkness (from 7 days to 3 months) has been explored before (Mock et al., 2017; Kvernvik et al., 2018; Kennedy et al., 2019; Lacour et al., 2019; Morin et al., 2020; Joli et al., 2024). It showed a drastically different pattern than acclimation to photoperiods or continuous light. While PSII remained functional (high F_V/F_M), the maximal (rETR_max) and effective (Φ_PSII) photochemical efficiency, and the production (P_max), were dramatically lowered. We propose this is due to a decrease in the number of active, and possibly total, PSII (Lavaud et al., 2016), as also suggested by the decrease in β-carotene which is mainly associated with PSII reaction centers in diatoms (Nagao et al., 2020; Nagao et al., 2022). However, F. cylindrus was able to preserve its light-harvesting capacity as illustrated by the stable contents in photosynthetic pigments (Chl a, Chl c and Fx). Nevertheless, most of the harvested light energy was diverted to unregulated passive energy losses (high but similar Φ_NO compared to photoperiods) and NPQ (more than doubled Φ_NPQ), which explains the low remaining fraction used for photochemistry (2–3 times lower Φ_PSII). Because of the slowed photochemistry, and whole metabolism (Mock et al., 2017; Kennedy et al., 2019), a sudden return to light with an intact light-harvesting capacity will likely generate harmful photooxidative stress (Lacour et al., 2019; Morin et al., 2020). To prevent such situation, a high photoprotection potential needs to be ensured (Lacour et al., 2019; Morin et al., 2020). This was illustrated here by the largest pool size of Ddx + Dtx which, together with a high DES, ensured a strong constitutive content of Dtx, only overpassed by continuous light acclimation. As a result, NPQ_max was the highest (with the 18L:6D treatment) and it was directly related to the increased Φ_NPQ (contrary to 18L:6D). Additionally, to ensure an efficient induction of NPQ for weak but potentially stressful irradiances (Lacour et al., 2019; Morin et al., 2020), E50_NPQ was lowered down (32.7 µmol photons m^-2 s^-1) close to the intensity to which the cells were acclimated to, i.e. 30 μmol photons m^-2 s^-1. Here too, but to a lower extent than under continuous light, NPQ_max/Dtx was <1 suggesting that probably many Dtx molecules did not participate to NPQ but served as ROS scavengers (Lacour et al., 2020).

These results well support the crucial role that NPQ plays in allowing diatoms to maintain their light-harvesting capacity and the integrity of their PSII under changing light conditions, an ability that is especially crucial when metabolic rates are slowed down by low temperatures (Lacour et al., 2020). This is a straightforward way to make the best out of the rapidly changing light conditions at the bottom horizon and underneath the sea-ice during the end of winter through spring in the Arctic Ocean (Croteau et al., 2022) (see Figure 1). Especially, it helps polar diatoms to resume their photosynthetic productivity and growth rapidly and efficiently under extremely low light at the end of the polar night (Kvernvik et al., 2018; Randelhoff et al., 2020; Hoppe, 2021).

Most of the photosynthetic parameters (photochemistry and photoprotection) we measured showed repeated daily variations when cells were exposed to a light:darkness alternation with stable low light intensity. It holds true for the photoprotective pigments Ddx and Dtx, and to a lesser extent for β-carotene, but not for the chlorophylls nor Fx. Because of the relatively short duration of the monitoring (52 h) and the low sampling frequency (every 4 h maximum), a reliable cycle analysis of rhythmic oscillations was not possible. Nevertheless, data visualisation (as in Figures 4, 5) shows that photosynthetic parameters values and Ddx, Dtx contents mostly vary, with a rhythm modelled onto the light:darkness alternation over a ca. 24 h period, and that for most of them the period of rhythm depends on the daylength reaching their maximum before/around ZT 6 h, i.e., corresponding to the middle of the “day.” However, due to the shorter daylength, several parameters did not follow this trend for the 6L:18D treatment, especially photoprotection parameters (Φ_NPQ, NPQ_max, Ddx, Dtx) were often controlled by the light dose in these conditions.

The photoperiodic rhythms shown by Φ_PSII, F_v/F_M and rETR_max indicate a fine modulation of photochemistry in F. cylindrus. Interestingly, NPQ_max oscillated in parallel with these parameters, suggesting a strong link between the daily regulation of photochemistry and photoprotection capacity. Moreover, many of the photoprotection parameters (NPQ_max, Φ_NPQ, DES, Ddx and Dtx contents, Figure 6, Supplementary Figures S13, S15) showed a maximum around the middle of the light period, with the noticeable exception of DES under 18L:6D, which maximum was always reached earlier (during the first quarter of the light period). These observations demonstrate that NPQ is not a binary “ON/OFF” mechanism. Instead, it is finely regulated through a specific rhythm that is modelled onto the photoperiod, i.e., both the light:darkness alternation and the light dose. However, when the daylength is shorter, it appears that the modulation of NPQ extent is shifted toward a more direct light dose-dependent regulation.

The circadian rhythmicity of photosynthesis in higher plants, when measured via dynamic chlorophyll fluorescence, reflects an underlying inner (“endogenous”) circadian clock regulation (Haydon et al., 2013; Dodd et al., 2015; Yarkhunova et al., 2018). In unicellular algae, a molecular circadian clock has been described only for a handful of species (Hu et al., 2017), including very few works dealing with rhythmic photosynthetic activity (Harding et al., 1981; Mackenzie and Morse, 2011). In the temperate model diatom Phaeodactylum tricornutum, there have been recent proofs for the existence of such circadian clock (Annunziata et al., 2019). With the present dataset, it is not possible to conclude on the existence of a circadian clock in F. cylindrus. Nevertheless, the question is open about the existence of an underlaying photoperiodic 24 h-based rhythmic regulation of the photosynthetic activity of F. cylindrus. Thanks to the recent sequencing of F. cylindrus genome (Mock et al., 2017), the analysis of F. cylindrus photoperiodic transcriptome will be essential to establish with certainty the nature of the observed rhythms, and the homology of the mechanisms in place in temperate and polar diatoms. Nonetheless, as shown here, non-destructive in vivo chlorophyll fluorescence appears to be an ideal tool for the integrative and comprehensive monitoring of photophysiology rhythms.