Widely rhythmic transcriptome in Calanus finmarchicus during the high Arctic summer solstice period

Summary Solar light/dark cycles and seasonal photoperiods underpin daily and annual rhythms of life on Earth. Yet, the Arctic is characterized by several months of permanent illumination (“midnight sun”). To determine the persistence of 24h rhythms during the midnight sun, we investigated transcriptomic dynamics in the copepod Calanus finmarchicus during the summer solstice period in the Arctic, with the lowest diel oscillation and the highest altitude of the sun's position. Here we reveal that in these extreme photic conditions, a widely rhythmic daily transcriptome exists, showing that very weak solar cues are sufficient to entrain organisms. Furthermore, at extremely high latitudes and under sea-ice, gene oscillations become re-organized to include <24h rhythms. Environmental synchronization may therefore be modulated to include non-photic signals (i.e. tidal cycles). The ability of zooplankton to be synchronized by extremely weak diel and potentially tidal cycles, may confer an adaptive temporal reorganization of biological processes at high latitudes.

Transcriptomic rhythms persist in C. finmarchicus during the midnight sun period Daily transcripts dominate in ice-free environments and at lower latitude Ultradian transcripts are revealed under sea-ice and at higher latitude Reorganization of the transcriptome might be an adaptation to under-ice tidal cues

Introduction
The day/night cycle structures biological processes from gene expression to physiology and behavior (Helm et al., 2017;Mermet et al., 2017). Organisms may respond directly to external stimuli (exogenous) or indirectly (endogenous) via the internal circadian clock. This molecular mechanism enables organisms to track changes in their environment by using the highly predictable light/dark cycle as a Zeitgeber (time-giver) although other clocks are known to synchronize to other monotonous cycles (i.e. tidal, lunar, and annual). Endogenous clocks are of adaptive advantage because they enable organisms to anticipate and prepare for predictable environmental changes by temporally organizing short-and long-term biological processes (Helm et al., 2017). However, it is still unclear how this temporal biological organization is facilitated in organisms inhabiting extreme photic environments. In high latitude marine environments without overt day/night cycles such as during the midnight sun period in the Arctic, the sun remains above the horizon for days or months (Abhilash et al., 2017;Bertolini et al., 2019;Bloch et al., 2013;Schmal et al., 2020). Entrainment of the circadian clock by light and associated rhythmic gene oscillations is therefore considered unlikely (Schmal et al., 2020). The persistence of daily rhythms is particularly questionable during the summer solstice, which represents the paroxysmal period of midnight sun, with the lowest diel oscillation and the highest altitude of the sun's position above the horizon (Schmal et al., 2020).
C. finmarchicus is a member of the ''Calanus Complex'', which constitutes up to 80% of the zooplankton biomass in the Arctic ocean (Søreide et al., 2008). Copepods provide a crucial trophic link between primary production and higher trophic levels, with significant impact on biochemical cycles via the biological carbon pump (Giering et al., 2014;Sanders et al., 2014;Søreide et al., 2008). Moreover, this key planktonic species has been shown to expand its habitat range poleward tracking isotherms as a consequence of climate change (Reygondeau and Beaugrand, 2011). As a consequence it will be exposed to greater annual photoperiodic ranges to which it has evolved (Reygondeau and Beaugrand, 2011), with unknown impacts on its phenology and fitness (Huffeldt, 2020;Saikkonen et al., 2012). A functional circadian clock has been described in this species under clear light/dark cycles (Hä fker et al., 2017), and a recent study revealed circadian clock gene transcript oscillation in C. finmarchicus in the high Arctic during summer solstice,

Results and discussion
Evidence of rhythmic transcriptomes during the summer solstice in the high Arctic We sampled copepods at 74.5 N (south station, ice-free) and 82.5 N (north station, ice-covered) (hereafter referred to South and North respectively) on the 30 E longitude, within nine days of the summer solstice ( Figure 1A). During the sampling periods, the sun remained above the horizon all day at both stations but still showed diel oscillations of altitude ( Figure 1B) and photosynthetic active radiation (Hü ppe et al., 2020). iScience Article North experienced lower diel oscillations of the sun's altitude due to the higher latitude and proximity to the summer solstice when compared to South. Furthermore North was under snow covered sea-ice, attenuating light irradiance and spectral composition of the water column (Wallace et al., 2010). Both stations exhibited semidiurnal tidal cycles (~12.4h), with slightly higher tidal amplitudes at North. Net sampling at both stations was performed at 4h intervals over 24 h and the gene expression of C. finmarchicus (CV stage) was analyzed on the transcriptomic level for each time point and station using RNA sequencing as described in   (Figure 1B).
Sequencing yielded a depth of 70.4 million reads per sample, optimizing the detection of rhythmic transcripts (Hughes et al., 2017;Li et al., 2015b) in C. finmarchicus which has a large genome (Choquet et al., 2019). For each station, temporal expression of 76,550 transcripts was obtained. These data were analyzed for significant periods at 20h and 24h (hereafter termed ''daily'') and 12h and 16h (hereafter termed ''ultradian'') periods, assuming that they are the results of an endogenous clock regulation or a direct response to environmental factors (Helm et al., 2017) (Table 1).
Our analysis yielded a total of 18 045 (23.6% of total transcripts) and 12 634 (16.5% of total transcripts) rhythmically expressed genes at South and North stations respectively (adjusted-p-value < 0.001, Table 1). The number of rhythmic transcripts achieved a total of 52 175 at South and 45 243 at North, by increasing the adjusted-p-value cutoff to 0.05, representing 68.2% and 59.1% of the total transcriptome respectively (Table 1). Representing the first in situ day-scale transcriptomic rhythm analysis in the Arctic Polar region, the results revealed a substantial temporal organization at the transcriptomic level in C. finmarchicus during the time of summer solstice, when daily changes in the sun's altitude are at a minimum, near the lowest anywhere on the planet. A comparable study from the Antarctic Polar region shows that about 600 genes (1.9 % of total transcripts tested) oscillated in Antarctic krill Euphausia superba during an Antarctic summer day (Pittà et al., 2013). In the current study, with higher sequencing depth and a more powerful sampling strategy (Hughes et al., 2017;Li et al., 2015b), we show that the number of genes rhythmically transcribed in C. finmarchicus in the absence of light/dark cycles is comparable to other marine invertebrates (Biscontin et al., 2019;Connor and Gracey, 2011;Pittà et al., 2013;Satoh and  Both stations revealed significant daily and ultradian cycling transcripts (adjusted-p-value < 0.001, Table 1 and Figure 1C). This result corroborates with the bimodal oscillations of circadian clock gene transcripts recently described by Hü ppe et al. (2020) and validated by the current transcriptomic analysis ( Figure S1), characterized by both daily and ultradian rhythms. Interestingly, a lower number of daily transcripts were detected at North, where they were 2.1 times less numerous than at South (Table 1 and Figure 1C) which may be attributable to lower diel oscillations of solar altitude and the presence of sea-ice cover, and corroborates with the lower amplitude of daily oscillation of clock, period 1 and timeless at North (Hü ppe et al., 2020). To increase understanding of daily transcripts at both stations, phase of gene expression with that of the environmental cycles was determined, i.e. the time of peaks of expression during the reporting period (Figures 2A-2D). Differences in phase of daily transcripts was observed between South (Figures 2A and 2C) and North ( Figures 2B and 2D). While daily transcripts at South could be defined as ''nocturnal'', as most peak expressions occurred when solar irradiance reached its daily minimum between 22h and 7h, most of the daily transcripts at North peaked between 14h and 23h, when solar altitude and irradiance was decreasing (Figures 2A-2D). A consistent phase shift of expression between stations was observed in the expression of the positive regulators of the circadian clock: clock peaking at 6h at South and 19h at North; cycle peaking at 22h at South and 19h at North (Hü ppe et al., 2020). In addition to being more numerous at South, the proportion of daily transcripts with high amplitudes (1.5-5; >5) is greater at South than at North ( Figure 2E).
In contrast to the daily transcripts, an increase of ultradian transcripts was observed at North, where they were 1.9 times more numerous than at South (Table 1 and Figure 1C). Thus, the rhythmic transcriptomes at South and North were characterized by different daily/ultradian ratios: 83.5% were daily and 16.5% were ultradian at South, while 55.6% were daily and 44.4% were ultradian at North (Table 1 and Figure 1C). These ratios reflect the pattern of circadian clock gene transcripts, for which an increase of ultradian oscillations was clearly observed at North (Hü ppe et al., 2020). Ultradian transcriptomic rhythms have been increasingly reported over the past decade in a wide range of species (Ananthasubramaniam et al., 2018;Biscontin et al., 2019;Connor and Gracey, 2011;Hughes et al., 2009;Pittà et al., 2013;Satoh and Terai, 2019;Schnytzer et al., 2018;Tarrant et al., 2019b;Westermark and Herzel, 2013;Zhu et al., 2018). These oscillations often cycle at different harmonics of the circadian rhythm, and among these, the~12h oscillation is most prevalent (Ananthasubramaniam et al., 2018;Hughes et al., 2009;Westermark and Herzel, 2013;Zhu et al., 2018). Further, ultradian transcriptomic oscillations of~12.4h, also called (circa) tidal oscillations, are observed in marine organisms under the influence of semidiurnal tidal cycles (Connor and Gracey, 2011;Mat et al., 2020;Satoh and Terai, 2019;Schnytzer et al., 2018). Here, ultradian transcripts phased to tides at both South and North, with two different phase patterns depending on the station . At South, most of the ultradian transcripts showed a peak of expression with low tides ( Figures  2A and 2C), while at North, most of the ultradian transcripts showed a peak of expression with high tides (Figures 2B and 2C). In contrast to circadian transcripts, the proportion of ultradian transcripts with high amplitudes (1.5-5; >5) is equivalent to (16h transcripts) or greater (12h transcripts) at North than at South ( Figure 2E).
To further compare South with North, we analyzed if the same genes were rhythmically transcribed at both stations. The results showed that a large proportion of the total rhythmic transcripts (adjusted-p-value < 0.001) were specific to each station (67.1% at South; 52.9% at North), with 12 101 transcripts being exclusively rhythmic at South and 6 690 exclusively rhythmic at North ( Figure 1D). These station-specific rhythmic transcripts might reflect differences between ice-free (South) and ice-covered (North) ecosystems, leading to differential physiological requirements. For example, in ice-covered areas where phytoplankton in the water column is scarce, the sea-ice algae community is a critical food source for copepods (David et al., 2015;Søreide et al., 2008;2013;Wallace et al., 2010), the nutritional quality of which differs to phytoplankton blooms in ice-free waters (Falk-Petersen et al., 1998). In contrast, the 5 944 transcripts that (E) Amplitude ranges distribution of daily (24h and 20h period lengths, adj-p < 0.001) and ultradian transcripts (16h and 12h period lengths, adj-p < 0.001) at South (green) and North (blue), expressed as percentage of rhythmic transcripts per period length and station. An amplitude of 0.5 means that the difference between the minimal and the maximal levels of expression is equal to 0.5 times the minimal level.

OPEN ACCESS
iScience 24, 101927, January 22, 2021 5 iScience iScience Article were rhythmic in animals sampled at both stations, could reflect common physiological requirements (Figure 1D). Of these genes, about half showed distinct changes in their period ( Figure 1E) with most (2 388) changing from daily at South to ultradian at North. We propose therefore, that the pattern of temporal regulation of common physiological processes is specific to each environment and that gene expression is re-organized according to an under-ice habitat.

Rhythmic biological processes of interest
Gene ontology (GO) analysis revealed that rhythmic transcripts at both latitudes were particularly common to metabolic and cellular process, signaling, response to stimuli, localization or biological regulation (Figures S2A,S3A,and S4A). To gain a better understanding of the rhythmic biology in C. finmarchicus, nine key biological processes particularly observed in rhythmic transcripts (Table S1) were selected based on enrichment analysis (results are presented Figures S2B, S3B, and S4B) to get a more detailed insight into their temporal regulation ( Figure 3). Examples of genes from functional groups presented in Figure 3 are provided in Table S2 and Figure 4. By analyzing the rhythmic status of genes involved in these nine key biological processes (Figure 3), we noted that: (1) a combination of daily and ultradian transcripts were observed for each process at both South and North, rather than daily-and ultradian-specific processes; (2) an increase of ultradian oscillations across all biological processes examined, except ''DNA repair'', occurred at North, explained by both station-specific and common rhythmic transcripts switching from daily at South to ultradian at North and; (3) the time of peaks of expression (phases) according to daily or tidal cycles is specific to each station.
Rhythmic patterns of ''cellular response to stimulus'' supported the observation of the persistence of a daily rhythmic environmental stimulus at both stations, despite the sun always staying above the horizon. Interestingly, over the common rhythmic transcripts between both stations, 128 present daily oscillations at both stations ( Figure 3A, both (sa.), full mustard yellow), while 87 switch from daily oscillations at South to ultradian ones at North ( Figure 3A, both (sw.), striped cyan). Finally, while 87% of rhythmic transcripts associated to ''cellular response to stimulus'' have daily oscillations at South, with peaks of expression between 22h and 7h ( Figure 3C), the proportion of daily transcripts decrease to 56% at North, peaking between 10h and 23h ( Figure 3E). In parallel, the proportion of ultradian transcripts increases from 13% at South ( Figure 3C) to 44% in the North ( Figure 3E), with clear peaks of expression around high tides at this station (North) ( Figure 3E). These results support the idea of the increasing response to an ultradian environmental stimulus (e.g. tides) at North. This trend is even more accentuated in the rhythmic transcription of genes involved in ''neurotransmitter metabolic process and transport'' (Figures 3B, 3D, and 3F), for which the proportion of ultradian transcripts is more evident than the one of daily transcripts at North. Indeed, the daily/ultradian ratio differs from 89%/11% at South (Figures 3D) to 42%/58% at North (Figure 3F). Neurotransmitters are involved in a wide range of processes comprising temporal organization, such as photic entrainment of the circadian clock, or transmission of clock outputs such as the circadian food anticipatory activity (Golombek and Rosenstein, 2010;Gotow and Nishi, 2008;Patton and Mistlberger, 2013). Among rhythmic transcripts involved in ''neurotransmitter metabolic process and transport'' in this study, 3 isoforms of Sodium-dependent nutrient amino acid transporter 1 (NAAT1) are identified (Figures 4A and Table S2B). This amino acid/sodium cotransporter that promotes absorption of essential amino acids has been shown to be expressed with a daily rhythm in circadian neurons of Drosophila (Abruzzi et al., 2017). In our study, one isoform of NAAT1 is rhythmically expressed exclusively at South, with a daily rhythm and a peak of expression at 6-7h ( Figure 4A and Table S2B). A second isoform shows ultradian oscillations at both latitudes, but showing a phase shift, the peak of expression being at rising tides at South and around high tides at North ( Figure 4A and Table S2B). Finally, a third isoform of NAAT1 is also rhythmically expressed at both latitudes, but with a daily rhythm at South and an ultradian rhythm at North ( Figure 4A and Table S2B). This result clearly illustrates the environment-dependent modulation of gene expression in terms of period and phase of rhythmic expression. In contrast, the ''DNA repair'' function, widely described to be under the control of the circadian clock in other species (Borgs et al., 2009), showed the majority of daily oscillations at both South and North. This result suggests that iScience Article the increase of ultradian oscillations at North may not be beneficial for all functions. However, despite the general common daily pattern for this function, a clear phase shift is noted between stations, with peaks of expression between 22h and 7h at South ( Figure 3C), and between 10h and 23h at North ( Figure 3E), highlighting again a clear environment-specific modulation of cellular processes. While chronobiological data are still scarce in marine organisms, some DNA repair genes have been identified to peak during nighttime in the mussel Mytilus californiaunus under temperate region's natural environment simulation in the laboratory (Connor and Gracey, 2020). Another function strongly represented by rhythmic transcripts is ''protein phosphorylation and ubiquitination''. The role of posttranscriptional mechanisms in rhythmic regulatory processes is increasingly demonstrated (Mauvoisin et al., 2015;Mermet et al., 2017). Thus, rhythmic transcripts involved in ''protein phosphorylation and ubiquitination'' suggests continued cyclic regulation at the proteomic level, with again a clear increase of the proportion of ultradian patterns at North. We also show clear rhythms in ''oxidation-reduction process'' and ''response to oxidative stress'' at both stations. The redox status of organisms involved in many cellular reactions, from respiration to metabolism, appears to be widely rhythmic in all species and all these functions are related to the endogenous clock (Biscontin et al., 2019;Eckel-Mahan and Sassone-Corsi, 2009;O'Neill et al., 2015;Pittà et al., 2013;Putker and O'Neill, 2016). While daily oscillations of redox markers are observed in terrestrial organisms, tidal oscillations, such as cytochrome oxidase subunits expression, have also been observed in marine organisms such as the crustacean Eurydice pulchra (O'Neill et al., 2015). In the current study, the increase of ultradian regulation at North can be illustrated by the switch of the rhythmic expression of cytochrome c oxidase subunit 2, from daily at South to ultradian at North ( Figure 4B and Table S2E). However, ultradian oscillations are not limited to the North station, as illustrated by the ultradian expression of an isoform of Peroxiredoxin-6 at both stations, peaking just after low tides at South, and just after high tides at North. . Indeed, while daily transcripts are in the majority at South, representing 84%, 76%, and 86% of rhythmic genes involved in ''carbohydrate metabolic process'', ''lipid metabolic process'', and ''proteolysis'', respectively ( Figure 3C and 3D); the proportion is much more nuanced at North, where the ultradian transcripts even become the majority, representing respectively 69%, 52%, and 52% of rhythmic transcripts at this station ( Figure 3E and 3F). These results indicate day-scale oscillations in energetic demands and nutrient supply, with clear modifications of period and phase of oscillations according to stations ( Figures 3C, 3D, 3E, and 3F). For instance, isoforms of Glucose-6-phosphate-1-dehydrogenase (carbohydrate metabolic process, Figure 4C and Table S2G), identified to be under the control of the circadian clock in Drosophila (McDonald and Rosbash, 2001), and Leukotriene A-4 hydrolase (proteolysis, Figure 4E and Table S2I), oscillate with a daily pattern at both stations, with clear phase shifts (peaking respectively at 6-7h and 2-3h at South and at 22-23h and 18-19h at North). In contrast enolase transcription is ultradian at both stations ( Figure 4C and Table S2G), with again a clear phase shift (peaking after low tides at South and at high tides at North). This gene encodes for a protein involved in glycolysis and has been shown to have a peak of expression during the night in the Antarctic krill in the lab under L/D exposure (no tides), corresponding to the high level of activity and oxygen consumption in this species in the field (Biscontin et al., 2019). Finally, isoforms of Trehalase ( Figure 4C and Table S2G), Calpain-7, Carboxypeptidase B and Aminopeptidase ( Figure 4E and iScience Article C. finmarchicus suggesting that the widely rhythmic transcriptome is tuned to the cyclic environmental conditions of the prevailing habitat.
Persistent cycling of C. finmarchicus transcriptome during midnight sun In high latitude marine environments without overt day/night cycles, entrainment of the circadian clock by light and associated rhythmic gene oscillations is considered unlikely (Bertolini et al., 2019;Schmal et al., 2020). However, even during months of permanent darkness or illumination, Polar marine regions remain rhythmic environments, with a persistence of the sun's oscillations below or above the horizon (Cohen et al., 2020;Hobbs et al., 2018;Wallace et al., 2010). Here, the widely daily rhythmic gene oscillations observed at two high Arctic latitude stations during summer solstice, illustrate that subtle daily changes of light intensity or quality are sufficient to synchronize daily molecular rhythms in the key zooplanktonic species C. finmarchicus, which is consistent with the high levels of visual acuity recorded in this species (Bå tnes et al., 2015).
Some studies show that several Arctic species exhibit daily activity rhythms in the absence of diel light cycles, while others become arrhythmic proposing that the absence of rhythms could be beneficial in polar environments (Abhilash et al., 2017;Bertolini et al., 2019;Bloch et al., 2013). In zooplankton, the most described daily rhythm is the behavior of diel vertical migration (DVM) to the surface at night in order to balance the need to feed close to the surface against the accompanying risk of predation by visually hunting predators (Hä fker et al., 2017). DVM has been frequently observed during autumn and spring in the high Arctic when the diel light/dark cycle is present (Dale and Kaartvedt, 2000;Fortier et al., 2001). However, data for synchronized DVM during the Arctic midnight sun are contrasting (Blachowiak-Samolyk et al., 2006;Cottier et al., 2006;Dale and Kaartvedt, 2000;Fortier et al., 2001;Wallace et al., 2010). While the persistence and the purpose of maintaining DVM during this period is under debate, the multitude of daily transcripts observed in this study, including those involved in circadian clock machinery, carbohydrate/lipid metabolism, and proteolysis, suggests that a daily temporal organization at the transcriptomic level could be an advantage for copepods, whether ex-or intrinsic (Abhilash et al., 2017).
However at extremely high latitudes and under sea-ice, gene oscillations become re-organized to include <24 h (ultradian) gene cycles. Entrainment of the circadian (or other) clock(s) and clock-controlled genes may therefore be modulated to include other, non-photic signals (i.e. tidal cycles) (Connor and Gracey, 2011; Mat et al., 2020;Satoh and Terai, 2019;Schnytzer et al., 2018). Interestingly, some genes belonging to eukaryotic translation initiation factors and heat shock proteins are shown to be ultradian in this study, while these genes families have been shown to present conserved harmonic oscillations (ultradian rhythms generated by the circadian clock) between fungi and mammals (Ananthasubramaniam et al., 2018) (Table  S3). Here, we propose that ultradian rhythms in C. fimarchicus may be entrained by ambient tidal cues such as potential current reversal, food availability, turbulence, salinity or temperature cycles caused by the tides (Massicotte et al., 2020;Oziel et al., 2019;Tessmar-Raible et al., 2011). While ultradian oscillations are also observed at the South sea-ice-free station, under-ice currents at North station could lead to important tidal cycles of food availability (from ice algae), salinity or temperature (Massicotte et al., 2020;Oziel et al., 2019). Thus, the tidal reorganization at North may facilitate for example ingestion rate (Conover et al., 1986;Ibáñ ez-Tejero et al., 2018;Petrusevich et al., 2020;Schmitt et al., 2011), as suggested by the increase of ultradian oscillations of key metabolic processes in copepods at this station.
The bimodal aspect of the C. finmarchicus transcriptomes, presenting both daily and ultradian oscillations, as well as a differential daily / ultradian ratio between stations, were in accordance with the bimodal oscillations of circadian clock transcripts ( Figure S1, Hü ppe et al., 2020). This corroborated the hypothesis of Hü ppe et al. (2020) that the circadian clock could be functional during summer solstice at high latitudes and, as proposed in other species (Enright, 1976;Tran et al., 2020), could be synchronized by both daily and tidal environmental cycles, with a balance between one or the other depending on their relative importance and the associated advantages to synchronize biological processes accordingly, relevant to each ecosystem. Moreover, the station-specific phases of daily and ultradian transcripts compliment the idea of the ability to adapt to site-specific changes of risks and opportunities related to the daily and tidal environmental cycles. Thus, the observed plasticity of rhythmic transcriptomes could be of high adaptive advantage to deal with the specificity of each habitat, and could allow C. finmarchicus to adapt to the high Arctic environmental cycles, unrestrained by photoperiod (

Limitations of the study
As discussed above, in a context of field study, organisms are exposed to environmental cycles. Thus we cannot rule out that the observed rhythmicity stems from a direct response to light, rather than a clockcontrolled regulation. However the consistency between the circadian clock genes expression and the transcriptomic patterns highly suggests a functional clock. Moreover, the endogenous clock(s) controls different layers of regulation to provide robust timing cues at the cellular and tissue level. Here we identified temporal patterns in periodic gene expression by measuring mRNA accumulation. However, the temporal regulation is a dynamic process, including regulation of posttranscriptional mechanisms such as translational efficiency or protein accumulation (Mermet et al., 2017). Thus, further studies at the proteomic or physiological levels are necessary to decipher the exact timing of key biological processes mentioned in this study.

Resource availability Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Laura Payton (laura.payton@uni-oldenburg.de).

Materials availability
This study did not generate new unique reagents.

Data and code availability
The transcriptomic data sets generated during this study are available in the NCBI Bioproject PRJNA628886 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA628886) and in the figshare collection 5127704 (https://doi.org/10.6084/m9.figshare.c.5127704). All the scripts supporting the current study are available from the corresponding author on request.

Methods
All methods can be found in the accompanying Transparent methods supplemental file. iScience Article the morphological identification of copepods and contributed to the manuscript; S.V performed the samples preparation for RNA sequencing; B.M. developed and coordinated the German part of the CHASE project, designed the study, and contributed to the manuscript. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Declaration of interests
We have no competing interests. finmarchicus) and stage (CV stage copepodites) at 2˚C under a stereo microscope using morphological characteristics. To distinguish C. finmarchicus from its closely related Arctic congener C. glacialis, the redness of the antenna, which has been shown to be a good indicator in the regions, was particularly used . Morphological identification method was validated by molecular species identification on a subset of samples from the same stations. DNA was extracted from individual copepods using the HotShot method (Truett et al., 2000), and the species-specific nuclear insertion/deletion (InDel) marker G-150 was amplified using a modified protocol from Smolina et al. (2014). Identification was done by accessing the size of the resulting amplicon via electrophoresis on a 2% agarose gel. 99 % of the individuals identified as C.
finmarchicus by the morphological identification method were also identified as C. finmarchicus by the molecular identification method (n=305 individuals).

Study sites characteristics
Sampling was conducted during Cruise JR17006 of the RRS James Clark Ross in summer 2018 at two stations along a latitudinal gradient. The South station was located in the southern Barents Sea (B13; 74.5°N, 30°E) and the North station in the Nansen Basin (JR85; 82.56°N, 30.85°E) (Figure 1). Water depth at South was 360 m and at North was 3700 m. During sampling times, the ice edge was located at about 81°to 82°N, roughly following the shelf slope north of Svalbard, thus South station (74.5°N) was ice-free, whereas North station (82.56°N) was located within the ice cover. The sun's altitude was always above the horizon but still showed diel oscillations of altitude above the horizon from 7.7°at midnight to 38.6°at midday at South, and from 16°at midnight to 30.9°at midday at North, at the times of sampling (local time, UTC +2). Sites were exposed to semidiurnal tide regimes, i.e., 2 tidal cycles per day, with a maximum amplitude of ± 0.36 m at South and ± 0.47 m at North at times of sampling. Atmospheric PAR measurements and additional physical characteristics of the sampling sites are available in the publication of Hüppe et al. (2020).

Field sampling time series
The sampling strategy was specifically designed for the detection of rhythmic transcripts. Sampling covered a complete 24h cycle at 4h intervals, resulting in seven time points per station. At each station, sampling was performed in similar times frames: 14-15h, 18-19h, 22-23h, 2-3h, 6-7h, and 10-11h (all times noted in local time (UTC+2)). Sampling at South station started on 30 th June (9 days after the summer solstice) at 14-15h and ended on 1 st July at 14-15h. Sampling at North station started on 18 th June (3 days before the summer solstice) at 10-11h and ended on 19 th June at 10-11h. Field sampling interval was conducted on a near semilunar cycle (12 days apart) to ensure tidal and solar cycles were in phase. At each timepoint the water column was sampled from 200 m to the surface with vertical hauls of a WP2 plankton net (opening ⌀: 57 cm, net length: 236 cm, mesh size: 200 µm) with a meshed bucket cod end (mesh size: 200 µm) at a speed of 0.5 m*s-1. Transferring the animals from the net into the stabilization solution was done within less than 12 minutes for all samplings. A 12h period of incubation at 2 -4˚C was allowed to soak the samples thoroughly with the RNAlater stabilization solution (Ambion, UK) before they were transferred to -80˚C for further transport and storage.

RNA extraction
Copepods were sorted at 2˚C and for each timepoint and station, 3 replicates of 15  (Lenz et al., 2014) and quantification matrix was generate thanks to samtools idxstat (Li et al., 2009) results. The percentage of mapped (single) reads is 95.5 ± 0.001 %, and the percentage of multimapped reads is 2.6 ± 0.05 %, attested for a good alignment of the dataset on the reference transcriptome. The 76 550 transcripts with more than 1 cpm were selected over the 206 012 transcripts for further analysis. The reference transcriptome was annotated with diamond v0.9.22 (Buchfink et al., 2015) against Swissprot, Trembl and NR. Only the best hits of each database were selected if i) the percent of the query length covered by the alignment was higher than 60% ; ii) the percent of the subject length covered by the alignment was higher than 40%; iii) the percent of identity of the alignment was higher than 40%.

Real-time quantitative PCR verification
RNA sequencing results were verified through RT-qPCR analysis ( Figure S1). Raw data of expression from Hüppe et al. (2020) of six circadian core clock genes (clock, cycle, period1, timeless, cryptochrome2, vrille) and 2 circadian clock-related genes (cryptochrome1 and doubletime2), relative to the geometric mean of the most stable reference genes (elongation factor 1-and 16s rRNA), were analyzed on the exact same samples analyzed in this study. In parallel, these clock and clock-related transcripts were identified in the transcriptomic results based on Christie et al. (2013). Profiles of expression were plotted and statistical rhythmic analysis were performed on both analyzes (RNA sequencing and RT-qPCR) and showed strong similarities ( Figure S1), comforting the relevance of the rhythmic transcriptomic analysis.

Environmental parameters
Information on the location of the sea ice edge at the time of sampling at North were obtained from ice concentration maps available from the meereisportal (Grosfeld et al., 2016). Modeled data of sun altitude were obtained from the United States Naval Observatory (https://aa.usno.navy.mil/data/docs/AltAz.php, USNO, USA) Information on the tidal dynamics have been drawn from the TPX08 model (Egbert and Erofeeva, 2002) by using the OTPS package (Tidal Prediction Software, http://www-po.coas.oregonstate.edu/~poa/wwwpo/research/po/research/tide/index.html), via the mbotps program (MB-System; Caress and Chayes, 2016).

Rhythmic analysis
Rhythmic analysis of transcripts over the 24h cycles was performed in RStudio (Version 1.2.1335, R version 3.6.3, R Core Team, 2013), using RAIN package. RAIN was specifically designed to detect rhythms in biological datasets independent of waveform by using a non-parametric approach (Thaben and Westermark, 2014). The 76 550 transcripts of each station were tested together (153 100 total transcripts). The false discovery rate of the p-values was corrected using the Benjamini-Hochberg method (Benjamini and Hochberg, 1995). The time series have been tested using the three samples per timepoint as replicates and the "independent" mode. Based on the sampling plan and RAIN algorithm terms of use, the following periods were tested: 24h, 20h, 16h and 12h. Period lengths of 20h and 24 h were in the circadian range, while period lengths of 12 and 16h were in the ultradian range. In this study, gene expression oscillating in a circadian or ultradian period range will be called "daily" and "ultradian", respectively, assuming that they are the results of an endogenous clock regulation or a direct response to environmental factors (Helm et al., 2017). For each period and each transcript, the waveform yielding the most significant result was selected. Then, for each transcript, the period yielding the most significant result was selected. The associated phase was determined by RAIN algorithm. When the period length was shorter than the time frame (24h), the second and third phases were deducted according to the RAIN model. Heatmaps were plotted in RStudio using heatmap3 package (https://www.rdocumentation.org/packages/heatmap3). Relative levels of expression were calculated for each transcripts by normalizing levels of expression to the median of the seven time points. For visualization purposes, relative levels of expression above 1 were binned to 1. For each heatmaps, transcripts were ordered by phases. Amplitudes of oscillating transcripts were calculated as ((maximum value -minimum value) / minimum value) (Hughes et al., 2012;. Normalization by the minimum value makes it possible to be exempt from the differences in the level of expression of each transcript. For the RT-qPCR verification of RNA sequencing (6 core circadian clock genes and 2 circadian clock-related genes, Figure S1), the period yielding the most significant result in each period range (circadian and ultradian) was selected, thus assuming two hypothetical significant period lengths per transcripts.

Rhythmic transcripts at both South and North
All transcripts Daily at both stations Ultradian at both stations Switch of period range at North Ultradian at both stations Switch to ultradian at North (daily at South) Daily at both stations Switch to daily at North (ultradian at South)