Photoperiodic adaptation of aanat and clock gene expression in seasonal populations of Daphnia pulex

ABSTRACT Changes in day-length entrain the endogenous clock of organisms leading to complex responses to photoperiod. In long-lived organisms experiencing several seasons this response of the clock to photoperiod is phenotypically plastic. However, short-lived organisms often experience a single season without pronounced changes in day-length. For those, a plastic response of the clock to different seasons would not necessarily be adaptive. In aquatic ecosystems, zooplankton species like Daphnia live only for some weeks, i.e. one week up to ca. two months. However, they often show a succession of clones that are seasonally adapted to environmental changes. Here, we found that 16 Daphnia clones per each of three seasons ( = 48 clones) from the same pond and year differed in clock gene expression with a homogenous gene expression pattern in ephippia-hatched spring clones and a bimodal expression pattern in summer and autumn populations indicating an ongoing adaptation process. We clearly demonstrate that spring clones were adapted to a short, and summer clones to a long photoperiod. Furthermore, we found that gene expression of the melatonin-synthesis enzyme AANAT was always lowest in summer clones. In the Anthropocene, Daphnia’s clock might be disturbed by light-pollution and global warming. Since Daphnia is a key-organism in trophic carbon transfer, a disruption of its clock rhythm would be devastating for the stability of freshwater ecosystems. Our results are an important step in understanding the adaptation of Daphnia’s clock to environmental changes.


Introduction
An important environmental factor influencing all kinds of organisms is light. On Earth, two rhythms of light exist: the daily rhythm which is caused by the Earth's rotation around its axis and the seasonal rhythm due to the Earth's rotation around the sun. During the course of a single day and the succession of seasons, all organisms experience environmental changes in conjunction with light including rhythms in temperature, resource availability and risk of exposure to predators, pathogens and parasites. In order to respond to changes in the length of day and night, organisms are forced to modulate their biology in a day-time or season dependent rhythm in order to adapt to photoperiod.
In temperate zones, the photoperiod changes with season. The information of photoperiod is often mediated by the duration of the signal by the annual time-keeping hormone melatonin (Lincoln 2006), although this seems not to be the case in Antarctic krill (Pape et al. 2008). In many other organisms however, when nights get longer in autumn, also the melatonin signal is longer. Melatonin is a molecule which can be found in all kinds of organisms from prokaryotes to eukaryotes and has also been detected in the freshwater micro-crustacean Daphnia (Markowska et al. 2009). In most animals, melatonin is produced at night, which is also the case for D. pulex (Schwarzenberger and Wacker 2015). In mammals, it is synthesized in the pineal gland in the animals' brains as a result of rhythmic transcription of genes of the circadian clock (Foulkes et al. 1997). Also in Daphnia the highest concentration of melatonin was detected in the nervous system (Markowska et al. 2009), which is in line with findings for several other invertebrates (Vivien-Roels and Pévet 1993). Until today, more and more functions of melatonin have been found in diverse organisms (reviewed in Poeggeler 1993): These functions include adjustment of the circadian clock to the environmental light regime and anti-stress effects. In Daphnia, melatonin has been shown to reduce the production of resting eggs (ephippia) in crowding conditions (Schwarzenberger et al. 2014) and to affect diel vertical migration (DVM; Bentkowski et al. 2010), which is probably under the control of the circadian clock (Cellier-Michel et al. 2003).
Adaptations to photoperiod have been observed both in terrestrial and aquatic ecosystems and can be found in all kind of organisms. Plants have season-specific flowering times (Jackson 2009). Animals migrate and breed following season-specific cues (e.g. Denlinger et al. 2017). Hibernation in vertebrates (Ruby 2003) or winter diapause and resting-egg production in invertebrate animals is dependent on photoperiod (e.g. Barbera et al. 2013;Denlinger et al. 2017;Schwarzenberger et al. 2020). In the cyclical parthenogenetic Daphnia also the asexual production of males -that are needed for sexual reproduction and ephippia production -is dependent on day-length and light colour (i.e. short days and red light; Toyota et al. 2015).
In order to respond to seasonal changes in photoperiod, two mechanisms are required: first a clock for measuring day-length and second a mechanism for counting the number of short days (a counter). Therefore -although controversially discussed (Bradshaw and Holzapfel 2010;Emerson et al. 2009;Ikeno et al. 2010) -the functional involvement of the circadian clock in seasonal time measurement is likely, but more complex (Saunders 2020) than originally suggested by (Bünning 1936). Indeed, in several insect species the involvement of genes of the circadian clock in photoperiodism was verified (e.g. Denlinger et al. 2017;Nunes and Saunders 1999;Pittendrigh et al. 1984Pittendrigh et al. ,1991Syrova et al. 2003), and only recently it has been shown that genes of the circadian clock play a role in diapause of Daphnia (Schwarzenberger et al. 2020). The production of diapausing ephippia in Daphnia is dependent on short photoperiods (and low temperatures; Carvalho and Hughes 1983). Tilden et al. (2011) found several clock genes in Daphnia that show a cyclic expression pattern over 24 h (Bernatowicz et al. 2016;Coldsnow et al. 2017;Cremer et al. 2022;Rund et al. 2016;Schwarzenberger and Wacker 2015;Schwarzenberger et al. 2021). Daphnia's clock is truly circadian since it persists in complete darkness for several days (Schwarzenberger et al. 2021). The cyclic expression of clock genes of Daphnia is responsible for the rhythmic production of melatonin and genes that code for the melatonin-synthesis enzyme aryl-alkylamine N-transferase (AANAT; Schwarzenberger and Wacker 2015). Melatonin is the transmitter of the signal of the circadian clock (Foulkes et al. 1997) which translates into the rhythmic expression of diverse (clock-related) genes (Cremer et al. 2022;Rund et al. 2016).
Daphnia's circadian clock is adapted to the photoperiod a clone originates from: the strength of expression of clock genes correlates with the clone's original day-length in the growing season, i.e. summer, and translates into the melatonin concentration necessary for the light responses to a certain photoperiod (Schwarzenberger et al. 2021). We hypothesized that seasonal adaptation to photoperiod of subpopulations from spring, summer and autumn from the same pond should be found in Daphnia. Juvenile somatic growth rate is generally a measure of fitness in Daphnia (Lampert and Trubetskova 1996) and it has been found that Daphnia differ in size between seasons (e.g. Brzezinski et al. 2010). Therefore, besides circadian clock and aanat gene expression and AANAT activity we also measured juvenile somatic growth rates, of 16 clones per season ( = 48 clones) in a middle-European summer photoperiod of 16 h light and 8 h darkness at 20°C.

Sampling site and Daphnia pulex clones
At three different points in time (29 th of April, 27 th of June, 9 th of October 2019; respective day-length: 14:05 h, 15:04 h, 11:00 h according to http://solartopo.com/ daylength.htm; respective mean daytime air temperature: 7.4°C, 27.8°C, 11.5°C as according to the German Weather Service, DWD, for the weather station 2712 in Konstanz) we sampled 16 clones per each of three seasonal populations ( = 48 clones) of D. pulex ("spring," "summer," "autumn") from a shallow pond in Konstanz, Germany (N 47.690976, E 9.156476). This pond lies in the middle of a wood and is therefore not affected by any artificial light sources. Across seasons, the pond showed a high density of Chaoborus and other insect larvae and contained many copepods in spring; in summer, also many Notonecta were present. In spring, no hatched/living D. pulex were present in the shallow pond. Instead, many ephippia floated on the surface probably due to their release from the sediment e.g. by digging waterfowl. This means that there was no newly established Daphnia population after the winter break, and the pond was in the state before selection of Daphnia individuals by environmental factors such as light. Therefore, we collected floating ephippia as these represent the potential establishing population rather than sediment-buried ones. Each ephippium was then individually transferred to 100 ml vessels (i.e. one ephippium per vessel) with filtrated (< 0.2 µm) Lake Constance water in a light-dark cycle of 16:8 h at 20°C.
From each ephippium a single D. pulex hatched after a few days. In summer and autumn we collected live clones that we equally separated in 100 ml glasses (i.e. one clone per vessel). In autumn only few D. pulex individuals were present in the pond. The 16 individual clones of the three seasonal populations ( = 48 clones) were fed with Acutodesmus obliquus ad libitum every second or third day. As soon as the first clutch was released, three to five neonates were transferred to new medium. The clones were cultivated for at least three generations before they were used in the experiments in order to exclude maternal effects. Only thirdclutch neonates that were born within 12 hours were used in the experiments.

Set-up of the experiment, gene expression and AANAT-activity
We grew ten new-born individuals per each of the16 clones of each season ( = 48 clones) in 200 ml filtrated (0.2 μm) Lake Constance water a 16:8 h light-dark cycle for four days at 20°C. They were fed with 2 mg C L −1 of A. obliquus every other day. The individual clones of each season served as biological replicates (i.e. 16 clones = 16 biological replicates per season). After four days, 4 animals per clone were dried for the determination of growth rate and one or two animals were frozen at −80°C for subsequent RNA extraction. Those animals were sampled one hour after the beginning of the night (the point in time when D. pulex gene expression should peak (Schwarzenberger and Wacker 2015). The residual four animals per clone were sampled and frozen (−80°C) four hours after the beginning of the night for the measurement of AANAT-activity (which should be the point in time with the highest AANAT-activity; Schwarzenberger and Wacker 2015).
For each clone we measured the juvenile somatic growth rate (as according to Schwarzenberger et al. 2021), the expression of four clock (clock, cryptochrome 2, period, timeless) and three melatonin synthesis genes (AANAT 1, 3 and 5) and AANAT-activity. The methods for AANAT activity measurement, RNA extraction, cDNA transcription and gene expression measurement via qPCR are described in Schwarzenberger and Wacker (2015).

Statistics
After verifying homogenous variances (Levene's test), the juvenile somatic growth rates and AANAT-activities were analyzed via one-way ANOVA and Tukey's HSD post-hoc tests. ANOVAs and the linear correlations between AANAT activity and aanat gene expression were calculated with the program STATISTICA (StatSoft, Inc. 2011, version 10.0, Tulsa, OK, USA). The level of significance was p < 0.05.
Levene's tests were applied to test for homogeneity of variances of gene expressions between seasonal populations. Since gene expression variances showed inhomogeneity and the gene expression differed immensely between the clones within each of these two seasons (see large standard deviations), calculation of ANOVAs was not feasible.
PCAs were conducted using the program R (version 3.6.1). Data points, i.e. individual Daphnia clones, were either dismissed from the analysis when gene expression data of one gene was missing, or the missing gene expression was substituted with the mean of the respective seasonal population. This was done to keep the original number of clones in the analysis, when in some cases only one single gene expression was missing. The clones of each season were analysed for deviations of their gene expression of single genes from the unimodal distribution by using the excess mass test of Ameijeiras-Alonso et al. (2019) and the R package "multimode."

Juvenile somatic growth rates and AANAT-activity
The juvenile somatic growth rates ( Figure 1) differed between seasons. In spring and summer, the growth rate was similarly high, whereas the autumn population showed a significantly lower growth rate (Tukey's HSD after one-way ANOVA F 2,44 = 245.3, p = 0.014). In contrast, AANAT-activity of the populations (Figure 1) did not differ between seasons (one-way ANOVA F 2,50 = 229.5, p = 0.696).

Gene expression and bimodality
When comparing the gene expression between the three seasonal populations, we found that the variances in summer and autumn were inhomogeneous and within each of both seasons the gene expression differed immensely between the clones. We conducted a principal component analysis (PCA, Figure 2) which separated both the summer and the autumn Daphnia population into two gene expression clusters, and exhibiting a bimodal pattern. This indicates that each of the seasonal summer and autumn populations, respectively, is composed of two subpopulations regarding clock and aanat gene expression. The first principal component (PC 1) explained 50.2% of the variation in the data and separated the spring from the two summer subpopulations, which were correlated with higher expressions of mainly the three aanat genes. The PC 1 additionally separated the two subpopulations in summer and    Figure 2. PCA across all 48 D. pulex clones and genes of the three seasonal populations with data points for which 11 missing expressions of single genes was substituted by the average of the respective seasonal (sub)population. Independent of substituting or excluding the mentioned data points or Daphnia clones, the separation pattern remained unchanged with the same bimodal gene expression pattern in summer and autumn (compare suppl. Figure S2). Clones belonging to apparent different subpopulations are indicated by filled and empty symbols.
autumn, respectively. The PC 2 explained 29.1% of the variation in the data, separated the autumn populations from the others roughly by higher gene expression of timeless and clock and lower expression of cryptochrome2 and aanat 5. The PC 3 explained further 16.8% of the variation, but did not add additional separation of the seasonal populations. In contrast to growth rates and enzyme activities, which all showed unimodal distributions throughout all seasons (suppl. Figure S1, Table S1), the PCA clearly shows that the gene expressions of the subpopulations of both the summer and autumn populations were significantly different (suppl. Figure S1, Table S1). The subpopulations within one season (summer and autumn) showed a bimodal distribution whereas the spring population was unimodal. The pattern of the PCA persisted if we dismissed 10 data points because of 10 missing expression data of single genes (in some cases the gene expression analysis failed which led to a clone number of lower than 16, see methods for detail, and suppl. Figure S2 for comparison).
To verify whether the deviation from a unimodal distribution was significant, we applied excess mass tests for the growth rates, enzyme activities and gene expressions of each single tested gene and for each of the populations isolated from the three different seasons (suppl. Figure  S1, Table S1). The spring population did not show a bimodal pattern in any of the genes, whereas both the summer and the autumn population were clearly bimodal for each gene (suppl. Figure S1, Table S1).
Both apparent subpopulations in summer showed lower gene expressions of the three different aanat genes than the homogenous spring population (Figures 2 and 3). This was also the case for clock. Half of the two apparent autumn subpopulations had a higher gene expression than spring or summer populations, whereas the other half had a gene expression intermediate between spring and the summer subpopulation with the higher gene expression (clock, aanat 1 and 3). In the autumn subpopulations aanat 5 gene expression was comparable with gene expression of the two summer subpopulations.
The period, timeless and cryptochrome2 gene expression of the spring population and one of the two summer subpopulations was comparable, whereas the other summer subpopulation showed lower gene expression than spring (Figures 2 and 3).
Concerning timeless, both autumn subpopulations showed higher gene expression than spring and summer, whereas the two autumn subpopulations showed a slightly lower cryptochrome2 gene expression than spring. The cryptochrome2 gene expression of one autumn subpopulation was comparable with one of the summer subpopulations, whereas the other one was intermediate between the two summer subpopulations. For period, one of the autumn subpopulations showed a gene expression higher than spring and summer, whereas the other subpopulation was intermediate between the two summer subpopulations.

Discussion
Clonal succession of Daphnia over time and seasons has often been observed in lakes and ponds (Pfrender and  Spaak 1996). This succession can be a result of seasonal changes in temperature and population density (Carvalho and Crisp 1987), anoxia (Geedey et al. 1996), the presence of vertebrate and invertebrate predators (Stibor and Lampert 2000), putatively the occurrence of protease-inhibitor producing cyanobacteria (Schwarzenberger et al. 2013), etc. One of the most pronounced seasonal changes is the change from short to long photoperiods and vice versa. It has been demonstrated that a marine Antarctic crustacean (i.e. krill) shows adaptations to seasonal photoperiod that are timed by adjustment of endogenous clock gene expression (Piccolin et al. 2018a(Piccolin et al. , 2018b. Similarly, a marine copepod has been demonstrated to differ in clock gene expression in the active (long photoperiod in summer) in comparison to the diapause phase (short photoperiod in winter; Häfker et al. 2018). To our knowledge, a seasonal adaptation of the endogenous clock has not been investigated in other crustaceans, e.g. the freshwater crustacean Daphnia.
It has been demonstrated that a marine copepod shows differences in clock gene expression when sampled at two different latitudes at close to the summer solstice (Hüppe et al. 2020). Only recently, we could show that also D. magna clones are adapted to the latitude they originate from (Schwarzenberger et al. 2021). Therefore, we hypothesize that within one pond the change in photoperiod over a year might similarly results in a succession of clones that possess a seasonally adapted clock. Another possibility might be that clones are phenotypically plastic in their clock's response to different photoperiods (and temperatures).
Here, the D. pulex clones that were sampled in three different seasons differed in growth rate when cultivated in a Middle European summer photoperiod at 20°C. Spring and summer clones had on average a higher growth rate than autumn clones. Interestingly, this finding differed from literature where Brzezinski et al. (2010) found that summer clones had a lower growth rate than clones from spring. This discrepancy might be a result from the use of different Daphnia species. Another possibility might be that the different freshwater systems did (Lake Roś; Brzezinski et al. 2010) or did not (our pond) contain fish predators. Fish have a high density in summer and fish cues lead to a lower size at first reproduction (e.g. Effertz and Von Elert 2014;Stibor and Lampert 2000). Furthermore, an adaptive smaller body size was observed in Daphnia clones derived from habitats in which fish were present (De Meester et al. 1999).
Since we could exclude maternal effects and all clones received the same food quality and quantity, the physiological difference between the three seasons demonstrates that a succession of clones over a year had actually taken place. However, it was unclear whether this succession also led to clones that were adapted to their respective seasonal photoperiod. Therefore, in the next step we measured AANAT activity and aanat gene expression.
We did not find a correlation between AANAT activity and aanat gene expression. As demonstrated earlier, Daphnia clones that are adapted to their specific latitude do not differ in AANAT activity when they are grown in the same photoperiod (16:8 h light-dark cycle, 20°C; Schwarzenberger et al. 2021). Nevertheless, they show differences in the underlying aanat gene expression with a lower gene expression in clones that originate from southern latitudes (e.g. longer photoperiods during Daphnia's growing season in summer; Schwarzenberger et al. 2021).
We could observe the same phenomenon in our seasonal clones: The gene expression of all aanat genes was lower in both summer subpopulations (long photoperiod) than in spring, whereas the AANAT activity remained constant in the light-dark cycle of 16:8 h.
Since aanat gene expression of Daphnia follows clock expression (Schwarzenberger and Wacker 2015), it was not surprising that the same lower gene expression in summer in comparison to spring was also found for cryptochrome2 and clock, and in one of the summer subpopulations for period and timeless. Interestingly, the other summer subpopulation showed a similar period and timeless gene expression as the spring population; this probably captures the shift of the spring population (e.g. through clonal succession) into a summer population adapted to a shorter photoperiod. Furthermore, this also explains the bimodal gene expression pattern of the summer population that is in the course of adaptation.
We expected a switch-back to higher gene expression (as in spring) from summer to autumn because photoperiods shorten after summer solstice. Actually, this could be observed for timeless for both autumn subpopulations, and for clock, period, aanat 1 and 3 in one of the autumn subpopulations while the other autumn subpopulation showed a similar or lower (period) gene expression as in summer. This not only means that we captured the switch from a summer population adapted to long photoperiods to an autumn population that is in the course to adapt to shorter photoperiods; we also found that the single genes adapt independent of each other (i.e. in different subpopulations) to a change in photoperiod. Since the change of seasons happens with a slow change in photoperiod lengths, also the succession of clones adapted to those different photoperiods is not a harsh selection process but is slow and takes place in intermediate stages (cf. the bimodal pattern). This also explains why the subpopulations of autumn and summer showed a gene expression pattern comparable to the summer population for cryptochrome2 and aanat 5 whereas the other clock genes already show an average gene expression higher than in summer.
We hypothesize that the gene expression of the autumn clones -at a later point in time -will be similar to gene expression of newly hatched spring clones. This is because ephippia are probably produced by the previous autumn population, which would lead to the observed pre-adaptation of the spring population with homogenous and higher than in summer gene expression.

Conclusions
Our findings for aanat and clock gene expression clearly demonstrate that summer clones showed a different gene expression pattern than the spring population from which they originated. The gene expression of summer clones was in most cases lower than in spring, which demonstrates that the summer population was seasonally adapted to a longer photoperiod. The spring population probably originated from pre-adapted parents since they showed a homogenous distribution of clock and AANAT gene expression, whereas the adapting summer and autumn populations are highly variable (with a range of bimodal gene expression pattern between clones).
Whether there is a phenotypically plastic response of the clock to different photoperiods and temperatures (in addition to seasonal succession of clones) could not be investigated. Unfortunately, because of the university's lockdowns in spring and winter 2020 due to the COVID-19 pandemic all clones died out and could not be tested in other photoperiods and temperatures.
In light of climate change, the effect of temperature on the endogenous clock will become highly important for adaptation of organisms. It has been shown that both the circadian light and temperature cycle are necessary for entrainment of behavioural and molecular rhythms of Drosophila melanogaster (Yoshii et al. 2009). Furthermore, it has been demonstrated that the fitness of northern populations of the mosquito Wyeomyia smithii that are adapted to a northern photoperiod declined harshly when transplanted into a southern photoperiod and a southern temperature regime (Bradshaw et al. 2004). Additionally, a mid-latitudinal photoperiod and temperature regime prevented the timely entry of southern populations into diapause. Therefore, the expected northern migration of hightemperature adapted organisms will depend on their clock's ability to rapidly adapt to a northern photoperiod.
Adaptation to photoperiod is challenging in the Anthropocene. Increasing anthropogenic light pollution affects both terrestrial and aquatic ecosystems (reviewed in Grubisic 2018). Light pollution, i.e. Artificial Light At Night (ALAN), disturbs molecular and behavioural light responses and affects species interactions. In case of Daphnia, it has been shown that urban light pollution reduces diel vertical migration (Moore et al. 2000), and that ALAN of different light colours interferes with antipredator defence, eyesize and the development of the body and the tail-spine length (Li et al. 2022). Although ALAN clearly changes clock gene expression it also enhances Daphnia's ability to cope with dietary cyanobacterial protease inhibitors by increasing its digestive protease activity (Cremer et al. 2022). It remains to be tested which effect of both light pollution and increased temperature will act on seasonally and locally adapted Daphnia populations.

Data availability statement
All data can be found in the Supplementary Table S2