In a set of behavior experiments, we assessed if the ontogeny of circadian rhythms in activity during post-emergence development of O. bicornis differs from the one in A. mellifera.

Honey bees were sampled after they were raised by the colony. For this, we reintroduced newly emerged honey bees (overnight in an incubator at 35°C and 60% RH) into the colony after marking them with paint on the thorax and sampled the marked honey bees at specific ages. Solitary bees were submitted to daily cycles of fluctuating temperature and light conditions (12 h of 25°C and light and 12 h of 15°C and darkness) and kept all together until sampling in a big flight tent (size: 60 cm × 60 cm × 56 cm) that was equipped with sugar syrup (APIinvert^®, Südzucker, Mannheim, Germany), pollen, and water ad libitum as well as tubes for providing hiding places and clay. We performed sampling in all groups of bees always at the same time of the day, which resembled the subjective morning (Zeitgeber time 0–1.5).

None of the honey bees in our experiments exhibited circadian rhythms during the first 2 days after their emergence (Figure 1). On the contrary, most of the solitary Osmia bees (mean for all experiments ± STE: 88% ± 0.1) showed circadian rhythmicity in their activity right after their emergence and this was the case under different temperature conditions and for both sexes (Exact binomial test and p < 0.05) (Figure 1 and Table 1). The few individuals that did not show strong rhythmicity died only a few days after their emergence.

The PDF-immunostaining of pre- and post-emerged bees of different ages showed that, in honey bees, the number of PDF neurons steadily increased with age (Kruskal Wallis rank sum test: 19.3, df = 4, p < 0.01; post-hoc test: pairwise comparison using Wilcoxon rank sum test) (Figure 2, left graph). In contrast, we could not detect such increment in Osmia, and the PDF cell number was approximately 14–15 neurons per hemisphere for all ages (Figure 2, right panels). Although we could see a slight increase in the PDF cell number from 2-day old mason bees to older mason bees (Figure 2, right graph) this turned out to be not significant (Kruskal Wallis rank sum test; post-hoc test: pairwise comparison using Wilcoxon rank sum test). Interestingly, in the two-day-old honey bee brains, the PDF cell bodies cluster closer together than it is the case in the older honey bees and Osmia of all ages (Figure 2, second row).

Several earlier studies suggested an important role of the social lifestyle of honey bees in circadian behavior and the normal development of the circadian system (Bloch and Robinson, 2001; Eban-Rothschild et al., 2012; Beer et al., 2016; Fuchikawa et al., 2016). Several studies discovered that the social environment plays an important role in the ontogenetic process of circadian behavior and bees kept in a colony environment during early adulthood could already exhibit circadian rhythms in locomotion at the age of 2 days (Eban-Rothschild et al., 2012; Beer et al., 2016; Fuchikawa et al., 2016). Recently, it has been shown that temperature plays a pivotal role in the ontogeny of circadian activity rhythms as well. Bees kept at 35°C, which corresponds to the core hive temperature, for the first 24 h of their life display rhythms already at the age of 1–2 days while the bees kept at typical laboratory temperatures of 25°C need a few days longer (Giannoni-Guzman, 2016; Nagari et al., 2017). This is not surprising, because the correct temperature is important for a normal brain development and affects behavioral performance in the adult bee (Tautz et al., 2003; Groh et al., 2004; Jones et al., 2005). The honey bee brain further matures during adulthood, which is indicated by an age-dependent plasticity in mushroom bodies (Farris et al., 2001; Groh and Meinertzhagen, 2010). In our experiments (performed at 30°C), all honeybees needed more than 2 days until they exhibited circadian rhythms in locomotion, with or without social contact to a mini-colony. This ontogeny of circadian rhythms is slightly later than reported in the above mentioned studies, that showed circadian ontogeny in young honeybees at the age of 2 days or earlier, when kept under social contact or hive temperature (Eban-Rothschild et al., 2012; Beer et al., 2016; Fuchikawa et al., 2016; Giannoni-Guzman, 2016; Nagari et al., 2017). Nevertheless, our results comply with other behavioral studies indicating that the circadian system is not fully matured in newly emerged honey bees (Moore et al., 1998; Toma et al., 2000; Bloch et al., 2001; Eban-Rothschild et al., 2012; Beer et al., 2016; Giannoni-Guzman, 2016). Maturation speed in young honey bees thereby seems to depend on at least two factors: temperature and colony environment (possibly indirectly because of colony thermoregulation).

The solitary mason bee on the other hand shows circadian rhythms in locomotion right after emergence, which led us to the conclusion that the circadian system of this bee is fully matured at emergence. In contrast to the honey bee, temperature seemed to have no effect on the ontogeny of rhythmicity in solitary bees and most bees were rhythmic in both temperature treatments, which adds to our interpretation that development is largely finished at the time point of emergence of solitary bees compared to honey bees. Similarly, newly emerged fruit flies and tsetse flies, as examples of other holometabolic insects, also showed circadian behavior in locomotion (Brady, 1988; Sehgal et al., 1992).

We found that the number of neurons expressing PDF increased moderately but significantly after honey bee emergence, which supports our hypothesis of an immature circadian clock in newly emerged honey bees. Similarly, Bloch et al. (2003) reported that the number of clock protein PERIOD (PER) expressing cells in the brain of foragers is higher than in the brain of nurses (Bloch et al., 2003). In spite of the fact that the molecular circadian clock ticks already in nurses (Fuchikawa et al., 2017), this early clock appears immature. For example, per mRNA levels are higher in forager brains than in nurses and show circadian cycling only in foragers, while in nurse bee brains, per transcript levels are not rhythmically produced (Toma et al., 2000; Bloch et al., 2004). The lack of cycling or cycling with a diminished amplitude of clock gene transcripts in nurse bees was later proven to be task related and consistent with the lack of rhythmic behavior in nursing bees (Shemesh et al., 2007, 2010). The growing clock network during aging of the adult honey bee, as we have observed it in our study, may promote the integration of new functions into the circadian system. It may be that the growing network also gains complexity during aging. Our recent analysis of the PDF network in nurses and foragers demonstrated that peptide levels oscillated with slightly different phases in different regions of forager brains, while they were highly synchronous in the brains of nurse bees (Beer et al., 2018). This indicates a higher level of complexity in the transfer of time-related information to brain areas with different functions in forager bees, which may relate to the complex tasks of foragers outside the hive. Interestingly, the number of positively stained PDF cell bodies stayed always high throughout the day in forager bees and was highest at the end of the subjective night/beginning of the subjective day in nurse bees (Beer et al., 2018). Therefore, we chose the subjective morning for quantification of the cell number in the differently aged bees in the here presented study. It is not clear, if the difference in cycling of PDF levels in different brain areas depends on developmental stage or task, but the maturation of the circadian clock seems to correlate with the age dependent polyethism in the honey bee. Due to the time dependent PDF peptide levels in PDF neuron arborizations observed in our recent study (Beer et al., 2018), we limited our analysis to counting PDF cell bodies, which was more consistent throughout different ages/tasks of the honey bee (at least for day times with high expression levels). Future studies, under controlled laboratory conditions and in the social hive environment, may dissect, which components of the honey bee clock are subjected to age- or task-related plasticity. Furthermore, the temperature dependency of clock system maturation in adult bees may be investigated also on the neuroanatomical level in the future.

Sociality has evolved in multiple groups, such as mammals (e.g., mole rats), insects (e.g., some bees, some wasps and ants) or even microorganisms, because of evolutionary benefits from group hunting, development of resistance against disease or predators and increased reproductive success (Alexander, 1974; Crespi, 2001). In bees, eusociality (full sociality) was found only in few bee species (e.g., Apis and Melipona species), but social behavior has been found in several other bee species in different nuances (e.g., primitively social bumble bees or facultatively social sweat bees) (Shell and Rehan, 2018). In case of the honey bee, improved colony defense is one clear benefit of sociality, but also the vast amount of offspring production in a colony (Breed et al., 1990).

We propose in our research another adaptation that accelerates colony growth: a differentially timed development in social and solitary bees that results in the emergence of social bees with an immature circadian system. This may provide an evolutionary advantage on both colony and individual level. Even with an immature circadian system, honey bees can survive being under the protection of a social community and they can undertake in-hive tasks in the colony. A mismatch of activity to daytime is not relevant for avoiding predators or foraging success, because the young honey bee is provisioned by the older foragers. An immature circadian clock may be even beneficial to the young honey bee, which takes care of the brood day and night, because honey bees have so far shown a lack of ill effects caused by arrhythmic behavior, while other animals are strongly affected (Bloch et al., 2013).

As we expected, the number of PDF neurons does not change during aging of the solitary bee in our study. This fits, like the activity rhythms in young solitary bees, to our hypothesis of an already fully matured circadian system in the newly emerged solitary mason bee. Similarly, Withers and co-authors found that A. mellifera showed continued brain development during aging, while Osmia lignaria did not (Withers et al., 2008). Proportions of mushroom bodies compared to Kenyon cell bodies in newly emerged Osmia were just as high as in foraging honey bees. Furthermore, the proportions did not change during aging in Osmia, while this was the case in Apis (Withers et al., 2008). Our findings point to a shift of emergence to an earlier developmental stage of the circadian system in the social honey bee as compared to the solitary bee. It is certainly true, that there are other differences in the life style of A. mellifera and O. bicornis and we want to state clearly that a causal association of sociality and difference in clock development cannot be drawn from our results. For example, O. bicornis is a univoltine species, but A. mellifera produces several generations per year (Raw, 1972). Nevertheless, sociality appears to be the most striking difference in life style of these bees and other adaptations may have co-evolved with the specific life style, like for example the almost continuous reproduction in honey bee colonies.

We limited our investigation of clock development in bees to one social and one solitary species out of technical reasons. Especially the clock in solitary bee species is so far extremely little investigated and future studies may show, whether our findings can be generalized to other social and solitary bees. There are a few studies on circadian plasticity in other social hymenoptera. In bumblebees, which display primitive sociality, circadian plasticity in activity and number of PDF expressing neurons has been associated with body size and task rather than age (Yerushalmi et al., 2006; Weiss et al., 2009). For ants, a similar dependency of circadian rhythms on task, like in the honey bee, has been postulated: forager ants display rhythmic behavior and cycling levels of clock gene transcripts, which was not the case in nurse ants (Ingram et al., 2009; Fujioka et al., 2017). In addition, age-dependent modulations of the circadian clock have been reported in ants. Ingram and coauthors (Ingram et al., 2009) determined an increased level of overall per transcript in old foragers of the harvester ant species Pogonomyrmex occidentalis, which is reminiscent of the situation found in A. mellifera (Toma et al., 2000; Bloch et al., 2004; Abreu et al., 2018). The modulation of circadian behavior via contact to other colony members seems to play a role in rhythmic behavior of ants as well, because young ants in contact with old ants showed only weak behavioral rhythms (Fujioka et al., 2019). Interestingly, the clocks of different mammals, including humans, showed progressed maturation after birth like the honey bee (Rivkees, 2003; Weinert, 2005). This raises the question, whether social life style (in particular the protection of the colony) is a prerequisite for evolution of emergence with an immature circadian system, or the other way around or these traits rather co-evolved in honey bees (and possibly other eusocial hymenoptera like ants). Further studies are needed to dissect the evolutionary background of maturation processes of the circadian system in different animals.

Summarized, our comparative study demonstrates a differentially timed clock development of the social honey bee and the solitary mason bee. While the circadian system of the mason bee seems completely matured at emergence, the honey bee clock shows continued maturation after emergence. The early emergence with an immature circadian system benefits the colony community of the social bee and this adaptation of the honey bee clock may have evolved along with this species’ social life style.