Bumblebee colonies are typically founded in spring by individual fertilised queens and are subsequently engaged in a race against time to maximise colony growth and ultimately the production of more queens and males while favourable conditions last 1 2 . In temperate regions, therefore, the first bumblebee workers typically leave the nest to forage as soon as it is bright enough, often even at dawn before the actual sunrise and the last foraging trips of the day are concluded around sunset 3 . The foraging activity of the workers is synchronized to the available photoperiod by endogenous circadian clocks 4 5 .

Given that foragers use the entire available photoperiod in temperate regions, one would expect that they would also fully utilise longer or even continuous photoperiods prevailing north of the Arctic Circle during summer to gather as much food as possible and thus maximise colony growth. Other animal species in that area, such as reindeer, become arrhythmic during the summer, when the sun is continuously above the horizon 6 7 . Previous studies have found an impressive sociality-dependent plasticity in the circadian rhythms of social insects, with the ability to alter the expression of clock genes to fit the demands of the colony 8 9 10 . Hence we might expect that arctic bumblebees would adapt to the foraging conditions that prevail under the midnight sun.

To answer the question whether bumblebee foragers utilise the available photoperiod under continuous daylight conditions, we conducted field experiments with free-flying bumblebees during midsummer in northern Scandinavia. The sun did not set for weeks during our experiments, giving the foragers the opportunity to forage around the clock. We used fully automatic radio-frequency identification (RFID) to monitor the foraging activity of hundreds of individual B. terrestris workers around the clock for several weeks and analysed their circadian foraging rhythms. Since B. terrestris is not native that far north, we also recorded the traffic at the nest entrance of a native B. pascuorum nest.

A total of 1,049 workers were radio-tagged in our experiments. The RFID readers never recorded 164 of these workers leaving the nest, indicating that these workers spent their time entirely with tasks inside the nest. The data generated by the remaining 885 workers were used for the analysis of the circadian foraging rhythms on the colony level. On the individual level, the data of another 504 workers were discarded from the circadian analysis because they did not meet the requirement of at least six consecutive days of activity, leaving a total of 381 workers for individual analysis.

Although bumblebees experienced 24 h of bright daylight in the field, foraging activity was restricted mainly to the times from 08:00 to 23:00, with almost no activity between 01:00 and 06:00 (Figure 1). Activity levels started to increase between 06:00 and 07:00, rising steadily during the morning and reaching the maximum during midday before they decreased again (Figure 1). Thus, the colonies exhibited a robust diurnal rhythm in their foraging patterns (Figure 2). The same was true on the individual level. A mean of 95.2% ± 1.2% (mean ± 1 SEM) of individual workers showed a robust diurnal rhythm in their foraging activity, with a mean period of 23.8 h ± 0.04 h (Figure 2). The foraging pattern of the native B. pascuorum colony was similar to the ones observed in the introduced B. terrestris colonies, with the foraging activity being restricted to the period between 06:00 and 22:00 and no recorded activity during the 'night' (Figure 3).

In 2007, light intensity was measured between 1 July 2007 and 8 July 2007. In that period, the hourly mean values varied between 2,670 lux ± 520 lux at 02:00 and 162,110 lux ± 6,490 lux at 13:00. Individual measurements reached a maximum of 275,560 lux on a sunny day and dropped below 1,000 lux only during one cloudy night with a minimum of 820 lux. Mean hourly ambient temperature in the same time period varied from 9.9°C ± 0.7°C at 02:00 to 18.4°C ± 1.3°C at 12:00, with individual measurements ranging from 6.6°C to 23.9°C and a mean daily fluctuation of 9.8°C ± 0.9°C.

In 2008, light intensity was measured from 5 July 2008 to 13 July 2008. The hourly mean light intensity varied between 1,280 lux ± 110 lux at 02:00 and 132,650 lux ± 17,750 lux at 14:00 (Figure 4a). Individual measurements stretched from 740 lux to 192,980 lux. Light intensity dropped below 1,000 lux during two nights only. Mean hourly ambient temperature ranged from 5.4°C ± 0.5°C at 02:00 to 13.2°C ± 0.8°C at 13:00 (Figure. 4b). Individual measurements varied from 3.6°C to 18.2°C, with a mean daily fluctuation of 8.3°C ± 0.5°C.

Our results show that bumblebee foragers do not utilise the 24-h foraging period under constant daylight conditions in northern Scandinavia but exhibit robust diurnal rhythms in their foraging patterns instead. In previous carefully controlled laboratory experiments 4 , foragers showed free-running circadian rhythms both in constant light conditions (LL) and constant darkness (DD), with mean free-running periods (τ) being significantly shorter in LL than in DD (LL: τ = 22.4 h at about 2,300 lux; DD: τ = 24.0 h). Free-running circadian rhythms in laboratory LL conditions have previously also been found in the related honeybee (Apis mellifera) 11 12 13 . The rhythmic behaviour of bees under LL conditions is in contrast to findings in the fruit fly Drosophila melanogaster, a model organism for chronobiological studies 14 15 16 , which become completely arrhythmic in LL 17 . In Drosophila, the blue-light receptor cryptochrome (dCRY) is largely responsible for the photic entrainment of the circadian clock 18 19 20 21 . The decoding of the honeybee genome revealed that honeybees lack a Drosophila-type CRY but encode a vertebrate-like protein (CRY2) instead 22 . CRY2 has also been found in other insect species, such as butterflies and mosquitoes 23 , as well as in the North American bumblebee Bombus impatiens 24 . In contrast to dCRY, CRY2 is not photosensitive 24 and thus might not function as a circadian photoreceptor, which poses the yet unanswered question how light enters the circadian system of bees.

The fact that the bees did not show free-running rhythms in the prevailing continuous daylight conditions indicates that some external cue, or Zeitgeber, must have entrained the molecular clock of the bees in the field, the most prominent cue being natural cycles in light intensity and associated temperature fluctuations. Although the sun always stayed above the horizon during our experiments, levels of light intensity varied drastically during the course of the day. Laboratory studies have shown that, on average, a minimum of 3.2 lux is needed for Bombus terrestris workers to be able to fly 25 and they have been observed leaving the nest at light intensities below 50 lux on summer mornings in Germany (RJS, unpublished data). During most '"nights"' of our experiments, light levels stayed well over 1,000 lux during the darkest part (Figure 4a) and fell below 800 lux only during the last week of the experiment. Thus it was never too dark for the bumblebees to forage, however the gradually decreasing light levels in the evening could have triggered the daily resetting of the circadian clock of the foragers and studies on Drosophila have shown that the absolute levels of light are less important than the relative light/dark light intensities for rhythm entrainment 26 27 .

In line with the changes in light intensity there were daily cycles in ambient temperature (Figure 4b). Although circadian rhythms are normally temperature compensated 28 , circadian clocks can be synchronised by temperature cycles. In D. melanogaster and the honeybee, daily temperature cycles of just 3°C and 6°C, respectively, are enough to entrain circadian rhythms in their locomotor activity 29 30 31 32 . Thus, daily temperature cycles during our experiments were in a range where temperature entrainment might be possible, although experiments on arrhythmic mutant Drosophila flies have shown that a functional circadian clock is indeed needed for the anticipation of daily temperature changes but is not essentially required for the generation of diurnal activity rhythms under such conditions 33 . The minimum temperatures recorded in our field experiments were as low as 3.6°C during exceptionally cold nights. However, Bombus terrestris workers have been observed foraging at ambient temperatures as low as 3°C during late winter in London, UK, and the foraging rates they achieved under these conditions were similar to those found during summer 34 . Therefore, the temperature during our experiments did not drop below a threshold that would have made it impossible for the bumblebees to forage and the foragers should have been able to gain a net energy profit in these temperatures.

Another possible Zeitgeber are daily cycles in light quality, i.e. in the spectral composition of light, especially in the UV range. Daily changes in ultraviolet radiation can be used to entrain and influence the circadian rhythms in other species 35 . Figure 4c shows the hourly means of UV-A radiation (320 nm to 410 nm) recorded at the field station during seven days of our experiments in 2008. Like other bees, bumblebees are UV-blue-green trichromats and can thus see UV light 36 37 . A recent study in bumblebees has found a new photoreceptor organ that expresses UV-sensitive opsins in a brain area that has been associated with circadian clock activity in other insects 38 , so it might be possible that the circadian clock of bumblebees can be entrained by daily cycles in UV radiation levels 39 .

Finally, daily variance in the availability of floral resources could have been another Zeitgeber. However, in that case we would expect that the foragers would carry on searching for food during the first few nights after being set up in the field before they gradually adjust their foraging patterns to the diurnal pattern of resource availability, which was not the case.

The native B. pascuorum colony observed at the field station showed the same behaviour as the introduced B. terrestris colonies, indicating that our results were not an artefact of using a temperate species in permanent daylight conditions. Field observations at flower patches during summer nights in northern Scandinavia also found a drop in foraging activity around midnight 40 . Thus it seems that even native bumblebees are not adapted to fully use the whole available photoperiod, which is surprising given that the vegetation period at these latitudes is quite short and colonies have relatively little time to grow. Reindeer (Rangifer tarandus) have adapted to these extraordinary conditions by loosing their molecular circadian clock 41 . But staying inside the nest during the nights, although it is bright enough to forage, could be an adaptive behaviour, especially for native bumblebees since colonies in the far north are generally quite small 42 and the foragers might therefore be needed inside the nest during the cooler nights to warm the brood. The cold nights might especially affect species like B. pascuorum, which build their nests on or just below the soil surface 43 . Another possible adaptive reason for extended resting periods might be positive effects on the memory of the foragers. Previous studies have found a mammalian-like sleep behaviour in honeybees 44 45 . Honeybees compensate for sleep deficit by intensifying the sleep process during the following night 46 , which makes it unlikely that they show this sleep behaviour just to save energy during the night and it also has been shown that sleep deprivation affects the consolidation of memory in bees 47 .

Our results show that bumblebee foragers do not use the whole 24 h foraging period under the continuous daylight conditions that prevail north of the Arctic Circle during summer. Foragers show robust diurnal rhythms under these conditions, which means that their circadian clocks must be entrained by some external cue. The fact that even foragers of bumblebee species native to the Arctic are inactive during the '"night"' indicates that this might be an adaptive behaviour. More studies are needed to identify the adaptive reasons for the observed foraging patterns and the Zeitgeber for synchronising the circadian clock of the foragers under continuous daylight. There might also be a synergism between several external cues as has been suggested previously for light intensity and temperature in honeybees 30 .

LC and RJS conceived of the study and designed the experiments. RJS carried out the experiments and data analysis. RJS and LC drafted the manuscript. All authors read and approved the final manuscript.