Honeybees are master thermoregulators, capable of maintaining nest homeostasis across fluctuating ambient temperatures. When workers must cool their nest, they use multiple thermoregulatory behaviors (e.g., fanning, collecting water), but bearding, where hundreds to thousands of workers evacuate their nest and form a bivouac outside, is relatively unexplored. Here, we (1) describe natural bearding patterns, (2) experimentally manipulate colonies to determine what impacts beard size and timing, and (3) explore how workers dissipate back into their nest. We show that bearding occurs daily in hot weather, but the largest beards consistently happen in the evening/night, between the hours of 1800 and 2400. Beards are located around the nest entrance, but workers bias their position toward the shaded side of the nest box. As colony size increases, beard size and duration also increase, but the proportion of the colony bearding does not increase with colony size. Colonies with and without brood still cast beards; brood presence/absence did not impact beard size or duration. After noticing that beards tend to dissipate at sunrise, we experimentally showed that beards induced in the afternoon dissipate within 1–2h, whereas beards induced in the evening remain overnight (10+h). Bearding overnight, however, does carry risks for developing brood inside, as nest temperatures dropped below the optimal range, until the beard dissipated at sunrise. What cues workers use to depart the beard remain unknown, but experimentally illuminating colonies at night did not induce beards to dissipate. Our results suggest that bearding is an individual decision, not one that is coordinated across the colony. Still, these individual actions result in a dramatic collective response that colonies employ to reduce the temperature of their nest. Here, we show how and when colonies use bearding, despite its risks.
]]>Homeostasis is critically important for biological function in a variable environment, and organisms use a combination of physiological and behavioral responses to maintain a consistent body temperature (Smith 1979;Clench 1966). For example, an organism can respond physiologically by sweating, but also behaviorally, by seeking shade to avoid sun exposure. This combination of physiology and behavior helps define the range of ambient temperatures in which individual organisms can successfully thermoregulate, and survive (Muñoz and Losos 2018). When organisms gather into groups, they can also incorporate social behaviors to thermoregulate (Sukhchuluun et al. 2018), which can be as simple as huddling together in cold weather (Le Maho et al. 1976). As groups become more integrated, so too can their suite of thermoregulatory behaviors.
Superorganisms are a unique level of biological organization, where a group of organisms form a cooperative unit to propagate their genes (e.g., colonies of ants, bees, wasps, and termites; Wheeler 1911, Hölldobler andWilson 2009;Szathmáry and Smith 1995). In these highly integrated societies, the collective actions of the individual workers can give rise to thermoregulation at the colony-level (Seeley 1995). Colonies of the Western honey bee (Apis mellifera) are a classic example of a thermoregulating superorganism, because their developing brood must be maintained between 33 and E. B. Rowe and P. Prathibha have contributed equally to this work.
36 °C (Jones et al. 2005). Despite their brood's narrow temperature range, colonies of A. mellifera have an incredibly wide thermal niche, and are capable of thermoregulation in ambient temperatures as cold as -40 °C and as hot as 50 °C (Johnson 2023). This wide thermal range is made possible by their collective behavior.
The behaviors that A. mellifera workers use to warm the nest are relatively simple-adult bees cluster around the brood, and generate heat with their thoracic muscles (Southwick 1983). To cool the nest, workers spring into action with multiple complementary behaviors. Workers fan their wings inside the nest and at the entrance, creating air currents to cycle out hot air, and replace it with cooler air from the environment (Gravish et al. 2015;Cook et al. 2016;Kaspar et al. 2018;Peters et al. 2017). Foragers collect water, which is spread throughout the nest for evaporative cooling, and even stockpiled in cells or the bees themselves (Kühnholz and Seeley 1997;Ostwald et al. 2016). The brood area, typically dense with workers, empties as workers move frantically throughout the nest (Jhawar et al. 2023). Some workers perform a behavior called heat shielding, where they use their bodies to absorb heat around the brood and will then disperse away from the brood (Starks et al. 2005). Finally, in a robust collective spectacle, workers evacuate the nest en-masse, aggregating outside the colony in a "beard", sonamed because the bees clustered around the nest entrance resemble a beard (Fig. 1a).
Although many honey bee thermoregulatory behaviors have been studied extensively (water collection; Kühnholz and Seeley 1997;Kleinhenz et al. 2003;Ostwald et al. 2016, heat shielding;Starks et al. 2005;Bonoan et al. 2014, fanning;Cook et al. 2016;Kaspar et al. 2018), bearding is relatively unexplored (Johnson 2023). Colonies form beards when exposed to a heat stress, particularly when water is unavailable (Ostwald et al. 2016), but general bearding patterns in unmanipulated colonies are largely unknown. We also know that a colony's internal state impacts thermoregulation broadly (e.g., colonies without brood have a more variable metabolic rate than colonies with brood; Southwick 1982), but it is unclear how colony state may influence bearding (e.g., colony size and brood status). Finally, given that bearding bees settle outside the nest, separated from workers and nest conditions within, it is unknown how bees in the beard know when to return to the nest (e.g., when brood temperature returns to normal).
The goal of this paper is to describe natural bearding patterns, and to experimentally manipulate colonies to understand what impacts the size and timing of beards. Our research investigations are: (1) How does beard size change throughout the day? (2) How does colony state (number of adult workers; presence of brood) impact bearding? (3) When do workers in the beard return to the nest?
We performed our experiments between May and August 2023 at the Smith Bee Lab at Auburn University (32.674 N,. Honey bee colonies (A. mellifera) were sourced from an apiary containing 20-25 colonies, all initiated as 10,000-bee packages with a mated queen (Gardner Apiaries, Baxley, GA). For each experiment, we matched colonies according to queen status (all queenright), distribution of worker ages (workers obtained from multiple frames), and similar nest contents (brood, pollen, and nectar/honey stores). All colonies were free from any visible signs of disease (e.g., chalk brood, American foulbrood, and deformed wing virus; Hansen 1987). Colonies housed in nest boxes (as opposed to observation hives, see below) were kept in one of two apiaries: the "concrete" apiary, or the "grass" apiary. In both apiaries, we positioned all colonies with nest entrances facing south, and fully exposed to the sun. We used the concrete apiary to raise ambient temperatures, so that colonies would be more likely to beard. When peak ambient temperatures were already high (> 40 °C), we kept colonies in the grass apiary, located 200 m from the concrete apiary. For any given experiment, we kept all colonies in the same apiary, in the same type of nest boxes (5-frame Langstroth "deep").
To study the dynamics of bearding behavior, we established a standard method to measure beard size. Bearding bees form a bivouac around the nest entrance, but this cluster can aggregate beneath the nest box, making it difficult to estimate the number of bees that have evacuated the nest and joined the beard (Fig. 1a). To accurately measure the number of bees in the beard, we modified our nest entrances with plywood, creating a flat vertical surface upon which the bees could aggregate (see Fig. S1). This also helped standardize beards across colonies housed in 3-dimensional nest boxes and colonies housed in observation hives.
To estimate the number of bees in a beard, we photographed the beard from four directions (facing the nest entrance, left side, right side, from above; with ruler for scale, see Fig. S1). We identified bearding bees by their behavior, which includes a characteristic stance (immobile, flattened upon the surface, often hanging in aggregations with other bearding bees). This makes them easily distinguishable from other bees located around the entrance (e.g., foragers quickly walk inside, guards stand alert with front legs lifted, and fanning bees have beating wings). For each group of bearding bees (e.g., a single colony may have bees clustered above and below the nest entrance, but also on the left and right sides of the nest box), we used ImageJ (Schneider et al. 2012) to measure beard volume (beard area × depth). We then divided beard volume by the average volume of a honey bee to estimate the total number of bees in the beard (average volume of a bee in a beard: 0.35 cm 3 ; length: tip of head to tip of abdomen, height: base to top of thorax, width: side to side of thorax). Bees on the nest entrance, but not clustered together, were counted from the photographs and added to the estimate. If the beard size was small and direct counting was feasible, we counted the number of bees in the beard in the field during our observation. Our cutoff for a colony having a beard was 25 bees, which does not include forager traffic, or guards at the entrance. If the number of bees on the entrance was below the 25-bee cutoff, we considered the number of bees in the beard to be zero. Note that, in some experiments, we are working with small colony sizes (~ 2000 bees), in which case a beard size of 25 bees represents over 1% of the colony's workforce.
We first investigated the daily pattern of bearding in unmanipulated colonies (7-10 June; 31 July-3 August 2024). For 12 total colonies (concrete apiary n = 6; grass apiary n = 6), we measured beard size every 3 h for 66 h. All colonies were kept in 5-frame boxes, with a standardized entrance (2 cm × 2 cm). We measured apiary temperature using J-type thermocouples and a HOBO 4-channel logger (Onset Computer Corporation, Bourne, MA) taped to the west side of a colony in the center of the apiary. We estimated colony size for each colony at the start of the observation period, and immediately after the final observation, using the Liebefeld method (Imdorf et al. 1987). At the start of this experiment (7 June 2024; timepoints 1200 and 1500), bees in the beard continued to cluster behind the modified entrances, making it difficult to estimate the total number of bees in the cluster (we could still record the presence/absence of a beard). We removed these two timepoints from the analysis involving beard size (but not beard presence/absence), and adjusted the entrances to prevent the clustering problem.
To experimentally test the impact of colony size on bearding, we established colonies with 2000, 6000, or 10,000 workers (12 colonies total; four of each colony size. These are the same colonies as used for Investigation 1). We kept all colonies in 5-frame boxes, with two frames of brood, two frames of resources (honey, nectar, and pollen), and one empty comb frame. All colonies were given 48 h to acclimate before we began measuring beard size every 3 h for 66 h ("measuring beard size" method above). We monitored brood temperature using broodminder temperature loggers (Broodminder, Stoughton, WI) placed in the center of the colony. We measured colony size again at the end of the experiment to account for any changes during the experiment (Imdorf et al. 1987), and used this final measurement as our estimate of each colony's size. One 2000-worker colony failed to thermoregulate its brood (hive temperature outside the viable 33-36 °C), so we removed this colony from our subsequent analyses as it was not indicative of a viable colony.
To experimentally test the impact of brood presence on bearding, we established colonies with and without brood. All colonies (n = 8; four per treatment group) were established in 5-frame boxes with 5000 workers. We gave colonies in the "brood" treatment group four bee-frames with a mixture of eggs, larvae, and pupae, and one frame of resources (honey, nectar, and pollen); colonies in the "nobrood" treatment group were given two frames of resources and three empty comb frames. All colonies were given 48 h to acclimate, after which we measured beard size every 3 h for 66 h. We then estimated colony size (Imdorf et al. 1987) to account for changes during the experiment.
To directly compare the effect of brood presence on beard size for each colony, we then swapped the brood status of each colony, and repeated the experiment ("brood" colonies became "no-brood" colonies, and vice versa). Colonies were given 24 h to acclimate before we resumed beard measurements (every 3 h for 66 h). After the final beard measurement, we re-estimated colony size, and used this estimate of colony size for the second round of beard observations (i.e., each of the 8 colonies is represented in both the "brood" and "no-brood" treatment groups, but we account for any changes in colony size between the two trials).
We next explored how heat stress occurring at different times of day (afternoon versus evening) impacted both the formation and dissipation of beards. For this experiment, we housed colonies in two-frame observation hives, so that we could reliably induce a controlled heat stress (each colony contained 2000 workers and a mated queen; observation hives as in Seeley 1995). To heat stress the colony, we positioned two 250 V heat lamps, one on either side of the observation hive, 38 cm from the glass walls. To maintain a consistent heat stress, and to avoid melting the wax combs, we turned the heat lamps off when the internal temperature rose above 42 °C (the heat lamps were turned back on once the temperature dropped to 38 °C). The total duration of the heat stress was 1.5 h. Temperature was measured using J-type thermocouples and an HOBO 4-channel logger (Onset Computer Corporation, Bourne, MA). We embedded one thermocouple in the center of the brood, and placed one thermocouple next to the nest entrance and one thermocouple inside the observation hive room.
We experimentally heat stressed colonies in the afternoon (start time: 1500) or the evening (start time: between 1830 and 1900, one hour before sunset, so that the heat stress concludes after sunset). Each colony (n = 5) received one heat stress in the afternoon and one in the evening, with the order randomized, and 48 h of rest between heat stress trials. We measured beards every 15 min throughout the 1.5 h heat stress and for 3 h post-heat stress. If the colony still had a beard 3 h after the end of the heat stress, we measured beard size hourly until 2400, again at 0300, and every 15 min from 0500 until the beard dissipated completely (0 bees left on the colony entrance). We increased our measurement interval from 0500 onward, because we noticed that beards often dissipated with the onset of sunrise (0530-0600). For these data, we report beard area instead of estimated bee count, because perfectly perpendicular photographs of the beard were not possible, making estimates of beard depth unreliable.
Our work showed that beards tend to dissipate rapidly with the onset of dawn, so we experimentally tested the effect of light on beard dissipation. We artificially illuminated the apiary at night, between 2300 and 0100, using four 1000-W lights (Halogen Free Standing Construction Lights) placed in the middle of the apiary, shining on the beards at the nest entrances (distance from light: 3.4 ± 0.6 m). We positioned the lights in the apiary across all three overnight trials ), but we only illuminated the lights from the 19-20th (experimental test), whereas the other trial dates served as a control (18-19th, 20-21st). Across all three trials, we measured beard size before sunset, throughout the night, and after sunrise (beards were measured at: 1800, 2100, 2300, 0000, 0100, 0500, 0530, 0600, 0630, and 0700; time of sunset in Auburn AL across 18-21 Aug 202318-21 Aug : 192118-21 Aug -1924; sunrise: 0608-0609). The colonies used for this experiment (n = 8), were housed in 5-frame boxes, and were previously used for the brood experiment (see above). One colony did not beard at all during our experiment, so we removed this colony from subsequent analysis (bearding was a necessary precursor to investigate beard dissipation).
We used linear mixed effect models to investigate the effect of colony size on the proportion of observations where bearding occurred, the daily peak beard size, and the colony proportion in daily peak beard (daily peak beard size divided by colony size). To control for differences between the two apiaries, we included it as a fixed effect, and colony ID as a random effect. To understand the impact of brood status on bearding, we ran linear mixed-effects models on the daily peak beard size, and the proportion of observations where bearding occurred, with colony ID as random effect. We also tested the effects of brood status on the internal nest temperature, and its variability, using a linear mixed-effects model with colony ID as random effect. For the experiment inducing beards in the evening versus afternoon, we ran three paired t-tests, predicting the time spent bearding till the beard dissipated completely (with 0 bees on the modified entrance), the peak beard size, and the time to cast a beard from the onset of the heat stress based on treatment group (beard induced in the evening versus afternoon). To test the effect of light on beard dissipation, we ran a linear mixed effect model (lmer) predicting the number of bees that dissipated and the proportion of beard dissipated (change in beard size from 2300 to 0100, divided by the initial size of the beard at 2300), with treatment group (illuminated versus dark) as the independent variable and colony ID as random effect. All statistical tests were done in R (Version 4.4.2), and the lme4 package (Bates et al. 2015;R Core Team, 2020).
To track daily beard patterns, we monitored naturallyoccurring beards in 11 colonies. During our observation period (7-10 June; 31 July-3 August 2024), all of our colonies cast a beard daily except one (this colony bearded on the first and third day of observation, but not on the second day). In one colony (1 of 11) a beard was present every time we observed them. For the majority of colonies (9 of 11), beards began in the afternoon (1200-1500), persisted past 2400, and dissipated by 0600 the following day (Fig. 1b). Beard patterns did not align with ambient temperatures, which peaked during the day between 1200 and 1400. Instead, we found that beard size peaked at night, between 1800 and 2400 (Fig. 1b). The number of bees in the peak beard size ranged from 78 to 5174 bees (peak beard size: 1483.64 ± 1526.84 bees; mean ± SD), or 3.7 to 53.3% of the colony (percent of colony bearding at peak: 24 ± 17.3%).
We also tracked the position of the beard upon the colony's nest box (Fig. 2). Across all timepoints and dates, the majority of the beard was located upon the south-facing wall of the nest box, where the entrance was located (94.5 ± 16.1%). However, at 1200, when the sun was positioned in the east and cast a shadow on the west-facing wall, 22.2 ± 38.1% of the bees in the beard were located the beard at each timepoint is divided by the daily maximum beard size; 1.0 denotes the largest beard for a given colony on a given day, whereas 0.5 is a beard half that size; see also Fig. S2 for raw data counts per colony). Lines show mean value, shading shows SD, and points show raw data (color figure online) Fig. 2 Beard location throughout the day. Illustration showing the position of the beard depending on time-of-day (moon symbol for nighttime; sun symbol for daytime, with rays indicating sun-exposed side of the nest box, in gray; nest entrance in black; all nest entrances faced south). Bees predominantly formed a beard on the south-facing wall (beneath/around the colony entrance), but also skewed toward the shaded side of the next box during the day (see also Fig. S1) (color figure online)
on the west side of the nest box (beard position at 1200; east: < 1%; south: 77.8 ± 38.1%; west: 22.2 ± 38.1%). Conversely, at 1800, when the sun was positioned in the west and cast a shadow on the east-facing wall, 8.2 ± 12.2% of the bees in the beard were located on the east side of the nest box (beard position at 1800; east: 8.2 ± 12.2%; south: 91.8 ± 12.2%; west: < 1%). At night, when the sun was not shining upon the colonies, the beard position returned to the south-facing wall (beard position between 2100 and 0600; south: 99.5 ± 1.6%; east: < 1%; west: < 1%). Therefore, while workers predominantly beard where the entrance is located (in our study, the south-facing wall), their position is also skewed toward the shaded side of the nest box.
Investigation 2: how does colony state impact bearding?
We experimentally manipulated colony size and found that larger colonies have more bees in their daily peak beard (n = 11, slope = 0.331, P = 0.027); for every 1000 additional bees in a colony, daily peak beard size increased by 331 bees (Fig. 3a). Larger colonies also cast a beard more often than smaller colonies (n = 11, slope = 0.00005, P = 0.009); for every 1000 additional bees, beards are observed 5% more often (Fig. 3b). However, we found no impact of colony size on the proportion of the colony bearding in the largest beard for a colony in a day (n = 11, P = 0.283). For each colony, its largest beard size represented 25 ± 40% of the colony, but this did not increase significantly with colony size (Fig. 3c). Therefore, while larger colonies do beard more often and with more workers than do smaller colonies, the proportion of the colony that exits the nest to form a beard is similar across large and small colonies.
To determine the influence of brood on colony bearding, we experimentally manipulated colonies, such that some had brood (brood colonies) and some did not (no-brood colonies; note that each colony was represented in both treatment groups, see Methods). Comparing brood and no-brood colonies, we found no difference in daily peak beard size (brood: 387.79 ± 415.05 bees; no-brood: 252.04 ± 352.62 bees; n = 8, P = 0.213, lmer), or proportion of observations bearding (brood: 0.34 ± 0.18; no-brood: 0.25 ± 0.13; n = 8, P = 0.111, lmer). No-brood colonies still thermoregulated their nests (Fig. 4), but we did find that no-brood colonies had lower and more variable internal nest temperatures than when the same colonies did contain brood (nest temperature in brood colonies: 35.61 ± 0.94 °C; no-brood colonies: 33.93 ± 1.74 °C; P < 0.01, lmer). A carefully thermoregulated nest is critically important for brood rearing, but colonies still cast beards regardless of whether or not their colony contained brood.
We next explored how heat stress occurring at different times of day (evening versus afternoon) impacted both the formation and dissipation of beards. Regardless of when beards were induced, we found no difference in peak beard size (evening beards: 301.15 ± 200.61 cm 2 ; afternoon beards: 162.74 ± 105.12 cm 2 ; P = 0.152, n = 5, paired t-test), or the time to cast a beard from the onset of the heat stress (evening beards: 27 ± 16.43 min; afternoon beards: 36 ± 8.22 min; P = 0.208, n = 5, paired t-test). We did, however, find that if colonies were heat stressed in the evening, the beards lasted significantly longer than if colonies were heat stressed during the day (duration of evening beards: 8.8 ± 4.2 h; afternoon beards: 1.2 ± 0.5 h, P = 0.013, n = 5, paired t-test). Four of the five evening stressed colonies bearded over the entire night; the bees in these beards only fully dissipated with the onset of dawn, 10+ h after the end of the heat stress. In contrast, all of the afternoon stressed colonies (5 of 5) had beards that fully dissipated within 2 h of the end of heat stress (Fig. 5a). Note that each colony was represented in both treatment groups (see methods), and so, the difference in beard duration is not due to the colony itself, but rather the timing of beard formation.
During the heat stress, as expected, brood temperature within the colonies increased. There was no difference between the two treatment groups in their temperature profiles pre-heat stress or during the heat stress (evening heat stress colonies, pre-heat stress: 33.35 ± 1.26 °C, during heat stress: 39.94 ± 2.39 °C; afternoon heat stress colonies, pre-heat stress: 33.38 ± 1.13 °C, during heat stress: 39.94 ± 1.96 °C; Fig. 5b). Immediately after the heat stress concluded, the brood nest temperature in both treatment groups dropped (Fig. 5b). However, whereas the temperature profile of the brood nest in the afternoon heat stress colonies returned to the optimal range (33-36 °C), the temperature profile of the evening heat stress colonies dropped below the optimal range (brood nest temperature during h 2-12 postheat stress: evening heat stress colonies: 30.88 ± 1.85 °C; afternoon heat stress colonies: 33.55 ± 0.91 °C; P < 0.001, n = 5, lmer; Fig. 5b). The brood nest temperature profile in the evening heat stress colonies only returned to its optimal range once the bearding bees had fully dissipated back into their nest the following morning (Fig. 5b). This shows that brood temperature falling back to its normal range, or even below, is not a cue for bearding bees to dissipate and return to their nest.
Given that beards dissipated in the morning (Figs. 1, 5), we next explored the potential role of illumination as a cue for workers to return to their nest, by shining bright lights on bearding colonies from 2300 to 0100 ("illuminated night"), and comparing that to the two adjacent nights without artificial lighting ("dark nights"). The number of bees that dissipated was not significantly different between nights with and without experimental lighting (illuminated night: 179.43 ± 173.65 bees returned between 2300 and 0100; dark nights: 128.15 ± 124.18 bees returned; n = 7, P = 0.242, lmer). The percentage of the beard that dissipated between 2300 and 0100 was also not significantly different (illuminated night: 80.02 ± 36.85% of the beard returned; dark nights: 49.11 ± 38.16% of the beard returned; n = 7, P = 0.07, lmer). Regardless of whether the apiary received artificial lighting or not, by 0530 (38-39 min before sunrise) only a few bees remained in the beard (illuminated night: 0 bees; dark nights: 4 ± 10 bees). To confirm that the initial beard sizes were no different between the illuminated and dark nights, we compared beard size at 1800 (i.e., before any experimental manipulation), and found no significant difference (beard size prior to illuminated night: 88.29 ± 133.06 bees; dark nights: 129.36 ± 194.6 bees; n = 7, 4 Nest temperature of colonies in the "brood" and "no brood" treatment groups. When colonies did not have brood, their nest temperature was consistently lower than when they had brood. Note that each colony is represented in both treatment groups (i.e., brood colonies become no-brood colonies, and vice versa). Dashed horizontal lines represent the range of optimal brood temperature. Solid lines show mean values per treatment group, shading shows SD, points show raw data (color figure online) P = 0.5213, lmer). Given our artificial lighting setup, we found no evidence that illumination is the cue that workers use to dissipate from the beard in the morning (Fig. 6).
Bearding is a thermoregulatory behavior to cool the nest, where hundreds to thousands of workers leave their nest and form a bivouac near the entrance. Here, we investigated natural bearding patterns, how colony condition impacts bearding, and when beards dissipate.
We observed colonies bearding throughout the day, but beard size consistently peaked in the evening/night, between 1800 and 2400; a 4-6 h lag behind the hottest daily temperature (Fig. 1b), and a 1-2 h lag behind peak nest temperature (Fig. 4). This lag provides support for the idea of bearding as a "last resort" thermoregulatory behavior, occurring in earnest once nest temperature rises, and other thermoregulatory behaviors have reached their maximum. The nocturnal shift in bearding behavior, which was unexpected and only observed due to round-the-clock data collection, has two likely causes. First, colonies have more workers in the nest overnight, when foragers are no longer in the field (Crailsheim et al. 1996), and we have shown that larger heat stress (1.5 h). In a, each individual colony is shown. In b, solid lines show mean values per treatment group, and shading shows SD. Dashed horizontal lines represent the range of optimal brood temperature (color figure online) illumination period 0.0 0.4 0.8 1.2 1800 2100 0 000 0 300 0 600 Time of day Scaled beard size Dark Illuminated
Fig. 6 Testing illumination as a cue for beard dissipation. Pattern of beard dissipation when colonies are exposed to artificial lighting at night ("illuminated", in orange), versus no artificial lighting ("dark", in purple). Colonies were illuminated from 2300 to 0100. Scaled beard size was calculated by dividing the beard size at a given time by the largest beard size for that particular day/colony. Lines show mean values per treatment group, shading shows SD, and points show raw data colonies have larger beards (Fig. 3a). Second, bearding has been shown to increase when water is unavailable (Ostwald et al. 2016); at night, workers cannot forage for water and so must instead rely on bearding to cool the nest (though workers can also store water in cells; Ostwald et al. 2016). This combination of fewer thermoregulatory tools (e.g., no access to water), and more dense colonies (i.e., all foragers at home), may explain the nocturnal pattern of bearding behavior. It also shows how superorganisms can combine different thermoregulatory tools across different ambient conditions (e.g., day versus night). Indeed, workers spending the night outside the nest may be more cost effective for the colony than collecting and storing water for nighttime evaporative cooling, which would have reduced efficiency in the higher humidity.
If bearding does not coincide with peak ambient temperatures, can it still be considered a thermoregulatory behavior? We define a thermoregulatory behavior as a basic motivated behavior for nest temperature homeostasis. Given that we, and others, have consistently shown that experimentally heat stressing a colony will induce a beard (Ostwald et al. 2016;this work), we do consider bearding to be a thermoregulatory behavior. However, it is part of an interconnected suite of behaviors, and how it contributes to thermoregulation is debatable. Whether bearding bees are responding to temperature per se, or some other correlate (e.g., increased movement; Jhawar et. al, 2023; CO 2 concentration; Seeley, 1974), is unknown. Bearding is likely linked to ventilation, as it reduces the number of bees inside the cavity, and presumably improves airflow. Bearding has also been observed during mite treatments, such as thymol, a vapor-releasing essential oil (Richards et al. 2021;ApiGuard FAQ 2021). If the primary function of bearding is to reduce the number of individuals inside the nest to improve ventilation, then this behavior may be employed in multiple contexts (e.g., during a heat stress, when exposed to volatiles). Similarly, other thermoregulatory behaviors can also be employed in multiple contexts (e.g., workers collect water for evaporative cooling but also brood rearing; Kühnholz and Seeley 1997;Ostwald et al. 2016).
Colonies formed beards regardless of whether or not their nest contained brood. We found no impact of a colony's brood status (brood versus no brood) on the size of their beard, or duration of bearding. Colonies without brood, however, did have a lower and more variable temperature profile than colonies with brood (Fig. 4). This shows that while brood is an important factor in nest temperature homeostasis, workers do not modify their bearding behavior based on the presence or absence of brood. Colonies without brood must still thermoregulate; the queen is sensitive to high temperatures (above 38 °C, she begins to lose sperm viability in her spermatheca; McAfee et al. 2020), and their precious wax nest begins to melt above 40 °C (Buchwald et al. 2008).
Bees that depart in a beard may simply be detecting their immediate thermal environment as being above a threshold, rather than modifying their behavior based on the contents of their nest. This implies a simpler decision-making process for bearding bees (e.g., "if hot, then beard" instead of "if hot, and brood present, then beard"). Social insects are exemplars of how simple individual responses can give rise to robust group-level phenomena (Lutz et al. 2021).
Beard formation was no different in colonies that were induced to beard in the afternoon or the evening, but beard dissipation was extremely context dependent (Fig. 5a); afternoon beards dissipated within 1-2 h, but evening beards persisted overnight (10+ h). Bees bearding during the day experience a different environment than bees bearding at night (e.g., illumination levels, ambient temperatures, and forager traffic), and these conditions may impact an individual's decision to return to the nest. We experimentally tested if illumination was the cue that bearding workers use to dissipate, but our artificial illumination did not significantly impact beard dissipation (Fig. 6). How workers in the beard determine to return to their nest remains an open question.
Inducing beards at different times of day also showed that bearding carries risks for colonies. When beards were induced in the evening, brood temperature dropped below the optimal range after the heat stress finished (Fig. 5b). This low-brood temperature persisted until the sun had risen and the bearding bees returned to their nest. Therefore, during a heat stress, bearding can cause colonies to "over-correct" the temperature profile of their nest. Brood temperature is also an unlikely cue for beard dissipation (e.g., if workers inside the nest would relay information to bees in the beard, but we did not observe this). These results suggest that there is limited information flow between internal nest conditions, and workers that have left the nest to join the beard outside. One caveat, however, is that using heat lamps to raise the temperature in an observation hive may not fully replicate how colonies naturally experience high ambient temperatures.
Here, we present three lines of evidence that suggest bearding is an individual decision, and not coordinated across the colony: (1) bearding patterns do not respond to a colony's brood status, (2) beards follow shade (but only slightly), and (3) individuals in the beard do not return to the nest even when the brood is at lower-than-optimal temperatures. Nevertheless, individual behaviors are the basis upon which collective patterns in the superorganism take form, and indirect coordination can produce impressive higherorder processes.
Thermoregulation is an important feature of organisms living in a variable environment, and cooperative group living allows for thermoregulatory feats that would be impossible for individuals to achieve. In superorganisms, thermoregulation can be reflected in how workers build their nests (King et al. 2015;Kleineidam et al. 2001), adaptive behavioral responses to social conditions (Cook and Breed 2013;Ostwald et al. 2016), or simply where individuals position themselves (Jhawar et al. 2023). Here, we show how workers use nest evacuation as part of their thermoregulatory toolkit, and how these patterns change with internal and external conditions.