Honey bee forager thoracic temperature inside the nest is tuned to broad-scale differences in recruitment motivation

Insects that regulate their flight muscle temperatures can fly and forage at lower ambient temperatures than insects without the ability to generate metabolic heat (Kammer and Heinrich, 1978). Such thermoregulation is widespread in Hymenoptera such as wasps (Coelho and Ross, 1996; Eckles et al., 2008), solitary bees (Baird, 1986; May and Casey, 1983; Stone, 1993) and social bees (Nieh et al., 2006; Nieh and Sánchez, 2005; Stabentheiner, 2001). This ability to thermoregulate is a valuable adaptation that has allowed honey bees to occupy wide altitudinal gradients (Heinrich, 1993) in which they are also important facilitators of plant gene flow (Kearns et al., 1998). Thus, their ability to regulate flight muscle temperatures is an important factor in their ability to pollinate.

Insect muscles must achieve a minimum temperature to generate sufficient force for flight (Heinrich, 1993), a force that is correlated, within a range, to muscle temperature and metabolic rate (Coelho, 1991; Harrison and Fewell, 2002; Josephson, 2006; Woods et al., 2005). Thoracic temperature (T_th) is also tuned to food source profitability and the colony's need for carbohydrate and protein (Stabentheiner, 2001). The T_th of foragers remains elevated once foragers have returned inside the nest under several conditions: when food has a high sucrose concentration (Stabentheiner and Hagmüller, 1991), is close to the nest (Stabentheiner, 1996), or flows at a high rate during food exchange (trophallaxis) (Farina and Wainselboim, 2001; Farina and Wainselboim, 2005). In the field and within the nest (intranidal), increased thoracic temperatures may assist flight readiness by decreasing warm-up times.

Honey bees recruit nestmates to a food source through the waggle dance and round dance (von Frisch, 1967). In both dance types, forager motivation to recruit and the number of nestmates recruited is positively correlated with the number of dance repetitions (number of waggle phases or number of round dance cycles) (von Frisch, 1967; Seeley et al., 2000). Recruitment motivation is also influenced by colony need, forager genotype, relative food inflow rate, food availability, food quality, and the distribution of environmental resources (Barron et al., 2002; De Marco, 2006; De Marco et al., 2005; Dornhaus and Chittka, 2004; Dyer, 2002; Mattila and Seeley, 2007).

Studies that measure honey bee forager temperatures over relatively brief periods of time inside the nest reveal an interesting, but largely unexplored, phenomenon that may result from the physiology of heat production and forager motivation. Forager T_th fluctuates inside the nest (Stabentheiner et al., 1995). Between foraging trips, forager T_th changed in cooling and heating cycles with peak-to-peak amplitudes of 1–2°C while the forager was inside the nest (Stabentheiner and Hagmüller, 1991).

Our goal was to determine how forager recruitment motivation affects average T_th and fluctuations in T_th. We tested the hypothesis that a forager's T_th inside the nest is positively correlated with its motivation to recruit (Stabentheiner, 2001). Forager motivation was measured on two scales. On a fine scale, we measured T_th, the number of waggle dance circuits (positively correlated with motivation) and waggle dance return phase duration (inversely correlated with motivation) (Seeley et al., 2000). On a broad scale, we compared the T_th of actively foraging bees that were waggle or round dancing (recruiting nestmates to a food source), tremble dancing (recruiting nestmates to assist in nectar handling inside the nest or communicating other food source conditions) (Seeley, 1992; Thom, 2003), simply walking around the nest, or remaining stationary after unloading their food and before leaving the nest. Bees that tremble dance, only walk around the nest or remain stationary after unloading their food do not recruit new nestmates (Seeley, 1992; von Frisch, 1967), although they can reactivate experienced foragers that experience a familiar food odor (Reinhard et al., 2004). To elicit a broader range of recruitment motivation, we used a range of food distances and food qualities that reliably elicited forager visitation at our site during the field season.

We used three colonies of Apis mellifera (Linnaeus 1758) sequentially placed in a temperature-controlled room (25°C) at the University of California, San Diego, La Jolla, CA, USA (09° 09.890′ N, 79° 50.201′ W) from September to November of 2003 and 2004. We conducted one trial per day, from 09:00 h to 13:00 h. In May 2010, we used a fourth colony for thermal calibration measurements (see below). We housed each colony in a three-comb (Langstroth, American Standard) observation hive (56.5×78.7 cm) with doors to keep the colony dark during non-observation. All colonies had approximately equal populations and stores of pollen and honey. For measurements, we covered two sides with clear, infrared-transmitting, plastic film (Polyolefin FDA grade 75 gauge film, catalog no. LS-2475, BCU Plastics, Temecula, CA, USA). This film reduced air-current disturbances and facilitated normal colony thermoregulation. Colonies had access to the outside through a 3.75 diameter, 0.5 m long vinyl tube exiting the lab wall.

To measure thorax temperatures (T_th), we used a Raytek PhotoTemp MX6 (close-focus model, accuracy of 1% of measured temperature, Raytek Corp., Santa Cruz, CA, USA) infrared (IR) thermometer equipped with True Spot laser sighting to delineate precisely the measured area. We adjusted the spot measurement size to the diameter of a honey bee thorax (Fig. 1A). The emissivity of the polyolefin film was measured and the IR sensor was calibrated as described by Mapalad et al. (Mapalad et al., 2008). Internal nest air temperatures (T_{air nest}, within 5 cm of the focal bee) were simultaneously measured with the PhotoTemp MX6 (100 cm long type-K thermocouple, 0.3 mm diameter tip). Ambient air temperatures at the feeder were recorded each 5 min with a Vantage Pro weather station (Davis Instruments, Hayward, CA, USA). To calibrate our IR measurements, we placed the thermocouple in direct contact with the thorax of a bee harnessed in a tube while we measured the uncorrected thoracic temperature through a sheet of our IR-transmitting plastic film placed 1 cm above the thorax, as it was in the bee colonies. Bees (N=15) were harnessed to allow direct, stable contact with the thermocouple. We then calculated a correction (T_th,corrected=0.8165×T_{th,IRrawmeasurement}+9.0404) that we applied to all IR measurements. Throughout this paper, we only report T_th,corrected values, referring to them simply as T_th.

We trained bees to an inverted-jar feeder (von Frisch, 1967) and used unscented sucrose solution (catalog no. 821721, Ultra Pure, ICN Biomedicals, Irvine, CA, USA) and added 20 μl of lemon scent (lemon extract 98-0554, Kroger, Cincinnati, OH, USA) each hour to filter paper on top of the feeder. We trained approximately 10 bees (at a time) to distances of 5, 25, 50 and 100 m north of the colony, randomly alternating at each location between three concentrations [1.0, 1.5 and 2.5 mol l^–1, equivalent to 31, 43 and 65% sucrose (w/w)] (Kearns and Inouye, 1993). The lowest concentration to which bees would dance at all feeder distances and seasons at our site was 1.0 mol l^–1. Floral nectars occur at a variety of concentrations, and generalist bee foragers collect nectars ranging from 10–70% sugar (w/w) (Roubik et al., 1995). We individually marked all feeder foragers with plastic tags (0.2 mm thick, 2.5 mm diameter, 2 mg; Bee Works, Orillia, ON, Canada) glued with cyanoacrylate. These thin plastic tags do not interfere with IR thoracic temperature measurements (Mapalad et al., 2008). All foragers were verified on their return to our focal colony as colony members.

We videotaped behavior inside the nest with a Canon XL-1 digital camcorder (Canon USA, Lake Success, NY, USA). An assistant measured focal forager (T_th) and ambient nest air temperatures (T_airnest) approximately each 3 s and recorded these as voice notes while filming forager behavior. Thus, forager thoracic temperatures could be correlated with behavior and nest visit time. When the focal forager was obscured by other bees or positioned its thorax away from the assistant, we recorded T_th when it returned to a measurable position. We measured T_th each 3.2±2.9 s (mean ± s.d.).

We used iMovie v4.0.1 on an iMac G3 (Apple Computer, Cupertino, CA, USA) to analyze videos. The time of each temperature measurement, nest visit duration (time spent performing a behavior inside the nest between foraging trips), and focal forager behavior were recorded. We only followed foragers who consistently made multiple feeder trips during the day and randomly chose among these for our focal foragers. We did not select focal foragers based upon which intranidal behavior they performed. Thus, our data are based upon a random selection of these behaviors.

We examined four behaviors: dancing (round or waggle dancing, as determined by distance to the food source) (von Frisch, 1967), trembling (forager performs the tremble dance), walking (forager walks on comb without trembling or dancing), stationary (forager does not move). Round dances are approximately circular motions that communicate the presence of resources close to the nest (generally <100 m away). Waggle dances are looping figure-eight motions in which the forager waggles its body to communicate the distance and direction of the food source (generally ≥100 m away) (von Frisch, 1967). Tremble dancing can occur when nectar inflow is high and foragers need additional nestmates to help process the nectar (Seeley, 1992). Tremble dancing can also be elicited by feeder conditions. Bees that return from crowded artificial feeders produce tremble dances and vibrational ‘stop signals’, behaviors that are highly correlated (Nieh, 1993; Thom, 2003). Stop signal recipients reduce their waggle dancing and recruitment is therefore inhibited (Kirchner, 1993; Nieh, 1993; Nieh, 2010; Pastor and Seeley, 2005; Thom, 2003). All of our foragers returned to the nest and unloaded their food, but stationary bees remained stationary until their departure. We will refer to round or waggle dancing as ‘dancing’ and tremble dancing as ‘trembling’. We also tested for the fine-scale effect of waggle dancer motivation on T_th by measuring the return phase (time between waggle phases), and the total number of waggle phases per nest visit. Return phase duration is inversely correlated and the number of waggle phases is positively correlated with dancer motivation (Seeley et al., 2000).

Previous studies have recorded T_th of a bee performing different behaviors during the same nest visit (Stabentheiner, 1996; Stabentheiner et al., 1995). We focused on foragers that only performed one of these four behaviors during their nest visit. To avoid pseudoreplication and provide independent data points, we recorded each forager's nest visit only once. For example, a forager that entered the nest, unloaded food and then walked around without dancing (round or waggle) or trembling was classified as a ‘walking’ bee. We recorded its T_th only when it was walking. Like other investigators (Stabentheiner and Hagmüller, 1991), we observed fluctuations in forager thoracic temperatures during a nest visit. To quantify this, we calculated the temperature range (T_range = maximum T_th – minimum T_th per nest visit) and variance (Zar, 1984) in T_th per nest visit (T_variance). Honey bees are poikilothermic and their body temperature is influenced by the surrounding air temperature (Heinrich, 1993). Thus, we calculated ΔT_th (=T_th–T_airnest, where T_airnest is the ambient air temperature at the center of the dance floor inside the nest) to provide a standardized way to compare forager thoracic temperatures.

We used JMP IN v4.0.4 statistical software to conduct analysis of variance (ANOVA). We log transformed (Zar, 1984) the following data: average T_th, nest visit duration, T_range and T_variance. All data met assumptions for normality as determined by residual analyses. For simplicity, we will refer to these transformed variables by their untransformed names. We used two different analysis models. First, we used standard least squares ANOVA and avoided pseudoreplication by using the average per bee for measures describing a complete nest visit (nest visit duration, average T_th, average ΔT_th, T_range and T_variance). We tested for a significant effect of colony (a random effect, EMS algorithm), sucrose concentration, feeder distance, and behavior (fixed effects). We tested the significance of all fixed-factor interactions, and then ran simplified models after removing non-significant interactions (Zar, 1984). We used Tukey–Kramer Honestly Significant Difference (HSD) tests for post-hoc analyses.

To examine the effect of time on forager temperatures, we used a one-way ANOVA repeated-measures model because temperatures were successively measured each 3 s with the same individuals. A major goal of our study was to examine thermal fluctuations in waggle dancers for comparison with data found in other studies. Thus, we analyzed the nest visits of foragers visiting the 100 m feeder providing 2.5 mol l^–1 sucrose (the only concentration that reliably elicited waggle dancing in our study). There was substantial variation in these nest visit times (39.9±34.4 s), and we therefore calculated temperature measurement times as a percentage of total nest visit time (applying the arcsine-square root transformation to normalize this data) (Zar, 1984). All averages are expressed as means ± s.d. Where appropriate, we applied the Sequential Bonferroni correction (Zar, 1984). Tests passing this correction are marked ‘*^SB’.

In total, we analyzed the nest visits of 186 foragers from three colonies. Throughout the trials, air temperatures inside the nest (34.8±1.2°C) and at the feeder (17.9±2.4°C) remained relatively constant. As expected, there was a strong linear correlation between average T_airnest and average T_th for each nest visit (F_1,184=240.0, P≪0.0001) such that T_th increased by 0.6°C for each 1°C increase in T_airnest (linear regression: y=0.60x+19.60, R²=0.56, data pooled from all trial conditions). Over all trials (pooling all colonies, distances, sucrose concentrations and behaviors), average T_th=38.5±1.6°C, average ΔT_th=3.8±1.4°C, T_range=1.9±1.1°C, T_variance=0.4±0.7°C, and nest visit duration=49.0±37.8 s. Fig. 1B shows representative examples of temperature fluctuations in the four behavioral categories at different distances and sucrose concentrations.

We first examined effects on average T_th. There was a significant effect of sucrose concentration (F_1,178=15.91, P<0.0001*^SB), such that average T_th was significantly higher for higher sucrose concentrations. There were no significant effects of distance (F_1,178=0.54, P=0.47) or colony (F_2,178=0.24, P=0.79). There were no significant interactions (F_2,175≤0.40, P≥0.67). There was no significant relationship between average T_th and nest visit duration (F_1,184=0.07, P=0.79). Foragers exhibiting different behaviors had a significantly different average T_th (F_3,178=4.91, P=0.003*^SB; Fig. 2A). Dancing and trembling foragers were significantly warmer than stationary foragers, but T_th among moving bees was not significantly different (dancing, trembling and walking, Tukey–Kramer HSD, Q=2.59, P>0.05). We therefore pooled moving bees to calculate that for each 1 mol l^–1 increase in sucrose concentration, there was a 1.0°C increase in average T_th (F_1,134=19.46, P<0.0001; Fig. 2B). However, ambient air temperatures were significantly colder when foragers choose to remain stationary after returning to the nest as compared to when they moved (F_3,182=8.23, P<0.0001; Tukey–Kramer HSD Q=2.59, P<0.05; Fig. 2A).

Because of the significant differences in T_{air nest} during different forager behaviors, we calculated ΔT_th. As with T_th, the following factors were not significant: sucrose concentration (F_1,178=2.79, P=0.10), distance (F_1,178=0.38, P=0.54) and colony (F_2,178=0.49, P=0.61). There were no significant interactions (F_3,169≤1.51, P≥0.22). We again found a significant effect of behavior (F_3,178=5.41, P=0.001*^SB). However, unlike T_th, a slightly different behavioral pattern emerged for ΔT_th. Dancing bees had a higher average ΔT_th than trembling or walking bees (dancing, 3.9±1.1°C; trembling, 2.8±0.6°C; walking, 2.5±1.5°C; pooled across all sucrose concentrations; Tukey–Kramer HSD Q=2.59, P<0.05). On average, dancers were, respectively, 1.4- and 1.5-times warmer than trembling or walking bees, relative to nest ambient air temperatures (Fig. 2C). For ΔT_th, dancers were not significantly different from stationary bees (Fig. 2C).

The temperature range (T_range; examples in Fig. 1B) during a nest visit was not significantly affected by sucrose concentration (F_1,177=0.29, P=0.59), distance (F_1,177=0.37, P=0.54), behavior (F_3,177=0.17, P=0.92) or colony (F_2,177=0.05, P=0.95). There were no significant interactions (F_1,170≤1.76, P≥0.19). However, T_range significantly increased with increasing nest visit duration (F_1,183=44.38, P<0.0001*^SB; Fig. 3A). For each 1 min increase in nest visit duration, T_range increased by 0.72°C. Similarly, the variance in T_th per nest visit (T_variance) during a nest visit was not significantly affected by sucrose concentration (F_1,177=0.79, P=0.38), distance (F_1,177=0.25, P=0.62), behavior (F_3,177=0.50, P=0.68) or colony (F_2,177=0.08, P=0.92). There were no significant interactions (F_3,177≤151, P≥0.21). Like T_range, T_variance significantly increased with increasing nest visit duration (F_1,183=6.27, P=0.006*^SB; Fig. 3B). Thus, foragers (all behavioral categories) who spent longer periods inside the nest exhibited a wider T_th range and greater variation in T_th.

On average, waggle dancers performed 6.4±4.7 circuits with an average return phase of 2.3±0.7 s. In waggle dancing bees (100 m feeder, 2.5 mol l^–1 sucrose), there was no significant relationship between the number of dance circuits per nest visit or return phase duration (measures of dancer motivation) and different measures of dancer thoracic temperature (average T_th, average ΔT_th, T_range or T_variance; P≥0.12; Table 1). There were no significant interactions (P≥0.33; see Table 1).

During waggle dancing, there was a slight but significant change in T_th over time (F_10,532=3.36, P=0.0003*^SB; Fig. 4A). There was also a significant effect of bee identity on T_th (F_62,532=34.31, P<0.0001*^SB). The interaction of time and bee identity was not significant (F_353,532=0.83, P=0.93). For ΔT_th, there was a significant effect of time (F_10,532=4.53, P<0.0001*^SB), such that ΔT_th increased slightly throughout each waggle dancer's nest visit (Fig. 4B). For ΔT_th, there was also a significant effect of bee identity (F_62,532=29.22, P<0.0001*^SB), and no significant interaction of time and bee identity (F_353,532=0.86, P=0.88). There was, therefore, significant individual variation in T_th and ΔT_th during a waggle dancer's nest visit. However, despite this individual variation, waggle dancers' thoracic temperatures increased slightly throughout their nest visit. Waggle dancer T_th and ΔT_th was estimated (from linear regression; Fig. 4) to increase by 0.18°C and 0.28°C, respectively, halfway through a nest visit (average nest visit duration of 30.3±22.6 s).

We examined T_th and fluctuations in T_th with respect to nest air temperature, time spent inside the nest, food quality, food location, and forager motivation to recruit while foragers were inside the nest between foraging trips. In general, average T_th was strongly correlated with T_{air nest} (confirming previous studies) (Esch, 1960). For waggle dancers, there was a slight increase in T_th and ΔT_th over time (0.18–0.28°C after 15 s, on average) that exhibits high variation (Fig. 4) and that is significant (P≤0.0003*^SB). Similarly, T_range and T_variance significantly increased over time for waggle dancers (Fig. 3).

As expected (Dyer and Seeley, 1987; Schmaranzer and Stabentheiner, 1988; Stabentheiner et al., 1995; Underwood, 1991), foragers had a higher average T_th (2.7°C higher) when returning from richer food (2.5 mol l^–1 sucrose) than when returning from poorer food (1.0 mol l^–1 sucrose, Fig. 2A). For moving bees (walking, trembling and dancing), we found a T_th increase of 1.0°C per 1 mol l^–1 increase in sucrose concentration (Fig. 2B), similar to the 1.5°C increase per 1 mol l^–1 sucrose increase shown by Stabentheiner et al. [calculated from log regression of their data on waggle dancers, pooled distances, estimating change in T_th from 1–2 mol l^–1 sucrose (Stabentheiner et al., 1995)]. However, we found no significant effect of sucrose concentration on ΔT_th, perhaps because we used relatively high sucrose concentrations (1.0–2.5 mol l^–1), which were necessary at our site and season to train bees and have them recruit for an artificial feeder. Bees generally find such sucrose concentrations quite rewarding (Balderrama et al., 1992).

At the relatively short distances used (5–100 m), we did not find an effect of distance on T_th. This is not surprising because Stabentheiner (Stabentheiner, 1996) reported a temperature decrease of between 0.5–0.8°C per 1000 m increase in distance (0.5 to 2.0 mol l^–1 sucrose feeders). Using these numbers, we would expect at most a 0.08°C decrease in average T_th from 5 to 100 m. In general, our foragers had an average T_th of 38.5±1.6°C (La Jolla, CA, USA), nearly identical to the T_th measured for honey bees collecting nectar from floral resources (38.0±2.2°C, Graz, Austria) (Stabentheiner, 2001).

Thoracic temperatures (T_th, uncorrected for nest air temperature) varied with forager behavior and were highest for recruiting foragers (dancers; Fig. 2A). Prior studies have demonstrated that the T_th of honey bees placed in boxes is positively correlated with activity (Stabentheiner and Crailsheim, 1999). Fuchikawa and Schimizu found that thoracic temperatures were elevated by 7–8°C during locomotor activity (Fuchikawa and Schimizu, 2007). Like Stabentheiner et al. (Stabentheiner et al., 1995), we did not find significant average T_th differences between dancing and walking bees. In our experiment, there was a significant variation in dance floor air temperatures, and stationary foragers were observed when the nest air temperature was on average 1°C and 2°C below that recorded when dancing and trembling foragers were observed (Fig. 2A). Lower ambient nest air temperatures may contribute, for some unknown reason, to stationary behavior between foraging bouts.

We explored the association between behavior and thoracic temperatures by using ΔT_th, a measurement that corrects for different ambient air temperatures because T_{air nest} exerts a strong influence on T_th (linear regression R²=0.56, P<<0.0001). Bees that were highly motivated to recruit nestmates (dancers) were significantly warmer (1.4- to 1.5-times higher average ΔT_th) than trembling or walking bees who simply foraged but did not recruit during their nest stay (Fig. 2C). This higher thoracic temperature may reflect general foraging motivation, but other explanations are possible. For example, cold stress within the colony increases thoracic flight muscle heat generation (Stabentheiner et al., 2010). This could explain why stationary bees had the second highest average ΔT_th (Fig. 2C), although this does not account for the average elevated ΔT_th of dancers, which was the highest in our study. Elevated ΔT_th could result from maintenance of flight muscle temperature in preparation for a rapid flight back to the food source. With respect to proximate causation, elevated ΔT_th could be a byproduct of waggle dancing because leg muscles have origins in the thorax and are active during waggle dancing (Stabentheiner, 1996). However, dancers were significantly warmer than trembling or walking bees (ΔT_th, Fig. 2A) Slight wing motions occur during the waggle phase that are not as consistently present in trembling or walking bees. However, video thermography of waggle dancing did not reveal increases in T_th during the waggle phase (wing motion and walking) when compared with walking-only return phases immediately before and afterwards (Stabentheiner and Hagmüller, 1991). The thermal contribution of muscles due to wing motions during the waggle phase is therefore likely minimal.

It is important to consider whether T_th fluctuations are artifacts from optically measuring the temperature of moving foragers. If so, then temperature fluctuations should be greater for moving than for stationary bees. However, there is no significant effect of behavior (stationary or moving) on the magnitude (T_range or T_variance) of these fluctuations (P≥0.68). In addition, data from investigators using real-time video thermography (which allows continuous tracking of bee temperatures) clearly demonstrate similar temperature fluctuations in forager intranidal T_th (Stabentheiner and Hagmüller, 1991). Such T_th fluctuations are also exhibited by guard bees, bees investigated by guards, and workers warming the nest (Kleinhenz et al., 2003; Stabentheiner et al., 2002; Stabentheiner et al., 2007). Although we used a different method of IR thermography, our average T_range of 1.9±1.1°C is within the T_range of 2.4±1.1°C obtained with continuous thermography (calculated from graphs of 18 dancing foragers collecting 1.0–2.0 mol l^–1 sucrose at distances of 60–1750 m from the nest) (see Stabentheiner and Hagmüller, 1991; Stabentheiner et al., 1995).

In summary, we did not find significant correlations between waggle dance parameters associated with fine-scale dancer motivation (number of waggle circuits or dance tempo) and T_th or ΔT_th. Thus, the effect of dancing motivation on T_th is limited to a broad-scale contrast between bees that recruit and those that only tremble dance or walk around inside the nest between foraging bouts. Elevated relative T_th and ΔT_th may be related to recruitment motivation, not only to heat generated by a moving bee, because tremble dancers and walking bees were cooler than dancers. Thus, three factors: the basic physiology governing thoracic temperature maintenance, the relationship between T_th and flight, and forager motivation likely play an important role in regulating intranidal T_th.

We would like to thank Constantine Lau, Shaza Hanafy, Renee Lafrenz, Eri Suzuki, Michelle Renner, and Melissa Mazzarella and Alon Orlitsky for their valuable assistance. We are also indebted to the comments of the referees who have significantly improved this manuscript.