Hot bumble bees at good food: thoracic temperature of feeding Bombus wilmattae foragers is tuned to sugar concentration

The modulation of body temperature by active or passive means at or near a specific set point is found widely among insects(Heinrich, 1993), particularly in the Hymenoptera (Himmer,1932). Researchers have reported thermoregulation in ants(Hölldobler and Wilson,1990), wasps (Stabentheiner et al., 2004), solitary bees [Anthophora(Stone, 1994)] and social bees(Bujok et al., 2002; Dyer and Seeley, 1987; Kleinhenz et al., 2003; Stabentheiner et al., 1990; Stone and Willmer, 1989). Here, we focus on the phenomenon of thoracic metabolic heat production linked to food quality.

Flight thermoregulation is widely found among bees(Stone and Willmer, 1989),including solitary bees [Xylocopa virginica(Baird, 1986)]. All bees in which thermoregulation has been examined are heterotherms(Heinrich, 1993) and can regulate higher body temperatures during foraging and attacks(Heinrich, 1979). Thermoregulation plays a role in the ability of bees to fly(Coelho, 1991; Esch, 1976) and thus to visit flowers under cold conditions (Corbet et al., 1993). Bombus terrestris, B. pascuorum and B. hortorum began foraging at lower temperatures than honey bees and were thus expected to forage earlier in the morning and later in the afternoon than honey bees (Corbet et al.,1993), a potential example of how thermoregulation allows niche specialization. Recently, bumble bees (B. terrestris) have also been shown to be able to associate warmth with floral color, preferring to visit warmer flowers in a flight arena (Dyer et al., 2006).

Several species of honey bees, including Apis mellifera, A. cerana, A. dorsata and A. laboriosa, are able to regulate their body temperature, exhibiting higher thoracic temperatures after feeding at rich sucrose solutions than at poorer sucrose solutions(Schmaranzer and Stabentheiner,1988; Underwood,1991). In general, honey bee thoracic temperature corresponds to the quality of the food source as perceived by sweetness(Stabentheiner and Hagmüller,1991), proximity to the nest(Esch, 1960; Stabentheiner, 1996) and nectar flow rate (Farina and Wainselboim,2001), with increasing net caloric intake per unit time giving rise to higher temperatures. In honey bees, metabolic expenditure increases with increasing nectar flow rate at an artificial feeder(Balderrama et al., 1992; Moffatt and Nunez, 1997). On a feeder, honey bees evidently traded off the amount of energy expended in keeping the thorax warm and thus in a state of immediate readiness for takeoff(Moffatt, 2001) versus net caloric benefits to the colony by expending less energy in thoracic heating for poorer food(Waddington, 1990). Honey bees are thus able to adjust their energetic expenditure and exploit relatively poor food when richer food is not readily available.

Recently, foragers of the stingless bee Melipona panamica(Meliponini) also have been shown to adjust their thoracic temperature according to food caloric value. Like honey bees, M. panamicaforagers maintained a thoracic temperature T_th excess above air temperature T_a (21.5-29.5°C) that corresponded to sucrose concentration when unloading food and recruiting nestmates inside the nest (Nieh and Sánchez, 2005). Thus the phenomenon of increasing thoracic temperature with increasing sucrose concentration may be common among the flying Hymenoptera.

In their study of bumble bee field foraging, Heinrich and Heinrich observed an interesting phenomenon in B. vagans, B. terricola and B. perplexus foragers and wrote: (p. 561) “flight activity by itself cannot account for the high T_th (>34°C) observed on raspberry, since bees often perched on flowers and resumed flight fully heated; they maintained an elevated T_th even while they were perched” (Heinrich and Heinrich, 1983b). Based upon recent findings that stingless bee foragers (Meliponini, Melipona panamica) can maintain elevated T_th temperatures tuned to food caloric concentration while feeding (Nieh and Sánchez,2005), and given that honey bees (Apini) do the same(Stabentheiner, 2001; Stabentheiner and Hagmüller,1991), we examined whether their close relatives, the bumble bees(Bombini), possess a similar ability.

Bombus wilmattae (Cockerell,1912) is restricted in distribution to the highlands of Chiapas and Guatemala, corresponding to that of the Bosque de Pino y Encino and the Bosque Tropical Caducifolio (Labougle,1990; Rzedowski,1978). Elevational distribution of this species, based upon this and additional studies in preparation (21 additional colonies, R.V.,unpublished) is between 1413 m and 2014 m. This is considered a tropical-to-subtropical species, which makes our study particularly interesting because examinations of bumble bee thoracic temperature have largely been restricted to temperate species.

We collected six wild colonies of Bombus wilmattae(Cockerell, 1912) around the towns of Unión Juárez, Talquian, and San José, in Chiapas, Mexico at altitudes ranging from 1110 to 1690 m (laboratory designations: BW5, BW8, BW10, BW12, BW13 and BW14). We used BW5 from November to December 2003 and other colonies from August to September 2005. Colonies were transferred to wood observation hives (30 cm ×30 cm ×30 cm for colony BW5 and 18 cm ×18 cm ×18 cm for all other colonies)covered with glass. Experiments were conducted inside a laboratory at El Colegio de la Frontera Sur, Tapachula, Chiapas, Mexico (14°53.191′N, 92°17.183 ′W) inside wood foraging arenas (120 cm×90 cm ×30 cm, L ×W ×H) covered with IR transparent plastic film [BCU Plastics, Polyolefin FDA grade 75 gauge film, catalog#LS-2475, protocol of Stabentheiner and Hagmüller(Stabentheiner and Hagmüller,1991)]. Each colony was connected to a single, separate foraging arena through a 3.5 cm diameter, 10 cm long, clear vinyl tube. All colonies were maintained inside the laboratory for at least 20 days before use in the experiments and provided ad libitum with 50% (v/v) honey/water. All trials were conducted inside a light and temperature-controlled room(24.4±1.9°C). We visually inspected the exposed honey pots of each colony before using them in experiments and found that the food stores (honey and pollen) of all colonies were similar.

Before beginning our experiments, we used paired feeders for a preliminary test of any potential sucrose concentration effect, and a resulting thermographic image is shown to better illustrate the effect of sucrose concentration on forager body temperatures. In our experiments, we measured the temperatures of bees foraging at a single feeder containing unscented 0.5,1.0, 1.5, 2.0 or 2.5 mol l^-1 sucrose solutions (double distilled water and reagent grade sucrose, Quimica Meyer, cat#6710-1000, Tapachula,Mexico). Beginning at 09:00 h on each experimental day, we presented the colony with a single sucrose concentration for 30 min, followed by a 5 min pause (when no sucrose was available) before providing the next sucrose concentration. We presented all five concentrations on each day, alternating between ascending and descending concentration series on subsequent experimental days. We thoroughly cleaned the feeder between concentration changes. Each colony was used for two complete concentration series and thus experienced both diminishing and increasing sucrose concentrations.

Foragers were individually marked either with Opalithplättchen honey bee tags (colony BW5) or with small acrylic paint marks on their abdomens (all other colonies). Every 5 min, we made an instantaneous count of the number of foragers collecting sucrose solution at the feeder. Thus we made six counts during each 30 min presentation of a particular sucrose concentration. To examine the proportion of colony foraging elicited by different sucrose concentration, we calculated the normalized forager census count(F_N) by dividing the census count by the largest census count obtained for that colony at any sucrose concentration. We used a circular grooved plate feeder [40 ml capacity, 10 cm diameter, 60 grooves,design after description by von Frisch(von Frisch, 1967)]. Foragers fed ad libitum (feeder was never exhausted during the 30 min feeding trial), and the feeder did not limit forager numbers because each feeder could accommodate over 35 individuals; we observed a maximum of 25 foragers feeding simultaneously. To determine forager size, we randomly selected 15 live foragers from each colony, weighed them in plastic vials and measured their length (from abdominal tip to anterior head) with calipers. Colonies were not fed outside of the experimental period, but were not food deprived because daily visual inspections revealed largely full honeypots in exposed areas of each colony.

We used infrared thermography to measure forager thoracic surface temperatures (Stabentheiner and Schmaranzer, 1987). We recorded bee temperatures 5 s after they had begun feeding on the feeder using a photographic infrared thermometer(Raytek PhotoTemp MX6, San Diego, CA, USA; close-focus model) equipped with True Spot laser sighting to precisely delineate the measured area (spot measurement size adjustable to the diameter of a bumble bee thorax). Each time we made a thermographic measurement, we measured air temperature 5 cm from the feeder (T_a) using a weather meter (Kestrel 3000, Chester,PA, USA). To calibrate the MX6 infrared thermometer, we inserted a type K thermocouple into a dead bee whose internal and external temperature had equilibrated to ambient, and then recorded its dorsal thoracic surface temperature through the IR transparent plastic film. Equipment emissivity values were then adjusted until both thermocouple and infrared temperature readings matched. We measured T_th and T_a, and calculated the difference between the thorax temperature and the ambient air temperature at the feeder[ΔT_th (Stone,1993)].

We used JMP IN v4.0.4 software to run ANOVA (standard least-squares model)after checking for data normality through residual analysis(Zar, 1984). Models with multiple effects were run first with all interaction terms and then run as simplified models without interactions if interactions were nonsignificant(Zar, 1984). Values are presented as means ± s.d. We considered results non-significant (NS) at the critical value P ≥0.05.

The majority of foragers walked from the entrance of the foraging arena to the food source, although some bees flew briefly before landing near the feeder. At the conclusion of feeding, the majority of foragers walked back to the nest, but some took off and flew for a short period of time before landing and entering the nest. At all sucrose concentrations, foragers exhibited abdominal respiratory movements with the exception of seven evidently ectothermic foragers (Fig. 1),who fed at the lowest sucrose concentration offered (0.5 mol l^-1)and whose respiratory frequency was too low to be detected by eye. All foragers either folded their wings over their abdomens or, in some cases,performed wing buzzing in which the wings fluttered rapidly in an arc of 10-85° relative to the body. On average, live foragers weighed 0.17±0.11 g and were 15.0±1.5 mm in length (N=90 total foragers, 15 randomly selected from each colony).

We first examined the effect of T_a on T_th at each sucrose concentration. In the overall model,both T_a and sucrose concentration are significant effects,but colony identity is not (two-factor ANOVA, F_2,1620=243.0, P<0.0001; individual effects: T_a, F_1,1620=1484.4, P<0.0001; sucrose concentration; F_1,1620=1491.2, P<0.0001; colony identity NS;interactions NS). We therefore pooled the data from all colonies and examined the relationship between T_th and T_a at each sucrose concentration. Foragers generally increased their thoracic temperature above ambient at all sucrose concentrations, and the slope of T_th on T_a is significantly less than one for all concentrations (slope confidence intervals: P<0.001). For clarity, Fig. 1 shows only the highest and lowest concentrations. Note that with only the exception of seven evidently ectothermic foragers (all at 0.5 mol l^-1), all other T_th values are higher than ambient air temperature(Fig. 1). These seven different bees (from colonies BW5, BW8 BW10, BW13 and BW14) also did not show any noticeable abdominal respiratory movements. On average, forager T_th increased by 0.53 to 0.66°C for each 1°C increase in ambient air temperature over the range of sucrose concentrations used.

We next examined the effect of sucrose concentration onΔ T_th (Table 1). The overall model for ΔT_th with sucrose concentration and colony identity is significant (two-factor ANOVA, F_6,1616=152.2, P<0.0001, R²=0.36, interaction NS). Sucrose concentration is a significant effect (F_1,1616=475.4, P<0.0001, SS_sucrose=2292.6). Colony identity is also a significant effect (F_5,1616=106.6, P<0.0001, SS_colony=2569.4). A detailed analysis by colony reveals that ΔT_th significantly increases with sucrose concentration with all colonies, with linear regression slope estimates ranging from 1.2 to 2.4 (Table 2). Thus foragers increased their thoracic temperature excess, on average, by 1.2 to 2.4°C for each 1 mol l^-1 increase in sucrose concentration.

There is a significant effect of colony identity and sucrose concentration on F_N (the normalized forager census count, two-factor model ANOVA, F_6,1615=241.4, P<0.0001, R²=0.47, interaction NS). Colony identity is a significant effect (F_5,1615=35.5, P<0.0001, SS_colony=4.5). Sucrose concentration is also a significant effect (F_1,1615=1082.2, P<0.0001, SS_sucrose=27.6), and accounts for approximately six times more of the variance in F_N than colony identity. A detailed analysis by colony reveals that F_N significantly increased (P<0.0001) with sucrose concentration in each colony. Linear regression slope estimates for the colonies fell into two groups: those with slopes of 0.16-0.18 (colonies BW5, BW8, BW10 and BW14: group 1) and those with slopes of 0.21 and 0.23 (colonies BW12 and BW13: group 2). We therefore present these data pooled into two groups(Fig. 2C), with sucrose concentration accounting for 40% of the variance in F_N of group 1 (ANOVA F_1,1120=739.1, P<0.0001) and 46% of the variance in F_N of group 2 (ANOVA: F_1,498=424.5, P<0.0001). On average, the proportion of feeding foragers increased by 18% with each 1 mol l^-1increase in sucrose concentration in group 1 and by 22% with each 1 mol l^-1 increase in group 2.

Bombus wilmattae foragers from six different wild colonies consistently increased their thoracic temperatures when feeding at food with higher carbohydrate rewards (Figs 1 and 2). For each colony, F_N also increased with increased sucrose concentration(Fig. 2C). Although colony effects were found for ΔT_th, each colony exhibited the same trends of increasing ΔT_th and increasing F_N with increasing sucrose concentration(Table 2). Thus the phenomenon of increasing thoracic temperature of foragers feeding while stationary upon food with increasing caloric value has now been shown to occur in the Bombini and may be homologous in all social bees in the Apidae.

A low T_th may be costly because it increases the time between exploiting different inflorescences by prolonging pre-flight warm-up or increasing escape times when avoiding predators. With a variety of bumble bee species and worker sizes, a T_th of approximately 29-30°C is required for flight(Heinrich, 1974). So maintaining a high T_th, and thus flight readiness, may allow workers to gamble on the chance of finding a new food source(Heinrich and Heinrich,1983a). For food of lower quality, less energy could be spent on maintaining higher thoracic temperature to optimize net caloric returns to the colony (Stabentheiner et al.,1995). Body temperature decreases when foraging on relatively poor quality food could have evolved in bumble bees as an energyconserving strategy under active physiological control or perhaps to a lack of direct energy supplies (Heinrich and Heinrich,1983a), although in honeybees there is no correlation between haemolymph sugar levels and body temperatures, except for a slight effect with highly dilute sucrose solution (Blatt and Roces, 2001; Blatt and Roces,2002). High T_th may also be adaptive on natural food sources to properly harvest the food, particularly pollen(Feuerbacher et al., 2003). For example, B. terricola workers may have maintained a high T_th to collect pollen by twirling around Spiraea latifolia inflorescences (Heinrich and Heinrich, 1983a).

Colonies were of different sizes and were collected at different times of year and thus potentially at different life stages (S.C., unpublished observations). Both of these effects may account for differences in the proportion of workers activated to forage at given sucrose concentrations, in addition to differences in colony food stores. We found colony differences in slope and y-intercept for the effect of sucrose concentration onΔT_th and F_N, although in all cases, there was a significant positive effect of sucrose concentration onΔ T_th and F_N(Table 2). In particular, our largest colony, BW5, exhibited the strongest correlations between sucrose concentrations and ΔT_th(R²=0.43). We speculate that worker thresholds for sucrose collection and thus the perceived quality of different sucrose concentrations may also have varied with either colony stage (whether the colony was at the beginning of its life cycle and growing or near the end of its life cycle and collecting relatively little food) or food stores. With our current data, we are unable to determine whether seasonality and colony life stage play a role in finely modulating the effects of sucrose concentration onΔ T_th and F_N, although this would be important to test in future studies.

It is relevant to consider how our experimental design may have affected our measurements. Floral nectars occur at a variety of sugar concentrations(Baker and Baker, 1982) and generalist neotropical bee foragers collect nectars ranging from 10%-70% sugar by mass [with means of 38%, 44% and 48% for Euglossini, Meliponini and Centridini, respectively (Roubik et al.,1995)]. We therefore used sucrose concentrations ranging from 0.5-2.5 mol l^-1 (16% to 65% by mass), with the highest sucrose concentration (2.5 mol l^-1) used to provide comparative data with foraging studies on other social bees.

We conducted our experiments with colonies foraging in foraging arenas, a standard experimental design for investigations of bumble bee foraging(Chittka et al., 2003; Dornhaus and Cameron, 2003; Dornhaus and Chittka, 1999; Dornhaus and Chittka, 2004; Dyer et al., 2006; Spaethe et al., 2001). In most cases, bumble bees walked to the food source, but some flew and we did not detect any differences between the T_th effect of sucrose concentration on bees that had walked or flown to the food source. Nonetheless, it would be desirable to verify the effect of sucrose concentration on T_th with flying bees using a large flight arena.

In B. wilmattae the average ΔT_thincreased by 4.2°C with each 1 mol l^-1 increase in sucrose concentration. This is a greater increase than is found in honey bees[1.5°C increase in thoracic temperature per doubling of the sucrose concentration (Schmaranzer and Stabentheiner, 1988)]. Foragers of the stingless bee M. panamica increased T_th by 0.9°C per 1 mol l^-1 increase in sucrose concentration(Nieh and Sánchez,2005). Body size may play a role in the effect of sucrose concentration on T_th excess, because the bumble bee foragers that we studied are larger than honey bees, which are in turn larger than M. panamica. This may be an interesting question to pursue in future comparative studies within and between bee groups.

In our study, the tropical species, B. wilmattae, maintained thoracic temperatures on average between 29°C and 32°C while foraging,similar to the range reported by Heinrich and Heinrich in a study of temperate Bombus terricola foraging on field spiraea (Spiraea latifolia) and goldenrods (Solidago sp.)(Heinrich and Heinrich,1983a). Because there have been no similar studies on the effect of sucrose concentration on temperate bumble bees foraging while perched,future investigations may look for differences between the T_th and ΔT_th of tropical and temperature bumble bees flying to food sources. For example, at 2.5 mol l^-1, we found an average T_th of 31.6°C at an average T_a of 24.7°C. Studies conducted under similar ambient temperatures and sucrose concentrations will be necessary to draw conclusions on potential differences or similarities between tropical and temperate bumble bees.

The forager abdominal pumping movements that we observed during our experiments are common in a variety of insects, and are generally accompanied by a rapid rise in thoracic temperature(May, 1979). In particular,such movements are associated with thoracic heat generation in a wide variety of bumble bee species (Heinrich,1993). Fibrillar muscles are responsible for the generation of bumble bee thoracic heat (Heinrich and Kammer, 1973), and contractions of B. impatiens thoracic flight muscles, particularly the dorsoventral muscle fibers, were associated with flight warmup (Esch and Goller,1991). In the shade (as in our experiments), the difference between ambient and thoracic temperatures is primarily a result of flight metabolism or thermoregulation through muscular activity during the times when they are perched on flowers (Heinrich,1972a).

Previous studies hint at a similar effect of food quality on T_th. When Heinrich added concentrated, viscous sugar syrup to fireweed flowers, he reported that feeding B. vagans foragers held their wings folded dorsally with rapid abdominal respiratory movements (as we observed in our study) (Heinrich,1972b). After 2 min of feeding, bees had an average T_th of 34.8°C, significantly higher(P<0.05) than those foraging on fireweed with normal,un-supplemented nectar at the same range of ambient air temperatures. There were significant differences in T_th based upon flower type: higher T_th in bumble bees foraging on raspberry flowers than on fireweed or spiraea(Heinrich and Heinrich,1983b). Based upon our results, we suggest that the quality of nectar reward may have contributed to some of these T_thdifferences in B. terricola. For B. terricola foraging in the field on Asclepias syriaca nectar, mean T_thwas relatively independent from T_a (regulated approximately near 36°C, depending upon exposure to shade or sun), unlike mean T_th of bees foraging from Spiraea latifoliaor Solidago canadensis, which were more strongly influenced by ambient air temperatures. Interestingly, individual A. syriacaflowers contained larger volumes of nectar than S. latifolia or S. canadensis flowers (Heinrich,1972a). These results suggest that volume of a reward may also have an influence on T_th, as is found in honey bees, where higher nectar flow rates resulted in higher T_th(Waddington, 1990). We hypothesize that bumble bee forager T_th will be correlated generally and positively with nectar sugar content and flow rate.

Nieh and Sánchez hypothesized that all corbiculate bees possess the ability to adjust body temperature based upon food rewards(Nieh and Sánchez,2005), and supporting data had been found in the honey bees and stingless bees (Nieh and Sánchez,2005; Stabentheiner,2001). The present results demonstrate that at least one species within the bumble bees increases its thoracic temperature with increased carbohydrate reward. Whether this effect can be observed in other groups of bees, such as some of the Halictidae, remains to be demonstrated, but we suggest that this is quite likely. The modulation of thoracic temperature in accordance with food value may be a widespread adaptation facilitating the rapid collection of good food in flying Hymenoptera.

We would like to thank Don Felipe Pérez Verdugo for his help with locating the colonies, and Daniel Sánchez and Omar Argüello for their logistical support and unfailing good cheer. The ORBS (Opportunities for Research in the Behavioral Sciences) program provided data analysis assistance. Sarah Corbet and anonymous reviewers provided valuable comments that have significantly improved our manuscript. This research was supported by NSF Grant IBN 0316697, NSF Grant OISE 0503478, and UC Mexus 2003-6552.