Roles of individual honeybee workers and drones in colonial thermogenesis

The nest-warming capacity of honeybee (Apis mellifera) colonies has helped allow this tropical insect to expand its range into cold-temperate environments. Although the behavioural and physiological responses of honey-bee colonies to cold are well characterized (Simpson, 1961; Seeley & Heinrich, 1981; Kronenberg & Heller, 1982; Southwick, 1985), the division of the labour of thermogenesis among individual bees has received little attention. In this study, I examine the thermogenic roles of individual workers and drones in a colony of brood-rearing honeybees, A. mellifera.

Colonies with brood regulate brood areas at a temperature between 30 and 35 °C (Seeley & Heinrich, 1981). Survival and normal growth of honeybee brood requires brood nest temperatures between 30 and 36°C (Himmer, 1927, 1932; Jay, 1963). In colonies with brood, environmental temperatures below 30°C stimulate an increase in colonial metabolic rate, and temperatures below 15 °C induce clustering centred on brood areas (Corkins & Gilbert, 1932; Free & Simpson, 1963; Southwick & Mugaas, 1971; Koeniger, 1978; Kronenberg & Heller, 1982; Ritter, 1982; Southwick, 1985).

It is not known how the increase in colonial metabolic rate is divided among individual bees. In winter clusters, it has been reported that only bees in the cluster centre activate their flight muscles for heat production (Esch, 1960). However, in swarms, resting metabolism suffices for heat production and the thermoregulatory pattern is consistent with a simple model where bees individually thermoregulate towards a preferred body temperature (Heinrich, 1980, 1985). A number of factors may influence whether an individual bee contributes to colonial heat production in a summer, brood-rearing colony. Young bees generally feed and care for larvae and are usually located near the colony centre (Seeley, 1982). However, workers increase thermogenic and thermoregulatory capacities with age and foragers have the greatest thermogenic capacities (Himmer, 1925; Allen, 1959; Nijland & Hepburn, 1985; Harrison, 1986). Himmer reported that worker temperature in a colony increases with age (Himmer, 1932). Drones are also present in summer, brood-rearing colonies. Drones have poorer thermoregulatory capacities than workers (Cahill & Lustick, 1976), and are generally thought to contribute little to colony work (Wilson, 1971). However, bumble-bee males have recently been shown to incubate pupae (Cameron, 1985).

In this study, I investigated the effect of age, foraging behaviour, location on or off the brood area, and sex on bee contribution to colonial heat production. I also examined the effect of age and sex on bee location within summer clusters.

The honeybee colony used in this study was started from a commercial 21b package (McCary Apiary; Buckatunna, MS) on 1 May 1983, and maintained at the University of Colorado, Boulder. Beginning on 15 June, 100 newly emerged bees were added to the colony each day until 13 August. Introduced bees were obtained from several donor colonies and were colour marked as described previously (Harrison, 1986). No differential mortality, measured using a dead bee trap (Gary, 1960), occurred among introduction days. The effect of age on foraging behaviour was characterized on 13, 14 and 15 August using a technique described previously (Harrison, 1986).

Thermoregulatory experiments were conducted with the four brood-containing frames from this colony during 9–12 August. Frames containing brood (both pupae and larvae) and approximately 1000 bees were removed from the colony at 12.00 h ± 0·25 h (range) and placed within a wooden single frame box. At 16.00 h ± 0·25 h the frame was transferred to a 15 °C ± 0-5°C temperature-controlled room. The box containing the frame was edged with clear Mylar cut with an 8x4 array of flaps which were taped shut. The flaps could be opened to allow access to the frame and bees. Throughout the time the frame was kept at 15 °C, brood surface temperatures were maintained above 30°C by the bees. After 1–1·5 h, the bees began to cluster. I studied the division of thermogenesis in the frames of bees during the 30 min prior to clustering. Temperature measurements were made of bees located behind the windows, and age and location of the bee were recorded. A bee touching an adjacent bee could be heated passively by its neighbour. Therefore, only bees not touching other bees were measured.

The thoracic surface temperatures of individual bees were measured with a Bailey MT-4 micro temperature probe which was slipped in through the windows in the Mylar edging to touch the dorsal surface of the bee. The microprobe used had a time constant of 0·25 s. When freshly killed bees were heated internally with a wire resistor, thoracic surface temperature was always within 0·5°C of internal temperature, indicating that surface temperature accurately reflected bee thoracic temperature. Internal thoracic temperature was measured with an implanted 0·25 mm copper-constantan thermocouple connected to a Wescor thermocouple thermometer. Bees feeding larvae and bees regurgitating food to one another were measured without sign of disturbance. Only measurements during which there were no behavioural signs that the bee was disturbed were used.

Although ambient temperature was 15°C, air temperature within the box containing the frame of bees varied relative to the location of the brood. Air temperature 1 cm above the comb in the frame centre averaged 28°C; while near the frame edge air temperature averaged 21 °C 1 cm above the comb. For this reason, bee thoracic surface temperature (T_th_s) was expressed relative to local air temperature 1 cm above the bee (T_a). This temperature was measured with a second temperature probe taped at a fixed distance (1cm) above the first probe. As T_ths−T_a could be affected by movement of the bee within the frame, only bees observed to remain in a given area for >30 s were measured. Only bees with T_ths above T_a will contribute to colonial heat production.

The four frames described above were maintained overnight ( 16 h) at 15 °C. In all cases, the bees were tightly clustered over the brood the next morning, with temperatures at the cluster centres averaging 33 °C. I recorded the age of marked bees in the outer layer of the cluster. Clusters were then dispersed by directing a heat lamp at the cluster. The clusters were 4–5 layers thick. The age and sex of marked bees directly over the brood (in the innermost layer) were recorded.

To evaluate the local thermal environment of different parts of the frame, I measured cooling curves of bees freshly killed with carbon dioxide at various frame locations under conditions similar to those described above. Wire mesh cylinders (1 cm radius, 0·3 cm mesh) were installed between the Mylar edging and the comb surface. These cylinders allowed measurement of cooling curves without interference from patrolling bees. Preliminary experiments demonstrated that cooling curves of dead bees measured inside a cylinder were identical to cooling curves of bees not enclosed within a cylinder. Cooling curves were measured at four sites over the brood near the frame centre, and at four sites away from the brood near the frame edge. Cooling curves were measured by implanting a 0·25 mm copper-constantan thermocouple into the thorax of a freshly killed bee. The bee was connected by 0·5 cm of thermocouple wire to a small (0·2 cm diameter) wooden splint.. The bee was heated with a microscope lamp until internal thoracic temperature (T_th) measured 40°C, and then the bee was quickly placed within a cylinder 0·5 cm over the comb. Local ambient temperature (T_a) was measured with a second thermocouple located 1cm above the bee and T_ths−T_a was continuously recorded on a Linear Instruments chart-recorder. Cooling constants of the thorax (min⁻¹) were converted to thoracic conductance (C,.cal g⁻¹ °C⁻¹ min⁻¹) assuming a specific heat of 0·8 calg⁻¹⁰C⁻¹.

Prior to cooling, bees appeared to be evenly distributed over the frame. During the preclustering period, bees on and off the brood area walked, cleaned cells, exchanged food, groomed or stood quietly. Individual bees moved on and off the brood area; however, the density of bees located over the brood area slowly increased until the cluster was formed after 1–1·5 h at 15°C.

Data from the four trials were not pooled as average bee age varied among trials (Table 1). Bees in the inner layer were significantly younger than bees in the outer layer of clusters in three of the four trials, and were close to being significantly younger in the fourth trial (t-test, P= 0·058, Table 1). One-day-old bees were all located at the centre of the cluster (four trials, 49 bees), as were drones (four trials, 52 bees).

The elevation of T_ths above T_a was quite variable among workers (Fig. 1). Of the unmarked bees 89 % of those probed had T_ths ⩾2°C above T_a, indicating that most bees contribute to colonial heating; average T_ths−T_a was 3·6°C (S.E. = 0·17, N= 145). When marked and unmarked bees are included in the calculation, T_ths averaged 4·1 °C above T_a. T_ths of bees over the brood averaged 7-0°C warmer than bees off the brood area, but T_ths−T_a did not vary with location (Table 2).

As the age of drones was not known, drone temperatures were compared to those of unmarked workers. T_ths−T_a for drones (x̄ = 3·48, S.E. =0-31, 2V = 41) did not differ from that observed for unmarked workers (x̄ = 3·61, S.E. = 0-17, N= 145; t-test, P = 0·594).

One-day-old bees had significantly lower T_ths−T_a than older bees (Scheffe’s a posteriori test, 0·05 level), and in many 1-day-old bees T_ths equalled T_a (Fig. 2).

However, T_ths−T_a did not differ among all bees 2 days of age and older (Scheffe’s test, 0’05 level). Peak foraging behaviour occurred in workers that were 30–37 days old (Fig. 2). T_ths−T_a of bees of foraging age did not differ from younger bees (Fig. 2).

Temperatures of dead bees equalled T_a when equilibrated, indicating that T_a accurately reflected operative environmental temperature. Thoracic cooling curves of dead bees were linear when plotted on a semilogarithmic plot. All linear regressions for the cooling curves were highly significant (all r²>0·98) and the slopes were not significantly different among sites (on the brood: x̄ = 0·447, S.E. = 0·036, N = 13, average T_a = 25·1 °C; off the brood: x̄ = 0·453, S.E. = 0·022, N= 12, average T_a=19·9°C; t-test, P = 0·894). When all data are pooled, the cooling constant averaged 1·036 min⁻¹, and the calculated thoracic conductance, C, was 0·829 cal g⁻¹ °C⁻¹ min⁻¹.

In honey-bee colonies containing brood, most of the individual bees contribute to colonial thermogenesis (Fig. 1). The age-related transition to foraging of worker bees is not associated with a significant increase in heat production on the comb (Fig. 2). All workers 2 days of age and older contribute equally to colony heat production under these conditions (Fig. 2). This is possible because of the relatively low cost of thermogenesis. The thorax-specific necessary for the average elevation of T_ths above T_a (4·1 °C), is 40 ml O₂g-thorax⁻¹ h⁻¹, only 15% of the thorax-specific maximal oxygen consumption rate of 4-day-old bees (Harrison, 1986). The calculated average metabolic rate of 58Wkg⁻¹ (assuming 115 mg average bee mass) is within one standard deviation of the average daytime metabolic rate determined for frames of honeybees at 15°C, as determined by Kronenberg & Heller (1982). This supports the validity of these indirect measures of honeybee heat production.

Despite a 7°C average difference in T_ths between workers on and off the brood, average heat production per bee is not affected by worker location. This result must be interpreted with caution, as it is unlikely that conductances will be identical in different parts of a honeybee hive as they are on the single frame. Thoracic cooling constants of dead bees on the frame average 58 % higher than those measured under standard conditions (Heinrich, 1980). Cooling constants (min⁻¹) are measured with respect to T_a, which equals box wall temperature under standard conditions. For bees on the cold-exposed frame, box wall temperatures are cooler than T_a, substantially increasing the rate of heat loss relative to T_a.

An important question is whether thermoregulation may be achieved simply with resting heat production rates as occurs in swarms (Heinrich, 1981). CO₂ production of honeybees engaged in moderate activity such as slow walking and cleaning appendages averages 10·1 ml CO₂g⁻¹ h⁻¹ (Louw & Hadley, 1985). Assuming a Q₁₀ of 2 and RQ = 1 (Southwick, 1985), at 21 °C will be 22 ml O₂g-thorax⁻¹ h⁻¹, and resting metabolism will provide roughly 90 % (on the brood) and 55 % (off the brood) of the heat production necessary for the 4·1 °C elevation of T_ths over T_a. However, in Louw & Hadley’s study, muscle activity was not measured. Bumblebees monitored to exclude the possibility of flight muscle activity have a resting VQ₂ of 2mlCO₂g-thorax⁻¹ h⁻¹ at 21°C (Kammer & Heinrich, 1974). If metabolism is similar in honeybees, then metabolism without flight muscle activity will contribute only 10 % (on the brood) and 5 % (off the brood) of the heat necessary to achieve the average T_ths−T_a. Thoracic temperatures of individual bees implanted with thermocouples and exposed to cooling frame temperatures exhibited intermittent rises and falls (Kronenberg & Heller, 1982). This supports the hypothesis that shivering thermogenesis by individual bees is necessary for thermoregulation in cold-exposed frames of bees prior to clustering.

Drone cooling constants may be estimated from the allometric equation for bees reported by May (1976; drone thoracic mass averages 85 mg, J. Coehlo, personal communication). If drone cooling constants are increased 58% when frames are exposed to 15 °C as found for workers, drone thoracic conductance will be 0·488 cal g⁻¹ °C⁻¹ min⁻¹, and the average heat production of a drone will be 0·145 cal bee⁻¹. Accordingly, each drone will contribute l·5 times as much heat as a worker to colonial thermogenesis, mainly because drones are larger.

It is unclear why drones should invariably be located within the centre of the clusters. Drones are inferior to workers in thermoregulation at temperatures below 25 °C (Cahill & Lustick, 1976), so perhaps their location within the cluster centre, where temperatures are higher, allows their continued contribution to colonial thermogenesis. Alternatively, drone presence in cluster centres may increase sperm viability. Honeybee sperm decreases in viability in vitro below 15°C (J. Harbo, personal communication). Males of the eusocial Hymenoptera rarely contribute to colony welfare (Wilson, 1971). This study and the observations by Cameron (1985) on bumble-bees demonstrate that drone contribution to colony heat production represents an exception to the general lack of altruism by male Hymenoptera. However, the energetic cost of this behaviour is low.

Individual bees seek out environmental temperatures near 35 °C (Heran, 1952). The pattern of thermoregulation demonstrated here is not compatible with a simple model where individuals attempt to maintain their own body temperature near the preferred temperature for individual honeybees, as has been hypothesized for swarms and clusters within the hive (Heinrich, 1985). Bees on the frame periphery allow their thoracic temperatures to drop to an average of 25 °C. However, the attraction of bees to the brood area may be partly due to behavioural thermoregulation. These experiments suggest that honeybee brood warming before clustering is due to relatively uniform heat production by individual bees combined with the attraction of bees to the brood area.

This project was supported in part by a Sigma Xi Grant-in-aid of Research and a Graduate Research Award from the University of Colorado to JMH, and by NIH grant AM 31243 to T. T. Gleeson. I thank my advisor, Dr T. T. Gleeson, for guidance beyond the call of duty, and J. Fewell for exceptional support, help and advice. M. Breed, J. Coehlo and three anonymous reviewers provided many helpful comments.