Performance benefits of stable, warm muscles are believed to be important for the evolution of endothermy in mammals, birds, and flying insects. However, thermal performance curves have never been measured for a free-flying endotherm, as it is challenging to vary body temperatures of these animals, and maximal flight performance is difficult to elicit. We varied air temperatures and gas densities to manipulate thoracic temperatures of flying honey bees from 29-44°C, with low air densities used to increase flight metabolic rates to maximal values. Honey bees showed a clear thermal performance curve with an optimal temperature of 39°C. Maximal flight metabolic rates increased by ~2% per 1°C increase in thoracic temperature at suboptimal thoracic temperatures, but decreased ~5% per 1°C increase as the bees continued to heat up. This study provides the first quantification of the maximal metabolic performance benefit of thermoregulation in an endotherm. These data directly support aerobic capacity models for benefits of thermoregulation in honey bees, and suggest that improved aerobic capacity likely contributes to the multiple origins of endothermic heterothermy in bees and other insects.

We manipulated body temperatures and assessed maximal capacities of bees by flying them in various air densities and temperatures (Table 1, 2). Foragers were collected in random order from three colonies of the Italian honey bee, Apis mellifera ligustica, maintained on the third-story balcony of the Interdisciplinary Science and Technology Building 1 at Arizona State University in Tempe, AZ, USA. Unloaded, outgoing foragers were captured when leaving the colony by holding an opened plastic bag (~950 mL) approximately 15 centimeters from the colony entrance. After a single forager flew directly into the opened bag, it was sealed and the bee was quickly transported to a room regulated at 23±0.5°C or 35±0.5°C (EGC, Chagrin Falls, Ohio, USA) and its flight metabolism was assessed at a single air density.

Substituting helium for nitrogen in air lowers its density, requiring bees to generate more lift in order to fly (19,22). This substitution will also increase heat loss rates because helium has a thermal conductivity about six-times higher than nitrogen (21,23,24). To further manipulate heat loss, we examined metabolic rates and body temperatures at two air temperatures, 23 and 35°C.  Heat loss rates are proportional to thermal gradient between an animal’s body and ambient temperature. Thus, we predicted that flight in gases enriched in helium at low air temperatures would induce the greatest heat loss rates and therefore the coolest body temperatures, whereas heat loss would be lowest in nitrox mixtures at 35°C air temperatures.

Metabolism during free flight was assessed in a cylindrical, transparent acrylic flight chamber (350 mL). After placing the bee in the chamber, it was sealed and covered with a dark cloth to encourage reduced activity of the bee and the chamber was flushed to completely replace atmospheric air and water with the desired gas mixture. Gas mixtures were created by using cylinders of pure O2, N2, and He, which were regulated at a total flow rate of 2 L·min-1 by a multi-channeled Sable Systems FB8 flow meter system (Las Vegas, NV, USA). Each flow meter was calibrated for the different gas densities using a soap-film bubble meter. The gases from the flow meters flowed sequentially through a CaSO4 and soda lime column to remove H2O and CO2, the reference cell of a LI-COR 6262 CO2/H2O analyzer (Lincoln, NE, USA), the respirometry chamber, a small column of MgSO4 to remove water produced by the bee, and then the sample cell of the LI-COR. Differential analog output from the LI-COR was digitized (Sable Systems UI-2) and recorded each second (Expedata, Sable Systems, Las Vegas, NV). The LI-COR was calibrated using 252 ppm CO2 and Ultra-Zero calibration gases at the same flow rate and pressure (761.5-761.8 mm Hg) as during the flight respirometry, and baseline recordings were taken before and after each measurement period.

Flight was then encouraged for two minutes by shining a 150W dual goose-neck Fiber Optical Illuminator (China) over the chamber. Bees that landed were immediately encouraged to fly by gently tapping and inverting the chamber. Flight behavior was categorized based on ability, duration, and control (i.e., quality; 25). Flight was categorized and ranked as: 1 – no flight, 2 – flapping wings with brief periods of flight (< 3 sec), 3 – intermittent flight characterized by frequent crashing (i.e., bee usually ends upside down), 4 – intermittent flight characterized by frequent controlled landing (i.e., bee gently lands on its feet), or 5 – continual, stable flight. Expedata (Sable Systems, Las Vegas, NV) was used to find and average the 10 seconds with the highest CO2 readings during each trial. Flight CO2 emission rates (mL·hr-1) were calculated by multiplying the decimal CO2 fraction times the STP flow rate through the flight chamber. After flight CO2 emission rates were measured, the bee was shaken into a plastic bag, which was flattened to restrict the bee’s movement. Thoracic temperature was then measured by inserting a Physitemp model MT29/1 hypodermic microprobe (Clifton, New Jersey, USA; 29-gauge, time constant=0.025·s) through the bag and into the center of the thorax. The temperature data were recorded with a Pico Technology USB TC-08 Thermocouple Data Logger (Tyler, TX, USA). Thoracic temperatures were measured within five seconds of cessation of flight, and the highest temperature reported by the thermometer was recorded. After measurement, the bee was weighed (±0.1 mg) using an A&D HR-120 Analytical Balance (Tokyo, Japan) and stored at -20°C.

Data were analyzed using R (3.6.2; R Foundation for Statistical Computing, Vienna, Austria). Two-tailed significance was determined at α = 0.05. We used a linear mixed-effects model to test the independent and interactive effects of air temperature and gas density on flight metabolic rate [i.e., milliwatts (mJ·sec-1)] and thoracic temperature, with hive included as a random effect. To determine the independent effect of gas density on thorax temperature, we ran a linear model for each separate air temperature. We also ran a similar model for the above independent variables, with body mass included in the model. We used an ordinal logistic regression model analysis to test the independent and interactive effects of air temperature, gas density, and thoracic temperature on flight quality. Models were chosen using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC).

Data were analyzed using R (3.6.2; R Foundation for Statistical Computing, Vienna, Austria).