Insect pollinators, including honey bees (Apis mellifera), are essential for agriculture and terrestrial ecosystem function, yet their responses to warming conditions and heat waves remain poorly understood. Honey bees have well-documented mechanisms to cope with heat exposure that could help them adapt to extreme heat. However, there have been no studies to date that have assessed how natural heat waves affect the capacity of honey bee colonies to thermoregulate and grow. To test the hypothesis that excessive heat impairs honey bee colony growth by exceeding the thermoregulatory capacity, we studied how variation in summer temperatures affected hive temperature regulation and colony growth during a desert summer in which maximal shaded air temperatures intermittently exceeded 40°C. We monitored the growth of nine colonies biweekly for three months and recorded temperatures at the center and edge frames of the brood nest, as well as on combs at the outer edge of the hive body.  Average temperatures in the brood center and edge were quite stable and within the optimal range of 34-36°C necessary for healthy brood development throughout the summer.  However, all hive locations exhibited cyclic, diurnal thermal fluctuations, and brood experienced considerable portions of each day (14% for the brood center, 33% for the brood edge) above and below the optimal temperature range. Higher maximal air temperatures and greater temperature fluctuations within the hive led to declines in colony population. These findings suggest that excessive heat, with maximal temperatures exceeding 40°C, can reduce colony populations by impairing the thermoregulation of brood or by exposing adults to temperatures that shorten their lifespans. If excessive heat periods occur more frequently as predicted due to climate change, this could limit regions where colonies can successfully survive the summer.

Colony establishment and maintenance

From May to August 2022, nine colonies of Apis mellifera ligustica were monitored at the Arizona State University Polytechnic campus. Before the experiment commenced, each colony was provided with a new queen (in April 2022), sourced from Pendell Apiaries Inc. in Stonyford, CA (39.376956, -122.558801). Initial colony sizes varied from 3,000 to 13,000 bees, with four double-box hives and five single-box hives.

Colony population assessments

The response of the colonies to climatic conditions was monitored through regular assessments of their growth and development.  We inspected every hive biweekly to estimate the number of adult worker bees and brood levels across all life stages from May 18th, 2022, to August 22nd, 2022. We took two sets of photographs of each side of every frame using a Canon® EOS Rebel T5 camera. The first set, taken with adult bees still on the frame, was analyzed using ImageJ (National Institutes of Health) to estimate the number of adult workers through “Multi-point” counting (Fisher II et al., 2022). For the second set, we gently shook and brushed the bees off the frame and placed a 877.2 cm² grid over the comb to measure how many cm² were occupied by the pupae. If a grid cell wasn’t completely filled, we calculated a fraction based on the number of filled cells out of 23 (the average number of cells in our grid cells of 6.45 cm²). Because eggs and larvae were difficult to detect in the photographs, an observer estimated their total abundance in the field by counting the number of grids and cells occupied by eggs or larvae within the 877.2 cm² grid.

Temperature Logging

Temperature was monitored in the hives at 6-minute intervals from May 18th to June 13th and at 30-minute intervals from June 14th to August 22nd using Telid® RFID temperature loggers (Microsensys, Erfurt, DE).  Each hive box was equipped with five loggers. Two loggers were positioned near the edges of the peripheral end frames, which were generally partially or completely filled with food reserves (usually honey) and lacked brood.  Three additional loggers were allocated to the brood frames, including one in the brood's center and two at the brood area's edge for each box. The locations of the loggers were adjusted during the population surveys, if necessary, to maintain their positions relative to the brood. The temperature data were extracted using a Telid® RFID sensor and software (Microsensys, Erfurt, DE).

Phoenix-Mesa Gateway Airport (Mesa, AZ: 33.307833, -111.655472) is located approximately 1.5 miles from the ASU honey bee laboratory, and its temperature readings serve as our source for shaded air environmental data. The data were obtained from the MesoWest website (https://mesowest.utah.edu/), developed by the University of Utah. Temperature is recorded every 20 minutes from 0 to 5 am; for the remaining time in a day, temperature is recorded hourly. Daily variability of temperatures in the hives was assessed through the diurnal temperature range (maximum minus minimum recorded for that day). All loggers were analyzed by logger positions: Outer Edge (OE), Brood Edge (BE), and Brood Center (BC). To examine the relationship between temperature and colony population size variation, we calculated the average diurnal thermal range over two weeks for temperature loggers, as well as the two-week averages of daily maximum and minimum temperatures. Additionally, to assess the amount of time that bees experienced potentially stressful temperatures, we calculated the duration brood spent at temperatures outside of the optimal thermal range of 34- 36 °C.