Optimization of Greenhouse Crop Pollination through Artificial Homeostatic Control of Bumblebee Hive Temperature

Final report for GNE19-222

Project Type: Graduate Student
Funds awarded in 2019: $13,931.00
Projected End Date: 05/30/2022
Grant Recipient: The Pennsylvania State University
Region: Northeast
State: Pennsylvania
Graduate Student:
Faculty Advisor:
Dr. Rudolf Schilder
Penn State University
Expand All

Project Information


Enhancing pollination service by insects will be vital to meet increasing crop harvest demands. Bumblebees are increasingly used to fulfill these pollination needs, as they provide a less labor-intensive process compared to mechanical pollination. Bumblebees are fantastic pollinators, but have colony demands that compete with foraging (and therefore pollination) and are associated with colony survival. One of these is the need to regulate brood temperature within a specific range to ensure proper development of larvae. Any worker occupied with brood thermoregulation will not forage, resulting in decreased pollination service. I hypothesized that lowering the need for brood thermoregulation experimentally will enhance the number of foraging workers and increase pollination service. I tested this hypothesis in Bombus impatiens bumble bees by experimentally manipulating colony temperature and examining the impact this had on foraging rates and thermoregulatory behavior. I found that fewer bees exhibit thermoregulatory behavior when their colony temperature is experimentally increased, but no increase in the number of workers foraging was observed, supporting the idea of a thermoregulatory reserve rather than task flexibility. I conclude that artificially controlling colony temperature is not advisable in applied settings, as there does not appear to be an increase in pollination.

Project Objectives:

Objective 1: Determine how artificial homeostatic control of colony temperature affects pollination services relative to a traditional commercial colony box.

Objective 2: Quantify how much time workers spend brood incubating in temperature-controlled colonies compared to a traditional commercial colony box.


Pollinator populations are declining worldwide, while food and crop demand continually increase. Therefore, growers have to rely more heavily on commercially available pollinator colonies. While honey bees are a popular choice, some crops can only be pollinated through buzz pollination, or need pollination service across a wider range of environmental temperatures than honey bees can supply. Commercially available bumblebees provide these features, and to help meet food and crop demand increases, there is therefore a need to examine ways in which we can optimize the efficacy and health of bumble bee colonies

Bumble bees are increasingly used to fulfill pollination needs, as they provide a less labor-intensive process compared to mechanical pollination. Bumble bees are highly effective pollinators, but have interests that compete with foraging (and therefore pollination) associated with colony survival. One of these is the need to regulate brood temperature within a specific range to ensure proper development of larvae. Bumble bees task switch between colony maintenance and foraging so any worker occupied with brood thermoregulation will not forage, resulting in decreased pollination service. My work looks to examine if by enhancing brood thermoregulation artificially, it will enhance the number of foraging workers and increase pollination service.


Materials and methods:

Colony treatments: Colonies were randomly assigned one of two treatment groups (temperature-controlled or control), with 3 colonies being used per group. Temperature controlled and control colonies were tested simultaneously. Colonies were placed inside of the growth chamber and allowed 3 days to acclimate and learn where the petri dish was before recordings began. Colonies were recorded for 4 days. During this time, recordings via the thermal camera were made twice daily for 15 mins.

Both colonies were placed on platforms (1.5” tall) inside their respective shade/temperature-controlled box which the colony rests on, but access to their sugar box is not obstructed. A growth chamber was used to manipulate ambient temperature. The ambient temperature was set to 24 °C. Lights in the growth chamber were set to a 12 hr day/night cycle.

Temperature controlled box: The temperature-controlled box was constructed of 1” insulation foam housed in 1/4” plywood (interior dimensions L x W x H = 18” x 16” x 12”). PEX pipe was run along the interior perimeter of the box to serve as a radiator. A circulating water bath pumped maintained the internal temperature of the box at 29 ± 0.2 °C by temperature-regulated water through the piping. The top of the temperature-controlled box had a layer of Shrink Film Insulation to allow for imagery of the colony.

Camera stand: A camera stand was constructed above the temperature-controlled box and around the control using 1” x 2” common board wrapped in black fabric to prevent white light from disturbing either colony. This stand allowed for both the Canon 7D EOS DSLR camera and the thermography camera to be mounted vertically 20-24” above each colony. Recordings with both cameras were made to analyze thermoregulatory behaviors: wing fanning and brood incubation. However, initial analyses showed minimal wing fanning in either the control or temperature-controlled colonies, likely due to the ambient temperature (24 °C) being below the preferred thermal window of B. impatiens (28-32 °C (Vogt, 1986)). Therefore, the digital recordings and the amount of bees’ wing fanning were not analyzed further.

Thermal recordings: For each flight arena setup, K-type thermocouples were placed in the following locations: inside the flight arena (x1), inside the shade/temperature-controlled box (x2), inside the colony within 2 cm of brood (x2), and inside the colony but wrapped in black electrical tape for thermal calibration (x1, see below). Thermocouples were plugged into two TC-08 Picologgers. Temperatures were logged once per second from 8:30 on the first day of the experiment until 17:00 on the last day.

Body temperature was recorded with an infrared thermography camera mounted on the camera stand. Emissivity was set to 0.97, and the camera recorded images at 1 frame/ 30 seconds. Obtained thermographs were analyzed with Xeneth v2.5 software. A K-Type thermocouple was attached to cardboard and wrapped in black electrical tape (Scotch), with the probe unobstructed. This thermocouple was then placed inside of the colony and used to calibrate the temperature readings in the Xeneth software. For each thermal image, the thoracic temperature of each visible bee was measured, and each bee was marked as being on or off brood.

Arena set up: Flight arenas (L x W x H = 36” x 18’ x 36”) were constructed of 1” x 2” common boards for the frame, 1/4” plywood on the base, and plexiglass on the top. The frame was wrapped with two layers of brown insect mesh on the sides. PVC tubing leading from the colony was mounted through 3/4” plywood on the colony-side of the flight arena. The opposite end of the arena had a hinged door that allowed for pollen to be placed on a 15” high stand made of 3/4” plywood supported by 1” dowels. This elevated platform was used to encourage flight (instead of crawling) and painted blue to help bees find the pollen and make analyzing the recordings easier. The mesh and the plexiglass lid allowed plenty of light inside of the arena so bees could see and orient themselves.

Colony boxes were connected to a flight arena via 1/2” I.D. 5/8” O.D. PVC tubing, with a 3D printed tube in the middle of the tubing in and out of the arena. As this tubing is the only way in and out of the colony, this helps keep all bees within focus of the digital and thermal cameras and prevents bees flying into them. The methods for the break beam sensors were based on the methods described in (Kevan et al., 2009). The 3D printed tube holds two adjacent 3mm IR break beam sensors. An Arduino mega board was programmed to recognize what and when each beam was broken and is logged via TeraTerm (v. 4.106). The Arduino code for both tracking the beams, and the subsequent R code for analyzing entry vs exit has been completed. It was determined that the break beams need to record to the milli-second, which was not done for these experiments, but is ready to be utilized for the future. They were therefore not analyzed for these results.

On the platform in the flight arena, a petri dish contained 6-7 g of honey bee-collected pollen from the spring of 2020 in central PA. The pollen was weighed daily before being placed in the arena and at the end of the day to assess how much pollen was collected per day. Whole colony mass respective sugar boxes were also weighed weekly to track colony development and energy usage. Above the flight arena, a GoPro HERO6 Black was placed which recorded the arena from 8:30 to 17:00. This was utilized to track the number of individuals on the petri dish over time, a proxy for pollination rates. The number of bees on the petri dish was counted every 15 mins for a total of 35 time points. The total number of bee visits was determined by the sum of the number of bees on the petri dish (divided by the total number of individuals counted in that colony). The maximum number of bees on the petri dish was measured as the highest number of individuals on the petri dish at a single timepoint (divided by the total number of individuals counted in that colony).

Research results and discussion:

The temperature controlled box maintained a consistent temperature inside of the colony. The percentage of bees performing brood incubation was higher in control colonies than in temperature controlled colonies. 

Linear models were used to determine how treatment affected the change in mass for each colony, the amount of sugar solution consumed and how much pollen was collected. A type II ANOVA indicated that there is not a significant difference between the changes in mass or sugar solution consumption for control and temperature-controlled colonies. For pollen collection, treatment was significant, but day was not significant.

For the maximum number of bees recorded at a single time point in a day, a linear model was used to determine the effect of day and treatment group. Day did not have a significant effect, but treatment was significant.

For the total number of bee visits recorded in a day, a linear model was used to determine the effect of day and treatment group. Neither day or treatment had a significant effect.


There is a trend of temperature-controlled colonies gaining more colony mass and consuming less sugar solution than control colonies, although it is not statistically significant. These results could be due to the limited sample size utilized in this study. However, the amount of pollen collected per day was significantly affected by treatment; control colonies collected more pollen than temperature-controlled colonies.

The amount of pollen collected was used as a proxy for pollination rates, but pollination rates are likely even higher under natural settings. There is a diminishing return such that the pollen removal per visit is negatively correlated with the probability of pollen grain dispersal (Larson & Barrett, 1999). As such, plants restrict the amount of pollen removed during a single visit (Harder & Thomson, 1989). In this study, as the pollen was readily available all at once, bees did not have to make as many visits to obtain the same amount of pollen. It could therefore be that when the amount of pollen in a single visit is restricted, the number of visits would also increase, which increases the amount of pollination occurring due to more flowers being visited. However, despite an overall increase in pollination rates, it would likely be scalable to what is demonstrated here, meaning the treatment differences would still be present. 

Interestingly, the total number of bees seen on a petri dish was not different between the treatments, but the maximum number of bees seen on a petri dish at a single time point was. However, as the total number of visits from each treatment was not statistically different, it is likely the two treatment groups would visit a similar number of flowers. Control colonies are therefore suggested to be better pollinators than temperature-controlled colonies, due to them collecting more pollen despite differences in visitation.


Kevan, P. G., Cooper, E., Morse, A., Kapongo, J. P., Shipp, L., & Khosla, S. (2009). Measuring foraging activity in bumblebee nests: a simple nest-entrance trip recorder. Journal of Applied Entomology, 133(3), 222-228. doi:10.1111/j.1439-0418.2008.01338.x

Harder, L. D., & Thomson, J. D. (1989). Evolutionary Options for Maximizing Pollen Dispersal of Animal-Pollinated Plants. The American Naturalist, 133(3), 323-344. doi:10.1086/284922

Larson, B. M. H., & Barrett, S. C. H. (1999). The Ecology of Pollen Limitation in Buzz-Pollinated Rhexia virginica (Melastomataceae). The Journal of ecology, 87(3), 371-381. doi:10.1046/j.1365-2745.1999.00362.x

Vogt, F. D. (1986). Thermoregulation in Bumblebee Colonies. I. Thermoregulatory versus Brood-Maintenance Behaviors during Acute Changes in Ambient Temperature. Physiological Zoology, 59(1), 55-59. doi:10.1086/physzool.59.1.30156090

Research conclusions:

The method developed for externally controlling the colony temperature was effective. However, this external temperature control resulted in less foraging behavior (i.e. fewer visits to the pollen dish, less pollen collected). These results were the opposite of those hypothesized, and suggest that current housing for bumble bee colonies provided by suppliers is optimal for promoting pollination in greenhouses. My recommendation to growers would be to follow manufacturers suggestions regarding bumble bee placement and housing.

Participation Summary

Education & Outreach Activities and Participation Summary

1 Consultations
2 Webinars / talks / presentations

Participation Summary:

3 Number of agricultural educator or service providers reached through education and outreach activities
Education/outreach description:

I consulted with Koppert (a major supplier of bumble bees in the United States) to give suggestions on what temperatures are outside of the colonies preferred temperature range (when not temperature controlled) and would therefore result in a decrease in pollination services. I also presented 2 research talks where I disseminated my findings to fellow peers and educators so if growers had questions they would be able to best address them. I plan on publishing the results from this work in a scientific journal in the near future. Once the paper is in review, I will take the results found here and further share them with commercial bumble bee suppliers. 

Project Outcomes

Project outcomes:

Commercial suppliers do not recommend altering bumble bee housing, and so far our results align with this. By not changing the recommendations, this has increased sustainability of using pollinators. Externally temperature controlling colonies is energetically expensive. It also incurs less cost for farmers, as they would not need to provide additional structures for housing beyond what is provided via the commercial supplier. This means that bumble bees, which are naturally a cost effect solution to providing pollination, do not have additional expenses beyond the initial purchase. 

Knowledge Gained:

The current recommendations provided by commercial bumble bee providers are in line with what was found in this study. Although improvements can be made in regards to how correlated external temperature is to internal colony temperature, it does not appear to be advantageous to provide external temperature control to colonies unless the temperature far exceeds their desired range. 

My future career direction is to go into industry, where I can help test and implement ideas to maintain sustainable agriculture.  

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.