Is Thermacell Safe For Bees? Examining The Impact of Mosquito Repellents

Because nontarget, beneficial species like insect pollinators can be accidentally exposed to insecticides, it is important to test how chemical controls affect the behaviors of insect pollinators in the field. We used a blinded, randomized, paired, parallel group trial to look at how a portable mosquito repellent that releases prallethrin, a pyrethroid insecticide, affects honey bees’ ability to find food and join new colonies. The insecticide that was released into the air didn’t have a significant effect on the number of times the birds foraged (our main outcome), the number of times they did waggle dances, or how long they stayed at the feeder (our secondary outcomes). However, another deposition study showed that the treatment device was working correctly. These findings could be helpful for people who want to keep mosquitoes away but are also worried about what might happen to good bugs like honey bees.

It is also possible for an insecticide to be released into the air (Lichtenstein and Schulz 1970; Holoman and Seymour 1983; Achee et al. 2012, Revay et al. 2013), which creates a chemical plume that then is used in the control of some flying insect pests. Devices that use this volatilization of insecticides are often used to keep mosquitoes away, like on patios and porches. But a chemical plume could also affect other flying insects, like good pollinators that might be foraging near the patio, and not much is known about how these devices might affect them. Prallethrin, which is a pyrethroid, is one such insecticide that may be volatilized. A well-known major class of synthetic pesticides is pyrethroids, which are chemical copies of pyrethrins (found in pyrethrum flowers). They work by blocking neuronal voltage-gated sodium channels in the central nervous system of insects (Soderlund and Bloomquist 1989, Casida and Durkin 2013).

We looked at the MR300 Portable Mosquito Repeller (Thermacell Repellents Inc.) in a blinded, randomized, controlled study. , 32 Crosby Dr. , Suite #100, Bedford, MA 01730) charged with mats that emit prallethrin on the behaviors of honey bees. In line with what most people do, we looked at (1) how the device affects honey bees’ foraging and recruitment in the late afternoon and early evening and (2) how it affects honey bees’ persistence the next morning when the device is turned off.

We looked at honey bees of different subspecies, mostly Apis mellifera ligustica, that lived in observation hives with glass walls at the Prices Fork Research Center in Blacksburg, Virginia, USA. Three American Standard Deep frames were used to build the observation hives. They were linked to the outside by a 3 x 30 cm plastic tube. We worked with one colony at a time and did three trials on days with good foraging weather from July 28 to August 4, 2022. Video analysis continued through the fall and winter of 2022–2023. We chose to do our experiments in the summer because it is easier to get bees to feed from feeders when there isn’t much forage around (Ohlinger et al. 2022b). Throughout the project, colonies were managed to keep them from getting too crowded and to make sure that the number of empty cells stayed the same. This is because it’s important to have room for more sucrose solution storage (Seeley 1989, 1995, Couvillon et al. 2015). The colonies were not given supplemental food for the duration of the field experiments.

Experimental Trial—Secondary Data Collection (Waggle Dance Propensity, Waggle Dance Frequency, Feeder Persistency)

The marked bees in the observation hive also did waggle dances that we recorded on video during the four-hour experimental phase. The waggle dance is a unique behavior that a bee doing when it comes back from foraging at a good food source lets the other bees know how far and in what direction it is from the hive to the food source (von Frisch 1967, Couvillon 2012). We used two Canon Vixia HF R82 cameras, one on each side of the glass-walled hive. We recorded to a SanDisk Extreme SD card and then uploaded the videos to a Google Team Drive to be analyzed.

During subsequent video analysis, we monitored for two dance metrics. First, we determined the proportion of successful foragers that made any dances at all during the experimental phase. We term this binomial (yes/no) outcome the waggle dance propensity during the experimental phase. Once a bee had made a dance, we monitored how often she then repeats a dance (i. e. , how many unique dances that bee will perform). We term this discrete, positive count the waggle dance frequency during the experimental phase.

The next morning (Day 1), we returned the tripods with platforms to the two positions (Fig. 1) and placed two, empty, unscented feeders for the persistency phase from 9:00 AM until 12:00 PM. The tripods for the Thermacell devices were also put back where they were the previous evening, but the devices were turned off because we were collecting data on their effects the next day. We counted visitations of marked bees to the empty feeders during the three-hour persistency phase. We term this discrete, positive count feeder persistency during the persistency phase. We collected these data as previous research has shown that bees are more persistent (i. e. , going back on Day 1 to see if the empty feeder has become rewarding again if the spot was highly rewarding before (Al Toufailia et al. 2013), making persistency another honey bee behavior that correlates with the resource quality. These last three response variables (waggle dance propensity, waggle dance frequency, and feeder persistency) were our secondary outcomes.

We can’t completely control sample sizes for each treatment because we need bees to come in, but we were able to get a pretty even distribution of bees between treatments in all three of our trials (Treatment = 67 bees, Control = 74 bees). For many behaviors, the correlation between bees in the same hive is very low, about the same as the correlation between bees in different hives (Schürch et al. 2016, 2019). Therefore, we consider each bee as an independent replicate. To make sure that the treated and untreated devices were equal in terms of feeder position, we did three trials and randomly assigned the treatment to a different feeder for each trial (see published code for randomization code).

To sum up, we used a blind, randomized, paired, parallel group trial with about equal numbers of treated (bees that were exposed to active compound) and control (bees that were not exposed to active compound) bees. We looked at how treatment affected our main outcome (the number of times the birds went foraging) and our secondary outcomes (the likelihood of waggle dances, the number of times they happened, and how often they continued to visit the feeders).

Deposition Study to Validate the Presence of Prallethrin

During one of our experimental trials, Thermacell’s Science and Research Division (C.B and B.E.M) concurrently performed a deposition study to confirm that the devices and repellent mats were functioning appropriately (see Supplemental Information online for full report). Briefly, this additionally involved setting up six data collection stations around each Thermacell device, both treatment and control. Each data collection station was equipped with three pieces of filter paper, secured by a ring stand, at 0 m (ground), 1 m, and 2 m, to act as a passive dosemeters. The three inner-most test stations featured an air sampling pump with appropriate media (see supplementary online for details). The six data collection stations around the control feeder were far enough away to avoid any potential cross contamination of the systems.

A typical Thermacell multi-insect repellent mat contains approximately 10. 86% prallethrin. In our experiment, the treatment mats were slightly overdosed so that the honey bees were exposed to as much of the compound as possible. This was done to show the worst-case scenario. Because insect neurons are sensitive to even small amounts of pyrethroids (Soderlund and Bloomquist 1989), we wanted to make sure that the honey bees that were out foraging would get the most exposure possible from our setup. As usual, a control mat containing no repellent was loaded into the other repeller device.

The deposition study ran for the entire four-hour experimental phase on Day 0. Afterwards, the concentration of the active ingredient, prallethrin, was determined via GC–MS for the filter papers and air sampling media. The Thermacell MR-300 Portable Mosquito Repeller utilizing the prallethrin treatment emitted the insecticide at a level consistent with previous experiments and shows that the device and mat functioned properly in the study: the filter paper recovery in the field (mean prallethrin Spike Recovery (%) = 96% at 10 minutes) was similar to the previously determined lab mean Spike Recovery Rate for the same time (97.4%, Table 6 of Supplementary Online Information). After four hours, the mean recovery was 90%. In contrast, the control mat demonstrated no detectable concentrations of prallethrin in any of the material tested (Supplementary Online.

All analyses were done in R 4. 2. 3 (R Core Team, 2020). We showed the raw data in the form of graphs and summed it up by treatment and trial phase (training, experimental, and persistency). We present both means (and 95%-CI) and medians (and lower and upper quartiles) in these summaries. Differences in treatments were derived from mixed-models. There were differences in the treatments that were pointed out, along with two-sided 95% confidence intervals (CI) and the corresponding P-values. In addition, we analyzed the bees according to the intention-to-treat principle (Gupta 2011). That is, we determined to which feeder a bee was trained, either treatment or control. Once a bee was trained, it rarely switched between feeders, so we didn’t need to do a per-protocol analysis (Ohlinger et al. 2022a). Lastly, we counted how many times a bee visited a feeder during the training phase and added that number to our models as a fixed effect to account for each bee’s unique commitment, no matter what treatment they were given. This is important because the bees that have been around a feeder longer will care about it more (Seeley 1995; Couvillon et al. 2015), which may impact the primary and secondary outcomes.

We used Poisson generalized linear mixed-effect models to look at count data for foraging frequency, dance frequency, and feeder persistency during the experimental phase. These models let us include trial-specific and individual-specific intercepts. These latter observation specific random intercepts were necessary because in model checks we detected over-dispersion. The data were modeled using the glmer function from the lme4 package (Bates et al. 2015). You can use a Binomial generalized linear mixed-effect model (function glmer, package lme4) to model the binomial outcome of dance propensity (Bates et al. 2015). We used the emmeans package to calculate treatment group contrasts and extract 95% CI (Lenth, 2021).

Finally, we discovered during our video analysis that a small proportion of our bee number tag combinations of font/background color made the font illegible, although of course the color, which should designate whether the bee was treatment or control (see above), was still visible. We dealt with this by assuming these numbers to be missing at random in our data sheet and ignored unidentified dancers for the dance outcomes analyses, which were two of our secondary outcomes (note: our primary analysis was not impacted by this). As an additional sensitivity analyses, where we could only identify the color, not the number, of the tags, for our waggle dance frequency analysis, we assigned the dances of the unidentified bees to the identified dancers in proportion to how many dances these positively identified bees performed. We then performed the same analysis as on the original data set where we assumed IDs to be missing at random. All data and analyses code are available at https://doi.org/10.7294/24147375 (a permanent DOI will be provided after final acceptance of paper).

THERMACELL ANY GOOD | OR DANGEROUS | REVIEW MR300

FAQ

Is Thermacell good for bees?

Is Thermacell safe for other insects?

Is allethrin toxic to bees?

Is Thermacell safe for wildlife?

Does thermacell harm bees?

According to Thermacell, the Environmental Protection Agency (EPA) and the World Health Organization (WHO) have concluded that there is a reasonable certainty that no harm will result to bees or other terrestrial wildlife if using the product according to the label instructions. What makes the Liv system unique?

Is a thermacell safe?

Although the Thermacell is EPA approved and uses allethrin (a synthetic version of a substance found in Chrysanthemums) instead of DEET, warnings include keeping the lantern away from uncovered food during operation and avoiding directly inhaling the vapor. Allethrin is also highly toxic to fish, bees and cats.

Is thermacell safe for cats?

The device is rated for outdoor use only and should be kept away from open food. Thermacell’s technology is independently tested and EPA reviewed for safety, but the manufacturer warns that the active ingredient is toxic to cats, bees, and fish.

How do you use a thermacell?

Key to thermacell usage is understanding that the deterrent effect is a round bubble only in windless conditions and to set it up appropriately. If there is any wind, the unit needs to be upwind of the people who need protection. Wind will make the protected area shift and with enough wind it is harder to know where that area is.

Does thermacell repel mosquitoes?

CEO of Thermacell Repellents Steve Hill says, “Thermacell has been effectively fighting mosquitoes for the past 22 years and we are thrilled to step into the smart home space with the first connected mosquito repellent system,” adding, “It’s now easier and more convenient than ever to enjoy the outdoors with fewer mosquitoes.”

Should you buy a thermacell E55?

As someone lives in an area with tons of mosquitos and is prone to bites, the tiny-but-mighty Thermacell E55 has come in so handy. I’d gladly pay full-price for the gadget, but thankfully, it’s on sale right now — time to grab some more!