A laboratory amitraz resistance assay for Varroa destructor may not be a reliable predictor of field-level amitraz efficacy

Amitraz is used by beekeepers to control Varroa destructor infestations within their colonies worldwide. Evidence for resistant mites continues to be reported and could be a major factor in widespread colony losses. Over 2 years, we investigated a diagnostic bioassay that tested our mites for susceptibility to amitraz. After diagnosing mites with the bioassay, we genotyped the mites searching for the Y215H mutation of the β2-octopamine receptor. We found that the diagnostic bioassay accurately predicted the resistant genotype 94.6% of the time, and it accurately predicted the susceptible genotype in 66.8% of mites. Additionally, we performed a standard treatment with the product Apivar ® (active ingredient amitraz 3.33%) on the source colonies of these mites. Colonies receiving the Apivar ® treatment (n = 32) were categorized by the expected efficacy of amitraz on their mites, and a group of colonies was designated as untreated controls (n = 12). Both high and low expected efficacy groups had significantly reduced mite infestations by the end of the treatment period (reduced by 98.2%) compared with control colonies, which did not have a significant change in mite numbers over the same period (reduced by 23.0%). The diagnostic bioassay was a strong predictor of the resistant genotype of V. destructor, but it was not an accurate predictor of field-level amitraz efficacy.

Beekeepers in the United States and around the world continue to suffer from high colony losses that can be caused by many biotic and abiotic factors (Noël et al., 2020; Boncristiani et al., 2021). One of the most impactful elements contributing to colony mortality is Varroa destructor (Guzmán-Novoa et al., 2010; van Dooremalen et al., 2012). These parasitic mites cause direct harm to individual honey bees by feeding on them as both adults and pupae (Bowen-Walker and Gunn, 2001; Lamas et al., 2023). They cause further harm by vectoring and transmitting viruses, resulting in several diseases and syndromes of the bee (Martin et al., 2012). Without effective control, V. destructor can lead to highly pathogenic virus loads (Thompson et al., 2014) or cause the collapse of entire honey bee colonies (Jack et al., 2023). While beekeepers have access to several products designed to control V. destructor, not all are practical or safe to use throughout the year, and these products can be costly and require multiple applications for effective control (Jack et al., 2024).

Historically, beekeepers in the United States had overused synthetic chemicals, like coumaphos and fluvalinate, in their efforts for V. destructor control, leading the mite to rapidly develop resistance toward these once-effective compounds (Elzen et al., 1999; Elzen et al, 2000; Elzen, 2002; Pettis, 2004). Although beekeepers in the United States have had access to amitraz through the registered product Apivar ® since 2013 (Vu et al., 2020) and have used off-label products containing amitraz like Taktik ® since the 1990s (Jack and Ellis, 2021; Oliver, 2014), reports of amitraz resistance in the country have been somewhat limited until recently (Rinkevich et al., 2023). Amitraz is a formamidine pesticide that has miticidal and insecticidal properties; it overexcites nerve cells, causing uncontrolled nervous system activity that leads to death of the mite (Hollingworth, 1976). Formamidines, in general, function by binding to octopamine receptors, mimicking the effects of octopamine in the arthropod endocrine system (Evans and Gee, 1980). One putative cause of recent significant honey bee colony losses in the United States has been amitraz-resistant V. destructor (Lamas et al., 2025). To our knowledge, six single-nucleotide polymorphisms (SNPs) or other base pair substitutions within the Oct-ß2 receptor gene have been identified globally within V. destructor (Hernández-Rodríguez et al., 2022; Rinkevich et al., 2023; İnak et al., 2024; Hernández-Rodríguez et al., 2025; Lee et al., 2025; Marsky et al., 2024).

Previous research has connected mutations to failure of amitraz treatment in the field, but this connection is often through beekeeper reports of treatment failure and not experimental conclusions (Hernández-Rodríguez et al., 2022). Hernández-Rodríguez et al. (2022) selected mites for sequencing from many sources, including apiaries that had recently been treated with amitraz and reported failure. In this study, we specifically aimed to investigate the Y215H mutation, as this is the sole mutation currently identified in North America linked to amitraz resistance (Rinkevich et al., 2023; Hernández-Rodríguez et al., 2022). We hypothesized that the bioassay described by Rinkevich et al. (2023) would effectively predict the presence of the Y215H mutation and a high prevalence of that mutation would lead to a treatment failure of the amitraz-based treatment Apivar ®. Herein, we present evidence of the frequency of this mutation in our Florida apiaries and correlation with field-relevant results.

Laboratory resistance bioassay

We used the methods described by Rinkevich (2020) to conduct an assay of amitraz resistance. Briefly, approximately 300 worker honey bees infested with V. destructor were collected from hives in December 2023 and again in December 2024 at a University of Florida field site in Gainesville, FL (29.734197, −82.275388). The samples were placed in the heated cab of a truck, as temperatures were low in December, while being transported to the University of Florida Honey Bee Research Laboratory. The bees were housed in a 0.946-L container (Contain Fresh, Jiangsu, China, part # A11). The container lid was fitted with a screen of metal #8 mesh (3-mm spacing) to allow mites to fall through and inverted over a 14-cm × 14-cm weigh boat (Research Products International, Illinois, USA, part #142502) coated with petroleum jelly. Within each container, a 4-cm × 4-cm piece of Apivar ® strip was glued to the bottom, allowing the bees and mites to come into contact with it for 3 h. After the 3-h exposure period to Apivar ®, the bees and any remaining mites were thoroughly washed in soapy water (~5 mL of Dawn ® dish soap in 1 L of tap water) and separated by rinsing over a sieve (2 mm allowing mites to pass, 0.5 mm to catch mites). The mites that survived the 3-h assay were considered phenotypically resistant to amitraz and were stored in 95% ethanol at 4 °C. The mites that fell onto the weigh boat during the assay were considered phenotypically susceptible to amitraz and were preserved as described before. The total number of bees in the cup was counted, and the mite infestations (mites per 100 bees) were assessed in all sampled colonies (N = 28 in 2023 and N = 31 in 2024).

Field trials

Field trials took place over the winter season in 2023 and again in 2024. In both years, colonies that we expected to have very high mite infestations were specifically targeted to provide adequate sample sizes for the resistance assay as well as genetic analysis. Colonies selected had not received a miticide treatment in the previous 6 months. In 2023, an entire apiary was used for the trial, with no surrounding colonies not participating. In 2024, some colonies within the same apiary were not included in the trial because their mite levels were below the treatment threshold and therefore posed low risk to trial colonies from reinfestation. After collecting the bees for the laboratory assay, colonies were treated with Apivar ® for 42 days. Colonies were all managed during this time in single, deep, 10-frame Langstroth boxes and received Apivar ® strips as directed by the label of one strip per five frames of bees or less. A control group was created in the second year of trials (2024), where 18 colonies were left untreated for the same 42-day period. An alcohol wash containing ~300 bees was performed at the end of the treatment period to assess treatment efficacy and V. destructor populations. During the treatment period, colonies were fed 1:1 sucrose syrup as needed with an inverted jar feeder.

Apivar ® was expected to be minimally effective if the results of the bioassay showed less than 80% mite mortality (Rinkevich, 2020). Using this metric, we compared the results of our 42-day Apivar ® treatment on our colonies.

Genetic analysis

For each colony, an equal subsample of mites was genotyped from the total resistant and total susceptible mites. In 2023, 10 mites of each phenotype from each colony were genotyped, and in 2024, three mites from each phenotype from each colony (or fewer if the total number of mites in the sample was less) were genotyped. All V. destructor specimens were placed into a 2-mL tube and rinsed off with phosphate-buffered saline (PBS) to remove any residual ethanol and petroleum jelly for approximately 1 h. The PBS was aspirated off, and each specimen was pulverized using a mini beadbeater at 3,450 rpm shaking speed (BIOSPEC Products Inc., Bartlesville, OK). Specimens were extracted using the Qiagen Blood and Tissue Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s protocol. Genomic DNA from the extraction was stored at −80 °C until genetic analysis. Genomic DNA (average concentration 14 ng/µL) from the mites was analyzed for the presence of the Y215H mutation of the Oct-ß2 receptor and genotyped using Y215H primers and probes (susceptible—Vd_Y215H_V, resistant—Vd_Y215H_M) described in Hernández-Rodríguez et al. (2022). A total reagent volume of 12.5 µL was used per sample, which included 7.5 µL of SensiFAST Probe Lo-ROX Kit (Meridian Life Science, Memphis, Tennessee), 1 µL of each primer (10 µM), 0.2 µL of each probe (5 µM), and 2.6 µL molecular grade water (Thermo Fisher Scientific, Waltham, Massachusetts). Each reaction contained 2.5 µL of genomic DNA. Molecular assays were run on a QuantStudio 5 (Thermo Fisher Scientific, Waltham, Massachusetts) with the following settings: 95 °C for 5 min followed by 39 cycles of 95 °C for 15 s and 60 °C for 1 min.

Amplification of susceptible samples was determined by an increase in fluorescence of the Vd_Y215H_V probe, while resistant samples were distinguished by their enhanced fluorescence of the Vd_Y215H_M probe, and heterozygous samples showed combined fluorescence of both probes. (Examples of amplification can be found in the Supplementary Materials) The QuantStudio Design and Analysis Software v.1.5.1 calculated Ct values using default settings.

Calculating ratios of resistant and susceptible genotypes in our populations

Based on the genotype results from a large portion of our mites included in the bioassay, we can calculate a reliability factor of how accurately our bioassays predicted the genotype. Using this reliability factor (Λ), we can calculate a more accurate estimation of the genotype of our entire population. Making inferences of genotype or phenotype based on a known sampling method reliability and of an assay has been used to extrapolate results (Jin et al., 2004; Lewis et al., 2004).

R Ex (resistant expected) is the number of mites showing a resistant phenotype in the bioassay, S Ex (susceptible expected) is the number of mites showing a susceptible phenotype in the bioassay, and Λ is the reliability factor calculated from our predictor-outcome tables, which were different each year.

Reliability factor Λ R = N R (genotyped)/ N Rex (phenotyped).

To extrapolate to the mites that were not genotyped, all phenotyped mites that went through the bioassay were categorized as either resistant (expected) or susceptible (expected).

The heterozygosity factor (H) is calculated from the number of mites genotyped as heterozygous out of the total mites genotyped. Each year had a different heterozygosity factor.

First, we applied the heterozygosity factor to the number of mites expected to be resistant:

N Rex * H = N H

Next, we subtracted this number of heterozygous mites from our N Rex group and applied the reliability factor Λ R.

(N Rex − N H) * Λ R = N R

Starting with our initial group of N Rex, we subtracted the more accurate N R and N H groups, leaving a new group that we recategorized as N S. These three groups are the updated, improved extrapolated numbers of resistant, heterozygous, and susceptible groups, respectively, in the larger mite population.

We completed the same procedure with the N Sex group of phenotyped mites, then totaled the accurately categorized mites at the end.

Finally, we added our newly recategorized mites to our calculated reliably categorized mites to get the estimate. To determine the true genotype frequency in our large population of mites that underwent the laboratory resistance bioassay, we can use their phenotype as an imperfect proxy and use a Rogan–Gladen adjustment to correct our misclassification bias.

Statistical analysis

To determine differences between groups in the field trials, we used a generalized linear model (GLM) with a negative binomial error structure (theta = 1) to compare the interaction between all groups and time points using the package MASS in R (version 4.5.1, RStudio 2025.09.2 + 418; R Core Team, 2021; glm.nb; Venables and Ripley, 2002). Residuals were checked for normality, homogeneity of variances, and evaluation of model assumptions using the DHARMa package in R (Hartig, 2026; see Supplementary Materials). Following the initial GLM to test the interaction, data were subset by both time to test for treatment effects and treatments to test changes over time. Post hoc comparisons were made using the “emmeans” package in R to get pairwise p -values comparing each group at each time point.

To determine the probability that a heterozygous genotype would result in a mite dying during the laboratory resistance bioassay, we created a linear mixed model in R with genotype as the predictor variable and phenotype as the response variable, with year as a random effect. The “emmeans” package was used again to calculate pairwise probabilities with standard errors.

Laboratory resistance bioassay

In both years, the results of the pretreatment bioassays were highly variable. In 2023, the expected Apivar ® efficacies from the 28 colonies ranged from 27.6% of mites dropping within 3 h to 88.0%, with an average of 48.5%. In 2024, the expected Apivar ® efficacies from the 31 colonies ranged from 21.4% to 94.7%, with an average of 57.4%. (See Data availability section for the link to raw data files).

Field trials

Based on the results of the pretreatment bioassays, we were able to predict the expected efficacy of the Apivar ® treatment. In 2024, we compared colonies with high and low expected efficacies as well as untreated colonies before and after the 42-day Apivar ® treatment. Overall, there were significant differences in the mite infestation levels in the interaction between treatment and time χ 22,82 = 83, p -value < 0.001). Prior to the application of treatment, treatment groups differed significantly (χ 22,41 = 18.3, p -value < 0.001). There was a significantly higher average mite count in the low expected efficacy group compared to the control group (p < 0.001). After the trial ended, treatment groups differed significantly (χ 22,41 = 55.84, p -value < 0.001). At the end of the trial, the low expected efficacy group showed a significantly lower mite count compared to the control group (p < 0.001). At the beginning of the trial, neither the low expected efficacy group (p = 0.136) nor the control (p = 0.377) differed significantly from the high expected efficacy group. Similarly, at the end of the trial, neither the low expected efficacy group (p = 0.981) nor the control (p = 0.071) differed from the high expected efficacy group. Both Apivar ® -treated groups of mite populations decreased significantly over the course of the experiment (high expected efficacy: χ 21,12 = 29.52, p < 0.001; low expected efficacy: χ 21,48 = 147.07, p < 0.001; Figure 1). The control group was not significantly different from its pretreatment infestation level (χ 21,22 = 0.67, p = 0.414). The expected efficacy of Apivar ® was not a good predictor of the 42-day treatment outcome.

Phenotype and genotype

In total, 525 mites were genotyped. Table 1 presents results from 2023 and 2024, showing the associations between a mite’s phenotype (result from the bioassay) with its genotype. Results are displayed in this manner to remain consistent with Rinkevich et al. (2023), who first described this mutation in the United States.

Total ratios of resistant and susceptible genotypes in our populations

When matching the genotype to the phenotype observed in the bioassay, we found the phenotypically resistant mites most closely aligned with their expected genotype at 94.6% agreement. These associations are outlined in Table 2. The prevalence and association of mites with a heterozygous genotype were noted in Table 1. In 2023, we only discovered 3 mites out of 367 with a heterozygous genotype. Of these three mites, two had a susceptible phenotype in the resistance bioassay, and one appeared resistant. The three mites came from two colonies, neither of which had an expected Apivar ® efficacy over 70%. In 2024, we found 11 out of 158 mites with the heterozygous genotype. These were split nearly equally between resistant and susceptible phenotypes in our bioassays (five susceptible, six resistant). Furthermore, these 11 mites were found across 10 colonies, and only two of these 10 colonies had an expected Apivar ® efficacy over 70% based on the mite fall in the bioassay.

Our analysis of pairwise probabilities showed that when mites had the heterozygous genotype, their probability of exhibiting susceptibility in the bioassay was 0.536 (SE = 0.146). The year was included in this analysis despite having 14 mites in total with a heterozygous genotype and equal numbers of phenotypically resistant and susceptible phenotype and unequal results within each year. See the Data availability section for the link to this raw data and analysis code. Based on these results, a heterozygous individual has a similar probability of being susceptible or resistant to amitraz.

In our study, we observed a lower frequency of the heterozygous genotype with 0.8% (3/367) in the first season and 7.0% (11/158) in the second season. Using our reliability factors Λ R and Λ S, we used our phenotype data from the bioassays and estimated the genotypes of all 1,867 mites that were collected (Table 3). This resulted in an estimated 70.5% of mites being resistant in 2023 and 48.2% in 2024.

Our research suggests that the Y215H mutation is indeed very closely associated with individual V. destructor that survives an amitraz exposure in the described Rinkevich (2020) bioassay. This work is further validation of the ability of the bioassay to detect this mutation. However, this mutation is not solely associated with field-level failure of control when using a standard 42-day treatment of Apivar ®. Our results reflect a similar scenario to the study conducted by Marsky et al. (2024), where they discovered that a previously described mutation by Hernández-Rodríguez et al. (2022) was present in their mite populations; however, it did not result in fewer mites being susceptible to the amitraz treatment in their field trials. Mites sampled in the Hernández-Rodríguez et al. (2022) study represented multiple samples: some were collected from colonies part-way through an Apivar ® treatment, while others were collected by a state apiculture worker after a beekeeper reported amitraz treatment failure.

Our study is somewhat limited in scope and scale, so future research should build on this and include more field colonies in more diverse locations. Florida presents a unique scenario where capped brood is present through our winter season. Temperatures are mild, and colonies highly infested with mites may still survive here compared with those in more northern U.S. states (Giacobino et al., 2025). The number of colonies containing mite populations where Apivar ® was expected to work well was small. In the future, more diverse groups of colonies, hopefully representing diverse mites, could make a more robust dataset. Additionally, a lack of untreated controls in our first field season means that the high efficacy of Apivar ® should be met with caution and more research should attempt to replicate the overall design of our second season of field trials that included control colonies.

A possible explanation that should be tested in the future is whether this mutation confers resistance to amitraz, but not to its breakdown products. If more time is allowed in the bioassay, the mites will continue to metabolize amitraz and possibly succumb to N-(2,4-dimethylphenyl)-N′-methylformamidine (DPMF), 2,4-dimethylformamide (DMF), or 2,4-dimethylaniline (DMA). This could explain the excellent control we found in our field colonies, despite many with very low expected control based on the bioassays.

Another possibility could be the season of treatment application. Jack et al. (2024) found that Apivar ® was only effective for V. destructor control during the winter season, which is when we conducted our studies in both years. Although we discovered that many V. destructor had the Y215H mutation in our 2023 and 2024 trials, it did not result in an Apivar ® treatment failure, possibly due to the advantageous treatment window. Future research targeting amitraz efficacy related to this mutation through different seasonal conditions would be valuable. More research should be conducted to identify other mutations that may appear in conjunction with the Y215H mutation or are otherwise associated with amitraz resistance in North American V. destructor.

The Y215H mutation has been observed in V. destructor populations across North America (Lamas et al., 2025; Rinkevich et al., 2023; Bahreini et al., 2025). However, the same mutation has not been discovered in European populations when searched for (Hernández-Rodríguez et al., 2022; 2025). The reasons for this mutation to be so widespread and common are not precisely understood.

A limitation of our current data is the number of V. destructor we were able to genotype. While 525 of the 1,867 (28.1%) were genotyped, this analysis was not a random sample of the mites that we could extrapolate to the whole. Using the bioassay results as a predictor of genotype and applying the reliability of that prediction to the remaining mites, we believe we have accurately estimated the prevalence of the Y215H mutation in our mite population. In 2023, that prevalence was estimated at 70.5% (665/943), and in 2024, we estimated the prevalence at 48.2% (445/924). This range is not unusual. In 2022, Canadian researchers found the prevalence of the Y215H mutation to be 95.67% in southern Alberta, 57.5% in central Alberta, and 31.75% in northwestern Alberta for an average of 58.78% that year compared to an average of 37.1% in Alberta in 2020 (Bahreini et al., 2025). In a recent study examining colonies that remained after high winter losses in the USA, the researchers genotyped 39 mites and found 100% of them to have this resistant genotype (Lamas et al., 2025).

We sought to analyze amitraz efficacy on three levels: phenotype observed in the bioassay, genotype of the mites, and colony-level effects on mite populations from a full-field treatment. A study that was conducted in Spain (Hernández-Rodríguez et al., 2025) also links field-level efficacy to an amitraz-associated mutation from that area. However, that link is made by the proportion of mites that have been killed 15 days into the amitraz treatment. No follow-up is reported after the full 6–10-week recommended treatment duration. This study involves the F290L mutation, which is different from the Y215H mutation thought to be the primary mutation that confers resistance in our study. Still, as with a lab-based bioassay, there is a chance that a short-term observation of mite death combined with a high prevalence of an Oct-ß2R mutation could be an incomplete account of the amitraz treatment on the mite population. Other studies that have attempted to link their genotype results to field-level efficacies have relied on beekeeper reports of less effective amitraz treatments (Bahreini et al., 2025). Beekeeper reports can be very useful. In Argentina, researchers used beekeeper reports of failed amitraz treatments to collect mites and perform LC 50 tests on them (Maggi et al., 2010). They found V. destructor populations with amitraz LC 50 values 35–39-fold higher than their baseline populations, an example of clear resistance, but these resistant Argentinian mites have not yet been linked to a specific mutation or genotype. So, while investigating reports made by beekeepers is necessary and valuable, controlled experiments are needed to verify the accounts.

Previously, V. destructor individuals with a heterozygous genotype were considered susceptible to amitraz (Rinkevich et al., 2023). The researchers concluded this because of the higher association of heterozygous mites with bioassays showing greater than 70% efficacy. This conclusion does not match our findings, where only two of our 14 heterozygous mites were associated with bioassays with greater than 70% efficacy. When considering the phenotype associated with the heterozygous genotype, our mites are divided equally with seven mites surviving the amitraz exposure in the bioassay and seven perishing. The low frequency of heterozygosity in our test populations is difficult to explain. On one hand, we do not expect the allele frequencies to be in Hardy–Weinberg equilibrium (HWE). Previous V. destructor studies have shown extreme deviation from the expected HWE allele frequencies (Dynes et al., 2016; Mendoza et al., 2020). Considering very high infestation levels and low honey bee brood areas throughout the previous months before testing in December, mites should have had ideal conditions for multiply infesting cells and higher occurrence of outbreeding (Beaurepaire et al., 2017). With both alleles being well represented in the population, we expected to see more heterozygous mites.

Our conclusions highlight the need to explore factors that can increase or maintain the effectiveness of amitraz treatments. These factors could include seasonality of mite susceptibility, metabolism of mites and bees leading to increased toxicity, and formulations that aid in the delivery of amitraz. Researchers also need to continue investigating the methods of bioassays used to give beekeepers reliable tools to predict the effectiveness of amitraz.