Reducing Pesticide Use in Honey Bee Colonies through Sound Sampling and Treatment Procedures

Final Report for LNC05-264

Project Type: Research and Education
Funds awarded in 2005: $150,000.00
Projected End Date: 12/31/2008
Region: North Central
State: Minnesota
Project Coordinator:
Marla Spivak
University of Minnesota
Expand All

Project Information


We developed a sampling plan to help beekeepers monitor Varroa destructor mite infestations in honey bee colonies. From a sample of 280 adult bees, collected using our novel sampling device, beekeepers can estimate the total mite density in a colony (mites on adults and those reproducing on pupae). Standard sampling will help beekeepers make educated treatment decisions. Field trials of bees bred for both Hygienic Behavior and Varroa Sensitive Hygiene had significantly lower mite levels compared to an unselected line of bees demonstrating that the hygienic trait can reduce mite loads and thus the frequency of pesticide application.


The North Central region of the US, particularly MN, ND and SD are the top honey producing states based on yield per colony, together producing over 30% of the total honey production for the nation. The majority of the commercial beekeepers in the Upper Midwest are “migratory,” meaning that they transport their colonies every winter either to southern states where they produce bulk bees and queens for sale, or to CA and other states where the bees pollinate almond orchards and other fruit and vegetable crops. The beekeepers transport their colonies back to the Upper Midwest for the summer to produce large honey crops.

The honey bee, Apis mellifera, population has been declining since introduction of the parasitic mite Varroa destructor in 1987. Varroa is a devastating pest. The mites weaken individual honey bees, increasing the bee’s susceptibility to the effects of disease and poor nutrition. Infected colonies die in 1-2 years. To prevent potentially large colony losses, many beekeepers have resorted to using pesticides (organophosphates and pyrethroids) within their colonies. Many treat all their colonies with pesticides once or twice a year, irrespective of mite level. The treatments add a great operating expense for the beekeepers, increase the risk of contamination of hive-products (Frazier et al., 2008), pose health risks to bees (Currie, 1999; Rinderer et al, 1999; Haarman et al., 2002; Collins et al., 2004), and the mites have developed resistance to the synthetic pesticides. To minimize these negative effects, beekeepers have expressed their desire to decrease the number of treatments. Many beekeepers have heard of the integrated pest management practice of treating only when economically necessary, but this requires sampling to determine the current mite levels and having a treatment threshold. Prior to this research, beekeepers lacked a standard, accurate and efficient way to sample mites in a honey bee colony. Our goal is to encourage beekeepers to sample their bee colonies for mite pest populations in a standardized way to help them make wise treatment decisions and reduce pesticide use. Such a sampling plan for a mite pest in an insect colony has never been attempted before.

Once a beekeeper samples to find the mite infestation, that beekeeper will need to make a treatment decision. Treatment thresholds of 10-12% colony mite infestation have been developed (Martin 1998, 1999; Delaplane and Hood, 1997, 1999). However, the thresholds were created based on mite levels in relatively small-scale beekeeping operations (hobby or side-line) in which the colonies are maintained in one location year round. Migratory beekeepers, those that move their colonies across the U.S. for honey production and to fulfill pollination contracts, would also benefit greatly from a treatment threshold. These commercial beekeepers will likely treat at a lower threshold because their livelihoods depend on it. Transporting bees en masse leads to increased horizontal transmission of diseases and pests, and nutritional stress, which may decrease the colonies tolerance for the mite. Bees with natural mite resistance due to hygienic behavior may have a different treatment threshold than susceptible colonies. There should be an absolute lowest threshold (e.g. 2-5%) where below that level no beekeeper should need to treat. Between the low threshold and the 10-12% threshold is a gray area where a beekeeper should make a decision based on the nature of their beekeeping operation, and their personal tolerance for colony loss. Monitoring mite loads is a first step in helping beekeepers understand their mite levels and dynamics. With this system and good record keeping, beekeepers can find the treatment level works best for them based on their own operation.

Project Objectives:

The objective of this project was to encourage beekeepers to reduce the use of in-hive chemicals by monitoring mite populations through sampling, helping them to make informed treatment decisions, and demonstrating the benefits of keeping bees with natural resistance to the mite. The reduction of treatments will help beekeepers avoid unnecessary costs, pesticide residues, and help slow the evolution of resistance of the mites to chemicals; thereby fostering more sustainable beekeeping practices, increasing the profitability of beekeeping, improving environmental quality, and promoting pollinators.

Objective 1. Develop a simple and standardized sampling plan for commercial beekeepers to help them determine the economic treatment level for Varroa destructor mites.

Objective 2. Develop published guidelines for migratory beekeepers on making educated treatment decisions for the mite based on the sampling plan.

Objective 3. Compare mite levels between our line of bees bred for both Hygienic Behavior (HYG) and Suppression of Mite Reproduction (SMR) with an unselected, commercial line of bees, to demonstrate that the use of resistant bees can reduce mite loads and thus, the frequency of pesticide application.


Click linked name(s) to expand/collapse or show everyone's info
  • William Klett
  • Gary Reuter
  • Darrel Rufer


Materials and methods:

Objective 1. There are two populations of mites: those on adult bees and those on developing pupae which are in cells under wax cappings. The mites hidden in pupal cells are the reproductive population. The adult bee mite infestation is easier to estimate than the pupae infestation, but a comprehensive sampling plan needs to include both the mites on adult bees and the mites on pupae. The sampling plan was developed to efficiently sample mites on adult bees and, using the adult bee infestation, make an inference about the colony infestation; i.e., the mites on adults and on pupae.

Mites on adult bees were sampled from colonies in 31 commercial apiaries with 24-84 colonies per apiary for a total of 594 colonies. The apiaries were owned by five different beekeepers in four states, Minnesota, North Dakota, California, and Texas. All the beekeepers were migratory and the operation sizes ranged from 1,000 to 20,000 colonies. The colonies were sampled in the years 2005, 2006, and 2007. All the commercial apiary sampling was done during March (CA and TX), May-June (MN and ND), and August-September (MN and ND). These times were chosen to coincide with when beekeepers would sample and potentially treat their colonies. Hand drawn maps were created to detail the apiary layout including the colony and pallet placement (commercial beekeepers usually maintain 4 colonies per pallet to facilitate transportation) and the colony numbers (each colony was arbitrarily assigned a number). A sample of 35 adult bees ± 15 (mean ± s.d.) was collected from each comb in 70% ethanol. Each comb was classified as having at least one of the following contents: honey, pollen, open worker brood (egg through uncapped 5th instar larva), sealed worker brood (developing pupae covered by a wax capping), sealed drone brood, or empty. In the lab, the mites were strained from the bees to determine the number of mites and adult bees per sample (De Jong et al., 1982). The infestation for each comb within each colony was recorded and used to determine if the mites tend to congregate within a colony and within an apiary.

In addition, 75 colonies were sampled more intensively to examine the relationship between the adult bee infestation (mites on adult bees) and colony-level infestation (mites on adult bees and worker pupae). From each colony, the following populations were estimated: the total number of adult bees and worker pupae, and the total number of mites on adult bees and on worker pupae. The adult bee population was estimated either using a calibrated image (MAFF, 1998) or by shaking the bees into a screened box and weighing them. The number of worker pupae was estimated using a wire grid to count the number of squares of sealed brood and using a conversion factor of the number of pupal cells found in one square. The mite population on adult bees was estimated by finding the infestation on ≈900 adult bees in 70% ethanol or by combining the infestations from the 35-bee unit samples from a colony. The adult bee infestations were multiplied the mite by the number of adult bees. To estimate the total number of mites on the pupae, a minimum of 200 individual sealed pupal cells were opened and examined individually for the presence or absence of mites. The infestation of the pupae was multiplied by the number of pupae to estimate the total number of pupal cells infested.

Objective 2. Based on the sampling plan developed in objective 1, we developed a sampling device to collect and monitor mites on adult bees in a standard manner. The aim is to provide beekeepers a chart enabling them to convert the adult bee infestation into the colony infestation (mites on adult bees and in worker pupae). We will publish guidelines for beekeepers on making educated treatment decisions for the mite.

Objective 3. M. Spivak and G. Reuter have been breeding honey bees for hygienic behavior since 1994. Hygienic bees are able to detect, uncap and remove diseased young bees, preventing the spread of diseases like chalkbrood and American foulbrood. Hygienic bees are also sensitive to V. destructor infested cells, opening the cell and removing the pupae, which interrupts the mite’s reproductive cycle. Another line of bees was bred by researchers at the USDA-ARS Bee Research Lab in Baton Rouge, LA. They bred for bees that reduced the mite loads over a period of time, and called this trait, “Suppression of Mite Reproduction” or SMR (Harbo and Hoopingarner, 1997; Harbo and Harris 1999a; b). During the course of this field study, we investigated the mechanism of SMR to determine how the bees were able to suppress mite reproduction. Originally, it was thought that mites in the colonies expressing this trait appeared to produce fewer offspring overall, fewer offspring that reached maturity, or failed to initiate egg-laying. However, our research, which was later confirmed by the USDA team, showed that the mites do not necessarily have a lower reproductive success. Instead, SMR bees tended to express hygienic behavior on the cells containing mites that are reproductively successful (Ibrahim and Spivak, 2006; Harbo and Harris, 2005). After this finding, the name SMR was changed to Varroa Sensitive Hygiene, or VSH.

We compared the mite levels of colonies selectively bred for both hygienic behavior and Varroa Sensitive Hygiene with colonies bred solely for hygienic behavior and unselected control colonies. Colonies were evaluated for strength, brood viability, removal of freeze-killed brood, honey production, mite loads on adult bees and within worker brood, and mite reproductive success on worker brood for two years in MN and ND.

Research results and discussion:

Objective 1. The sampling plan developed for beekeepers to help them estimate mite levels on adult bees was developed by applying sampling statistics normally used in plant and pest agricultural systems. We first examined possible sources of variation in mite infestation at the level of apiary, pallet (grouping of 4 colonies), individual colony, box, comb, and comb content (combs with brood verses combs without brood). A nested analysis of variance showed that the colony explained the most variation in mite prevalence, followed by pallet. The analysis also showed apiary can be a large source of variation, meaning that the mite levels were different among apiaries within a single beekeeper’s outfit. This finding indicates beekeepers should sample and make a treatment decision based on each apiary instead of assuming all the apiaries have a similar mite level.

To examine within colony variance, the mite infestation on combs with brood were compared to combs without brood. A logistic regression showed combs with brood tended to be 1.6 times more likely (95% confidence interval =1.44-1.74) to be infested compared to combs without brood (chi-squared test=90.73, d.f=1, P <0.0001).

The sampling approach, Resampling for Validation of Sampling Plans (Naranjo and Hutchison, 1997), was used to determine the optimal number of adult bees to sample from each colony and the number of colonies to sample within an apiary. Results showed that the best method is to sample 280 bees per colony and eight colonies per apiary to achieve a precision level of 0.25 for apiaries with 24-84 colonies. For the infestation of a single colony, beekeepers should sample 280 adult bees to obtain a precision of 0.25 if there are ≥ 4 mites per 100 bees. If there are ≤ 4 mites per 100 adult bees, the precision decreases and the reliability of the sampling will be ± 1 mite.

The next step was to factor in the number of mites parasitizing the worker brood (developing pupae) to estimate total mite loads within a colony. We compared the adult bee infestation to the colony infestation (adult bees and worker pupae) using a linear regression. The regression showed that multiplying the adult bee infestation by a factor of 1.3 estimates the colony infestation with an R²=0.75.

In application, this analysis will allow beekeepers to sample mites on adult bees and using a conversion factor to estimate the total colony density of mites. If a beekeeper finds 10 mites on 280 adult bees, they can divide 10 by 280 and multiply by 100 to find 3.6 mites per 100 adult bees, then multiply by 1.3 to account for the mites in the brood to obtain a total infestation of 4.6 mites per 100 adults and pupae. To make this conversion more accessible, it will be provided in a simple chart that allows beekeepers to count the number of mites in a 280 adult bee sample or from eight 280 adult bee samples (2240 bees) and readily find the colony or apiary infestation.

We will recommend beekeepers estimate the apiary infestation by sampling 280 adult bees from a brood comb from each of eight colonies or sample 280 adult bees to estimate colony infestation. Ideally beekeepers would sample eight randomly chosen colonies to estimate apiary infestation, but to simplify to recommendation and to assure multiple pallets will be sampled we will recommend beekeepers sample every 5th colony starting at one end of the apiary until they sample a total of eight colonies.

The same statistics were applied to develop colony and apiary level sampling plans for researchers. These plans were developed to achieve a higher level of precision, since researchers are more interested in data gathering instead of making a treatment decision. To estimate the apiary-level mite prevalence for a precision of 0.10 researchers should sample 280 adult bees from each colony, where the number of colonies to sample depends on the apiary size. To find the number of colonies to sample, researchers can apply the formula y=0.403x+10.41, where y is the number of colonies to sample and x is the number of colonies in the apiary. To estimate the mite level on adult bees in a single colony, researchers should sample 17.2 (18 rounded up to an integer) 35-bee sample units to achieve a precision of 0.10 when there are ≥ 5 mites per 100 adult bees. If there are ≤ 5 mites per 100 bees, the reliability will be ± 0.5 mites. To convert the adult bee infestation to the colony infestation, researchers should estimate the adult bee and pupae populations and include the relative number of pupae compared to all bees (adults and pupae) for an R²=0.88.

Objective 2. It is not feasible to expect beekeepers to count out 280 adult bees in a sample; however consistently sampling the correct number of adult bees is important in estimating the mite infestation. To aid in sampling the correct number of bees we have developed a sampling device that measures out the bees by volume. The creation of a sampling device will allow beekeepers to sample 280 adult bees from a colony with both ease and accuracy in a standardized manner. The body of the device is a jar with a wire mesh cap that can hold up to four 280 bee samples at a time. Once the bees are in the device, the mites will be dislodged from the bees using powdered sugar (Macedo et al., 2002). Bees are coated with powdered sugar, let set for a minute, and then the jar is inverted and shaken so the loose powdered sugar and any mites fall through the wire mesh into a white dish. The dislodged mites can be counted, and the bees released into one of the sampled colonies or outside the entrance unharmed. The simplicity of the device will increase its use and the use of the sampling plan. Once beekeepers tally the number of mites, they can use the simple chart to estimate the colony infestation.

The development of the sampling plan allows for beekeepers to monitor the mite levels in their colonies. Sampling and recording mite levels within apiaries over time can result in beekeepers determining the mite level they need to treat to make it until the next treatment window and the levels they can forgo a treatment. If beekeepers nationwide monitor mite levels in their colonies, a much clearer picture of treatment thresholds and mite dynamics will emerge and can be investigated in future studies.

Objective 3. Our line of bees bred for both Hygienic Behavior (HYG) and Varroa Sensitive Hygiene (VSH) had significantly lower mite levels compared to an unselected line of bees demonstrating that this line can reduce mite loads and thus the frequency of pesticide application (Ibrhaim et al., 2007). By autumn, the HYG/VSH colonies had significantly fewer mites on adult bees and in worker brood compared to the control colonies and the HYG colonies had intermediate mite populations. There were no differences among the lines in mite reproductive success.


Collins, A.M., Pettis, J.S., Wilbanks, R., and M.F. Feldlaufer. 2004. Performance of honey bee (Apis mellifera) queens reared in beeswax cells impregnated with coumaphos. Journal of Apicultural Research, 43: 128-134.

Currie, R.W. 1999. Fluvalinate queen tabs for use against Varroa jacobsoni Oud.: efficacy and impact on honey bee, Apis mellifera L., queen and colony performance. American Bee Journal, 139: 871-876.

De Jong, D., De Andrea Roma, D., and L.S. Gonςalves. 1982. A comparative analysis of shaking solutions for the detection of Varroa jacobsoni on adult honeybees. Apidologie, 13: 297-306.

Delaplane, K. and W.M. Hood. 1997. Effects of delayed acaricide treatment in honey bee colonies parasitized by Varroa jacobsoni and a late-season treatment threshold for the southern USA. Journal of Apicultural Research, 36: 125-132.

Delaplane, K.S. and W.M. Hood. 1999. Economic threshold for Varroa jacobsoni Oud. in the Southeastern USA. Apidologie, 30: 383-395.

Fraizer et al., 2008 M, Mullin C, Frazier J, and S. Ashcraft. 2008. What have pesticides got to do with it? American Bee Journal, 148: 521-523.

Haarmann, T., Spivak, M., Weaver, D., Weaver, B., and T. Glenn. 2002. Effect of fluvalinate and coumaphos on queen honey bees (Hymenoptera: Apidae) in two commercial queen rearing operations. Journal of Economic Entomology, 95: 28-35.

Harbo J.R., Harris J.W. 1999a. Selecting honey bees for resistance to Varroa jacobsoni, Apidologie 30, 183-196.

Harbo, J.R. and J.W. Harris. 1999b. Heritability in Honey Bees (Hymenoptera: Apidae) of Charateristics Associated with Resistance to Varroa jacobsoni (Megsostigmata: Varroidae). Journal of Economic Entomology, 92: 261-265.

Harbo J.R., Hoopingarner R. (1997) Honey bees (Hymenoptera: Apidae) in the United States that express resistance to Varroa jacobsoni (Mesostigmata: Varroidae), J. Econ. Entomol. 90, 893-898.

Ibrahim A, Reuter GS, Spivak M. 2007. Field trial of honey bee colonies bred for mechanisms of resistance against Varroa destructor. Apidologie 38: 67-76.

Ibrahim A, Spivak M. 2006.The relationship between hygienic behavior and suppression of mite reproduction as honey bee mechanisms of resistance to Varroa destructor Apidologie. 37: 31-40.

Macedo, P.A., Wu, J., and M.D. Ellis. 2002. Using inert dusts to detect and assess varroa infestations in honey bee colonies. Journal of Apicultural Research, 40: 3-7.

MAFF. 1998. Varroa jacobsoni: monitoring and forcasting mite populations within honey bee colonies in Britian. MAFF Publications PB 3611 pp.12.

Martin, S.J. 1998. A population model of the ectoparasitic mite Varroa jacobsoni in honey bee (Apis mellifera) colonies. Ecological Modelling, 109: 267-281.

Martin, S.J. 1999. Population modelling and the production of a monitoring tool for Varroa jacobsoni an ectoparasitic mite of honey bees. Aspects of Applied Biology, 53: 105-112.

Naranjo, S. E., and W.D. Hutchison. 1997. Validation of arthropod sampling plans using a resampling approach: software and analysis. American Entomology, 43: 48-57.

Rinderer, T.E., de Guzman, L.I., Lancaster, V.A., Delatte, G.T., and J.A. Stelzer. 1999. Varroa in the mating yard: The effect of Varroa jacobsoni and Apistan on drone honey bees. American Bee Journal, 139: 134-139.

Research conclusions:

The immediate goal of this research was to develop a sampling plan to help beekeepers estimate the infestation of V. destructor per apiary or per colony based on well-documented sampling statistics. The long-term goal was to help beekeepers use the sampling plan to make informed treatment decisions so they can reduce pesticide application to avoid unnecessary costs, pesticide residues, and evolution of resistance of the mites to chemicals.

It can also help beekeepers find colonies with the desirable traits hygienic behavior through monitoring colonies to examine which tend to have fewer mites. These colonies can be bred from those colonies to increase the frequency of colonies with natural defenses.

The sampling device will allow beekeepers to determine mite levels and make a treatment decision in a single trip to an apiary. Because this device is standardized, beekeepers can compare their colony or apiary infestation levels with other beekeepers in a meaningful way.

The most important outcome of this project is providing beekeepers with a way to acquire critical information about mite populations in their colonies. Beekeepers will finally be able to reliably monitor mite populations and dynamics within their own colonies.

Economic Analysis

Monitoring mite populations through sampling must be economically advantageous for beekeepers to implement this practice into their operations. The advantages can be communicated by showing that sampling is less expensive (both time and cost) than blanket treating, even if sampling reveals that a treatment is needed. For a hobby beekeeper, testing each colony will probably take 8-10 minutes. This includes the time it takes to open the colony, find the first brood frame, check for the queen, take the sample, and perform the powdered sugar roll. For a single person in a commercial operation testing four colonies at a time, sampling bees with the sampling device should take between three and five minutes per colony, depending on the level of experience. This includes the time it takes to perform two powdered sugar rolls (one for each set of four colonies). Sampling a single apiary should not take over 30 minutes. The time can be reduced by using two sampling devices or having two people sample. Beekeepers can integrate sampling into normal management regimen, such as while giving their colonies nutritional supplements. Since beekeepers are already in the colonies, taking eight samples would add little extra time. The use of this sampling plan gives beekeepers an immediate estimate of the mite population, permitting them to make a treatment decision while they are already, reducing travel time.

Monitoring mite levels can be more cost effective than treating without sampling. According to the current prices for the miticide treatments formic acid, thymol, tau-fluvalinate and coumaphos, the cost of treating an apiary with 30 colonies is $78 ($2.60 per colony, two fluvalinate strips per colony). These prices do not include any extra equipment, time and labor for the application, or the superfluous expense if the treatment is not warranted. The treatments for 100 apiaries (3,000 colonies assuming 30 colonies per apiary) would have an approximate cost of $7,800. In contrast, monitoring an operation of 100 apiaries would take approximately 50 hr (assuming 30 minutes per apiary). If labor costs $30 per hr (an employee paid $15 per hr, plus $15 in benefits), then the cost to monitor is approximately $1500 or about 19% of what it would cost to treat with fluvalinate. If 19 out of 100 apiaries are deemed to be below a treatment threshold, then sampling is cost effective. Even when apiaries are tested and found to have high enough mite levels to warrant a treatment, sampling would still be beneficial. Beekeepers can learn about the growth cycle of the mite in their colonies, leading them to determine the best times of year for them to sample and potentially treat

Farmer Adoption

M. Spivak, K. Lee, and G. Reuter have presented the sampling plan and device to beekeepers at meetings and in on-on-one conversations and we have received enthusiastic interest from the majority of beekeepers. Beekeepers have expressed a desire to reduce treatments for years. They understand the need to sample, but until now they have lacked a method to estimate apiary infestation. We anticipate a quick acceptance by many beekeepers since they are eager to bring this mite under control.

To increase the exposure of beekeepers to the plan, we will publish the sampling plan and guidelines on the University of Minnesota webpage and in the American Bee Journal, we will continue giving presentations at local and national beekeeper meetings, and we will have the sampling device produced along with instructions. We are seeking to have the sampling device manufactured by a beekeeping supply company, which would make the device available to beekeepers across the country and world.

Participation Summary

Educational & Outreach Activities

Participation Summary:

Education/outreach description:

M. Spivak presented these findings to over 25 different professional and public meetings of beekeepers, scientists and the general public in 11 states across the US including 12 talks to groups within MN. She also presented in Peru, Chile, Argentina, and Nordic-Baltic countries (including Norway, Denmark, Sweden, Finland, Estonia, Latvia, Lithuania). K. Lee presented her findings to the American Bee Research Conference (Sacramento, CA, January, 2008), the International Union for the Study of Social Insects in Puerto Rico (September 2008), the MN Hobby Beekeeping Association (April 2008 and April 2007), Entomological Society of America (December 2007), and the Minnesota Honey Producers (December 2006).

There are two papers by A. Ibrahim, G. Reuter and M. Spivak resulting from objective 3, and K. Lee will be submitting two for publication on Objectives 1 and 2. In addition, we will be writing at least two extension papers on the results of Objectives 1 and 2.

Ibrahim A, Reuter GS, Spivak M. 2007. Field trial of honey bee colonies bred for mechanism of resistance against Varroa destructor. Apidologie 38: 67-76.

Ibrahim A, Spivak M. 2006.The relationship between hygienic behavior and suppression of mite reproduction as honey bee mechanisms of resistance to Varroa destructor Apidologie. 37: 31-40.

Project Outcomes


Areas needing additional study

We will follow up with beekeepers that implement the sampling plan to collect feedback and address any issues. We will provide access to the sampling device and encourage beekeepers to monitor their colonies and keep records of the mite levels and the success of the apiary, which can be complied to better determine economic thresholds for commercial beekeepers.

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.