December 2010 1151

eeping bees healthy requires bee-keepers to address mite problemsthrough management techniques

such as breeding, colony manipulation, ortreatment with miticides. While breedingand manipulation are the first and most im-portant lines of defense, many beekeepersstill need to use chemical treatments to keeptheir colonies alive. No beekeeper enjoysapplying chemicals to their hive because itis expensive, time consuming, and may ad-versely affect bees. However, many bee-keepers apply miticides once or twice a yearto reduce mite infestation, often without firstchecking to see if the mite level is highenough to warrant treatment. Reducingtreatments to only those necessary is imper-ative to keeping costs down, reducing hivecontamination, and slowing the develop-ment of mite resistance to new miticides.

Here we describe an efficient samplingmethod for Varroa. This method will allowbeekeepers to make treatment decisionsbased on knowledge of actual infestationlevels of mites on adult bees and workerbrood in an individual colony or entire api-ary. Many sampling methods have been de-veloped previously and include dislodgingthe mites from adult bees with alcohol, pow-dered sugar, or ether; monitoring the naturalmite fall with a sticky board; or samplingbrood with either a capping scratcher to pullout drone brood or examining individualworker pupae. These methods have typicallybeen used to determine if mites are presentor absent. If these methods are used to quan-tify mite levels, it is often unclear how thenumber of mites in the sample translates into actual numbers of mites in a colony orapiary.

To develop an easy and standardized wayof sampling adult bees for mites, we ad-dressed two questions. First, what is the best

method of sampling adult bees to determinemite infestation? Second, can the colony in-festation level (i.e. mites on both adult beesand pupae) be estimated from a sample ofadult bees? We chose to focus on samplingadult bees because sampling worker broodis cumbersome, sampling drone brood haswide variability, and the use of sticky boardsrequires two trips to the apiary, specialequipment, and at least several days to get agood estimate of natural mite fall. In this ar-ticle, we describe a sampling plan for bee-keepers. A method suitable for researchers,with detailed sampling statistics, is pub-lished online in the Journal of EconomicEntomology8.

Sampling to Determine Mite Infestationof Adult Bees

To determine the most efficient way forbeekeepers to sample adult bees for Varroa,we needed to understand the distribution ofmites within a colony and an apiary. To ob-tain these data, we sampled a total of 954colonies in 31 apiaries owned by five com-mercial migratory beekeepers. The opera-tions were sampled in Minnesota, NorthDakota, California, and Texas. The sizes ofthe five operations ranged from 1,000 to20,000 colonies, and the number of coloniesat sampled apiaries ranged from 24 to 84.Sampling was done in March (TX and CA)in 2006, May-June and August-September(MN and ND) in 2005, 2006, and 2007. Wecollected approximately 35 adult bees in al-cohol from each frame in each colony andrecorded the following information for eachsample: date, beekeeper, yard, pallet,colony, frame location, and comb contents(e.g. open brood, sealed brood, pollen, nec-tar/honey, or empty). The 35-bee sampleswere taken back to lab, where we countednumber of bees and mites in each. In 142 of

the colonies, one sample of approximately300 adult bees was taken to compare themite infestation to the multiple 35-bee sam-ples. We found no difference between the in-festation of the large sample and thecombined 35-bee samples from the samecolony, indicating a single large sample isadequate to estimate adult bee infestation.

Using data from the 954 commercialcolonies, we wanted to know how manyadult bees needed to be sampled to accu-rately estimate mite infestation on all adultbees? We found that a sample size of 300bees per colony is adequate to determine themite infestation level of adult bees in acolony. The recommendation of sampling300 bees confirms previous recommenda-tions9, 10, but this is the first time this num-ber of bees has been associated with aprecision level.

To develop an apiary level sampling plan,we used a computer program11 to determinethe number of adult bees in a colony andcolonies in an apiary to sample. We firstcombined different numbers of the 35-beesamples within each colony to achieve dif-ferent sample-unit sizes (i.e. bees to sampleper colony, ranging from 35 bees to 280bees). To estimate the infestation in an api-ary, the computer program randomly selectsthe inputted sample-unit sizes for each api-ary until it reaches the set precision for thenumber of colonies per apiary to sample.Thus, if 35 bees are sampled per colony, 16colonies would need to be sampled in anapiary, and if 280 bees are sampled, only 8colonies would need to be sampled. Since itis easier to sample more bees per colonythan fewer bees from more colonies, we rec-ommend sampling 300 bees from each of 8colonies to estimate apiary infestation. Wechose 300 bees (rather than 280) to err onthe side of obtaining a better estimation.

American Bee Journal1152

We next determined if mites congregatedon brood frames (frames with eggs, larvae,or sealed pupal cells). We found frames withbrood had significantly more mites thannon-brood frames, with 2.4 mites per 100bees on frames with brood comb and 1.8 per100 bees on frames without brood. We rec-ommend beekeepers sample from a framewith brood.

Then, we wanted to examine how mitesare spatially distributed among colonies inan apiary. We were not surprised to find thatsome colonies had higher mite levels thanothers. However, we wanted to know if di-rection of the hive entrance or location in anapiary (e.g. colonies on pallets at the end ofa row in an apiary compared to colonies inthe middle) contributed to higher mite loads.Our analyses suggested that mite levelswere independent of colony direction or lo-cation.

Mite levels were sometimes highly vari-able among apiaries in beekeeper operationssampled at the same time of year. Thismeans that beekeepers should make treat-ment decisions on an apiary-by-apiary basisand should not assume that all apiaries havesimilar levels of mite infestations.

Relationship Between Mites on AdultBees and Mites in Brood

We developed a simple “correction fac-tor” to account for the proportion of totalmites in the colony that are on pupae (i.e.sealed brood) by intensively sampling broodin 62 colonies from two commercial bee-keepers in MN and ND. The colonies weresampled in May-June or August-September.These are times of year when many bee-keepers normally treat for Varroa. In eachcolony, we estimated the population of adultbees and sealed worker brood, and the miteinfestation of adult bees and sealed workerbrood. We also estimated the number ofdrone pupae and mite infestation on dronepupae in seven intensely sampled Universityof Minnesota colonies.

Drones were not included in the correc-tion factor because the number of mites indrone brood was dwarfed by the number ofmites found on adult bees or in workerpupae. In this study, an average colony had24,500 adult bees and 14,000 worker pupae.Drone brood comprised, on average, only3.2% of the total pupae. An average 6.8% ofall mites were on drone pupae, while an av-erage of 45.6 % were on worker pupae, andthe remaining 47.7% were on adult bees.These results suggest that unless there is anabnormally high amount of drone brood inthe colony, the number of mites in dronebrood contribute little to the total number ofmites in the colony.

We examined the relationship betweenthe adult bee infestation and the colony in-festation (density of mites on adult bees andworker pupae). We included two factors thatwe predicted would influence the relation-ship: the time of year the colony was sam-pled and the ratio of worker pupae to adultbees. Although both factors can affect the re-lationship between adult bee infestation and

colony infestation, the statistics indicatedonly adult bees need to be sampled, and acorrection factor applied, to estimate thetotal colony mite density. We calculated thiscorrection factor by plotting the adult bee in-festation against colony infestation (miteson adults and in brood) to find the slope ofthe line, which was 1.8. Thus, the number ofmites on adult bees can be multiplied by 1.8to correct for the number of mites in workerbrood. To simplify and err conservatively onthe side of over-estimation, we recommendusing a correction factor of 2, or doublingthe adult bee infestation level to estimate themite infestation in a colony. If there is nobrood, then no correction factor is needed.If there is an abnormally high amount ofworker or drone brood relative to adult bees,there is a possibility the correction factorcould lead to an underestimate of total miteload.

Sampling Plan Recommendations forBeekeepers

Based on our results, we provide the fol-lowing recommendations for beekeepers toestimate the mite infestation level:

Colony1. Sample 300 adult bees from one frame

containing brood (i.e. eggs, larvae or pupae). 2. Use Table 1 to apply the correction fac-

tor to convert the number of mites on adultbees to the colony infestation level (i.e. totalmites on adult worker bees and in workerpupae). Or divide the number of mites foundin a sample of 300 bees by 3 and multiply

the result by 2 to estimate colony infestationlevel.

Apiary1. Sample 300 bees from one brood frame

from each of 8 colonies. Sample every fifthcolony in an apiary until 8 colonies are sam-pled. This plan is valid for apiaries with 24to 84 colonies.

2. Use Table 1 to apply the correction fac-tor to convert the number of mites on adultbees from 8 colonies to the apiary infesta-tion level. Or divide the total number ofmites from adult bees from 8 colonies by 12.

How to Sample Adult BeesCounting out 300 bees for each sample is

impractical, but there are a few ways to sam-ple by volume since 300 live bees occupyabout 0.42 cups or 100 ml. We realize that0.42 cups of bees is a strange volume, how-ever bees are small so small variations in thevolume can mean large variations in thenumber of bees in a sample. For example,1/3 cup averages just under 200 bees, 0.4cups is about 275 bees, and 1/2 cup is justunder 400 bees. It is important to accuratelymeasure out the correct volume to sample300 bees.

To make your own measuring cup, add0.42 cups of water (1/3 cup + 1 tablespoon+ 1 ¼ teaspoon) to a cup that preferably hasa smaller diameter relative to height, andmake a mark at the water line. Add a handleto the cup to make sampling easier. To sam-ple, rap bees off of a brood frame into a 5gallon bucket or plastic wash-dish container,

Table 1. Number of mites found in a sample of 300 adult bees and thecorresponding colony mite density after the correction factor is ap-plied, and the number of mites found in eight 300 adult bee samplesand the corresponding apiary mite density after the correction factoris applied.

December 2010 1153

then use the marked cup to scoop out 300bees (Figure 1). Rap the cup on a hard sur-face to make sure the bees are at the 0.42cup line (add or subtract bees as needed).Keep the bucket or wash-dish from becom-ing coated in nectar, since mites can stick tothe nectar. If your cup is rectangular (suchas those that come in some powdered laun-dry detergent boxes), then you can use themarked cup by running it gently over thebacks of bees, causing them to tumble downinto the cup (Figure 2). Again, be sure tomeasure the bees at the 0.42 line. One fur-ther sampling method is to use a devicecalled “Gizmo” that was designed by GaryReuter to measure 300 bees (Figure 3).Gizmo is sold by the Walter T. Kelley Bee-keeping Company, or you can make yourown using the plans online at the Universityof Minnesota Bee Lab website (www.exten-sion.umn.edu/honeybees). The Gizmo de-

vice can be more accurate, but if you con-sistently measure bee at the 0.42 line, thenthe cup method works just as well.

Once the bees are measured, we recom-mend dislodging mites from the adult beesusing the powdered sugar method12 (Figure4). It is quick, easy, and gives a adequateestimate of the mites in the sample. Dumpthe 300 bee sample into a jar with a size 8hardware mesh top and add about 2 Table-spoons (or a hive tool scoop) of powderedsugar. Add more sugar if the bees don’tlook ghostly. Let the jar set at least oneminute in the shade so the bees don’t over-heat, then shake vigorously for one minuteinto a white dish. Be sure to shake hard.Some bees may lose a leg or two, but you’llwant to get as many mites off the bees aspossible. After shaking, add a touch ofwater to the dish to dissolve the powderedsugar, and count the mites. Replace the

sugar-coated bees to their colony wherethey will be groomed by nestmates. Inareas with high humidity, the powderedsugar may not work well because the sugarclumps in the jar so that some mites are notdislodged from the bees. Dislodging mitesusing alcohol and then straining them ismore accurate13, but it kills the bees. If youprefer the alcohol wash, Dr. Medhat Nasrmade a handy device (discussed in theAmerican Bee Journal, August 2010) oryou can make a strainer with size 8 hard-ware cloth to separate the bees and mites.

Treatment DecisionsOnce you sample a colony or an apiary to

determine the mite infestation level, howcan you use the information to help make atreatment decision?

Stationary colonies (e.g. beekeepers thatkeep their colonies in one location yearround)

Researchers have found treatment thresh-olds for colonies in a stationary apiary to be 10-12% colony mite infestation in autumn14, 15, 10.

However, the threshold may be differentin different regions, so these thresholds maynot apply to other locations. There are manyfactors that can influence the density ofmites a honey bee colony can tolerate, in-cluding number of neighboring colonies,length of brood rearing season, nutrition, hy-gienic behavior, and disease and parasitelevels. We highly recommend that ALL bee-keepers sample their colonies for mites inearly spring and late summer, and comparemite levels with other beekeepers in thesame area. It would be very beneficial forgroups of beekeepers to keep records of mitelevels in their regions. In this way, regionalpatterns could emerge and show the level ofmite infestation that warrants treatment tokeep colonies alive, and what level does notwarrant treatment. While other factors (i.e.colony strength, presence of diseases) affectcolony survival, having ongoing records ofmite levels, before and after treatment, indifferent regions at different times of yearwould be extremely useful for developingregional treatment thresholds.

Figure 1. Rapping a frame in a plastic wash-dish container, then using a cup that holds 0.42 cups to meas-ure 300 adult bees.

Figure 2. Using a rectangular cup, marked inside at 0.42 cups, tomeasure 300 adult bees. Gently run the cup down the backs of thebees, causing them to tumble in. Rap the cup until the bees reach the0.42 line.

American Bee Journal1154

Transported colonies (e.g., commercial mi-gratory beekeepers)

There are no reported estimates of treat-ment thresholds for migratory beekeepers, asthe thresholds will vary depending on region,season, and migratory path. However, thesame principle holds that ALL migratory bee-keepers should sample their apiaries for mitesprior to “treatment windows,” or periods oftime that treatments can be safely applied. Atreatment window could be in the spring be-fore honey supers are placed on colonies, orin late summer just after the honey supers areremoved. The idea is to treat ONLY if yourbees will not survive until the next treatmentwindow. Keeping records of mite levels be-fore and after treatment, over several years,will help beekeepers understand the mite lev-els that colonies can tolerate before the nexttreatment window. A few things that couldlower the treatment threshold are if a bee-keeper has many colonies situated in areasdense with other beekeepers within flightrange of the bees, moves and feeds coloniesto stimulate continued brood rearing for muchof the year, and if the bees have high virus anddisease levels. There will not be a singlethreshold for all beekeepers. Again, monitor-ing colonies and keeping good records canhelp beekeepers to find the infestation levelthat requires treatment in their specific oper-ation. Since the method to find the infestationlevels is standardized, beekeepers can sharetheir levels with each other in a meaningfulway and potentially help control the mite lev-els in the surrounding area.

With sampling, beekeepers have the poten-tial to decrease the use of miticides, reducechemical contamination in the hive, and savetime and money. Importantly, monitoring mitelevels is an important tool in the selection ofcolonies with few mites for breeding. Breed-ing queens from colonies whose bees havelower mite levels compared to coloniesaround them can increase the prevalence ofnatural mite resistance.

References1 vanEngelsdorp, D., Evans, J. D.,

Saegerman, C., Mullin, C., Haubruge.,E., et al. 2009. Colony Collapse Disor-der: A Descriptive Study. PLoS ONE4(8): e6481. doi:10.1371/journal.pone.0006481

2 De Jong, D., De Jong, P. H., and L. S.Gonsalves. 1982a. Weight loss andother damage to developing worker hon-eybees from infestation with Varroa ja-cobsoni. J. Apic. Res. 21: 165-167.

3 Yang, X. and D. L. Cox-Foster. 2005.Impact of an ectoparasite on the immu-nity and pathology of an invertebrate:Evidence for host immunosuppressionand viral amplification. Proc. Natl.Acad. Sci. USA 102: 7470-74.

4 Schneider, P. and W. Drescher. 1987. Theinfluence of Varroa jacobsoni Oud. onweight, development, of weight and hy-popharyngeal glands and longevity ofApis mellifera L. Apidologie 18: 101–110.

5 Chen, Y. P., and R. Siede. 2007. HoneyBee Viruses. Adv. Virus Res.70:33-80.

Figure 3. Gizmo, a device that can be used to measure 300 adultbees by volume. To operate 1) shake a frame of bees onto a pieceof flashing (or cardboard, newspaper, election sign, etc.) as longas a frame and bent into a V-shape, then 2) dump the bees into thetop of Gizmo. 3) There is a volume inside Gizmo that measures300 bees. 4) Rap Gizmo onto a hard surface three times, then turnthe handle to release the bees into the jar. 5) Remove the jar andscrew on the mesh lid, then 6) add powdered sugar to dislodgethe mites.

Figure 4. How to dislodge mites from adult bees using powderedsugar using a jar with a size 8 hardwire mesh top. 1) Add about 2Tbsp (or a hive tool scoop) of powdered sugar to the jar, or moreif the bees are not ghostly. Roll the jar to coat the bees. 2) Imageof ghostly bees. 3) Let the jar set for one minute in the shade, then4) Shake the jar hard into a dish. 5) Add a touch of water to dis-solve the powdered sugar, and 6) count the mites.

December 2010 1155

6 Cox-Foster, D. L., Conlan, S. Holmes,E. C., Palacios, G., Evans, J. D, et al.2007. A metagenomic survey of microbesin honey bee colony collapse disorder.Science 318: 283-287.

7 Johnson, R. M., Evans, J. D., Robinson,G. E., Berenbaum, M. R. 2009. Changesin transcript abundance relating to colonycollapse disorder in honey bees (Apismellifera). Proc. Natl. Acad. Sci.106:14790-14795.

8 Lee, K.V., Moon, R.D., Burkness,E.C., Hutchison, W.D., and M. Spivak.2010. Practical Sampling Plans for Var-roa destructor (Acari: Varroidae) inApis mellifera (Hymenoptera: Apidae)Colonies and Apiaries. J. Econ. Entomol.

103(4): 1039-1050.9 Delaplane, K. S. 1997. Strictly for the

hobbyist: varroa how and when to treat.Am. Bee J. 137: 571-573.

10 Strange, J. P. and W. S. Sheppard.2001. Optimum timing of miticide ap-plications for control of Varroa destruc-tor (Acri: Varroidae) in Apis mellifera(Hymenoptera: Apidae) in WashingtonState, USA. J. Econ. Entomol. 94: 1324-1331.

11 Naranjo, S. E., and W. D. Hutchison.1997. Validation of arthropod samplingplans using a resampling approach: soft-ware and analysis. Am. Entomol. 43: 48-57.

12 Macedo, P. A., Wu, J., and M. D. Ellis.

2002. Using inert dusts to detect and as-sess Varroa infestations in honey beecolonies. J. Apic. Res. 40: 3-7.

13 De Jong, D., De Andrea Roma, D., andL. S. Gonςalves. 1982b. A comparativeanalysis of shaking solutions for thedetection of Varroa jacobsoni on adulthoneybees. Apidologie 13: 297-306.

14 Delaplane, K. S. and W. M. Hood. 1999.Economic threshold for Varroa jacobsoniOud. in the Southeastern USA. Apidolo-gie 30: 383-395.

15 Martin, S. J. 1999. Population modellingand the production of a monitoring toolfor Varroa jacobsoni an ectoparasitic miteof honey bees. Aspect. Appl. Biol. 53:105-112.