Midnite Bee-Beekeeper's: Reports REPORTS  
A Rapid Field & Laboratory Method to Detect Varroa jacobsoni in the Honey Bee (apis mellifera)
*Department of Applied Zoology
P.O. Box 27(Viikki C)
SF-00014 University of Finland
Tel: +358-9-19158393 or +358-40-5536791 GSM
fax: +358-9-70858463
Tel.&Fax.(home): +358-9-3881451

Revised manuscript received for publication March, 2000


The most serious disease in apiculture is varroasis caused by the mite Varroa jacobsoni Oudemans. Immediate diagnosis of varroasis is not possible, especially in spring time when bee colonies are purchased. A new method was developed to detect ecto- and endoparasites of honey bees focussing on the varroa mite (n=129 separate examinations). The method incorporates washing and centrifugal force to separate mites from bees in a closed container. Four rotational speeds: 6342, 5718, 5076 and 4752 rpm, were tested, for 10, 30 and 60 seconds, with two concentrations of detergent solution. Both live and frozen bees were examined. The high detergent level as well as the highest speed were significantly better (p<0.05) than the low detergent and lower speeds respectively. The duration of the centrifugation is not critical. The qualitative detection efficiency (detection % of all infested samples) of the method is 100% (n- 93) when the infestation level is >3 mites/100 bees, and at 1 mite/100 bees the efficiency is >85% (i.e. 13 positive detection out of 15 samples). The quantitative detection efficiency (% of mite individuals detected in a sample) of the method under optimum conditions is >90% when the infestation level is >3 mites/100 bees. The detection rate when using frozen bees was significantly better (p<0.05) than when live bees were used.


Varroajacobsoni Oudemans is an external parasitic mite of honey bees. It was first described in Apis cerana, in Java (Oudemans 1904). The problem started when beekeepers transported Apis mellifera to Asia (Goncalves et a1.1985). Today varroa is found world wide, with some exceptions (Bradbear 1988, De Jong 1990). The mites attach to the adult bee between the abdominal segments or between body regions (head, thorax, abdomen) and are therefore difficult to detect (Ritter 1981, Shimanuki and Knox 1991). Generally, varroa remains undetected up to three or four years after a hive first becomes infested. Clinical symptoms become apparent when several thousand mites are present in a colony (Ritter 1981, Robaux 1988).

A laboratory diagnosis method is to shake bees in a liquid using a rotary shaker for 30 minutes. Various recommendations have been made for using different liquids, such as hot water (25deg;,40deg;,60deg; and 100deg;C) (De Jong et a1.1982), 96% ethanol (De Jong and Goncalves 1981) 70% alcohol (ethyl or isopropyl) (USDA 1987) 25% alcohol (De Jong et a1.1982) detergent solution, hexane, gasoline or diesel fuel (De Jong et a1.1982, USDA 1987).

One of the field methods to detect the mite is to uncap the brood cells and remove the pupae with forceps, a sharp knife or capping scratcher (De long 1979, USDA 1987, Szabo 1989). Throughout the year, the ether-roll method is used in the quick field diagnosis of varroa. Some 450-1000 live bees are collected into a wide mouth jar (Ellis et a1.1988, Herbert et al. 1989, USDA 1987, Gruszka 1988) by using a short burst of ether. The bees are then rotated in a jar for a few seconds and the mites adhere to the surface of the jar. The average detection efficiency of ether roll is 35% quantitatively (individual mites detected in a sample) (Herbert et a1.1989). and 12.5% qualitatively (detection rate of all infested samples) (Ellis et al.1988). However, the latter value would have been much higher in a survey with higher infestation levels. Since smoke is used in the apiary, the high flammability of ether is a safety concern when using the ether roll method.

Brood examination is a tedious process and is not possible in the absence of a capped brood (Herbert et a1.1989). When colonies are sold in spring, beekeepers and apiary inspectors need an accurate technique for the rapid detection of varroa. The decline of colony population affects the practicality of taking brood and/or substantial adult bee samples for ether rolling. The main objective of these studies was to develop a highly efficient method of detecting varroa mite on a small sample (100-150) of adult bees. Another objective was to determine the practicality under field conditions.

Figure 1. The Langstroth comb behind the prototype device is for the size comparison.


A new device has been designed and developed over four years by the Department of Applied Zoology of the University of Helsinki. The prototype device was built in Finland by Farmcomp Ltd. (figure 1). Varroa mite detection research was conducted at the Agriculture Research Centre in Jokioinen, Finland. The device was tested in the U.S.A. for the detection of Acarapis woodi, in cooperation with the U.S. Department of Agriculture, Iowa State University and Dadant and Sons Inc. However, the detection method for the internal mite is still under investigation and is out of the scope of this article.

The method incorporates centrifugal force to dissociate varroa mites from bees in a closed container. Bees and a shaking solution were placed in a rotating chamber. Attention was given to the consistency of the load (bees and liquid) during experimentation to have a steady rotational speed. consequently a constant centifugal force. Four rotational speeds (6342, 5718, 5076 and 4752 rpm.) were examined for 10, 30 and 60 seconds. In a preliminary research the cheapest and most efficient liquid was found to be a solution of detergent which has no hazardous side effects. Washing detergent (Leijona) and water was used. Adding the detergent to a container of water avoids unwanted foam. No annoying foam was formed during the analysis of the samples. Two concentrations of detergent solution were investigated: low (one ml in three liters ) and high (one ml in one liter). Sample collection

Bees were sampled from two apiaries in Partalansaari, southeastern Finland. A pair of samples (each 0.5 deciliter mean 117 bees) were collected from 36 bee colonies every three weeks from

Figure 2. The inner structure of the closed rotational chamber.

May to August. Equal numbers of colonies of Apis mellifera mellifera, A. mellifera carnica and A. mellifera ligustica were used in this investigation. Collected bee samples were divided randomly into two groups for live (117+;23x+SD range 61-171) and frozen bee (118+25 range 60- 165) analysis. Bees were shaken from a comb into a container; then 0.5 dl of bees were measured and placed in a plastic bag. The live bee samples were analyzed within fifteen minutes. At the end of the day the bee samples were carried to a laboratory and frozen at approximately -20degC. The samples were defrosted at room temperature prior to examination. Initial examination

A bee chamber within the rotational chamber consisted of pillars, 3 mm apart, to filter the bees (figure 2). The solution of water and detergent, without bees, (150 ml) was measured by the use of the rotational chamber and poured into the plastic bag, over a sample of bees (frozen or live). The solution and bees were then poured back into the rotational chamber and deposited into the centrifugal device. The remaining mites, if any in the bag, were also added to the chamber with a paint brush. By having a wet cluster of bees it becomes easier to place them into the rotational chamber. Research demands accuracy, however in practice bees can be brushed directly into the chamber. The motor for rotating the sample can be run by an automobile 12 volt battery. After centrifuging the cylinder at various rotational speeds and times, mites were counted on the bottom of the transparent rotational chamber and bees were removed. The removed bees were placed in a rotary shaker for 30 minutes in 70% ethanol (USDA 1987), a method which removes 100% of the mites (De Jong et a1.1982). The bees and any fallen mites were then washed under a shower of tap water over 2 sieve screens (3mm and 0.5mm in pore-size) for 2-3 minutes (Fakhimzadeh 1993). The mites and the bees were then counted (n=129 separate examinations). Experimental design and statistical methods

Experiment 1. The effect of concentration of the detergent solutions

This was 2 (3) factorial design with fixed effects of detergent concentration (low and high), speed 6342 and 5718 rpm for 30 and 60 seconds. Two levels of each factor were chosen based on previous experiments giving a total of 8 treatment combinations. Samples of live bees n= 26 (0.5d1= 100.6+;12.3x+SD ranging 76-120 bees) were randomly assigned to each treatment and inspected for mites blindly (the examiner not knowing the history of a sample), 4-6 replications were used. The bees were infested with 13.9+;14.1x+;SD ranging 0.9-49.6 mites per hundred bees and 8 samples had < 3% mite infestation.
Experiment 2. The use of frozen bees vs. live bees

This was also a factorial design with fixed effects. The factors were freezing/not freezing, speeds 5718, 5076 and 4752 rpm and times 10, 30 and 60 seconds, giving a total of 18 treatments (n=103 out of which 28 samples had <3% infestation). The low concentration of the detergent solution was used. Randomization and inspection of the samples were as in experiment 1. The analyses of both experiments were accomplished with SAS (Statistical Associates Software) General Linear Modelling (GLM-procedure). Both experiments were conducted simultaneously.

The effects of detergent and speed in experiment 1 were both significant (p=0.026 and p=0.005, respectively). In experiment 2, where different speeds were used, none of the effects were significant (n=103). However, the effect of freezing was further analyzed by excluding samples with low infestation levels (< 3.0%) (19 from 69 live samples and 9 from 34 frozen samples). A significant difference (p<0.05) was detected in favour of frozen bees. The live bees (n= 69) were infested with x+SD 8.7+8.4 range (0.7 to 38.2) mites per hundred bees. Frozen bees n=34 were infested

Figure 3: The quantitative detection efficiency based on experiment 2 (n=100 separate examinations). Frozen bees, live bees and combined data are shown by an angle, circle and dash, respectively. Categorical number of observations of infestations <5 to >40 were 43, 24,13, 9, 3, 5, 3, respectively. Notice that the optimum speed (6342 rpm) and high concentration of detergent solution were not used in this experiment.


Figure 4. The qualitative detection efficiency based on both experiments (n=65 separate examinations) The categorical observations were 15, 13, 13, 11, 8, 5, respectively.


Figure 5. The effect of detergent concentration on the mite detection (quantitative). Means+SE. N=12 in each concentration.
Figure 6. The effect of freezing the bees on the mite detection (quantitative). Means+ SE. N=50 for live bees, 25 for frozen bees.

Figure 7. The effect of freezing the bees on the mite detection (quantitive). Means + SE. N=50 for live bees, 25 for frozen bees.
with ztSD 15124.8 range (0.7- 105.1) mites per hundred bees. The quantitative detection efficiency of the method is >90% (i.e. over 90% of mites in the sample are detected) when optimal conditions are in use (figures 3, 5 and 6). The qualitative detection efficiency of the method is 100% when infestation level is >3 mites/100 bees (n=93), at 51 mite /100 bees the efficiency is >85% (figure 4).The effects of single factors are displayed in figures 5 to 7.

The effect of detergent concentration is quite clear (figure 5) and becomes obvious considering the washing and centrifuging mechanisms that are utilized.

Speed was significantly different only in experiment 1 since in experiment 2 only the lower speeds were used. In fact, 6342 rpm seems to differ from 5718, 5076 and 4752 rpm (figure 6).

Freezing seemed to have no effect but this was in fact due to the large variation; it rose mainly from low infestation samples and could be minimized by excluding samples of <3% infestation level. Again a logical explanation to the freezing effect can be found that freezing at -20 C may dislodge the mites in the period before actual freezing occurs (figure 7).

The duration of centrifugation is not critical; even 10 seconds of rotation seems to be enough especially when optimal levels of other factors are chosen. The maximum velocity, causing the highest centrifugal force, was obtained in less than ten seconds. The mites are either dislodged from the bees (e.g. when on top of the bees' thorax) with the maximum centrifugal force, or remain in a safer position between the bees' segments. Since the duration of centrifugation does not increase the force, it is logical that after 10 seconds the duration does not play any significant role.


The qualitative detection efficiency (% detected in infested samples) is 100% for the infestation level above three percent (n=93). Even at one percent infestation level the efficiency is more than 85% (i.e. 13 positive detections out of 15 samples). This decline is due to high fluctuation in detection rate in low infestations. The detection rate (0%-100%) is based on only one mite in the sample. The quantitative detection efficiency (% detected in an infested sample) under optimum conditions was also >90% for the infestation levels above. three percent. while in the ether roll method it is 35% (Herbert et al.1989). Figure 3 shows the general level of the quantitative detection efficiency in relation to frozen and live bees, performed by lowest speeds (5718, 5076 and 4752 rpm) and low detergent concentration. which were less efficient for the detection. Out of nine observations in the category 15-20 mites per 100 bees, five have been performed with the poorest speed (4752 rpm) and time, hence the cause of the declination in the detection efficiency. The true quantitative detection efficiency of the method should be performed with the best speed 6342 rpm and high detergent concentration with frozen and live bees and higher number of observations in each infestation category. In spite of the high accuracy of the proposed method, nevertheless, as in ether roll method or brood method, subsampling of a colony is a limiting factor in low infestation levels. On the other hand, infestation levels under 3% may not be critical. The speed of 6342 rpm along with the higher concentration of the detergent solution was the most effective in mite detection. However, operating at 6342 rpm was difficult with the prototype device as the lid was not securely locked and had to be held by hand. Hence most of the experiments were conducted at velocities of 5718, 5076 and 4752 rpm. A preliminary research at 7272 rpm was examined and rejected since the excrement of bees necessitated filtration of the liquid to view the mites.

Non-flammability and low costs were advantages of the detergent solution over other methods (e.g. ether roll method, shaking methods). Mites were counted on the bottom of the transparent rotational chamber. Unlike the hand-shaking method where foam disturbed the operation (De Jong et a1.1982) in mechanical centrifuging, no annoying foam exists. A preliminary observation using the hand-shaking method, produced foam which hid and prevented mites from sinking. Often mites were encountered on bees after removing the bees from the rotational chamber.

The high efficiency of this method with its ten seconds of processing time suggest that it may serve as a better laboratory method for varroa detection. It is easy to apply at the apiary or laboratory, especially in the spring time when bees are vulnerable to sampling. It is not necessary to examine the bees immediately like in the ether roll method (USDA 1987) and the efficiency of the stored sample in the freezer is even higher. The impact of immediate freezing on mite detection still needs research, if of interest.

In this method with the low detergent concentration, even after 60 seconds of rotation, more than half of the bees will revive if placed in the sunshine or a warm place. It may be better to rinse the bees with water and place them in a dry warm place. This is true for mites as well, so be careful not to replace them in the colony.

I dedicate this Finnish invention to beekeepers and apiary inspectors and encourage mass production of the device and will answer all communication if more information is required. Farmcomp Ltd. in Espoo, Finland is a potential producer of the device and able to produce a few hand made devices at high costs or volume production if needed. However, as the device works with the car battery, its use is limited for the apiaries that are near a road. I hope other manufacturers can solve this problem by using rechargeable batteries or extension cords, even though this is not essential.

I thank H. Hokkanen, A-L. Varis, J. Helenius, K. Heliovaara, A. B. Mukherjee, J. Junttila and F Gates at the University of Helsinki, S. Korpela at the Agriculture Research Centre Jokioinen Finland, and R. Cox at the USDA - Iowa for their valuable comments and cooperation in the research. I am thankful to Farmcomp Ltd for manufacturing the prototype device. I am grateful for the supports of the Foundation for Finnish Inventions (Keksintosaatio), and Finnish Cultural Foundation (Suomen Kulttuurirahasto). REFERENCES

Bradbear, N.1988. World distribution of major honeybee diseases and pest. Bee Wld 69: 15-39.

De Jong, D. 1979. Field identification of Varroa jacobsoni, a parasitic mite of honey bees. Glean. in Bee Cult. 107: 639-640, 644.

De Jong, D. 1990. Mites: Varroa and other parasites of brood, in Morse R.A. and Nowogrodzki R. (Eds.) Honey bee pests, predators, and diseases, Comstock Cornell University press., pp. 474.

De Jong, D. and Goncalves, L.S. 1981. The varroa problems in Brazil. Amen Bee J. 121: 186-189.

De Jong, D., De Andrea Roma, D. and Goncalves L.S. 1982. A comparative analysis of shaking solutions for the detection of Varroa jacobsoni on adult honeybees. Apidologie 13: 297-303.

Ellis, M., Nelson, R. and Simonds, C.A.1988. Comparison of the fluvalinate and ether roll methods of sampling for varroa mites in honey bee colonies. Amen Bee J. 128: 262-263.

Fakhimzadeh, K. 1993. Unpublished data.

Goncalves, L.S., De Jong, D. and Morse, R.A 1985. The truth about varroa in Brazil in: The 300' international apicultural congress of apimondia. Nagoya-Japan. 171-173.

Gruszka, J. 1988. Evaluation of the ether sampling method to detect low level infestations of Varroa jacobsonL Beelines, (5) 8-14. Herbert, E.W., Witherell, P .C., Bruce, W.A. and Shimanuki, H.
1989. Evaluation of six methods of detecting varroa mites in
beehives, including the experimental use of acaricidal smokes
containing fluvalinate or amitraz. Amen Bee J. 129: 605-608.

Oudemans, A.C. 1904. Acarological notes xiii. Entomologische berichten uitgegeven door de Nederlandsche entomologische vereeniging 1: 169-174.

Ritter, W. 1981. Varroa disease of the honeybee Apis mellifera. Bee Wld. 62: 141-153.

Robaux, P 1988. Varroa jacobsoni: problems with diagnosis and control in Europe, in: Needham, G.R., et al., (Eds.) Africanized honey bees and bee mites. Ellis Horwood Ltd., W. Sussex, England, 370-372.

Shimanuki, H. and Knox D.A., 1991, Diagnosis of honey bee diseases. U.S.D.A., Agricultural Handbook No. AH-690, pp. 53.

Szabo, T.I. 1989. The capping scratcher: a tool for detection and control of Varroa jacobsoni. Amer. Bee J. 129: 442-403.

USDA, 1987. Varroa jacobsoni-Detection techniques. Amer. Bee J. 127: 755-757.