Apis mellifera is worldwide the most widely distributed honeybee species (van Engelsdorp and Meixner 2010). Recently elevated losses in managed honeybee colonies have been documented across the Northern hemisphere (Neumann and Carreck 2010). The performance of A. mellifera may be compromised by a range of diseases and environmental threats. One of the prime organisms challenging A. mellifera is the ectoparasitic mite Varroa destructor. Its direct damage through sucking hemolymph and its function as vector of viruses have made V. destructor the most detrimental honeybee pest (Genersch 2010). Attempts developed so far to control this ectoparasite have not yet been fully satisfactory and entail various disadvantages concerning bee, user and consumer (Genersch 2010; Rosenkranz et al. 2010). An alternative approach to established strategies, such as treatment with organic acids or synthetic acaricides, is the use of natural enemies, such as entomo-pathogenic fungi (Kanga et al. 2002; Rosenkranz et al. 2010; Meikle et al. 2012). According to Chandler et al. (2001), who evaluated the prospects of biological control of V. destructor, entomo-pathogenic fungi appear to be the most promising functional group of natural enemies. Fungi showing promising potential as biological control agents comprise primarily Beauveria bassiana, Hirsutella thompsoni and Metarhizium anisopliae (for review Chandler et al. 2001; Meikle et al. 2012), with the latter being the focal fungal species in our study. The risk of non-target effects of M. anisopliae on bee hives seems low to negligible (for review Meikle et al. 2012) but more investigations are needed (Bull et al. 2012). The lethal effects of M. anisopliae on V. destructor were observed both in the laboratory (Shaw et al. 2002; Kanga et al. 2002, 2007, 2010; García- Fernández et al. 2008; Rodríguez et al. 2009) and in the field (for review Meikle et al. 2012; Ahmed and Abd-Elhady 2013). However, nothing is known about possible sub-lethal and/or behavioral effects of M. anisopliae on V. destructor.

Here, we investigated behavioral changes of V. destructor in its phoretic phase when exposed to adult bees carrying fungal spores of Metarhizium anisopliae var. anisopliae BIPESCO 5. During the mites’ phoretic phase the adult female mites stay and feed on adult bees. Riding on adult bees, the mites can be transported to brood cells or can be distributed by foraging and swarming bees (Kuenen and Calderone 1997; Rosenkranz et al. 2010). For having reproductive success it is crucial for V. destructor to choose a suitable host, not only for their reproductive phase, but also for their phoretic phase. Therefore, the mites show distinct host preferences in both phases and are able to distinguish the age and task of adult worker bees. Nurse bees, at an age of 3 to 12 days, are the most infested group in the hive and are preferred by V. destructor over older forager bees and freshly emerged bees (Kuenen and Calderone 1997). Choosing nurse bees as their hosts ensures the successful transport of V. destructor to brood cells.

Varroa mites are able to perceive differences in light, vibration and temperature to find their hosts, however, most important are chemical cues (Rosenkranz et al. 2010). Several studies have focused on the influence of semiochemicals in Varroa host-finding behavior, for example the repellent effect of the forager bees' Nasanov pheromone (Hoppe and Ritter 1988) and various cuticular hydrocarbons of the bees (Del Piccolo et al. 2010). Changes in patterns of cuticular hydrocarbons by age have been found in drones, workers and larvae (Rosenkranz et al. 2010). Aumeier et al. (2002) showed that there are clearly distinct patterns in the relative amounts of cuticular hydrocarbons in the fourth and fifth larval stage and proposed that these patterns could function as signals in the host-parasite interaction between the bees and the mites. Further investigations on the signaling role of these hydrocarbons have been made (Del Piccolo et al. 2010) with the conclusion that V. destructor is very likely able to distinguish between different cuticular compounds, enhancing its host choice decisions. Here, we conducted two behavioral assays, choice and no-choice, to determine whether host finding and acceptance decisions of V. destructor are influenced by the application of fungal spores of M. anisopliae on bees.

Apis mellifera nurse bees were collected randomly from 12 Dadant hives established in 2013. Honeybees were considered as nurse bees when observed repeatedly visiting brood cells and tending larvae (Del Piccolo et al. 2010). The colonies belonged to the subspecies Apis mellifera carnica. All collected bees were killed by deep-freezing and stored at -20°C. No measures to control V. destructor were taken in the apiary used for bee and mite collections. Phoretic mites were collected randomly from the hives. To this end, visibly healthy adult bees infested with V. destructor were picked from the combs. They were then ice-chilled to facilitate transport and removal of the mites. The mites were picked up with a camel’s-hair brush and placed in a petri-dish, containing wet filter paper to prevent desiccation, until they were used in the experiment (Dietemann et al. 2013).

Two kinds of behavioral assays, choice and no-choice test, were conducted. Before use in the behavioral assays, all petri-dishes were cleaned with the non-polar solvent pentane. In both experiments, we used dead instead of live bees to avoid any inadvertent influence of different moving activities and cleaning behaviors of the bees on the mite’s host choice behavior. For the choice test (experiment 1) two dead adult bees were placed in a petri-dish (60 mm diameter) at opposite sides facing each other. A solution was prepared consisting of conidial M. anisopliae var. anisopliae BIPESCO 5 spores and the detergent 0.1% TWEEN®20. Using a pipette, one of the bees was inoculated with 30 μl of spore solution at a concentration of 3.8 x 10⁸ CFU/ml, the second bee was inoculated with 30 μl of 0.1% TWEEN®20. An adult female mite in its phoretic phase was placed in the middle of the petri-dish, which was then placed in a climate cabinet (35°C, 70% RH). In regular intervals over 3 h, the position of the mite was categorized as follows: (i) climbed on bee (fungus-treated or control bee), (ii) closer to a given bee (shorter distance between mite and fungus-treated or control bee), and (iii) undecided (mite at equal distance to either bee). Further, mite activity (moving or stationary) was recorded. Observations were made every 10 min in the 1^st h, and every 30 min in the 2^nd and 3^rd h. Fungus-treated and control bees were placed alternately on each side, to minimize any inadvertent positional effects of external factors. The experiment was replicated 60 times; no petri-dish, bee or mite was used more than once.

In the no-choice test (experiment 2), an adult female mite in her phoretic phase was presented one adult bee, either inoculated with fungal spores and the detergent 0.1% TWEEN®20 (fungus treatment) or only treated with the detergent (control treatment), in a petri-dish (60 mm diameter). Fungus-treated bees were randomly placed on the right or the left side of the mite to minimize any inadvertent influence of external factors on the behavior of the mites. For each group, control and fungus treatment, 20 replicates were conducted. Each petri-dish, each bee and each mite were used only once. The mites’ activity and position were recorded over 3 h, at the same time intervals as in the choice test, with the only exception that positions were only categorized as ‘on bee' or ‘not on bee'.

All statistical analyses were carried out using IBM SPSS Statistics 22 (IBM Corp., Armonk, NY, USA). In experiment 1, the choice of the mites for control or fungus-treated bees and their activity (moving/stationary) over time (used as inner subject variable) were analyzed by separate generalized estimation equations (GEE; binomial distribution, logit link function). In both analyses, AR-1 was used as working correlation matrix, as the repeated observations of single mites were correlated over time. Choice decisions were scored when mites were either observed directly on bees or when they were clearly closer to a given bee. Accordingly, GEE for choice was run twice, with and without including the score “closer to a given bee”. The number of mites deciding to climb on a fungus-treated or control bee within 20 min was compared by a binomial test, assuming equal probability (0.5).

In experiment 2, we used generalized estimating equations (GEE; binomial distribution, counts of events) to analyze whether the number of mites sitting on bees over time differed between mites presented control bees and those presented bees treated with M. anisopliae. The activity patterns (on bee, moving or stationary) over time were compared between mites presented control bees and those presented fungus-treated bees by GEE (multinomial distribution, cumulative logit link function). In both analyses, AR-1 was used as working correlation matrix, as the counts and observations of individual mites were correlated over time.

Mites given a choice between nurse bees with fungal spores (fungus treatment) and nurse bees without fungal spores (control treatment) preferentially orientated towards control bees (GEE: Wald ӽ₁² = 6.064, P = 0.014), which did not significantly change over time (Wald ӽ₁² = 0.035, P = 0.85) (Figure 1A). This was also true when excluding the option ‘close to bee' and only scoring mites that had climbed on bees (Wald ӽ₁² = 8.660, P = 0.003). However, in the latter analysis also time had an influence (Wald ӽ₁² = 5.863, P = 0.015) (Figure 1B). The number of mites sitting on treated bees slightly increased over time, while the number of mites sitting on control bees stayed at the same level over time (Figure 1B). 22 out of 60 mites were observed sitting on a control bee at least once in the course of the experiment, in contrast to 13 mites sitting at least once on a fungus-treated bee.

The mites decided more quickly to climb onto control bees than on fungus-treated bees. Decisions for the control bees were made by 18 of 60 mites (30%) within 20 min while decisions to climb onto fungus-treated bees took longer; only 4 of 60 mites (7%) took such a decision within 20 min (binomial test; P = 0.004). 32 of 60 mites (53.33%) were observed at least once on a bee, 28 mites (46.67%) were never observed on a bee. Activity (moving/stationary) of the mites did not change over the course of the experiment (GEE: Wald ӽ₁² = 0.673, P = 0.41) (Figure 2).

The number of mites sitting on bees differed significantly between mites presented control bees and mites presented Metarhizium-treated bees (GEE; Wald ӽ₁² = 8.395, P = 0.004) but progressed similarly over time (Wald ӽ₂² = 1.083, P = 0.58). Across time, more mites climbed on control than Metarhizium-treated bees (Figure 3).

The activity patterns differed between mites presented fungus-treated and control bees (GEE; Wald ӽ₁² = 5.316, P = 0.021) but progressed similarly over time (Wald ӽ₂² = 3.060, P = 0.22). Mites on control bees showed a higher activity, especially from minute 30 to 60, than those on fungus-treated bees (Figure 4).

For highly specialized parasites the chance to locate a suitable host of the correct taxon is often low, even when large numbers of offspring are produced (Gunn and Pitt 2012). Living inside honeybee hives, V. destructor is perfectly adapted to its host and is perpetually surrounded by bees, making finding the right taxon not a problem. Nevertheless, the mites have to choose suitable individuals among a large number of possible hosts and their reproductive success largely depends on this decision. Disrupting the orientation behavior towards their hosts could reduce reproduction of the mites. Therefore, understanding the mechanisms of host choice decisions and examining which substances and chemical compounds influence these decisions are highly relevant topics regarding the future prospects of Varroa control.

Several substances have been tested for their effects on Varroa orientation behavior in similar experiments as done in our study. Investigations of compounds that are produced by the honeybee itself like the Nasanov pheromone (Hoppe and Ritter 1988) and cuticular hydrocarbons (Del Piccolo et al. 2010) increased the knowledge about how the mites recognize their favored host, the nurse bee. The Nasanov pheromone is produced in larger abundance by older forager bees (Picket et al. 1980). This pheromone exerts a repellent effect on the mites, as does its main component geraniol (Hoppe and Ritter 1988). Del Piccolo et al. (2010) found that forager and nurse bees differed in production of (Z)-8-heptadecene, which also acted as repellent. Kraus (1990) discovered the repellent function of 1-octanol, a honeybee alarm pheromone.

Repellent properties of compounds that do not naturally occur inside the hive have been detected in substances derived from plants like various essential oils (Imdorf et al. 1999; Ruffinengo et al. 2005), ethanolic extracts (Damiani et al. 2011) or the oil extracted from neem seeds (González-Gómez et al. 2012). Especially essential oils and their single components have been found to influence the mites’ behavior (Imdorf et al. 1999). Most of them exhibit repellent properties, for example oil of lavender or citral (Kraus 1990), while only a few are attractive to the mites, for example oil of clover or eugenol (Kraus 1990). To successfully disturb the reproduction of V. destructor the repellent properties must be rather strong. Only few substances were efficient in field trials (Ruffinengo et al. 2005). Combined with toxicity, however, repellency, as sub-lethal effect, can be a useful asset for Varroa control (Damiani et al. 2011) and is already exploited by several acaricidal products (Rosenkranz et al. 2009).

In our experiments, the application of the entomo-pathogenic fungus Metarhizium anisopliae had a significant repellent influence on the host selection behavior of Varroa destructor. The choice test revealed a clear preference of the mites for nurse bees that were not in contact with the fungal spores: the number of mites staying on, or in close proximity to, control nurse bees was significantly higher than the number of those staying on or nearby bees treated with fungal spores. At the end of the choice experiment, the number of mites on control bees somewhat decreased because some mites had switched to fungus-treated bees. This was probably due to the fact that dead nurse bees, which do not represent optimal hosts for the mites, were used in the experiment. Thus, after their initial decision, the mites probably dispersed again to determine whether there was an alternative host available. The preference for untreated control bees was also evident in the no-choice experiment. Significantly more mites decided to stay close and/or climb onto control than Metarhizium-treated nurse bees. The first host choice decision was made by all mites within 20 min. During the initial phase of the experiment the mites showed a higher moving activity, resembling the initial phases of host searching behavior, which later declined. However, the moving activity differed between mites offered control and Metarhizium-treated bees. The activity of mites presented control bees increased up to minute 50 while the initial host search of mites presented Metarhizium-treated bees lasted only up to minute 20 and then intensified again during the last hour of the experiment. This activity pattern indicates that the Metarhizium treatment may have masked the bee-derived cues used by the mites to recognize nurse bees rather than acting as repellent inducing active avoidance behavior. It seems that, due to the lacking perception of bee-derived cues, the mites terminated earlier searching for a suitable host, but resumed searching in the last part of the experiment when the masking effect of Metarhizium started to fade. Actively searching for a host is energy demanding and thus non-random (Gunn and Pitt 2012).

The fungal concentration used in the behavioral assay was 1.14*10⁷ CFU/bee. It could be that the observed effects of the fungus on the mites decrease when also the density of spores on the bee cuticle decreases. In addition, bees remove foreign particles from their body by grooming themselves (auto-grooming) or each other (allo-grooming) (Evans and Spivak 2010). Grooming likely reduces the sporal loads of individual bees and could, under field conditions, weaken the effect of the fungus on the mite’s host choice observed in our study.

In conclusion, our study suggests that M. anisopliae as biocontrol agent may, additional to its lethal effect on V. destructor, exert sub-lethal behavioral effects by influencing the mites’ orientation behavior towards, and preference for, its hosts. To examine how strong and persistent these effects are under field conditions, further investigations are needed, especially with regard to the concentration of the fungus (dose-response tests) and its persistence under variable abiotic conditions.

We thank Ms. Sc. Valentino Giacomuzzi and Dr. Hannes Schuler for their help in the laboratory. This study was supported by an intramural fund of the Free University of Bozen-Bolzano. Lara Reinbacher was partially funded by an Erasmus Mundus stipend.

Ahmed A.A., Abd-Elhady H.K. 2013. Efficacy of two fungus-based biopesticides against the honeybee ectoparasitic mite Varroa destructor. Pak. J. Biol. Sci., 16: 819-825. doi:10.3923/pjbs.2013.819.825

Aumeier P., Rosenkranz P., Francke W. 2002. Cuticular volatiles, attractivity of worker larvae and invasion of brood cells by Varroa mites. A comparison of Africanized and European honey bees. Chemoecology, 12: 65-75.

Bull J.C., Ryabov E.V., Prince G., Mead A., Zhang C., Baxter L.A., Pell J.K., Osborne J.L., Chandler D. 2012. A strong immune response in young adult honeybees masks their increased susceptibility to infection compared to older bees. PLOS Pathogens, 8: e1003083. doi:10.1371/journal.ppat.1003083

Chandler D., Sunderland K.D., Ball B.V., Davidson G. 2001. Prospective biological control agents of Varroa destructor n. sp., an important pest of the European Honeybee, Apis mellifera. Biocontrol Sci. Technol., 11: 429-448.

Damiani N., Gende L.B., Maggi M.D., Palacios S., Marcangeli J.A., Eguaras M.J. 2011. Repellent and acaricidal effects of botanical extracts on Varroa destructor. Parasitol. Res. 108: 79-86. doi:10.1007/s00436-010-2043-3

Del Piccolo F., Nazzi F., Vedova G.D., Milani N. 2010. Selection of Apis mellifera workers by the parasitic mite Varroa destructor using host cuticular hydrocarbons. Parasitology, 137: 967-973.

Dietemann V., Nazzi F., Martin S.J., Anderson D.L., Locke B., Delaplane K.S., Wauquiez Q., Tannahill C., Frey E., Ziegelmann B., Rosenkranz P., Ellis J.D. 2013. Standard methods for Varroa research. In: Dietemann V., Ellis J.D. (Eds.). The COLOSS BEEBOOK, Volume II: standard methods for Apis mellifera pest and pathogen research. J. Apic. Res., 52: 1-54.

Evans J.D., Spivak M. 2010. Socialized medicine: Individual and communal disease barriers in honeybees. J. Invertebr. Pathol., 103: S62-S72. doi:10.1016/j.jip.2009.06.019

García-Fernández P., Santiago-Álvarez C., Quesada-Moraga E. 2008. Pathogenicity and thermal biology of mitosporic fungi as potential microbial control agents of Varroa destructor (Acari: Mesostigmata), an ectoparasitic mite of honey bee, Apis mellifera (Hymenoptera: Apidae). Apidologie, 39: 662-673. doi:10.1051/apido:2008049

Genersch E. 2010. Honeybee pathology: current threats to honey bees and beekeeping. Appl. Microbiol. Biotechnol., 87: 87-97. doi:10.1007/s00253-010-2573-8

González-Gómez R., Otero-Colina G., Villanueva-Jiménez J.A., Pena-Valdivia C.B., Santizo-Rincón J.A. 2012. Repellency of the oily extract of neem seeds (Azadirachta indica) against Varroa destructor (Acari: Varroidae). Exp. Appl. Acarol., 56: 261-270. doi:10.1007/s10493-012-9517-1

Gunn A., Pitt S.J. 2012. Parasitology: an integrated approach. Chichester, West Sussex: John Wiley & Sons. pp. 442.

Hoppe H., Ritter W. 1988. The influence of the Nasanov pheromone on the recognition of house bees and foragers by Varroa jacobsoni. Apidologie, 19: 165-172.

Imdorf A., Bogdanov S., Ibánez Ochoa R., Calderone N.W. 1999. Use of essential oils for the control of Varroa jacobsoni Oud. in honey bee colonies. Apidologie, 30: 209-228.

Kanga L.H.B., James R.R., Boucias D.G. 2002. Hirsutella thompsonii and Metarhizium anisopliae as potential microbial control agents of Varroa destructor, a honey bee parasite. J. Invertebr. Pathol., 81: 175-184. doi:10.1016/S0022-2011(02)00177-5

Kanga L.H.B., Adamczyk J., Patt J., Gracia C., Cascino J. 2010. Development of a user-friendly delivery method for the fungus Metarhizium anisopliae to control the ectoparasitic mite Varroa destructor in honey bee, Apis mellifera, colonies. Exp. Appl. Acarol., 52: 327-342. doi:10.1007/s10493-010-9369-5

Kanga L.H.B., Jones W.A., Gracia C. 2007. Efficacy of strips coated with Metarhizium anisopliae for control of Varroa destructor (Acari: Varroidae) in honeybee colonies in Texas and Florida. Exp. Appl. Acarol., 40: 249-258. doi:10.1007/s10493-006-9033-2

Kraus B. 1990. Effects of honey-bee alarm pheromone compounds. Apidologie, 21: 127-134.

Kuenen L.P.S., Calderone N.W. 1997. Transfer of Varroa mites from newly emerged bees: preference for age- and function-specific adult bees. J. Insect Behav., 10: 213-228. doi:10.1007/BF02765554

Meikle W.G., Sammataro D., Neumann P., Pflugfelder J. 2012. Challenges for developing pathogen-based biopesticides against Varroa destructor (Mesostigmata: Varroidae). Apidologie, 43: 501-514. doi:10.1007/s13592-012-0118-0

Neumann P., Carreck N. 2010. Honeybee colony losses. J. Apic. Res., 49: 1-6. doi:10.3896/IBRA.1.49.1.01

Pickett J.A., Williams I.H., Martin A.P., Smith M.C. 1980. Nasanov pheromone of the honey bee, Apis mellifera L. (Hymenoptera: Apidae). Part I. Chemical Characterization. J. Chem. Ecol., 6: 425-434. doi:10.1007/BF01402919

Rodríguez M., Gerding M., France A., Ceballos R. 2009. Evaluation of Metarhizium anisopliae var. anisopliae qu-M845 isolate to control Varroa destructor (Acari: Varroidae) in laboratory and field trials. Chil. J. Agric. Res., 69: 541-547. doi:10.4067/S0718-58392009000400009

Rosenkranz P., Aumeier P., Ziegelmann B. 2010. Biology and control of Varroa destructor. J. Invertebr. Pathol., 103: S96-S119. doi:10.1016/j.jip.2009.07.016

Ruffinengo S., Eguaras M., Floris I., Faverin C., Bailac P., Ponzi M. 2005. LD₅₀ and repellent effects of essential oils from Argentinian wild plant species on Varroa destructor. J. Econ. Entomol., 98: 651-655. doi:10.1603/0022-0493-98.3.651

Shaw K.E., Davidson G., Clark S.J., Ball B.V., Pell J.K., Chandler D., Sunderland K.D. 2002. Laboratory bioassays to assess the pathogenicity of mitosporic fungi to Varroa destructor (Acari: Mesostigmata), an ectoparasitic mite of the honeybee, Apis mellifera. Biol. Control, 24: 266-276. doi:10.1016/S1049-9644(02)00029-4

Van Engelsdorp D., Meixner M.D. 2010. A historical review of managed honey bee populations in Europe and the United States and the factors that may affect them. J. Invertebr. Pathol., 103: 80-95. doi:10.1016/j.jip.2009.06.011