**4. Discussion**

Varroa mites, acting simultaneously with other stressors, are known to be a main cause of honey bee colony weakening and collapsing [5,29]. The *V. destructor* mite population requires regular control, correctly and in a timely manner, of honey bee colony management. Five different VMPs, only authorized in Croatia in 2014, were tested at five apiaries to test their performance in different field conditions because control of varroosis by using VMPs is widely acknowledged to be an essential part of beehive managemen<sup>t</sup> [30].

The 2014 active beekeeping season was generally very weak in different aspects. There were more days with rain, very low average year air temperatures, along with a mild winter, and consequently, a very early honey bee brood development was noticed, even in January. As one replication cycle of *V. destructor* mites is about 13 days, multiple replications of mites can be expected to have occurred [17]. Weather circumstances were also very favorable for a fast-increasing growth trend of *V. destructor* mite populations in honey bee colonies with early reared brood [31]. Additionally, lack of natural food in the environment effected a decreasing of immunological status of the honey bee colonies and induced stress caused by hunger. All these detrimental impacts of environmental factors, high *V. destructor* mite infestations, and weak main pastures ended very early, so in most regions, the first summer varroacidal treatment was essential.

A different number of *V. destructor* mite drops was detected during the pretreatment period, which was probably a consequence of a different level of parasitic colony invasions at different EAs. Meeting the demand of no varroacidal treatments that year, before the start of the described study means that they were treated the last time in the winter preceding the experimental period. This may have implicated a situation where establishing new summer generations in all honey bee colonies coincided with a high level of *V. destructor* mite infestation, which may have damaged colonies in each experimental group. Similar findings were published before [32].

Because of extremely low amounts of stored honey in hives, at EA2 and EA5, the honey extraction was not done for the whole year.

Treatments against varroa mites started at di fferent dates for each EA because the blooming of plants that make up the main bee pastures finished at di fferent times. The best relative varroacidal efficacy was seemingly achieved at EA1 (59.24%), but if an analysis of the survival of treated colonies is included, then it is clear that 30% of them collapsed before the next active beekeeping season. Most of the lost honey bee colonies were treated with VMPs based on thymol as active ingredients. Here it must be stressed that first summer treatments on EA1 started only towards the end of July, which is relatively late but still in accordance with advised varroosis control schedules in Croatia.

Although relative average treatment e fficacy at EA2 (47.31%), EA3 (36.75%), and EA4 (48.33%) was pretty low and below expectations, all honey bee colonies survived and successfully overwintered (except one honey bee colony at EA4, treated with a thymol-based VMP).

Comparing the relative e fficacy of the used varroacides with di fferent active ingredients (Figure 6a), the best e ffect was achieved for honey bee colonies from an A experimental group, and lowest for colonies from a B experimental group, where statistically significant di fference was a ffirmed (*p* < 0.05; F = 6.93). Although the organic acaricides have certain advantages after repeated use, their e fficacy may be inconsistent and more variable compared with synthetic acaricide formulations [24,33–36], which is also confirmed with our results.

Changes in honey bee colonies' strength trends were di fferent between apiary locations, but within the expected ranges under the study and environmental conditions, as well as beekeeping practices (EA1, EA2, EA3, and EA4).

At location EA5 at the beginning of active beekeeping season, honey bee colonies were very weak with low numbers of adult bees (Figure 2). Then, in the middle of the active season, there was an opposite situation: a high number of adult bees without brood or very few comb cells were sealed with development stages of bees. At the same time there was non food in nature, and honey bees ate away almost all food storages in their hives. After emergency varroacidal treatment, which produced higher e fficacy compared to earlier treatments because larger proportion of phoretic mites being on adult bees [37], and transport at island Vis (EA5), they came in with much better environmental circumstances with plenty of natural food. In the next few weeks, the strength of colonies increased quickly, but because of serious damages, bees go<sup>t</sup> high varroa infestation and insu fficient e fficacy of VMPs treatment during the summer treatment (except A group—47.52%), most of the colonies from other treated groups died before the next spring (63.4%). The e fficacy of the VMP treatments was under expectations, probably due to fast reproduction of the surviving mites, but also because of possible reinfestations during experiments [35].
