**4. Discussion**

#### *4.1. Survey Data and Overall Losses*

Compared to previous years, honey bee colony winter mortality in Austria 2018/2019 was at an average [26]. The individual loss rates in the survey are not normally distributed (Figure 1C), which is one of the reasons why a GZLM was used for the analysis [15]. More than half of the participants (69.7%) suffered no losses at all or a loss rate lower than 20%. Therefore, it should be feasible for the majority of beekeepers with lower losses to restock the lost colonies by themselves over the next summer season [13]. A multiple winter analysis revealed that, after winters with high losses, more new colonies are created over the summer season and low loss winters result in lower net gain of new colonies in the following season [26].

The variation of loss rates in different states does not seem to be a phenomenon restricted to Austria but can also be observed in the Czech Republic or the USA, where different regions show contrasting loss rates [6,26,27]. In other studies, this was explained by the fact that different regions consist of a divergent composition of landscape which could influence honey bee colony development and winter mortality [20,28]. In addition to this, weather effects have been found to influence colony survival [19]. Weather conditions also influence the hive managemen<sup>t</sup> practices and varroa treatments examined in this article, but this lies beyond the scope of this study.

Different loss rates with respect to the elevation above sea level of the main winter apiary locations could be explained by more than one factor (Figure 3). One of those could be the climatic difference between elevation groups. In a previous study conducted in Austria, colder mean temperatures in September did result in lower overwintering loss rates [19]. Colder temperatures could infer longer winters and, therefore, a shorter breeding season, which results in lower varroa mite pressure. As mentioned before, the change in landscape could also play a role [20]. The latter study showed that semi-natural areas, pastures, and coniferous forests had a positive effect on colony survival in Austria. The lower loss rates with wintering colonies above 600 m elevation could further be explained by fewer honey bee colonies in higher regions and, connected to this, lower spread of viruses or other bee pathogens [29–31].

## *4.2. Queen Management*

Queen problems after winter and winter colony loss rate are influenced by many factors, such as biological causes or beekeeping managemen<sup>t</sup> practices [2,32,33]. Compared to previous years, the special case of winter colony losses due to unsolvable queen problems (living colonies without a laying queen or a drone laying queen) seem stable within 3.6–4.4% [26] and are similar to many other countries participating in COLOSS surveys, with 4–5% [17]. Only 7.8% of the participants experienced more queen problems in their colonies during the 2018 foraging season compared to what they usually observed (Figure 5A). More queen-related problems seemed to go hand in hand with a higher queen-related loss rate after winter (Figure 5B). A study from the USA identified "queen events", i.e., colonies with emergency or supersedure queen cells, as a significant negative factor for colony survival [34]. Possible causes for more queen problems could be led back to neonicotinoids [35] or even package transport of queen bees [36]. These findings and the given results underline the importance of queen bees to colony success [33].

One of the biological influences is likely the age of queen bees going into winter. Several studies found that old queens lower the chance of colony survival [11,12,18,37]. Interestingly, we found significantly lower queen-related losses and winter colony losses (excluding queen-related losses) when the participants exchanged more than one-fourth of their old queens with younger ones in the season before winter (Figure 4C,D). This is in accordance with References [11,18], where each percentage of new queens resulted in a small increase of colony survival. Thus, we conclude that a healthy young and well-mated queen is an important factor for overwintering survival and colony health, probably because a younger queen can build a stronger colony than an older queen [33,38]. Therefore, to replace old queens each year seems to be practical and could lower the winter loss rate [11,37].

#### *4.3. Hive Management Practices*

Beekeeping managemen<sup>t</sup> practices and operational factors are directly influenced by the individual beekeeper [39]. In international analyses of COLOSS surveys, large sample sizes are used to identify beekeeping practices that reduce colony mortality [7]. The power of big data sets (with varying participation from different countries) may obscure regionally important results. We therefore use the opportunity to investigate the efficacy and importance of hive managemen<sup>t</sup> practices for Austria.

We found a lower winter colony loss rate for migratory beekeeping operations compared to non-migrating operations (Figure 6B). The reasons for this could be that migratory beekeepers are more experienced and that migrated colonies have access to better foraging sources [18]. In Austria, colonies are mainly migrated to harvest special honeys, while migrating colonies for paid pollination service is rarely utilized. However, this effect was not consistently found in Austria for some previous investigations [13,40]. In a multi-country analysis of the winter season of 2016/2017 [16], most countries, including Austria, did show no significant difference between migrating and non-migrating beekeepers; however, in the following year, an effect could be found [17]. US beekeepers that migrated their colonies into almond fields experienced a higher total loss rate for the winter 2007/2008 [6]. In contrast to this, two years later, the results indicated significantly lower total losses [27].

Due to the fact that most beekeepers in Austria are hobbyists or sideline beekeepers, the operation size of most participants is relatively small [13]. Hence, most survey participants own a small amount of the total colonies whereas a small number of participants own a significant amount of colonies in Austria (Figure 1A,B). Significant different loss rates among operation sizes were already demonstrated before [12,13,17,41]. Beekeepers with a smaller number of colonies are at a greater risk of losing their colonies (Figure 8). Beekeepers who manage more colonies are likely to have better training and more experience than hobbyist beekeepers, a hypothesis supported by another Austrian study [12].

Over the years of conducting this survey, beekeepers requested to further investigate effects of some operational factors, which they thought might mitigate colony losses. Of these operational factors, only one could be statistically verified, underlining the importance of wax for honey bee health. If participants did purchase wax from outside their own operation, they had a higher loss rate than beekeepers using only their own wax. Bees are exposed to various pesticides, and beeswax, due to its chemical character, is the most contaminated beehive matrix and a bio accumulator of acaricides, fungicides, and insecticides over years [42]. Residues in the wax are frequently found in multiple countries [43,44]. A study in which beeswax foundations were artificially contaminated with pesticide resulted in no negative effect on colony survival [45] but could still influence the survival of bees in addition with other factors. Viruses and spores have also be found in beeswax, and it can therefore pose as a possible viral reservoir [46]. In the cited study, the removal of viral pathogens from old frames did not make a significant difference in the probability of colony survival, and varroa and its varroa-transmitted viruses were considered the greater problem [46]. However, removing pathogens still lowered the chance of infected emerging broods. Commercial wax producers do commonly heat the wax to eliminate American foulbrood spores, which also can remove other pathogens, but this has no effect on pesticides [44]. Thus, the reason why participants who purchased wax from outside their own operation had a higher loss rate is not entirely clear to us. It is possible that participants who bought wax from outside their own operation are beginners who do not possess the equipment or resources for executing their own wax cycle. The result of our study suggests that the factor "buying-in wax" may be a proxy for a certain managemen<sup>t</sup> or degree of professionalism. We sugges<sup>t</sup> further investigation of wax quality and origin as a risk factor for honey bee colony losses. Overall, colonies with new combs are healthier than colonies with old combs [47,48]. Nevertheless, the sourcing and

previous treatment of foreign wax could be an important factor to lowering the amount of residue in colonies near agriculture fields.

We found no difference in winter loss rate between colonies on natural comb or on a foundation of wax (Figure 6H). Replacing old brood frames had no significant influence on colony survival but showed a tendency for lower loss rates with exchange rates above 30% (Figure 7). Comparing this year to the previous years in Austria (2013/14, 2014/15, 2015/16, and 2016/17) two of the years showed a significantly lower loss rate while the other two did not [40]. One possible explanation for this fact could be the varroa treatment strategy "other biotechnical methods" and the lower loss rate associated with this method (Figure A3C), which often includes removal of old brood frames as a side effect.

Certificated organic beekeeping or nonorganic operations had no different probability of honey bee colony winter loss (Figure 6A). The European organic regulation (EC No. 834/2007, 889/2008) is the minimum standard for other organic authorities in the EU. The main restrictions for organic certified beekeepers are the mandatory use of organic certificated sugar/syrup for feeding, the prohibition for the use of synthetic treatments against mites or other pests, and the compulsory use of comb wax from organic beekeeping operations [49]. The location of organic certified apiaries could influence the winter loss rate but is not well defined in the organic regulation and can be differently interpreted by organic control bodies; the same can be said for wax which should be free of contamination by substances not authorized for organic production [49]. As the number of survey participants using synthetic treatments was quite low (Figure 10L) [26], we conclude that feeding and wax quality are the only major differences between conventional and organic beekeeping in our study.

The type of hive (hives fabricated from synthetic materials, insulated hives, or open screened bottom board in winter) had no negative or positive influence on the loss rate (Figure 6D–F). Previous analysis of these factors in Austria came to the same conclusion [40]. A study in Spain examined the temperature and humidity in hives with open screened bottom boards and without, but there seems to be no crucial difference between the hive types regarding colony health [50]. Naturally, the lowest outside temperature in this experiment was around 7 ◦C and is not comparable to cold winter periods in Austria.

Breeding lines with queens bred from varroa tolerant/resistant stocks had no different colony mortality than others (Figure 6C). Such breeding lines are often selected based on the amount of removed damaged brood, for example, via freeze-killed brood assay. In field studies from the US, such lines showed reduced mites in worker brood and adult bees [51]. Beekeepers from the US with varroa resistant bee stocks also experienced lower loss rates than those without [52]. It is probably difficult for the individual beekeeper to check if most of their honey bees have traits favourable for survival, but it shows that more research on this topic is needed.

It should be discussed for future surveys if beekeeping managemen<sup>t</sup> questions, which obviously do not influence the colony survival over winter in multiple years and countries, should be removed to minimize the amount of time spent by the participants to conclude the survey. However, these factors could still have an effect in combination with other factors. On the other hand, most of the questions discussed in this section resulted from participatory processes and reflect the interest of beekeepers.

#### *4.4. Forage as Risk Factor*

Colonies reported to be foraging on maize showed significantly higher winter loss rates compared to colonies that did not (Figure A1B). Maize does not produce nectar but pollen, which can be directly collected by bees; however, it is not a preferred source for honeybees [53,54]. Still, some cases of colonies collecting large amounts of maize pollen have been documented in Austria [55]. A possible explanation for the higher loss rate could be pesticide contaminated pollen, residues in guttation water, or other indirect ways of getting into contact with agriculture chemicals used in maize fields [18,54,56]. Other reasons for our result include poor landscape for foraging with a lot of maize fields nearby, which presumably increases honey bees collecting poor-quality maize pollen or guttation water as well as the lack of nutritive pollen [18,57]. Maize pollen foraging is difficult to assert by participants, but this causality is not needed to explain our results, as this may only describe the quality of maize growing landscapes versus environments without maize. This was also proposed as potential risk in the multi-country analysis from Gray et al. [17].

Late catch crop likewise caused a higher probability of loss (Figure A1D), which at least was not observed in winter 2017/2018 for Austria [17]. Possible reasons could be an extended brood period due to late honey and pollen flows or, again, contact with pesticides in these fields [17].

Honeydew had no influence on colony survival over winter, but participants experiencing a melezitose forage, which often comes in areas with honeydew, had significantly higher loss rates (Figure A1E,F). Melezitose sometimes appears in July or in autumn and has been a familiar problem for beekeepers for a long time. Melezitose fills the brood and honey frames with hard to remove crystalline honey, and bees can invert only a small percentage of the collected melezitose sugar in comparison to sucrose [58]. Furthermore, colonies overwintering on it are often affected of dysentery, which is attributed to the high mineral content [59]. The time when melezitose forage occurs could play a crucial role and is not the same each year. Though melezitose honey is poorly studied, it is a well-recognized problem for beekeepers located in dedicated areas of Austria.

Sunflower and oilseed rape forages are often discussed as risk factors for honey bees. In this study, bees foraging on sunflower could not be linked to raised colony losses (Figure A1C), although this was found in previous years [17,40]. Observed oilseed rape foraging could also not be associated with higher loss rates (Figure A1A). Contrary results are published for Austria and other European countries in winter 2013/2014 [18] and 2017/2018 [17], though some countries experienced the opposite. Oilseed rape provides abundant nectar and pollen for bees, but insecticidal treatments might affect colony development and survival of bees [60–62]. Finally, similar as discussed for maize, diversity of forage due to monoculture fields in such areas could be low [63].

#### *4.5. Varroa Control*

*Varroa destructor* is regarded as the greatest threat to apiculture. Beekeepers need to efficiently treat their colonies or they might face their collapse within 3–4 years [10,12]. The majority of participants stated to monitor varroa infestation levels. Those beekeepers experienced a significantly lower loss rate (Figure 9A). Though this monitoring alone does not decrease varroa levels, such practices can be promoted as good beekeeping practices.

Observations of bees with crippled/deformed wings were associated with a higher loss rate (Figure 9B). Crippled bees can emerge due to cold or viral damage, but they are most often connected to the mite transmitted deformed wing virus (DWV). Therefore, if beekeepers see such bees, this can be interpreted as an alarm signal and counter measures must immediately be taken. The need to monitor DWV load and varroa mite infestation was demonstrated in southern Spain, where high DWV load and high varroa counts resulted in weaker colonies and a higher probability of losing colonies but did not show a significant correlation between DWV symptoms and viral load [64]. This supports the notion that mite-related damage strongly influences the winter loss rate and is one of the crucial factors for colony losses in Austria [12].

The spectrum of varroa control methods applied in Austria is rather limited compared to other countries [26]. To treat colonies, most participants follow the recommendation from the Austrian Agency for Health and Food Safety (AGES) [65] to evaporate formic acid after honey harvest and trickle or sublimate oxalic acid products in winter (Table 2C–F). The benefits of organic acids are a low risk of resistance, a low risk of residues, and a good efficacy against *V. destructor* [10]. Beekeepers in Austria do not commonly apply synthetic acaricides (Figure 10L). Therefore, we pooled the few applications of different types of synthetic acaricides (e.g., Amitraz, Coumaphos, etc.). These agents could lead to wax residues or pollution of honey [10]. The residues in wax could account for the various complications in the bee brood stage [45]. Nevertheless, the synthetic acaricide Amitraz did result in lower loss rates than other varrocide products (synthetic and organic) in a study from the US [52].

In spring (April–May), the most common control method in Austria was drone brood removal (Figure A2A). Participants who performed this method had a significantly lower loss rate the upcoming winter period. If drone brood removal was done in spring and summer (June–October), there is a trend to lose less colonies compared to only removing drone brood in spring or summer (Figure 11). This result was also observed in Austria in other years [40]. In a field study from the US, frequent removal of drone brood resulted in lower mite infestation [66]. However, in comparison to participants who did not remove drone brood at all, there is no significant difference in loss rates. More research on this method under local environmental conditions needs to be done to possibly enhance the positive effect.

In summer (June–October). significantly lower loss rates were observed for participants applying "other biotechnical methods", e.g., trapping comb or complete brood removal to control mites (Figure A3C). These effective methods are often labour intensive [10]. In spring, biotechnical methods (excluding drone brood removal or hyperthermia) are not often used in Austria but could be considered as good practice to fight *V. destructor* with rising temperatures and a prolonged brood period in the future. This was shown in a study in Italy (Reggio Emilia, Po Valley) where the caging of the queen in spring produced no negative impact on honey harvest or brood amount but resulted in a lower mite infestation rate [67]. This could also encourage beekeepers aiming for late honey flows, which they would otherwise miss because the mite population is already too high. Participants applying oxalic acid by trickling in summer experienced high winter losses (Figure A3H). Possible reasons for this could be multiple tricklings and a negative effect on bee health or remaining broods in the colonies which leads to insufficient treatment success [68]. Therefore, oxalic acid trickling is recommended to be performed only once in the broodless period [10,69].

In winter (November–January), oxalic acid represents the dominant choice of treatment with no differences between application by trickling or sublimation (Figure A3A,B). Both treatments have already been evaluated to be very effective in the broodless period, and there is currently no alternative available [10,69].

So far, we solely discussed single treatment methods, but integrated varroa control strategies are comprised of combinations of different treatments. We therefore identified the most common combinations of treatments (Table 2, Figure 12). This allows, for example, further examination of the shown negative effect of oxalic acid—trickling in summer—in combination with other treatments. We found high loss rates for the combination (L) formic acid—short term in summer—and oxalic acid—trickling in summer and winter (Table 2L) and the combination (P) only oxalic acid—trickling in summer and winter (Table 2P). We conclude that trickling in summer is either not effective or causes negative effects on bee health. In combination with formic acid—long term in summer (Table 2I), this resulted in average losses. This might be due to the positive impact of formic acid—long-term evaporation.

Two frequently applied combinations were long-term evaporation of formic acid in summer and oxalic acid trickling or sublimation in winter (Table 2A,C), which both resulted in average colony losses. Formic acid is the only allowed organic acid in Austria which is effective against phoretic and

reproductive mites [10]. Our results hence support the current recommendation from AGES to use formic acid in summer and oxalic acid in winter [65]. Long-term evaporation should be preferred, as short term may not be as efficient in mite reduction. Short-term evaporation in combination with oxalic acid in winter (Figure 12B,D) and, additionally, the double application of formic acid—long and short term in summer—with oxalic acid—trickling in winter (Figure 12E)—had higher loss rates than the single application of formic acid—long term in summer—and oxalic acid—trickling in winter (Figure 12A). We assume that either the application of both formic acid variants could lead to brood damage or the double application is an emergency measure because of high varroa counts.

The combination of biotechnical methods and short-term formic acid evaporation in summer plus oxalic acid—trickling in winter (Table 2S)—showed one of the lowest loss rates. Care must be taken in interpreting these results, as only 15 participants exercised this combination. Further, the biotechnical methods asked for in our survey could include a wide range of various procedures. To learn more about efficacy of different biotechnical methods, further studies should specify which biotechnical methods are used. Though this would lower sample size, discrepant results with biotechnical methods in combination with other methods could be better understood.

The combination of oxalic acid—sublimation in summer and winter (Table 2F) resulted in low loss rates. This method seems to be efficient to treat mite infestation and sublimation in comparison to trickling in summer is favourable. The sublimation of oxalic acid does not reduce reproductive mites [10,69]. Therefore, the colony must be broodless or sublimation is repeated multiple times in summer. This combination could offer a potential to represent a reliable method but would require a more in depth analysis with field studies on the amount of oxalic acid, on frequency of application, and the different sublimation equipment.

There was only one frequent combination with thymol. Preparations with the essential oil (in summer) and oxalic acid—trickling in summer and winter (Table 2T)—resulted in high loss rates but with a wide confidence interval. In an Australian study on thymol and the beneficial effect on hygienic behaviour, researchers found inconsistent results [70]. They proposed that different factors play a role for this differential outcome, such as environmental or genetic differences. In the loss rates presented here for Austria, thymol was combined with oxalic acid—trickling in summer. As we already discussed that oxalic acid trickling in summer should not be performed, we cannot recommend this treatment combination. This is no general recommendation against thymol usage, as we are lacking reliable data. In this study, we identified some unexpected combinations of varroa control methods applied in Austria. We recommend further research to better understand the motivation of beekeepers behind those and direct them to reasonable and effective treatment plans.
