**4. Discussion**

The geographical distribution of the honeybee death events (Figure 1) coincides with the areas in Italy in which intensive agriculture is mainly practiced (such as apple and citrus orchards and vineyards mainly other than maize cultivations).

In honeybees (Tables 1 and 2), the most frequently detected active substances were insecticides with tau-fluvalinate having the highest prevalence. Tau-fluvalinate is a pyrethroid insecticide authorized both as a PPP and for the control of *Varroa* mite infestation of honeybees in Italy. Miticides have already been found by di fferent studies [2,18,19] to be the most frequent residues in honeybee samples around Europe. The pyrethroid permethrin, the second most frequently identified active ingredient, also has the highest detected concentration. It is a contact insecticide which has not been approved for use in the EU as a PPP, due particularly to its acute toxicity to aquatic organisms. Chlorpyrifos was the third most commonly determined pesticide in honeybees and, being an active ingredient highly toxic to bees, it could represent an important factor a ffecting colony health. Chlorpyrifos has already been identified as one of the most commonly detected insecticides in bees [20–22]. Neonicotinoids, mainly imidacloprid, were also frequently identified. In Italy, the use of three neonicotinoids, namely imidacloprid, clothianidin, thiamethoxam, and fipronil, was restricted in 2008 due to evidence of their negative e ffects on honeybee health. In 2013 the EU definitively banned the use of these active ingredients for seed treatment, soil application and foliar treatment of plants and cereals attractive to bees (but use in greenhouses is allowed) [23,24]. However, fipronil and all three neonicotinoids (including the restricted ones) screened in our study were detected in our samples. Therefore, despite the current limitations on the use of PPPs containing these active ingredients, according to the present monitoring results, honeybees are still exposed to potentially harmful levels of these pesticides, as already observed in previous studies [25–27]. Fungicides were also often detected (39.3%) with a wide variety of active ingredients, the most frequently found being penconazole and pyrimethanil. Although there are no restrictions on the use of fungicides on crops during blooming, various studies have shown that the impact of fungicides on honeybee health can be harmful, both due to their direct negative e ffects on honeybee health [28,29], and through a synergistic action between fungicides and other types of pesticide [30–32]. Our results partially agree with those obtained in a previous study carried out by our laboratory [33], which assessed the presence of pesticides and viruses in dead honeybees following mortality incidents in northeastern Italy in 2014. Compared to this study, in which imidacloprid was the most frequently detected active ingredient, there has now been a reduction in the presence of neonicotinoids, probably due to limitations imposed on their use by the European Commission [23,24]. Tau-fluvalinate and chlorpyrifos were instead confirmed to be among the most frequently identified active ingredients.

With the exception of 2018, the year in which we observed close correspondence between honeybees positive to pesticides and bee kill incident reports, percentage positivity stood at around 44% in the other monitoring years: a value probably influenced by various factors, as the speed of reporting and the subsequent sampling intervention. The concentration of pesticides in dead honeybees can rapidly decrease within just hours of the poisoning event and, if not properly stored at −20 ◦C, samples can reach a level close to environmental residue before being analyzed in the laboratory [20]. The analysis results may be also a ffected by the severity of the poisoning event (in terms of the active ingredients involved, their concentration, method of administration) and the presence of other bee parasites or stressors (such as viruses and *Varroa* mite) that can contribute to the weakening of colonies and predisposition to mortality events, even with sublethal concentrations of pesticides [28,34,35].

The honeybee is certainly the most important matrix to be analyzed in case of honeybee mortality incidents, as residues detected in honeybees reflect their exposure both to direct contact with PPPs, biocides, or even veterinary drugs, and to the consumption of contaminated nectar and pollen. However, the analysis of other matrices related to the same incident can help us to better understand the mortality event. For example, bee bread can supply useful data on any PPPs application occurring in the areas surrounding the beehive, while beeswax comb can provide information on exposure over a period of time. Unlike other beehive products, beeswax can remain in the hive for many years, thus resulting in an accumulation of various non-polar active ingredients applied in beekeeping and agriculture [19,35]. In the present study, most other analyzed matrices (67%) were positive for at least one active ingredient (Table 3), and again the acaricide tau-fluvalinate represented the most commonly found active substance, followed by the insecticide methiocarb and the synergist piperonyl butoxide. The most represented matrix was beeswax, with an average of 72% (70 out of a total of 97 samples) of the samples proving to be contaminated with pesticides, mainly tau-fluvalinate. Bee bread showed 74% positivity (25/34), and in this case too, the most commonly detected active ingredient was tau-fluvalinate. Being stored inside the beehive, bee bread can be a ffected by both beekeeper and agricultural activity. For these reasons, however, in the case of a honeybee killing event, we cannot rely on toxicological information provided by beeswax. Vegetable matrices (most frequently leaves, corn seedlings, maize) were contaminated in 70% of cases (12/17), with the widest variety of active ingredients (27), despite being by far the least numerous matrix received. Honey was also received as a matrix related to honeybee incidents and proved to be contaminated with pesticides in only 20% of cases; but in three samples the detected pesticide concentration exceeded the limits imposed by the EU (methiocarb 0.05 and 0.7 mg/kg and tau-fluvalinate 0.05 mg/kg) [36]. These results should draw attention to the fact that mortality events are harmful to honeybees, but consumers' health should also be considered. The risk of contamination of edible beehive products, as honey and pollen, but also beeswax, which can then be reused and lead to the transfer of contaminants to honey, cannot be ruled out [37].

It is also worth mentioning the detection of some active ingredients that are no longer authorized but in the past were present in both PPPs and veterinary medicinal products. Authorized active ingredients used against varroosis [38] were among the main sources of honeybee and hive matrices contamination, but so were old apicultural and agricultural acaricides that are now banned, such as bromopropylate (both), chlorfenvinphos, and rotenone (agricultural). The pyrethroid insecticide permethrin, which is highly toxic to honeybees and authorized as a biocide [39], was frequently detected, even in high concentrations, in both honeybees and other matrices. The same applies to the potent multi-purpose pyrethroid insecticide tetramethrin, registered in 1968 and often used to control insects presenting risks to public health, but which is highly toxic for honeybees and has never been authorized for use in crop protection. The insecticide thiodicarb was detected in a few honeybee samples. This insecticide and molluscicide is used to control Lepidoptera, Coleoptera, slugs, other pests of fruit, vegetables, and many other crops, with moderate or high toxicity to honeybees, depending on whether the administration is contact or oral [36,40].
