First detection of bee viruses in hoverfly (syrphid) pollinators

Global declines of insect pollinators jeopardize the delivery of pollination services in both agricultural and natural ecosystems. The importance of infectious diseases has been documented in honeybees, but there is little information on the extent to which these diseases are shared with other pollinator orders. Here, we establish for the first time the presence of three important bee viruses in hoverfly pollinators (Diptera: Syrphidae): black queen cell virus (BQCV), sacbrood virus (SBV) and deformed wing virus strain B (DWV-B). These viruses were detected in two Eristalis species, which are behavioural and morphological bee mimics and share a foraging niche with honeybees. Nucleotide sequences of viruses isolated from the Eristalis species and Apis mellifera were up to 99 and 100% identical for the two viruses, suggesting that these pathogens are being shared freely between bees and hoverflies. Interestingly, while replicative intermediates (negative strand virus) were not detected in the hoverflies, viral titres of SBV were similar to those found in A. mellifera. These results suggest that syrphid pollinators may play an important but previously unexplored role in pollinator disease dynamics.


Introduction
Emerging infectious diseases (EIDs) are a global problem for biodiversity and human health [1]. Their occurrence has been associated with anthropogenic pressures, such as the global transport of managed animals and plants [1,2], which introduce diseases into novel hosts and alter natural disease dynamics [3]. EIDs can be particularly problematic for small and declining populations where 'spillover' from large managed populations can occur repeatedly, potentially resulting in the eventual extinction of the native population [3].
The positive-stranded RNA viruses found in managed honeybees (Apis mellifera and Apis ceranae) represent a key complex of potential EIDs that are shared with other wild bee pollinators [4,5]. These viruses have been implicated in the declines of wild bee populations, leading to concern for the economic and ecological value of associated ecosystem services [6,7]. Viruses originally thought to be honeybee-specific are now known to occur in and infect a wide range of wild bee species [8]. Interspecific transfer of these viruses, and other parasites, is thought to occur when individuals forage at the same flowers [4,9,10]. While many other taxa commonly share floral resources with bees, information on the presence of these diseases in taxa other than bees is poor [11]. To understand and manage disease pressure on pollinator populations, the role played by other taxa of flower visitors in the transmission of 'bee' viruses needs to be evaluated.
Hoverflies (Diptera: Syrphidae) regularly share flowers with bees and are important providers of pollination services [12,13]. Here, we investigate whether four abundant taxa of hoverflies act as hosts or potential vectors for six common bee viruses.  The presence or the absence of six common bee viruses (acute bee paralysis virus, ABPV; black queen cell virus, BQCV; deformed wing virus strain A, DWV-A, and strain B, DWV-B; slow bee paralysis virus, SBPV; sacbrood virus, SBV) was determined by RT-PCR (electronic supplementary material, primers in table S1). Positive samples identified by the amplification of the correctsized product were verified by amplification in an independent RT-PCR reaction and subsequent Sanger sequencing (by Source Bioscience, Cambridge) to confirm they mapped to the virus of interest in the National Center for Biotechnology Information (NCBI) database. All amplicons of the correct size showed high sequence identity to the virus of interest (electronic supplementary material, table S2). All sequences are available at NCBI Genbank with the accession numbers MG737448-MG737473.

Material and methods
Viral titres of SBV and BQCV were quantified using qRT-PCR (see electronic supplementary material, primers in Table S1). To detect the negative strand of SBV and BQCV, which is indicative of virus replication, the protocol of de Miranda et al. ( [14]; section 10.2.8.1) was followed using Superscript III (Invitrogen). A combined exonuclease and restriction digest was carried out on tagged cDNA to reduce the chance of false-positives and non-specific priming during PCR (electronic supplementary material).

(c) Statistical analyses
Analyses were carried out in R v. 3.4.1 [15]. Viral titres were compared between Apis and hoverflies using Welch's t-tests following log-transformation. To compare virus incidence among species, we used x 2 -tests in the coin package [16]. An approximated null distribution using 9999 replicate Monte Carlo simulations was used to account for zero/low counts.

Results (a) Detection of bee viruses by RT-PCR
Viruses were detected in both A. mellifera and hoverflies (table 1 and figure 1). When considering positive results verified by independent amplification and sequencing (electronic supplementary material results), the most commonly detected virus in our samples was BQCV. BQCV was detected significantly more frequently in A. mellifera samples (13/20 samples) than in the hoverfly samples, Er. tenax (2/20) and Er. arbustorum (2/20; approximate test for differences among species: x 2 ¼ 42.2, p , 0.001). BQCV was not detected in P. albimanus or Ep. balteatus, but there was no evidence that the proportion of samples with BQCV differed significantly  When assaying for the DWV complex, results from hoverfly samples were highly inconsistent for most sets of primers, and we were unable to verify the presence of DWV-A in our samples using two different primer sets (see electronic supplementary material; figure S1). DWV-B results were also difficult to verify, so detection of this virus in only one hoverfly sample may underestimate its true incidence.

(b) Variation in BQCV and SBV sequences
Analysis of a 345 bp section of SBV capsid gene from A. mellifera and hoverfly sequences indicated that the strains of virus present in these individuals were highly similar (ranging from 95 to 99% nucleotide identity between hoverfly sequences and A. mellifera sequences; electronic supplementary material, table S3). Similarly, analysis of a 696 bp section of BQCV RNA-dependent RNA polymerase gene from A. mellifera and hoverfly sequences indicated high virus similarity (87-100% nucleotide identity of hoverfly sequences to A. mellifera sequences; electronic supplementary material, table S4).

(d) Evidence of replication of BQCV and SBV
Negative strand-specific RT-PCR of BQCV-and SBV-positive samples indicated possible replication of BQCV in 2/13 A. mellifera workers and replication of SBV in 3/6 A. mellifera workers. Replication intermediates of SBV or BQCV were not detected in any hoverfly samples (five and four individuals, respectively), suggesting lack of viral replication.

Discussion
Our study is the first to detect bee viruses in hoverfly pollinators. By contrast, an earlier study found no evidence for the presence of DWV in four hoverfly species [17]. Our results add further evidence that viruses traditionally considered 'bee' diseases are not restricted to Hymenoptera [11], and highlight the importance of understanding the role of nonbee pollinators in pathogen transmission. Interestingly, bee viruses were only detected in hoverfly species in the genus Eristalis, which mimic A. mellifera in both morphology and behaviour [18]. This presumed foraging niche overlap between Eristalis and A. mellifera may have increased the probability of exposure to bee pathogens via shared floral resources. By contrast, Ep. balteatus and P. albimanus are both generalist floral visitors that do not mimic bees [19].
Only viruses that were detected in co-foraging honeybees were detected in our hoverfly samples, and these were always at higher or equal prevalence in honeybees. Combined with high sequence similarity between isolates, this is consistent with spillover of these viruses into hoverflies, as has previously been suggested for bumblebees [4,5]. However, the detection of bee viruses in a sample does not imply infection and could be explained by vectoring. There was no evidence of viral replication for either BQCV or SBV in the hoverflies. But, given the low titres detected and subsequent likelihood of false negatives, we cannot rule out that these were true infections. While BQCV viral titres were much higher in honeybees, SBV titres in Eristalis were similar to those in honeybees, suggesting that Eristalis may potentially be acting as a host to SBV.
Regardless of whether hoverflies are active hosts or passive vectors of the pathogens [10], our results suggest that

Apis mellifera
Eristalis spp. Apis mellifera Eristalis spp. rsbl.royalsocietypublishing.org Biol. Lett. 14: 20180001 hoverfly flower visitors may play an important but previously unexplored role in pollinator disease networks. As abundant flower visitors sharing resources with both honeybees and wild bees, hoverflies may be capable of moving these pathogens around the landscape, facilitating transmission between susceptible bee species. Eristalis tenax is capable of extensive, long-distance migration [20], suggesting the potential for supra-national networks of pathogen transmission among pollinators. This is particularly concerning for emerging pathogens such as DWV-B, a recently discovered, highly virulent strain of the deformed wing virus [21]. Further work is now needed to investigate the role of hoverflies as both hosts and vectors for a wider range of pathogens, and the extent to which use of shared floral resources leads to spillover and transmission among species.
Ethics. Work complied with local ethical requirements. Data accessibility. Supporting datasets can be found in: https://doi.org/