Here the authors investigate a critical issue in honey bee health, assessing the impact of the Varroa destructor mite on the viral landscape of the western honey bee, Apis mellifera.

While it is already well established that varroa is a vector of Deformed wing virus - the main virus associated with the global spread of mites - this study also identifies trends in other common honey bee viruses that are not known to be directly vectored by varroa, such as the widespread Black queen cell virus and Lake sinai viruses. The authors propose the idea of such viruses being ‘mite hitchhikers’, a useful concept that nicely summarises the conclusions of the study.

The authors collate evidence from four separate locations to support the idea that when varroa is present, virus diversity and level increases in bees, making use of historical samples from 2010-2013 for these analyses. While previous studies have found similar results, these have all been either focussed on a single or few viruses (eg. DWV), or centred in a particular geographic region. The strength of this study is nicely summarised at line 117: “Sampling from four independent regions of the world enabled us to circumvent any geographic idiosyncrasy so that we could detect general patterns in the virome response to varroa invasion.”. Varroa is now well established across much of the globe, however there are still locations (eg. Australia) which are new to the parasite, therefore the findings of this paper continue to hold relevance and impact.

This is an excellent paper in general, well written and interesting, with the results clearly explained. Some specific minor comments are listed below.

Line 39, 41, 43, and elsewhere: Titre has been used frequently to describe virus levels, while load has been used at some places (eg. line 228). It was my understanding that a virus ‘titre’ refers to the lowest concentration of a virus that infects cells, and often the method of assessing virus titre is using a plaque assay in cell culture. Instead, a more appropriate term for qPCR data could be ‘level’ or ‘load’. You might consider replacing titre (eg. to load for viruses). In addition, when used to describe reference genes at line 166 “The absolute titres of the internal reference genes” – level would be more suitable here.

Line 42: Consider removing ‘unsustainable’ – I don’t think it adds much. If you retain it, probably should say ‘unsustainably high’

Line 57: ‘referred as’ should be ‘referred to as’

Line 212: The result describing the prevalence of DWV-B vs DWV-A: does the prevalence of DWV-A differ in the regions where DWV-B was also surveyed? Maybe a better way to ask – what is the co-occurrence of DWV-A and DWV-B like, in the UK (as the single sample in Canada is probably not informative)?

Line 229: I am confused by the SEM values in brackets here, eg. the BQCV SEM is as large as the mean (both are 9x10^9), is this a typographical error? It does not appear to match the very small SEM bars in the associated figure.

Figure 3: The scatterplots and PCA analysis is particularly compelling, these highlight the data very nicely. Would it be possible for the authors to colour code the dots with location? While I think it is important that the use of four sites can highlight global trends rather than local quirks, it is still of value for the reader to see whether any regional clustering illuminates trends.

Line 289: “The recently emerging variant DWV-B was seldomly detected in our historical samples.” – two points here, I suggest removing ‘recently’, as emerging implies that it is recent already. Second point is that was DWV-B frequency lower becaue it was only examined in two locations? Perhaps the sentence could be rephrased: “In the two locations where the emerging variant DWV-B was surveyed, it was seldomly detected, relative to DWV-A”.

Supplementary figure 1: If the journal allows, I would suggest elevating this figure to the main text, it helps to contextualise the breadth of sampling in the different locations.

Supplementary table 1: This study relies on the ability of qPCR primers to detect viruses, despite the potential for strain nucleotide variation. Is there evidence of polymorphisms in primer binding sites, based on known viral variants, around your amplified regions?