Tuesday, 5 January 2016

From the past: The EID paper referees comments from 30/1/2015. For the future: monitoring birds in Russia.

Reviewer: 1
Comments to the Author
In their manuscript, ‘The European and Japanese outbreaks of H5N8 derive from a single source population that has been dispersed along the long distance migratory bird migratory flyways’, Dalby and Iqbal use Bayesian coalsescent methods to infer ancestral of previously detected H5N8 subtype influenza A viruses and estimate time since divergence for isolates derived from recent Eurasian outbreaks.  Furthermore, the authors provide generalized and anecdotal informal on bird migration patterns in Eurasia to gain inference on possible origins and dispersal patterns.

First, I would like to applaud the authors for investigating a topic of great interest and importance to both human and animal population health.  Few studies combine genetic data with information on bird migration which can be a useful methodology for understanding the global dispersal of avian pathogens. I personally believe that such studies are relatively rare on account of the difficulty of combining disparate types of data and in obtaining relevant information for wild birds at locations along migratory pathways.  Regrettably, this is where I think this particular investigation falls short.

Aside from a few potential (minor?) issues (see specific comments below) I do not have any problems with genetic analyses included in this manuscript and the conclusions drawn therefrom per se; however, I do think that the authors ‘epidemiological data’ on migratory birds falls short of supporting conclusions.  For example, in the abstract alone, I would argue that the following claims are not sufficiently supported by empirical data: ‘traced to a single source population, which has been spread by migratory birds’, we can show when and where the outbreak originated’, ‘This population was located in the Siberian summer breeding grounds of long-range migratory birds’.  Because recent outbreaks of H5N8 influenza A viruses also have occurred in poultry throughout Eurasia, I feel that one could use the same genetic analyses conducted by the authors and provide similar generalized/anecdotal information regarding Eurasian poultry trade patterns to reach the conclusion that this virus has been dispersed through bird trade.  That is, the authors fail to provide any convincing evidence to demonstrate that wild birds have been solely responsible for viral dispersal as implied.

Unfortunately, I feel as though I cannot be more supportive of the authors’ current submission to Emerging Infectious Diseases at this time.  Given what I believe to the critical flaw in the manuscript as written, I might suggest that the authors shorten their submission by focusing on ancestry of H5N8 viruses and time of divergence.  By formulating a short communication (i.e. Dispatch) on this more focused topic, I think that a few speculative sentence could be included in the discussion in support of the authors’ thesis, that wild birds are dispersing H5N8 viruses throughout Eurasia.  In hopes that the authors will pursue a revision in the future, I’ve appended numerous specific comments below which I hope will ultimately prove useful towards this end.

Specific comments:

Lines 28-29: By ‘cases’, do you mean ‘detections’?  There have certainly been undetected cases, no?
Line 43: By ‘the virus’, which specific strain are you referring to?  Different strains of HP H5N8 probably have differences in pathogenicity and host adaptation.
Lines 46-47: (‘Ducks and…’) I’m not sure that this is supported as written.  If you are referring to the laboratory study conducted by Kang et al., specify (e.g. ‘in a laboratory challenge study…’) and provide citation.
Line 51: Have antibodies specific to this strain been demonstrated?  If so, I presume this was through experimental challenge?  Clarify.
General comment: Considerable text is included in the introduction presenting information from very distantly related viruses of the same serotype (H5N8).  Would the introduction be better focused on the evolution of the reassortant HP H5N8 viruses currently causing poultry outbreaks?

Materials and methods
Line 62: I don’t believe that ‘flu resource’ is correct nomenclature.  Change throughout.
Lines 74-75: By ‘different reassortment events’ do you mean ‘different ancestral lineages’?
Lines 79-80: Please justify why this model and molecular clock assumption.
Lines 82-83: How long were runs?  Burn-in period?
Lines 88-92: Methodology here is insufficient for evaluating contributions of wild birds in viral dispersal.

Results and discussion
Lines 96-97: Is support (i.e. posterior probability values) presented on trees?  If not, I cannot assess support for the topology presented.
Line 101: It is a bit confusing as to results for which dataset you are referring here.  Some clarification for the reader in this section would be helpful.
Line 102: Add ‘gene segment’ after ‘hemagglutinin’.
Line 115: ‘in Korea in Korea’;
Lines 199-121: This scenario is plausible but is not demonstrated by results.
Lines 123-124: Couldn’t one argue that wide dispersal is reflective of the extent of poultry trade in Eurasia?
Lines 126-128: Please provide genus/species for bird species.
Line 128: By ‘carrying the virus’, do you mean ‘infected with H5N8 viruses’?
Lines 131-132: Please provide genus/species for bird species.
Lines 133-134: Prevalence for these species has been reported as being much more variable than implied here.  Also, I’m not sure birds ‘are funneled’; rather, birds ‘congregate’.
Lines 138: Add ‘of Eurasia’ after ‘flyways’ to reflect recent detection in North America.
Lines 140-143: This anecdotal information provides weak support for your thesis.
Line 145: Is this referring to H5N1?
Lines 150-153: Speculative statements here need appropriate caveats.
Lines 157-159: I tend to disagree with this statement.  Estimated evolutionary divergence does provide information on host, location, or ‘events’.
Lines 167-169: (‘However the small…’) This statement is not supported.
Line 175: Define ‘vigilance’.  By ‘avian flu’ do you mean ‘HP H5N8’?
Line 185: Define ‘vigilance’.  Are you referring to surveillance?
Lines 185-187: How will this help if infections are asymptomatic in wild birds?  If the thesis put forward re. potential for transmission to humans in the prior paragraph holds true, might your recommendation here actually increase the probability for bird to human transmission?

Reviewer: 2
Comments to the Author
The manuscript by Dalby and Iqbal reports a phylogenetic study performed on sequences of influenza A virus (IAV), H5N8 subtype. The analysis is timely and has the potential to provide significant information to the understanding of the H5N8 IAV outbreak. The manuscript, unfortunately, suffers from a lack of clarity in both the methodology and the presentation of the results. Below is a list of both minor and major issues that the authors may want to address:

- The article has been submitted as a “full research manuscript” but is very short and could have rather been formatted as a “dispatch”.
- The use of “seroptype” is not appropriate; the use of “subtype” for the HA and NA gene segments is widely used and I recommend the author to do so in their manuscript to avoid confusion for the readers.
- line 28: “was identified in a wild bird”. More information on the bird species is needed as well as the context in which it has been found (active/passive surveillance, outbreaks in poultry in the same area, with other H5 virus subtypes, etc.).
- line 47: “mallards are often asymptomatic but can still be carriers of the virus”: What virus: H5 in general ? HP/LP ? Please be more specific.
- line 57: “this study identify the source of the outbreaks”. Define “source” and the specific objectives of the study. It is somewhat vague as the phylogenetic analyses provide information on the relatedness and evolutionary history of isolated viruses, but not on the exact identification of the donor of the virus circulating in Europe in November/December. Indeed, the lack of sequence available between the spring/summer and the winter of 2014 precludes conclusion on the exact origin of the virus.
- line 62: “all of the available H5N8 sequences were downloaded”. Provide details on the number of downloaded sequences per segment, whether if they were nucleotide or protein sequences, etc. Also, why not including H5 sequences of non-H5N8 virus subtypes ? This would have given more power to the analyses and strengthened the conclusions regarding the global evolutionary history and origin of the H5N8 virus. Since IAV gene segments have an – almost – independent evolution, it would have been more appropriate to not restrict the data to segment that belong to the H5N8 subtype only, in particular for the internal gene segments. I understand that it may not has been the initial objective of the study but I believe that including only H5N8 sequences gives biased results and a limited understanding of the global evolutionary history of the virus genes (reassortment events, etc.).
- line 69 “no editing at the 3' end”. Why that ? Why not trimming the sequences to the stop codon ?
- lines 71-75: I'm confused by this part. Based on the results it looks like it had affected the estimates of the TMRCA. Sequence selection procedure overall needs more clarity.
- lines 77-80: Why use a strict molecular clock and a nucleotide-based substitution model while it has been shown that more appropriate models exits for IAV, in particular for Bayesian analyses performed with BEAST ? (Shapiro et al. Mol Biol Evol. 2006;23:7–9; Bahl et al. Virology 2009;390:289-297; etc.). Where tips dates coded by year only, or years/month/day ? Was that the case for all sequences ? How did you dealt with missing information ?
- lines 82-84: Where several runs combined ? What were the chain lengths and sampling frequency ?
- line 96: “produces a consistent gene tree”: How that is consistent ? Also, there is a major information missing on the phylogenetic trees: the posterior probabilities. It is overall difficult to evaluate the methodology but since no posterior are indicated on the trees, it also makes the interpretation of the results somewhat complicated.
- line 103 and throughout the manuscript: “outbreaks diverged between 1.58 and 5.53 months ago”. I suggest to use exact dates (e.g. 2014.XX) rather than some time in the paste, given that the time reference used is not clearly stated (December 2014 ?).
- line 119: “this result indicate”. This statement requires additional support. The fact that two events occurred at the same time does not supports that the two events were related... I agree that it is likely that these particular two events were linked, but it would be welcomed to have a more comprehensive discussion rather than a statement that is not fully supported by the data and analyses.
- line 124: “that migratory birds”. This is a very general statement and it is very unlikely that migratory birds, in general, were the source. Please be more specific (species involved or not, etc.). Overall, the manuscript critically lacks strong arguments and discussion of the ornithological aspects.
- line 133: “Mallard... high prevalence”. Prevalence of infected ducks strongly depends on the time of the year and geographic location. This again is a general statement that does not provide strong support to the conclusion. I suggest the author to have a look at recent publications on the ecology and epidemiology of IAV in wild ducks to strengthen the discussion (e.g. Latorre-Margalef et al. 2014. Proc B.281:20140098)
- lines 134-138: How would you explain this absence the virus spread in other migratory flyways ? Southwards ?
- lines 141-143: Not all bird species migrate at the same time, even for closely related ones (e.g. ducks). Again, although this information is interesting and valuable, concluding that the two events were related needs more support or at least to be better discussed.
- lines 149-153. Is there any molecular evidence of this change in virulence and selection ?
- lines 155-159: “the evolutionary events responsible for the European and Japanese cases must have occurred in migratory birds”. This statement is very vague. What evolutionary events ? In which migratory birds species and populations ? Where ? How ? etc.
- Line 169: Based on which segment ? If changes in population sizes were investigated then the results should be presented. Also, why not using a Bayesian Skyline priors to investigate such changes instead of exponential population growth ?
- line 178: “the longer outbreaks in wild birds and poultry persist...”. Not sure we can use “outbreak” for wild birds if they are asymptomatic. Is it expected that the H5N8 virus will persist in wild birds ? Are there other evidence of poultry-origin virus spillover to wild birds with subsequent long-time maintenance ?
- line 186: “bird watchers”. Bird watchers I doubt, especially if birds are asymptomatic. There is certainly a need for better surveillance and for the implication of trained ornithologists and veterinarian but this is somewhat different to bird watchers.
- Figures 1-8: The trees are poorly formatted (very difficult to read). The red clade is missing on Fig. 1. As they all seem to provide the same information I would suggest to place them together in an online supplementary file, and select one tree (e.g. HA) for detailed presentation in the manuscript. Slight editing (colors, simplified taxa names, etc.) could also help to read the figures.
- Figure legend: double check figure numbers: looks like there is a Fig. 9 missing.
- Figure 10. The map is not informative as presented. It shows very general migratory flyways and could have rather focused on the migratory routes of the species that are implicated in the spread of the H5N8 IAV subtype. Birdlife international and other websites provide maps that could be much more informative that the one presented in Figure 10: http://www.birdlife.org/datazone/species/factsheet/22680317

Reviewer: 3
Comments to the Author
In this manuscript, authors describe the recent H5N8 outbreaks of Europe and Japan might be a single source population, which has been spread by migratory birds by combining genetic methods and epidemiological data. It seems to be probable theory to explain current H5N8 outbreaks in European countries and Japan at the same time by similar viruses. Although the topic is of interest, the manuscript will need revising.

1. Line 36-39: Reference is inadequate. The contents of this sentence were derived from reference 17 rather than reference 4.
2. Line 44-45: The H5N8 virus of Ireland was genetically quite different from recent Asian H5N8 viruses, so “the original H5N8” is improper expression. These were just same serotype of viruses.  
3. Line 45-47: The authors have to identify a quotation.
4. Line 115: Delete the duplicated word (in Korea)
5. Line 149-152: The authors have to identify a quotation.
6. Figure Legends and Figures: All figure legends are same except for each gene name. However, there were different scale in figures between NA gene and the other genes. Please correct the scale in figures or rewrite figure legends.

 The bits highlighted in red are a fairly standard reviewers tool. First you give the praise. This is really good work but ...

Then the but kills off the paper. The but here is that the data could fit with the movement of poultry between farms. Except they cannot. For that to happen British, Dutch, And German farmers would have to all have obtained eggs from Korea. This would then have to infected a limited number of farms and spread from the farms to the local wild bird populations in small enough numbers not to the detected by a screen but in large enough numbers to give sporadic cases. This is absolute nonsense. A simple likelihood model says that the wild bird transmission is much more likely than these multiple domestic bird submissions. This referee has another agenda. Blocking the paper until their own paper is ready for the same journal or use this idea to write a paper for Science. EID went on to publish three more papers about the H5N8 outbreak and its spread. This extra data included cases from Russia which I had predicted must be there, but which were unavailable at the time.

The summer breeding grounds in Russia are an essential for monitoring bird influenza epidemics. We need to have a well funded international effort to get the best possible sampling of the virus so that we can predict future outbreaks.

No comments: