Evidence for the Widespread Occurrence of Bacteria Implicated in Acute Oak Decline from Incidental Genetic Sampling

Round 1

Reviewer 1 Report

Dear Authors,
I thoroughly enjoyed reading the manuscript. Because it was a well-written script with detailed descriptions of all the processes used in the study. Therefore, I've made a few minor comments (please see attachment).

What's lacking from this manuscript are your thoughts on what turns these potentially endophytic bacteria into pathogens.

- What is/are the catalyst/s for this shift in lifestyle? AOD bacteria were found in both symptomatic and asymptomatic oak trees. Simultaneously, stress-associated beetles were also frequently observed on many of these symptomatic trees. Therefore, I think some sort of stress is the triggering factor. A similar trend has been seen for fungi from the family Botryosphaeriaceae. These fungi are also a common endophyte of Eucalyptus. However, become pathogenic when the trees are stressed. As a result, it would be great if you could add a paragraph to the discussion to speculate on why these bacteria become pathogenic? 

Best wishes

Comments for author File: Comments.pdf

Author Response

1. We have made clearer the fact that abiotic factors cause stress to the tree, which may allow the beetle to colonise trees more easily. We have added in information about a current project that is investigating a possible metabolite link between the beetle larvae and the triggering of necrogenic activity in B. goodwinii.
2. Phytophthoras are a cause of sudden oak decline in the USA, but are not currently pathogenic in the UK.
3. We have corrected some minor typographical errors and given the full bacterial names at first mention in the main body of the text.
4. We have removed the sentence about the possible route of bacteria from litter into the tree.
5. We have included the full kit name for the DNEasy kit.
6. We have removed the table of most numerous phyla and added the information to the text.

Reviewer 2 Report

Review of: Evidence for the widespread occurrence of bacteria implicated in acute
oak decline from incidental genetic sampling

This manuscript uses molecular methods to document bacterial communities implicated in the emergence of an important forest disease.  The authors make a convincing case that the pathogens implicated in causing the disease are found throughout health and disease-impacted stands on a variety of materials (such as leaves, which, as I understand, are not impacted in this disease system).  The manuscript is very well written with clear messages and strong backing data. 

Also:

1. There are a few typos in the manuscript which the authors should ferret out

2. Most importantly, the figure quality is poor. The axis titles are very small, as is the text of the legends and the symbology.  This needs attention even though it’s a straightforward set of problems to change. It is, important to do this. The reader deserves to be able to read what you are trying to show and the quality is poor at present

3. Lastly, the PCA analysis.  I know why this is relevant, but its an odd choice. Why use PCA when one can just as easily use NMDS (nonmetric multidimensional scaling – or NMS). This addresses well known bias in PCA calculations, provides statistics that inform model appropriateness (the stress statistic), and provides a basis for applying a multivariate anova (the ‘adonis’ test / command). At present, the PCA is not very strongly hypothesis driven but it could be. How important is this to the authors? I would not stand in the way of publishing this paper because of the use of PCA, but its an odd choice given how well tested the alternatives are.  Also, it must be noted that the PCA could simply be removed. The various trees make the case that the bacterial communities appear at each site and that their presence alone is not enough to cause disease. I realize the authors are probably trying to go further and elucidate any shifts in bacterial community structure. That’s a good thing to do and it could probably be done with this dataset, but I don’t have confidence that the final say can be found in the PCA, its got well known problems.  I include this comment last, because while I think it’s a problem, there is a simple solution (remove the PCA, I don’t think it helps that much) or go full on and use NMDS with the hypothesis test (adonis).  In terms of my role as a reviewer, the last thing I want is to delay publication of a good manuscript. The ‘easy’ option of deleting the PCA is just that, but the authors should have a think about it and do what’s in the best interest of their goals. With a little clean up, this manuscript will be on its way to informing disease management.

Author Response

1. We have corrected the typographical errors that we found.
2. We have improved the quality of the figures and the size of text axes and legend data
3. We have replaced the PCA analysis and replaced it with NMDS and PERMANOVA analysis with the use of PERMDIST. A table of results and plots added for this analysis and the methods, results and discussion amended accordingly.