Referee 1's review:

This is an interesting manuscript that is addressing a very interesting and important questions about the structure of plant-pathogen communities. Evaluating the structure of plant-parasite networks will provide a much stronger evolutionary basis for understanding the impacts of diseases in natural communities. The authors take advantage of an unusual data set from French forests to attempt to address questions of compartmentalization and nestedness that has not really been possible before.

It is noteworthy that the introduction was excellent - I thought this was a very clear description of why we should care about species interaction networks, and a review of important measures.

Unfortunately, I think there are some significant failings in the analysis and interpretation as the manuscript currently stands. I think this could probably be a very significant paper, but as written there are too many ambiguous areas left unexplored. I detail those areas below.

1. Are the experiments, statistics, and other analyses performed to a sufficient technical standard?
I am familiar with, but not expert in analysis of interaction networks. I do not feel qualified to comment on the specifics of the clustering algorithms and nestedness analysis used here. I do, however, have some concerns about assumptions and applications of those analysis.

It seems very odd that the analysis of phylogenetic signal in the network architecture was conducted looking at the phylum and subphylum level (Figure 3). There is a lot of structure in the tree that was presented, and it seems like using readily available phlogenetic trees and distances among the hosts would make much more sense. Calculating such trees and distances from the APG structures is pretty straightforward using Phylocom / Phylomatic softward, available at www.phylodiversity.net.

Similarly, the analysis from the plant side is presented, but the fungal analysis are not shown because of lack of space. This really seems to be throwing away half the story, and I don't see any real reason to do this.

There is some discussion about the potential importance of the geographic segregation between Conipherophyta and Magnoliophyta, but I find it very strange that there is no real analysis of this. The analysis is focused on phylogenitic effects, but the author's own discussion of distribution patterns of the Rosaceae to me suggest that the entire pattern may be driven much more strongly by geography than by plant phylogeny. As such, since this is not analytically explored, I find the results and conclusions potentially misleading.

2. Are the conclusions presented in an appropriate fashion with speculations and hypotheses identified as such?

The discussion that "fungal host ranges should reflect the relative abundances of tree species at the time at which the specialization events occurred" was not convincing to me. Density dependent spread of pathogens is a fast process, and local patterns of community structure could easily be shaped over the course of just a few years. I really found this argument to be a stretch.

3. Have the techniques used been documented in sufficient detail to allow replication?
The description of how the data were collected by the DSF is far too sparse. "Skilled forestes evenly covering ... forests of the country" is not enough. Also, in my experience in forest pathology (albeit not in France) I know of few people who would spend equivalent time looking for or recording the pathogens that do not cause significant wood decay or tree mortality, and these are likely to be much less reliably reported than are major pathogens. This is likely to have a strong impact on the structure of the fungal community side of the interaction networks, but there is not enough information presented to evaluate potential biases, nor does it appear that the authors have made this effort. Similarly, the authors note the potential for foresters to be more likely to encounter rare fungi on common hosts than on rare hosts, but they don't really evaluate how that bias would affect the analytical results.

4. Is the report presented in an intelligible fashion and written in English?
Yes, although there are a few unclear sentences.
Bottom p4: "... sampling and life-history strategies and their position in the network architecture"
Middle p6: "Of the 51 tree taxa, 45 correspond.... Genetic continuum"
Results on Compartmentalization: This whole section could be much better focused. It isn't clear what we should get from the details of what is in each of the splits.

Fig 2B says tha t C1, C4, C5, and C6 are highlighted in gray, but it is very difficult to see which compartments correspond to which parts of the graphs.


5. Have the exact results reported been published elsewhere?
Not to my knowledge.

6. Does the research meet all applicable standards with regard to the ethics of human / animal experimentation, participants' consent, and research integrity? Please contact us if you have any concerns.
No concerns.

7. Has the report adhered to the relevant community standards for research conduct, presentation of results, and deposition of data, where appropriate?
Yes.

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.