In this interesting study, the authors used published empirical networks, machine learning techniques, and spectral graph theory to assess how well spectral properties (e.g., spectral radius, Fiedler value) can predict the dynamics of pathogen spread in animal social networks, compared to other structural attributes (e.g., modularity). They found that both the spectral radius and Fiedler value were important predictors of a population’s vulnerability to disease outbreak. Furthermore, the authors provide a user friendly web application to assess the vulnerability of any given network to pathogen spread, which should prove to be incredibly useful across a variety of contexts. This paper and its associated web application represent a major contribution towards better understanding the relationships between animal social networks and pathogen spread. The manuscript is incredibly well written and the material is clearly presented. I strongly recommend publication following a few very minor revisions which I have detailed below.

General comment: The raw data is not present in the supplementary file and needs to be attached per journal requirements. The github link provided is not well organized, and it (at least from what I can see) does not contain all of the network data.

Lines 44-133: The Introduction is exceptionally well written. The authors do a fantastic job of setting the stage for their work.

Line 138: Please describe the process you used to find the other comparable published networks (i.e., those not in ASNR). In other words, what was your literature search process?

Line 140: Please briefly state why you decided not to include networks of farmed domestic animals in your analysis, especially because in the text, examples are given of how spectral properties can be used to predict vulnerability of farmed animal networks to disease (e.g., cattle to bovine brucellosis). I am not convinced that networks of farmed domestic animals should have been removed if the goal is to have a broadly applicable web application.

Line 142: This is an impressive sample size!

Line 142: Were mating networks used? If not, the authors might want to clarify this point. Also, if not, why not? What about sexually transmitted diseases?

Line 160: Should “uniform random” just say “random”?

Line 314-316: “We trained our surrogate decision tree on the predictions of the RF model rather than the network observations directly”. Could you briefly remind the readers why you chose to do this.

Line 455: Please replace “significant” with another word such as “substantial”.