Reviewer 2's Review

-----

The authors have suggested that the time course of pathogenesis might scale with body size. They cite the well-known fact that metabolic rate scales with body size as a starting point for explaining why pathogenesis might scale. This is a worthwhile idea to discuss. However, for reasons listed below, I am unpersuaded by the idea that metabolic role plays a special role. The work would be improved after correction of the points listed below, especially the major comments.

Major comments:

1) General: The authors have left out a very significant point, namely that *body size* scales with body size. In other words, even if metabolic rate played no role in pathogenesis, one might imagine that the length of the body could play a role in determining the times t_s and t_d because it takes time for pathogens to spread through a body. Although it is not clear what slope would be predicted by body size-based scaling (number of cells? then slope=1; body length? then slope=+1/3), these alternate hypotheses must be mentioned. Note that slope=1 pretty clearly fails, slope=+1/3 less clearly so. These ideas must be mentioned prominently.

2) In Table 1, the "All diseases" rows do not quite make sense. If it is the slope and intercept of a fit to all the data pooled, this is incorrect owing to the phenomenon of grade shifts. Each dataset could follow the same power law (in this case +0.25), yet if they have different intercepts then pooling the data leads to a larger slope. See Gould SJ Approaches to Primate Paleobiology pp. 244-292, Barton and Harvey in Nature, and other sources.

The correct thing to do would be to calculate a weighted average of the slopes. If this is done then the average for t_S is approximately 0.21 with a confidence interval (estimated) of about 0.17-0.25; t_D, 0.19 (0.12, 0.30); t_s vs. t_D, 0.93 (0.6, 1.4). The space for the intercept would be left blank as there is no good way to combine the data.

3) A significant limitation in the analysis is the small number of orders of magnitude (OOM) spanned by the data. The search for allometric relationships is subject to the pitfall that when data span only a few OOMs, many monotonic relationships can be fitted by a power law. This must be discussed. The number of OOM should be listed as a column in Table 1.

The authors could also use their exhortation on pp. 10-11 to remind people that for allometric studies a wide range of body sizes is quite valuable. An obvious point worth making is that what would be especially valuable is extension to very small animals such as shrews, for the practical reason that such studies would be quicker and cheaper than studying disease in whale.

4) It would be helpful for a critical reader to go through the manuscript for errors of spelling and grammar. Starting from the beginning, three visible examples are page 2 line 11 "Principle" and line 13 "When the timing of pathogenesis are scaled..." and page 3 line 3 "Most emerging infectious diseases that currently or can potentially..."

5) The supplementary information list of species should include actual numeric values and give references so that people can use the numbers and/or go find the data for themselves and launch their own investigations.

6) Figure 3: The ratio c1/c2 seems quite variable, with a long tail to the right. Is this of interest? How would one evaluate whether any of these outliers were worth investigating as possible exceptions that can teach us something about pathogenesis? Please discuss!

-----

N.B. These are the general comments made by the reviewer when reviewing this paper in light of which the manuscript was revised. Specific points addressed during revision of the paper are not shown.