This manuscript reviews what is known about virulence evolution in SARS-CoV-2. It covers the timeline of the evolutionary emergence of key (phenotypic) variants of concern, expectations for virulence in the long term, and explanations for the observed increase in virulence with the VOCs. Given the wild speculation about virulence evolution that has been going on in the public sphere (and disagreement between scientists) for the last ~15 months, I think there is a need for an authoritative, synthetic review like this. I was excited to read and review this paper. In its current state, I find some of the ideas in this paper under-developed (or at least under-described), and would like to see them strengthened so they’ll be more accessible to the JEB audience. Below I outline the areas that in my opinion need more clarity or description.

Section 1
1. l. 29-31. What does it mean that virulence "has an evolutionary dimension" but those other metrics do not? For example, in as much as pathogenicity is a consequence of cellular interactions between host and pathogen, it will be subject to variation in underlying host and pathogen traits, no? Does this not also suggest "an evolutionary dimension"?

1. l. 31-32. It’s not clear here what is being predicted. This paragraph doesn’t introduce the question of "how much (or even why) should a pathogen harm its host?", so "prediction" is a surprise. (Unless "estimate" or "inference" is meant here?)

2. l. 35. What "problems" need to be solved? I gather these first three paragraphs are making the point that it’s hard to define the severity of infection imposed by a pathogen and constrain it to measures that are largely controlled by pathogen genetics, but I don’t think that message is sufficiently clear.

3. Further, Figure 1 is super cool, but not explained in enough detail in either the caption or the text. I assume Figure 1B represents estimates that the authors have calculated themselves, using the IFR data in Fig 1A and the UN demographic data, but it doesn’t actually say this anywhere.

Section 2
5. I appreciate the desire to want to keep distinct the genetic lineage and the phenotypic variant, but can it be made clear somewhere around line 70 that the 501.V1 variant is (or is related to) the B.1.1.7 lineage? I think this is done well for B.1.351 and 501.V2 in the next paragraph.

1. l. 113-114. I don’t know what it means to "set a new stage for the SARS-CoV-2 fitness landscape" and it seems pretty important given that a similar sentence+reference comes up two more times (l. 159 & 200). Is the key point ehre that this variant has some key features against which other variants are going to be competing? Or something else?

Section 3
7. I feel like I’m missing something, because I find the arguments around the Lavine et al. paper unsatisfying. So, we might expect the evolution of low virulence for a pathogen with increasing age-specific IFR, but high(er) virulence for something like MERS that has high IFR at low ages. Surely the evolutionary question then just shifts to why is virulence so high (or low) in hosts of young ages?

1. l. 143-147 presents the "tempting" idea of increasing affinity for ACE2 "as an evolution towards virulence". This makes sense (in a straw man sort of way) given the description of upper and lower respiratory tract infection described above. However, this isn’t at all how I have been thinking about increased binding affinity, which seems like it would more directly increase transmissibility (greater per virion chance of establishing an infection in a host?) and virulence (ditto, but in additional cells within a host, rather than in a new host). While the next section talks about different transmission/virulence phenotypes, it doesn’t connect those phenotypes to these changes in ACE2 affinity, which seems like a missed opportunity.

Section 4
9. This seems like one of the most important sections to me, but I don’t think there’s enough details for the casual interest reader. I would like to see a description of WHY "more virulent strains can be temporarily favoured" (l. 195), which I think should only take a sentence or two. Similarly, above that, when describing how "harming the host is not very costly" (l. 167) it would be useful to state the corollary here: if there is a benefit to be gained by increasing transmission (and there may especially be benefits early in an epidemic as per the arguments around l. 195), then any countervailing cost is likely to be too small to negate selection for increasing transmission. These ideas are all implicit in this section, but they could be made clearer for the reader. I also wonder if connecting this to the language of evolutionary theories of aging could be helpful to the JEB audience (e.g., Day 2003)?

1. Figure 2 could also use more description. I think this figure is crucial and really important for driving home a bunch of points. Specifically:
2. What does panel A bring to the table? I guess it makes the point that in the absence of a virulence-transmission tradeoff we might expect the evolution of avirulence. This could be stated explicitly in the text and caption.
3. The figure could be used to make the point that "there is no rule for maladaptation" (l. 151), i.e., a hypothetical virus could emerge that is to the right of the red dot and so we’d expect to see the evolution of decreased virulence.
4. Is there a way to illustrate the point contained in l.189-190?
5. In the caption, should the phrases "virulence is not adaptive/completely adaptive/partly maladaptive" be "increases in virulence in VOCs is not adaptive/..."? Should it say "Dashed blue lines show hypothetical transmission-virulence relationships"? Finally, the red dots look brown to me. Maybe use a different shape (more different than the variant blobs) to avoid confusion?

Section 5.
Two fairly minor points in this section.
11. l. 212-214. Are there data on how often we are reinfected by such coronaviruses? I guess I’m wondering if viruses can avoid having to generate escape mutants if immunity wanes sufficiently quickly?

1. l. 220-221. What are the conditions under which baseline immunity would have been higher at the time?

Minor edits.
- l. 6-7. It would be helpful to be explicit about what is meant here (e.g., "...the same evolutionary trajectory towards low virulence as other...")
- l. 26. Should this say "including with the use of therapeutics" or something?
- Table 1. Is it worth adding references/links to nextstrain, etc?
- l. 150. I think this point will be unclear to a lot of readers who don’t know anything about vaccine production or its history. Might be worth spelling this out in more detail.
- l. 151. the bit after the colon feels like it needs a "for example", given that there are a variety of ways for a pathogen to be maladaptively too virulent (e.g., infecting tissues more likely to lead to host death)