This paper comprises a new comparative study of diving ability and blood / muscle oxygen carrying functions. The investigation is well justified, novel, and has potential to illuminate fundamental evolutionary questions about matching of physiological traits with foraging niches.

The major result as stated in the abstract, is “Our results indicate that the independent evolution of diving in these two lineages of ducks led to an increase in O2 carrying capacity in the blood and muscles, and the extent of the increases in muscular O2 storage in the locomotory muscles has had a strong bearing on an individual species’ dive capabilities.”

Having read this paper and examined its findings, I somewhat agree with these assertions; however, I do not think the assertions about direction of causation can be justified. Was it the “evolution of diving” that led to an increase in O2 carrying capacity? Or are the O2-related traits to some extent reflecting plasticity? Have the increases in muscular O2 storage “had a strong bearing” on dive capabilities? Or has diving behavior potentially caused high muscular O2 storage, whether by plasticity, evolution by natural selection, or both? I don’t believe that it’s possible to distinguish between these alternatives based on the information presented in this paper, nor do I think it is necessary to do so for this paper to be successful. Fortunately, simple changes of wording can allow these assertions to become more conservative, pointing out the links or associations between traits and clades, or traits and behaviors, without claiming directional causation.

Generally, I was somewhat confused in trying to match the analyses described in the Methods to the individual results reported in Results. It would help in this respect if the Methods section were more clear about what question(s) was being asked with each analysis, and then have the results section follow the same order of reporting and using same terminology. The connections between analyses and questions became more clear once I examined the tables and figures, and this also led to my point in the next paragraph.

The PIC analysis seems to be redundant with the PGLS models reported in Table S1. At first I thought that perhaps the PIC was just to visualize those results. However, PIC plots are not all that useful in my opinion because of the non-interpretable axes. My suggestion here would be to drop the PIC plots (or make them a supplement), and plot the raw data in each case, while color coding points by clade as you did effectively in Fig. 2. You can superimpose plots of the phylo-corrected regression lines with each one (caculated using PGLS in phytools). Plotting the original data this way would improve another aspect of this paper that I found frustrating which is that the dive times were not reported or visualized anywhere (presumably those data would also become available when data are archived before publication as well?).

Specific comments, objections, suggestions, etc:

Fig. 1 and 3 are different styles, but it’s not at all clear to me why. I much prefer the cleaner style in Fig. 1, and I suggest these figures could be combined as well, as a 5 panel figure.

Line 34: I respectfully object to characterization of these traits as adaptations… “adaptations to increase” should be “traits related to” in my opinion. This occurs again on lines 51-52. I think this is not a study of ‘adaptations’ but rather of ‘trait associations’.

Line 50: “most of which are also found in Eurasia”; I suggest deleting this vague statement, as it also needs qualification, and doesn’t really add anything anyway.

Lines 58-60: Again, I would back off these claims somewhat. You’re looking at a trait-behavior association across species, but the species trait can be caused by several things, not just adaptive evolution for diving, but also phylogenetic affinities, plasticity, and unmodeled aspects of behavior or environment.

Line 64 and elsewhere: I suggest using ‘clade’ instead of ‘lineage’

Line 69: give the range.

Line 121: it’s not the ‘closeness’ of relationships, but rather that some are more closely related than others that justifies the use of phylo comparative methods – a subtle change of wording can fix this.

Line 130: what “differences”? between what?

Line 131-133: “Next, we used model selection and phylogenetic generalized least squares regression to investigate which parameters impacted dive time most significantly.” I’m confused as to which parameters were used here, and that you are trying to model dive time. Why didn’t you use PGLS to model the blood and myoglobin parameters?

Line 146: delete “increases in” (?).

Line 151: among instead of between.

Line 159: You should really be using the more conservative AICc here considering the modest sample sizes. Clarify please.

Line 222: I would adjust this assertion to make it more specific: “during longer dives”. If it were advantageous during diving always, then all divers would have higher Mb.

Line 228: change “importance of” to “association with” and edit the rest of sentence accordingly.

Line 243: optima should be optimum.

Line 253-4: suggest rewrite as: “adding to our growing understanding” (rather than to the ‘literature’)

Consider adding a figure with one or more timetrees that map key traits in color. Also quantify phylogenetic signal in each trait.

Supp Table 1: First off, I think this table should be in the main manuscript if at all possible, considering that it is central to the results. I would suggest that using AICc would be well justified improvement here, though perhaps this is a subjective assertion. I would also suggest reporting delta AIC(c) in order to make it easier for the reader to compare models.

The authors have done an excellent job with the revision. The revised manuscript is easier to follow and better highlights the central results, which are interesting and novel to the best of my knowledge. I note that two species of ducks are misspelled in Table 1 (but not in Table 2!). Regarding Table 3, it's interesting to note that the top model is a nested model of models 2-5, which really reinforces its status as the only model worth considering (see Arnold's classic 2010 paper about uninformative parameters). Finally, what are the lines in 3 panels of Fig. 5? Something looks off with those.