Reviewer #2's Review

N.B. These are general comments were made on the initially submitted version of this manuscript. The manuscript as published was revised extensively in light of these and other specific points which are not shown here.

“Calhim et al present and test an interesting idea. They suggest that the strength of postcopulatory sexual selection should be negatively correlated with the extent of variation in a trait. While appealing, there are flaws in their logic and their methods.

1) Logic: There is no clear prediction on how variation should relate to sexually selected traits - Kirkpatrick and Ryan suggest CV will be negatively associated, but others (like Schluter et al, Rowe & Houle) suggest the opposite because the traits could or should be condition dependence. Even the relationship between allometry and sexually selected traits is now questioned (Bonduriansky & Day's recent paper). It seems to me that the only clear prediction is that CV will go down with stabilising selection - a situation that does not appear to apply to the tests here. Directional selection changes the mean but there is no clear prediction on associated changes in variance. Stabilising selection changes variance. See Lande & Arnold for a description of the expectation for how selection changes different moments of the distribution.

2) Beyond this, I am unsure how to interpret the use of CV as a trait in comparative studies. This strikes me as a very odd use of CV.

3) The traits: The authors use total sperm length as one measure, then "sperm design" as another. But these are highly autocorrelated as the total sperm length is simply the linear combination of the three traits used to generate sperm design. Thus, they do not have independent tests of their hypothesis. If CV of sperm length shows a relationship, it is necessarily true that CV of sperm design will also show the relationship. PC would not solve this.

4) Sperm design is the mean CV of three CV's - again, inappropriately treating CV as a trait.

In the end, then, it is a result but it is a result that can't be interpreted. Not everything that is statistically significant is biologically significant.”