Referee 3's Review:

This paper investigates the hypothesis that genetic divergence is related to the number of ecological niche dimensions that are subject to divergent selection. The authors perform a series of field experiments on herbivorous Timema stick insects to test the hypothesis with regard to the host plant related niche dimensions crypsis and feeding performance. The tests consists of comparing pairs of ecotypes and closely related species with the hypothesis that different species should differ in more niche dimensions than ecotypes of the same species.

I find the idea as such interesting. However, one must say that the way it is put forward here, it is not a very strong test in a popperian sense. It is hardly surprising that you will find that interspecific divergence is larger, or along more dimensions, than intraspecific divergence. The authors already knew that one of the species in the species pair comparison (T. chumash) does not utilize one of the hosts, so they knew in advance that the treatment that transplants these insects to the non-host will result in poor performance. They also knew in advance that T. podura have the capacity to use both hosts. Hence, even if the principle makes sense, and is indeed interesting, it strikes me as a very blunt test. Haven't they just shown that individuals that are more closely related to each other are more similar than individuals that are more distantly related?
In my view, the merit of the paper lies in the general discussion of the problem, but I don't see that the authors have really tested the hypothesis at all. As the authors say, the outcome must not necessarily be as found, but it would have been rather surprising if it wasn't. A more proper test, in my view, would be either a comparative study of multiple species that vary in phylogenetic distance from each other (it would increase the n-value), or better still, a comparison of multiple ecotypes (or just populations) where genetic distance is measured independently.

Some additional comments:
First, I find the text to be unnecessarily hard to follow at times, and it could still do with some polishing. For example, the word "divergence" is used abundantly in the text and it is not always clear what type of divergence is meant. Take for example the second paragraph in the introduction, where I count to nine occurrences of various forms of the word (plus one "differentiation"). Sometimes the word refers to genetic divergence, sometimes to ecological, geographic or phenotypic divergence.
Another example is on page 13, line18: "Thus, although increased dimensionality of niche divergence may play a causal role in driving divergence ..." There are many more such examples and I recommend a thorough run through the manuscript to see if other wordings can be used, or to add more modifiers such as "ecological".
The authors claim to have measured "actual selection" (e.g. p 13, line 27), but I think they have measured "potential selection", as there is no information on the response to the selection. If, for example, there is no genetic variance for the trait, relative gene frequencies in the population will not change no matter how hard the selection pressure is.
Page 3, line 25: "the extent to which divergent drives ..." Insert "selection"
Page 14, line 12: "... different types of phenotypic traits that have diverged between the species pairs (e.g., water depth, distance from shore in the lake ..." Water depth and distance from shore are not phenotypic traits.
Page 16, line 21: "Lifetime fecundity was the number of eggs within an enclosure after the final census." Is this measure of fecundity not dependent of survival?

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.