This is a very interesting and brave study aiming to understand vocal complexity in suboscine birds, a taxon that does not learn song and that is traditionally characterised by limited vocal plasticity. The key finding is that sexual selection does not predict song complexity, whereas social selection does. I find the results really interesting and intriguing, but I have some doubts about the analytical procedure and the interpretation.

1) I see the pertinence of considering social selection separately from sexual selection, but sexual selection is a form of social selection (wee West Eberhardt’s original formulation or Lyon and Montgomerie 2012 PRBS), so it would be nice that this nuance is included in the introduction and discussion. In addition, social and sexual selection by definition cannot be directly, but using proxies for them. So it would be nice to see that this is expressed in the MS, for instance by using “sexual selection as measured by degree of polygyny” or “social selection as measured as territoriality”. You do not need to say it each time, but it would be honest to remind the reader from time to time what your proxy is. Because the interpretation should be very nuanced: certainly territoriality is key to obtaining resources, a clear case of social selection operating, but is it not sexually selected also? I mean territorial behaviour peaks at the fertile period in many species, would you not call this sexual selection? Don’t take me wrong, I see the point the authors are making, but I just find the all/nothing approach not very rigorous.

2) As for the song analysis, I congratulate the authors for the manual scoring of the information. However, the way that within-individual variation is measured, by adding standard deviations of 3 variables, seems a rather primitive way of doing things. This is in contrast to their using sophisticated Bayesian phylogenetic inference! Those 3 variables must have some common covariance, so this method of putting them together seems crude and unjustified statistically.

MINOR COMMENTS
Abstract. Please define at least briefly what temporal and structural complexity are. Also mention here what is your metric for sexual selection and state that territoriality is your proxy of social selection.
41-43. “it seems highly unlikely…” because? I see your point in the paragraph after but this sentence comes out of the blue and the logic is missing.
24-25. Social selection is a wide umbrella, and as far as I see it, it also encompasses sexual selection, so the idea of simply pitching one against the other is not justified.
57-58. It mases sense to predict that the repertoire of calls may be influenced by social complexity but why song complexity? Evidence in chickadees suggest different meanings to different calls, but a complex nightingale or mockingbird song is not expected to convey complex semantics for a social audience. This prediction seems poorly substantiated to me.
138-151 Some suboscines (ie Eleania spp) have dawn songs which are remarkably different from songs produced later in the day, what was the methods here when selecting songs? Also, I have analysed suboscine songs myself, and I was mortified at times trying to tell calls from songs. What do you do with a given species has two vocalizations equally simple? Which is the song and which the call?
193-195. Frequency variation. Adding up standard deviations seems at first glance rather unorthodox stats? Is this procedure backed up by statistical literature?
200-210. The distinction within-song versus within-bout would be of use here.
211- is EPP available for all species??
356. Multiple traits may play a role here, perhaps relevant to consider framework by Gil and Gahr 2002 (TREE). Or not, this is up to you to decide.
362-365- You cite elsewhere Sierro et al (2023) but I think it is relevant here to consider how there might be a balance between selection for variation and selection for repetition (i.e. precision).