The influence of environmental change over geological time on the tempo and mode of biological diversification, revealed by Neotropical butterflies

The investigation of events in the past is a challenging feat, but one that must be attempted with the best tools available in order to understand basic questions of neotropical diversity. To do this, various informed models have to be proposed and then tested with the best fit model that explains current diversity adopted until more data or better methods prove otherwise. The authors looked at the Andean uplift and the ensuing retreat of the Pebas ecosystem and the effect that the resulting landscape turnover had on neotropical butterfly radiation within the tribe Ithomiini. To do this they inferred a time calibrated phylogeny from molecular data set consisting of 87% of the known species (340 of 393) within the tribe, then tested several models of diversification rates, and finally to use the best fit models to infer the historical biogeography for the tribe.

The English and grammar of the paper appear adequate, although, it should be noted that depending on the journal, British Standard or American Standard may have to be adopted. I have no other comments on grammar other than to point out that on page 3 line 20 "During the late Miocene, during the Andean uplift" may better read with "and" instead of a comma. Also on page 6 line 21 that reads "One was at the root of and the core-group" seems to need modification as it seems the statement includes an error. It is recommended that both these statements be revisited.

After going through the introduction, I was expecting to go straight into the material and methods and had to search until they were found after the author contributions at the end of the paper. Due to the length and often times technicality of the cutting edge methods used in these complex analyses, journals often ask that they be lessened in length which often times dilutes them to useless banter. It is commendable that they authors chose to include them in the body of the paper, albeit in an usual position, but it is being suggested that a general paragraph that names the methods used without going into the details be included in the normal position, and that the more detailed discourse be included as one of the appendices. This would greatly reduce the length of the paper and those of us who are more technically minded will give the paper enough scrutiny to find the relevant information where we are directed to look.

With regards to the appendices, it is being suggested that they be reviewed and numbered correctly. It was a bit confusing to see two S1s and S8s, no S2, S3, S6, S7 or S9 and four S10s. I do realize that the missing information, for the most part, is contained within the file labelled Supplemental Information, however it is suggested that the file be named based on all the supplemental information that is contained therein.

On to more technical aspects, the authors should state categorically how they avoided much readily available data on species-rich subtribes from influencing (biasing) the calculation of the diversification rates. It is commendable that there was a good link made with food-plant evolution in this group as the retreat of the Pebas most certainly influenced food plant radiation; this was one the first questions that was raised in my mind and it is good that the authors included this in their model developments. The inclusion of multiple sources of data as well as as much of that data as possible makes for robust analyses and it was evident that the authors spent considerable time building a framework that validated the conclusions made.

Caution should be exercised, however, in making a general statement about the neotropical region and perhaps limit these statements to the specific group that was investigated. In the Discussion on page 11 lines 6 - 8 when the authors state that "dynamics of multiple landscape transformations ............. have determined the dynamics of speciation ........ in the Neotropical region." It is suggested that they limit statements to the group that they investigated as the events outlined may not have had the same effect, if any, on other groups of organisms. A modification of that statement to reflect this is suggested.

In general, I thought the paper was well written, the analysis, though quite technical, seemed robust enough to tease out a model that most likely explains the diversity of the tribe Ithomiini in the Andes and Amazon regions. The tables and diagrams are clear and adequately present the information that their legends say they do and are tastefully enough for the manuscript.

This manuscript presents the first species-level phylogeny of the Ithomiini (clear-wing butterflies). They then use this phylogeny to investigate the potential role of geophysical changes in South America in the Miocene and Pliocene (specifically the filling and draining of the Pebas wetland in present-day Western Amazon and the uplift of the Andes) in broad-scale macroevolutionary patterns in ithomiines. The authors examined this question thoroughly using several different biogeographic and macroevolutionary modeling frameworks. However, I have several concerns I would want to see addressed before I would consider recommending this manuscript, in addition to suggesting another editing pass be made to address grammatical errors and to capitalize the "g" in "BioGeoBEARS".

First, the phylogeny used in this study was time-calibrated using secondary calibration points and host-plant divergence times due to a lack of in-group fossil data. While such a practice is fairly standard practice among studies of fossil-poor taxonomic groups, it compounds uncertainty in date estimation. The authors would do well to report the error bars from their time calibrations and consider the implications of this uncertainty in their discussion and conclusions. Furthermore, estimating what can be interpreted as extinction rates from lineages without fossil representation can prove methodologically problematic (see Rabosky, 2010 in Evolution), and the authors should also consider this in their discussion as a probable source of error.

Second, the authors use a hybrid macroevolutionary modeling method (implementing Medusa, Morlon et al 2011, and TreePar) to identify and quantify diversification rate shifts to account for differnces in assumptions among the models. They do this in order to infer the number and position of rate shifts, as well as model how rates may vary between shift points. This study would be greatly improved if it were compared to the results of modeling these processes using BAMM (BAMM 2.5.0; Rabosky 2014), which can both infer likely rate shift points as well as quantify those shifts. While it is true that BAMM has come under criticism lately (Moore et al., 2016 in PNAS), many of these concerns have been addressed in subsequent literature (i.e. Rabosky et al 2017 in Systematic Biology).

Finally, the authors state that species were classified as Andean or non-Andean base on GPS coordinates and elevations, but do not state what the sources of this information were, nor what the criteria for classification were. The repeatability of this study would be greatly enhanced by the inclusion of this information.