This article on the evolution of head structures in lower Diptera is an important contribution to furthering our understanding of morphological evolution within true flies. The study collates information from several sources and provides new data using cutting-edge techniques such as microCT with 3D reconstruction and confocal laser scanning microscopy. The study of the selected larval and adult head structures is very exhaustive and the character descriptions in Appendix 1 and the discussion of the results (“Potential apomorphies of dipteran clades” p. 7) are very complete. They will be very useful for future studies on the morphological evolution and phylogeny of Diptera. The text is well-written.
The main weakness of the study is in my view the phylogenetic analysis, presentation of the phylogenetic results, and availability of the data matrix. I will here summarize my points:
1. The data matrices are not very accessible in the Appendix 1 and 2 as these are Word files. The matrices should be provided in Nexus file format (or even in WinClada format) so that future researchers can examine, include, or re-analyze the data. The matrices, phylogenetic methodology, and resulting trees should furthermore be submitted to TreeBase (http://treebase.org/) /DRYAD (http://datadryad.org).
2. I suggest to provide the specific matrices that result in the trees shown in Figs 1a, b, c. For example, the complete data-set (adult and larval characters) has fewer taxa than are coded for each partition because it appears that all of those taxa that are missing one or the other data partition are excluded from the analysis. To make the matrices, analyses, and results testable by others the specific matrices that were analyzed for the three trees in Fig. 1 need to included as separate appendices in addition to the two existing ones.
3. I was unable to find the tree length mentioned in the text for any of the resulting tree topologies. This information needs to be stated as it is the only measure to compare two trees resulting from the analysis of the same matrix.
4. I am not sure whether the tree search strategy (“Rachet [sic] Island Hopper, 1000 replicates” in WinClada, p. 3) is exhaustive enough to sample tree space enough. I have compiled a “complete data” matrix (adult and larva) from Appendix 1 and 2 in order to run a tree search in TNT (http://www.lillo.org.ar/phylogeny/tnt/). I combined the two Siphonaptera taxa in order to have complete data (adult and larva) for the resulting chimera and used it as the outgroup (root). All taxa that are missing one or the other data partition are excluded. (I am happy to share this matrix with the authors although it might contain mistakes introduced by me.) With a very exhaustive tree search resulting in 131,943,554 re-arrangements (mult1000; xmult=rss hits 20 fuse 2 drift 3 ratchet 10 multiply consense 10;) I obtained three trees of 505 steps length. The strict consensus tree is much better resolved than Fig. 1a (although the matrices used by the authors and the one used by me could be different) and I am happy to share it with the authors. (Somehow I cannot attach it to my review.) It is likely that the other two analyses are also the result of a tree search that did not exhaustively explore tree space.
5. I would like to ask the authors to state explicitly which taxon they used as the outgroup (the root) in each of the analyses. Please also include the outgroups in the tree topologies shown in Fig. 1.
6. A Bremer support analysis would provide a measure of support for each tree and would give the reader a better understanding of the relationships.
7. Fig. 2 maps the employed characters on the phylogenetic hypothesis of Diptera of the FlyTree project. In addition to the tree figure shown, I would have liked to see a tree on which all characters are individually mapped in Mesquite to give the reader some sense of character numbers supporting certain nodes. The figure caption is so long that it is very difficult to extract this information for a node/taxon of interest.
8. The Asilidae exemplar should be Stilpnogaster aemula and not “Silpnogaster”.
9. Nixon 1999 (The Parsimony Ratchet, a New Method for Rapid Parsimony Analysis. Cladistics 15: 407–414. doi: http://dx.doi.org/10.1111/j.1096-0031.1999.tb00277.x ) described the Ratchet algorithm and should be cited on page 3. Note spelling of Ratchet algorithm.