Abstract

Pacific robins exhibit one of the most complex range-wide mosaics of sexual dichromatism and monochromatism. The evolutionary origins of this geographic mosaic remain poorly understood despite long-standing interest from ornithologists, and its influential role in the development of Ernst Mayr's theories on speciation and the 'Biological Species Concept'. One factor limiting our understanding of the evolution of sexual plumage variation in Pacific robins is a lack of well-resolved taxon boundaries and phylogenetic relationships in the group. Here, we use primarily historical museum specimens to obtain dense sampling of mtDNA, nuclear DNA, plumage color and morphometrics from all named taxa in the radiation in order to infer taxon boundaries and relationships. We use these data to test hypotheses about colonization history, plumage evolution and reduced island dichromatism. Our data show that the Pacific robin radiation comprises four distinct lineages, which warrant recognition as separate species - the previously recognized Norfolk robin P. multicolor and red-capped robin P. goodenovii, and two new species we propose to name: 'Solomon robin' P. polymorpha Mayr, 1934 for the populations on Solomon and Bougainville Islands, and 'Mayr's robin' P. pusilla Peale, 1848 (in honor of Ernst Mayr's detailed work on the southwest Pacific robins) for the populations on Vanuatu, Fiji and Samoa. Our data suggest that the common ancestor of the entire Pacific robin radiation was most likely sexually dichromatic and that the radiation-wide mosaic of sexual plumage color arose via repeated losses of elaborate plumage in males and gains of elaborate plumage in females on separate islands.


Authors

Kearns, Anna M.;  Joseph, Leo;  Austin, Jeremy J.;  Driskell, Amy C.;  Omland, Kevin E.

Publons users who've claimed - I am an author
Contributors on Publons
  • 1 author
  • 1 reviewer
Followers on Publons
  • pre-publication peer review (FINAL ROUND)
    Decision Letter
    2020/01/31

    31-Jan-2020


    Dear Dr. Kearns:


    It is a pleasure to accept your manuscript entitled "Complex mosaic of sexual dichromatism and monochromatism in Pacific robins results from both gains and losses of elaborate coloration" in its current form for publication in Journal of Avian Biology.


    The comments of the Subject Editor who reviewed your manuscript are included below.


    Recommendation by the Subject Editor:
    The authors have made significant revisions based on the reviewers’ comments and improved the manuscript. You did a good job outlining the changes that were done. For structure I agree that what you have is good (as an aside, I would agree with K=4 based on your plots). I do agree with the original comment from Reviewer 2 about the Ima and sample size, however, respect your choice to keep it in.

    Decision letter by
    Cite this decision letter
    Endorsed by
    Ongoing discussion (0 comments - click to toggle)
    Author Response
    2020/01/23

    REVIEWER 1


    1
    Essentially, I found the plumage analyses a bit difficult to follow. There may be some extraneous analyses that can be removed (e.g., comparing within the dichromatic lineages) and some that can be more focused on (e.g., comparing within the drab V lineages).


    Response: We've worked on simplifying the plumage analyses to be more targeted to our key questions. One main way we've tackled this is to divide the results into two sections that cover all results pertaining to (1) taxonomy/species distinctiveness and then (2) plumage evolution. Within the plumage evolution results section we've divided results by the question/predictions that were outlined in the Introduction, and then this is repeated again in the Discussion. The other main change is to try and reduce the number of variables being compared and discussed by using MANOVA and LDA on all variables for each sexual plumage mode.


    2
    There also needs to be a clearer connection between theoretical and statistical predictions about color (a table might help).


    Response: We tried but the table idea was a bit too clunky. We've attempted to make the connection clearer in the predictions/questions section of the Intro, and also in how results are presented and discussed.


    3
    Also, organization of the patch (back, crown, throat) and color variable (brightness, hue, saturation) needs to be improved. For example, there are no figures depicting hue, and it’s unclear why Figure 4 focuses only on brightness.


    Response: See response 1. We've also added boxplot figures that cover each plumage patch, subspecies and color variable (see new Fig 4 and Supp Figs 3 & 4).


    4
    I think the data can be reorganized without any new analyses, but I might also suggest the authors explore color space analyses (e.g., R package pavo) that would help collapse some of the variables. For example, they could plot these colors in color space and quantify distance/overlap accounting for both hue and saturation simultaneously. Brightness analyses would still be separate. These methods also have the added benefit of accounting for the avian visual system, which may be relevant to such slight differences in color. I don’t think those analyses are necessary per se, but are worth investigating.


    Response: Certainly worth investigating for future work, however, we feel exploring the avian visual system and going down the rabbit hole of trying to fit these models to our non-model system is a little bit too much for our paper here. In an effort to reduce complexity, we've re-analysed the data using MANOVA and LDA (an approach used in other spectrophotometry papers, e.g., Cibois et al. 2019)


    5
    L89: Change “hereditary” to “heritable”?
    Response: Done


    6
    L105: You introduce the “SOL” acronym here but in Figure 1 use “SI”. I would suggest choosing one and staying consistent here and using only the acronym (for this and the other lineages) thereafter whenever possible. Throughout the ms you go back-and-forth between e.g., “Vanuatu/Fiji/Samoa” and “VFS”.


    Response: Changed all figures to SOL not SI. And where appropriate lineages are referred to as SOL and VFS throughout the manuscript.


    7
    L129: You need to expand on prediction (b) here. The plumage comparisons end up comprising a huge chunk of your results and are interesting and important, but need to be better justified here. This is related to my comment below from the Methods- matching up your specific theoretical predictions to statistical predictions. Exactly what comparisons are you going to do and why would certain results support the independent loss/gain hypothesis? This could be much clearer throughout.


    Response: The reviewer made an excellent point. Therefore, we have made revisions throughout the manuscript to address this issue. In particular adding a new suite of analyses, reformatting the sequence of results, highlighting which question is being addressed in the Results section, and reformatting the Discussion of plumage evolution to specifically address each major question and evidence for/against each prediction that appeared in the Introduction. Hopefully these changes help to make the plumage analyses clearer throughout.


    8
    L129 cont: It also seems like a potentially relevant prediction here would be differing “levels” of monochromatism (i.e., among monochromatic populations, some would have more identical sexes than others).


    Response: Good point. We've now added "levels of monochromatism" as a distinct prediction to Q2 in the Intro and also as a subsection of the Results. We tested for significant differences between the sexes and also used cohen's D to compare standardized differences in means between subspecies within and between the three sexual plumage modes.


    9
    Use active voice throughout unless there is a convention/formatting requirement by the journal.


    Response: Adjusted where possible. But not all sentences work in active voice.


    10
    L165: Decide on a capitalization scheme for “Robin(s)” and stay consistent throughout. To this point I assumed you were using lowercase “robin” to refer to the entire radiation and uppercase to refer to identified species, but here you seem to be referring exclusively to pusilla but use lowercase.


    Response: Yes, captialized for named species and lowercase for the radiation. Fixed here, and checked other instances throughout.


    11
    L194: Would be good to be explicit here about which populations are omitted from theses analyses. You mention earlier that you got no new tissues from NI and S but can re-state that here. Explains why you set Kmax at 4 not 6.


    Response: Ok. We've added such a line here, and we've also included a better justification about sampling to the first paragraph of the "Multilocus tests of taxon boundaries and gene flow" Methods paragraph starting line 205


    12
    L195: Delete “agnostically”.
    Response: Done


    13
    L222: Delete comma after “Note that”.
    Response: Done


    14
    L244: Change “1” to “one”.
    Response: Done


    15
    L245: Change “we are unable” to “we were unable”.
    Response: Done


    16
    L288: Your morphometric PCA is useful for visualization but not really for quantifying whether morphometrics distinguishes the lineages. A DFA might be more useful here where you can get classification error rates.


    Response: True. We added LDA and MANOVA analyses now, which have helped to better explore which, if any, variables pull apart the lineages.


    17
    L297: The plumage color analyses are complex and somewhat confusing. I think you need to frame your specific tests better in terms of your a priori predictions. In other words, what information are you getting from specific color comparisons? What statistical results would support what predictions. Maybe a short table could be useful here or in the results matching the plumage comparisons and results to your predictions.


    Response: See responses above, especially #1.


    18
    L306: Unclear why you don’t compare among the three monochromatic lineages on V. This could strengthen your independent evolution argument if there are differences, could it not? The taxonomic scale is a bit different from your main comparison between SOL and VFS, but it still seems relevant to the independent evolution argument even if its among lineages in the same archipelago.


    Response: We did not initially compare the Vanuatu subspecies, because this was already done in detail in Kearns et al 2015 (as stated in the Methods). The reviewer has a good point however that these data could be useful for supporting our hypotheses, so we've added MANOVA and LDA for the dull monochromatic subspecies, along with the other plumage modes.


    19
    L320: I think you may have missed an opportunity to do a more “nuanced” ASR considering you have quantitative info on plumage color (spec data). There are models that can place continually-varying trait data on a tree in an ASR-type framework. That might give you more power and would be a better use of your hard-earned plumage data rather than collapsing the dichromatism data into subjective categories. I understand the three plumage modes are certainly qualitatively different, and your current ASR does give you some relevant info, but using a continuous scale might give more insights and shouldn’t be too onerous. This could be an issue when including the outgroups for which you don’t have spec data though.


    Response: We opted not to do such analyses here.


    20
    L362: Use subscripts to refer to each panel of Figure 3 when presenting corresponding results.
    Response: Ok


    21
    L363: Include DeltaK and Ln(P)K figures in supplement?
    Response: Each peak in these plots shows the K value mentioned in the text. We believe that is sufficient.


    22
    L386: Not sure you can call this the VFS lineage here as there is no S in this analysis.


    Response: This issue is now highlighted and discussed in the first paragraph of the "Multilocus tests of taxon boundaries and gene flow" Methods paragraph starting line 205. And also in the second sentence of the "Species distinctiveness of SOL and VFS lineages" section of the Results. A compromise we've settled on for ease of reference is to refer to the lineage as VF(S) when making inferences based on nuclear data.


    23
    L407: Use subscripts to refer to each panel of Figure 4 when presenting corresponding results. Also you discuss morphometrics first but it is the fourth panel in the figure.
    Response: Our new figure 4 doesn’t have panel #s.


    24
    L408: As mentioned above, you could do a DFA or other clustering test here. Although your PCA certainly looks like there is no morphometric signal of divergence.


    Response: For simplicity we've removed PCA analyses in the revision, and have replaced with LDA. Like PCA, these also show no morphometric signal of divergence.


    25
    L428: It’s not clear how quantifying color differences among the three modes is relevant to your predictions. Brown and black are obviously different. It seems like the more important test is among either drab or elaborate lineages, are males and females different?


    Response: Yes, agreed. We've removed discussion of the former, and have concentrated on the latter two.


    26
    L439: Same comment as above- what do we gain by comparing within the dichromatic taxa? You may be able to simplify things (including your figures) by only focusing on comparisons among the monochromatic lineages.


    Response: We still think it's valuable to compare among the dichromatic taxa. If anything by showing there are/are not differences between dichromatic taxa, it puts in context the degree of differences/or lack of differences we see in the monochromatic taxa.


    27
    L442: You refer to Figure 4 here and several other times when discussing hue and saturation but this figure only shows brightness. Perhaps an identical supplemental figure displaying hue and saturation variation? Either way, you also need to better justify why Figure 4 focuses only on brightness.


    Response: Supplemental Figs have been added, and now we have three different figures (Fig 4, Supp Fig 3 & 5) which show either brightness, hue or saturation for each plumage patch divided by sexual plumage mode.


    28
    L444: Not sure “lighter hue” is informative. Hue = color. “Light vs dark” to me connotes brightness variation, not hue.


    Response: Ok. Most of these kinds of comparisons and descriptions are deleted from the revision, focusing on broad trends and significances instead.


    29
    L449: Separate figures for hue and saturation are especially necessary to evaluate these results. Also- you focus in the text only on back when it appears from Figure 4b and c that there are differences in the other patches as well. Finally, don’t you also want to compare these values among the dull lineages on V (soror, feminina, and cognata)?


    Response: See response #27 & #18


    30
    L460: Specify what you mean by “lighter”. Higher brightness values? And do you have stats for these comparisons? Some of the boxplots in Figure 4 don’t look all that different (e.g., drab V M vs F back brightness).


    Response: Ok. Most of these kinds of comparisons and descriptions are deleted from the revision, focusing on broad trends and significances instead.


    31
    L497: Can’t really say this re: VFS since there are no intron data from S.


    Response: We do have mtDNA for Samoa, and that's the data the comment about VFS being reciprocally monophyletic was referring to. We've modified the following statement about the nuclear data to be more explicit that we're referring to Vanuatu and Fiji and not Samoa. Generally, our discussion of VFS comes with the already stated caveat that we lacked nuc data for Samoa, but are hypothesising they'll fall with Vanuatu and Fiji. We feel this is a very reasonable hypothesis, and that it is not necessary to repeat this caveat every time. A compromise we've settled on is to refer to the lineage as VF(S) when making inferences based on nuclear data.


    32
    L526: Delete “support”.
    Response: Done


    33
    L530: Delete comma after “While”.
    Response: Done


    34
    L533: Refer specifically to Figures 5c and 5d where appropriate here.
    Response: Done


    35
    L588: Insert a line space before this paragraph.
    Response: Done


    36
    L1037: Change “2” to “two”.
    Response: Done


    37
    L1048: Insert a space in “boxplotsdepict”.
    Response: Done


    38
    Figure 3a: Use a color other than green for the third population emerging from K = 3. I have been associating green with the Samoa population up to this point.


    Response: The reviewer makes a reasonable point, but the problem is that there are only so many colors and shades in the rainbow. We already used blue, pink, orange, purple and green to represent an archipelago or subspecies. The best we can do is to choose a shade of one of these colors that hasn’t been used previously. Red and yellow have not been used but creating a spectrum incorporating intermediate oranges would be confusing because of the VAN. We've opted to change the K colors to be a spectrum from yellow to dark green. Since yellow hasn't been used at all and Samoa is not in the nuc dataset we think this will be the least confusing option.


    39
    Figure S4: This figure is really tough to decipher. I know there are tons of comparisons to make, but I think perhaps a series of boxplots might be better (ala Figure S3). Also, there are no hue data here.


    Response: This figure has been deleted and replaced by boxplots with brightness, hue and saturation each featured. In addition we added biplots showing LDA analyses.


    REVIEWER 2
    1
    Overall the methods are adequate but my main concern is the low number of individuals sequenced by taxa. I believe this low number actually impedes the power of several aspects of the proposed analyses, and in particular the Ima2 estimation. I would suggest to remove this analysis or, at least, to be more cautious in the interpretation of the results.


    Response: Small sample sizes certainly does impede our power and limit some of our interpretations. But, for a species group that is this widespread across so many different Pacific islands, the sample sizes and complete taxon sampling that we have for mtDNA and phenotypic datasets is actually very impressive. Even Reviewer 1 admits "I appreciate that sample sizes for these types of taxa are precious!". We strongly believe that our study represents a valuable contribution to our knowledge of not just Pacific robins but also island evolution in general. As next-gen and ancient DNA approaches improve, we're sure we (or someone else) will revisit this species complex and do a thorough study of plumage evolution and colonization history--however given the degradation of DNA in our samples this won't be feasible for some time. For now, we believe that our modest nuclear dataset is enlightening, and well worth publication and dissemination. Even with the small sample size, our IMa2 estimation tests and shows that there is no rampant gene flow between the SOL and VFS lineages. If this were occurring, even our modest dataset would detect it.


    2
    Page 4, line 57: I don’t think that the term “ancient DNA approaches” apply to this use of museum specimens. No particular methods seems to have been used, apart from managing degraded DNA.


    Response: True. Its not an ancient DNA study, but we did employ many wet lab methods tailored to maximizing DNA yield from degraded historical museum specimens, and these are the same approaches you use when dealing with true ancient DNA. We've removed reference to it.


    3
    Page 5, first paragraph: it might be worth mentioning here the previous results from DNA data that led to the recognition of the three species (i.e. the paragraph currently on page 6, “recent molecular studies…”°


    Response: We have added the following "Sixteen distinct taxa are recognized in the Pacific robin species complex and recent molecular studies have identified three, or possibly four, distinct species...etc..."


    4
    Page 8, the acronym SOL and VFS should be explained when first used in the text (even if it’s not difficult to understand)


    Response: We did do this at first mention in the second and third line of Pg 7 (now Pg 5 in revised version).


    5
    Page 16, last paragraph: the definition of “marked”, “dull” and “elaborate” plumage patterns should be given here (currently the last one two are described page 18)


    Response: Definitions for these appeared much earlier in the Introduction when first introduced, and also in Figure 1. But for clarity we have now also added the following to this paragraph-- "—i.e., comparing all subspecies with marked dichromatism (elaborate males, dull females; 6 subspecies from Solomon Islands, Fiji and Vanuatu) versus all subspecies with dull monochromatism (dull ‘feminized’ males and females; 3 subspecies from Vanuatu) versus all subspecies with elaborate monochromatism (elaborate ‘masculinized’ males and females; 5 subspecies from Solomon Islands, Vanuatu and Samoa) (see Figure 1 for further details)."


    6
    Results, page 19: the description of the nuclear network does not fit closely with the suppl fig 1. In CLOCK FT, FK and Va shared common haplotype; PCBD and CSDE show a single difference between SOL and VFS.
    Response: Ok


    7
    Page 21 and 22: as already mentioned, the Ima2 analysis based on such small number of individuals probably does not have the power to achieve convergence. I would suggest to remove this section. The Beast tree can be calibrated using the same mutation rate to infer a time of divergence between the taxa.


    Response: We'd like to keep the analysis, but better discuss its limitations owing to small sample sizes. For the record, and as stated in the methods, the models were run for a long period of time, and all showed good convergence metrics and consistency between independent replicates.


    8
    Page 23, figure 4: “dull” is used in the text, whereas “drab” is used in the figure
    Response: Fixed


    9
    Page 24: the figure 4 is mentioned several time but it only shows brightness, not hue or saturation. This section is a bit difficult to follow and I would suggest to add a Table (in sup. Material) summarizing the different ANOVA analyzes.
    Response: Agreed. These issues have now all been rectified through the addition of boxplots featuring hue and saturation, and also by using MANOVA. See responses to Reviewer 1 for more details.


    10
    Page 30, the discussion on the molecular evidence should include the fact that the results are based on a small number of individuals.
    Response: Agreed, we've added more dicussion of this issue now.


    11
    Page 31, Page 587: the radiation of Pacific reedwarblers might be as recent as Petroica diversification (Cibois et al. 2011 J. biogeo)
    Response: Added.



    Cite this author response
    Endorsed by
    Ongoing discussion (0 comments - click to toggle)
  • pre-publication peer review (ROUND 1)
    Decision Letter
    2019/12/18

    18-Dec-2019


    Dear Dr. Kearns:


    Manuscript ID JAV-02404 entitled "Complex mosaic of sexual dichromatism and monochromatism in Pacific robins results from both gains and losses of elaborate coloration" which you submitted to Journal of Avian Biology, has been reviewed. The comments of the reviewers and the recommendation by the Subject Editor are included below.


    The reviewers have suggested some minor revisions to your manuscript. Based on the recommendation by the SE, I invite you to respond to the comments by the Subject Editor and reviewers and to revise your manuscript accordingly.


    Recommendation by the Subject Editor:


    We received comments from two reviewers. Both have done a good job reviewing the manuscript and point out a few areas where revisions are required. There are a number of revisions required, but they mostly require reorganization of the data and figures, not any new analyses.




    Reviewer(s)' Comments to Author:


    Reviewer: 1


    Comments to the Author
    General Comments:


    What a fascinating system. The authors do a nice job of trying to make sense of the taxonomic situation in these robins. The genetic boundaries are more straightforward and clear to me than the issue of independent evolution of monochromatism, so I focus mainly on the latter.


    Essentially, I found the plumage analyses a bit difficult to follow. There may be some extraneous analyses that can be removed (e.g., comparing within the dichromatic lineages) and some that can be more focused on (e.g., comparing within the drab V lineages). There also needs to be a clearer connection between theoretical and statistical predictions about color (a table might help). Also, organization of the patch (back, crown, throat) and color variable (brightness, hue, saturation) needs to be improved. For example, there are no figures depicting hue, and it’s unclear why Figure 4 focuses only on brightness. I think the data can be reorganized without any new analyses, but I might also suggest the authors explore color space analyses (e.g., R package pavo) that would help collapse some of the variables. For example, they could plot these colors in color space and quantify distance/overlap accounting for both hue and saturation simultaneously. Brightness analyses would still be separate. These methods also have the added benefit of accounting for the avian visual system, which may be relevant to such slight differences in color. I don’t think those analyses are necessary per se, but are worth investigating.


    Introduction:


    L89: Change “hereditary” to “heritable”?


    L105: You introduce the “SOL” acronym here but in Figure 1 use “SI”. I would suggest choosing one and staying consistent here and using only the acronym (for this and the other lineages) thereafter whenever possible. Throughout the ms you go back-and-forth between e.g., “Vanuatu/Fiji/Samoa” and “VFS”.


    L129: You need to expand on prediction (b) here. The plumage comparisons end up comprising a huge chunk of your results and are interesting and important, but need to be better justified here. This is related to my comment below from the Methods- matching up your specific theoretical predictions to statistical predictions. Exactly what comparisons are you going to do and why would certain results support the independent loss/gain hypothesis? This could be much clearer throughout. It also seems like a potentially relevant prediction here would be differing “levels” of monochromatism (i.e., among monochromatic populations, some would have more identical sexes than others).


    Methods:


    Use active voice throughout unless there is a convention/formatting requirement by the journal.


    L164: I appreciate that sample sizes for these types of taxa are precious!


    L165: Decide on a capitalization scheme for “Robin(s)” and stay consistent throughout. To this point I assumed you were using lowercase “robin” to refer to the entire radiation and uppercase to refer to identified species, but here you seem to be referring exclusively to pusilla but use lowercase.


    L194: Would be good to be explicit here about which populations are omitted from theses analyses. You mention earlier that you got no new tissues from NI and S but can re-state that here. Explains why you set Kmax at 4 not 6.


    L195: Delete “agnostically”.


    L222: Delete comma after “Note that”.


    L244: Change “1” to “one”.


    L245: Change “we are unable” to “we were unable”.


    L288: Your morphometric PCA is useful for visualization but not really for quantifying whether morphometrics distinguishes the lineages. A DFA might be more useful here where you can get classification error rates.


    L297: The plumage color analyses are complex and somewhat confusing. I think you need to frame your specific tests better in terms of your a priori predictions. In other words, what information are you getting from specific color comparisons? What statistical results would support what predictions. Maybe a short table could be useful here or in the results matching the plumage comparisons and results to your predictions.


    L306: Unclear why you don’t compare among the three monochromatic lineages on V. This could strengthen your independent evolution argument if there are differences, could it not? The taxonomic scale is a bit different from your main comparison between SOL and VFS, but it still seems relevant to the independent evolution argument even if its among lineages in the same archipelago.


    L320: I think you may have missed an opportunity to do a more “nuanced” ASR considering you have quantitative info on plumage color (spec data). There are models that can place continually-varying trait data on a tree in an ASR-type framework. That might give you more power and would be a better use of your hard-earned plumage data rather than collapsing the dichromatism data into subjective categories. I understand the three plumage modes are certainly qualitatively different, and your current ASR does give you some relevant info, but using a continuous scale might give more insights and shouldn’t be too onerous. This could be an issue when including the outgroups for which you don’t have spec data though.


    Results:


    L362: Use subscripts to refer to each panel of Figure 3 when presenting corresponding results.


    L363: Include DeltaK and Ln(P)K figures in supplement?


    L386: Not sure you can call this the VFS lineage here as there is no S in this analysis.


    L407: Use subscripts to refer to each panel of Figure 4 when presenting corresponding results. Also you discuss morphometrics first but it is the fourth panel in the figure.


    L408: As mentioned above, you could do a DFA or other clustering test here. Although your PCA certainly looks like there is no morphometric signal of divergence.


    L428: It’s not clear how quantifying color differences among the three modes is relevant to your predictions. Brown and black are obviously different. It seems like the more important test is among either drab or elaborate lineages, are males and females different?


    L439: Same comment as above- what do we gain by comparing within the dichromatic taxa? You may be able to simplify things (including your figures) by only focusing on comparisons among the monochromatic lineages.


    L442: You refer to Figure 4 here and several other times when discussing hue and saturation but this figure only shows brightness. Perhaps an identical supplemental figure displaying hue and saturation variation? Either way, you also need to better justify why Figure 4 focuses only on brightness.


    L444: Not sure “lighter hue” is informative. Hue = color. “Light vs dark” to me connotes brightness variation, not hue.


    L449: Separate figures for hue and saturation are especially necessary to evaluate these results. Also- you focus in the text only on back when it appears from Figure 4b and c that there are differences in the other patches as well. Finally, don’t you also want to compare these values among the dull lineages on V (soror, feminina, and cognata)?


    L460: Specify what you mean by “lighter”. Higher brightness values? And do you have stats for these comparisons? Some of the boxplots in Figure 4 don’t look all that different (e.g., drab V M vs F back brightness).


    L497: Can’t really say this re: VFS since there are no intron data from S.


    L526: Delete “support”.


    L530: Delete comma after “While”.


    L533: Refer specifically to Figures 5c and 5d where appropriate here.


    Discussion:


    L588: Insert a line space before this paragraph.


    Figures:


    L1037: Change “2” to “two”.


    L1048: Insert a space in “boxplotsdepict”.


    Figure 3a: Use a color other than green for the third population emerging from K = 3. I have been associating green with the Samoa population up to this point.


    Figure S4: This figure is really tough to decipher. I know there are tons of comparisons to make, but I think perhaps a series of boxplots might be better (ala Figure S3). Also, there are no hue data here.


    Reviewer: 2


    Comments to the Author
    This manuscript presents new data and analyses on the Pacific robin complex. This work complements previous studies led by Kearns and colleagues on this species complex. In the present work, they add new nuclear sequences to previous data sets (mostly ND2 sequences) and they provided now analyses of morphological and plumage coloration data. Overall the methods are adequate but my main concern is the low number of individuals sequenced by taxa. I believe this low number actually impedes the power of several aspects of the proposed analyses, and in particular the Ima2 estimation. I would suggest to remove this analysis or, at least, to be more cautious in the interpretation of the results.
    Page 4, line 57: I don’t think that the term “ancient DNA approaches” apply to this use of museum specimens. No particular methods seems to have been used, apart from managing degraded DNA.
    Page 5, first paragraph: it might be worth mentioning here the previous results from DNA data that led to the recognition of the three species (i.e. the paragraph currently on page 6, “recent molecular studies…”°
    Page 8, the acronym SOL and VFS should be explained when first used in the text (even if it’s not difficult to understand)
    Page 16, last paragraph: the definition of “marked”, “dull” and “elaborate” plumage patterns should be given here (currently the last one two are described page 18)
    Results, page 19: the description of the nuclear network does not fit closely with the suppl fig 1. In CLOCK FT, FK and Va shared common haplotype; PCBD and CSDE show a single difference between SOL and VFS.
    Page 21 and 22: as already mentioned, the Ima2 analysis based on such small number of individuals probably does not have the power to achieve convergence. I would suggest to remove this section. The Beast tree can be calibrated using the same mutation rate to infer a time of divergence between the taxa.
    Page 23, figure 4: “dull” is used in the text, whereas “drab” is used in the figure
    Page 24: the figure 4 is mentioned several time but it only shows brightness, not hue or saturation. This section is a bit difficult to follow and I would suggest to add a Table (in sup. Material) summarizing the different ANOVA analyzes.
    Page 30, the discussion on the molecular evidence should include the fact that the results are based on a small number of individuals.
    Page 31, Page 587: the radiation of Pacific reedwarblers might be as recent as Petroica diversification (Cibois et al. 2011 J. biogeo)

    Decision letter by
    Cite this decision letter
    Endorsed by
    Ongoing discussion (0 comments - click to toggle)
    Reviewer report
    2019/12/13

    This manuscript presents new data and analyses on the Pacific robin complex. This work complements previous studies led by Kearns and colleagues on this species complex. In the present work, they add new nuclear sequences to previous data sets (mostly ND2 sequences) and they provided now analyses of morphological and plumage coloration data. Overall the methods are adequate but my main concern is the low number of individuals sequenced by taxa. I believe this low number actually impedes the power of several aspects of the proposed analyses, and in particular the Ima2 estimation. I would suggest to remove this analysis or, at least, to be more cautious in the interpretation of the results.
    Page 4, line 57: I don’t think that the term “ancient DNA approaches” apply to this use of museum specimens. No particular methods seems to have been used, apart from managing degraded DNA.
    Page 5, first paragraph: it might be worth mentioning here the previous results from DNA data that led to the recognition of the three species (i.e. the paragraph currently on page 6, “recent molecular studies…”°
    Page 8, the acronym SOL and VFS should be explained when first used in the text (even if it’s not difficult to understand)
    Page 16, last paragraph: the definition of “marked”, “dull” and “elaborate” plumage patterns should be given here (currently the last one two are described page 18)
    Results, page 19: the description of the nuclear network does not fit closely with the suppl fig 1. In CLOCK FT, FK and Va shared common haplotype; PCBD and CSDE show a single difference between SOL and VFS.
    Page 21 and 22: as already mentioned, the Ima2 analysis based on such small number of individuals probably does not have the power to achieve convergence. I would suggest to remove this section. The Beast tree can be calibrated using the same mutation rate to infer a time of divergence between the taxa.
    Page 23, figure 4: “dull” is used in the text, whereas “drab” is used in the figure
    Page 24: the figure 4 is mentioned several time but it only shows brightness, not hue or saturation. This section is a bit difficult to follow and I would suggest to add a Table (in sup. Material) summarizing the different ANOVA analyzes.
    Page 30, the discussion on the molecular evidence should include the fact that the results are based on a small number of individuals.
    Page 31, Page 587: the radiation of Pacific reedwarblers might be as recent as Petroica diversification (Cibois et al. 2011 J. biogeo)

    Reviewed by
    Cite this review
    Endorsed by
    Ongoing discussion (0 comments - click to toggle)
    Reviewer report
    2019/12/03

    General Comments:

    What a fascinating system. The authors do a nice job of trying to make sense of the taxonomic situation in these robins. The genetic boundaries are more straightforward and clear to me than the issue of independent evolution of monochromatism, so I focus mainly on the latter.

    Essentially, I found the plumage analyses a bit difficult to follow. There may be some extraneous analyses that can be removed (e.g., comparing within the dichromatic lineages) and some that can be more focused on (e.g., comparing within the drab V lineages). There also needs to be a clearer connection between theoretical and statistical predictions about color (a table might help). Also, organization of the patch (back, crown, throat) and color variable (brightness, hue, saturation) needs to be improved. For example, there are no figures depicting hue, and it’s unclear why Figure 4 focuses only on brightness. I think the data can be reorganized without any new analyses, but I might also suggest the authors explore color space analyses (e.g., R package pavo) that would help collapse some of the variables. For example, they could plot these colors in color space and quantify distance/overlap accounting for both hue and saturation simultaneously. Brightness analyses would still be separate. These methods also have the added benefit of accounting for the avian visual system, which may be relevant to such slight differences in color. I don’t think those analyses are necessary per se, but are worth investigating.

    Introduction:

    L89: Change “hereditary” to “heritable”?

    L105: You introduce the “SOL” acronym here but in Figure 1 use “SI”. I would suggest choosing one and staying consistent here and using only the acronym (for this and the other lineages) thereafter whenever possible. Throughout the ms you go back-and-forth between e.g., “Vanuatu/Fiji/Samoa” and “VFS”.

    L129: You need to expand on prediction (b) here. The plumage comparisons end up comprising a huge chunk of your results and are interesting and important, but need to be better justified here. This is related to my comment below from the Methods- matching up your specific theoretical predictions to statistical predictions. Exactly what comparisons are you going to do and why would certain results support the independent loss/gain hypothesis? This could be much clearer throughout. It also seems like a potentially relevant prediction here would be differing “levels” of monochromatism (i.e., among monochromatic populations, some would have more identical sexes than others).

    Methods:

    Use active voice throughout unless there is a convention/formatting requirement by the journal.

    L164: I appreciate that sample sizes for these types of taxa are precious!

    L165: Decide on a capitalization scheme for “Robin(s)” and stay consistent throughout. To this point I assumed you were using lowercase “robin” to refer to the entire radiation and uppercase to refer to identified species, but here you seem to be referring exclusively to pusilla but use lowercase.

    L194: Would be good to be explicit here about which populations are omitted from theses analyses. You mention earlier that you got no new tissues from NI and S but can re-state that here. Explains why you set Kmax at 4 not 6.

    L195: Delete “agnostically”.

    L222: Delete comma after “Note that”.

    L244: Change “1” to “one”.

    L245: Change “we are unable” to “we were unable”.

    L288: Your morphometric PCA is useful for visualization but not really for quantifying whether morphometrics distinguishes the lineages. A DFA might be more useful here where you can get classification error rates.

    L297: The plumage color analyses are complex and somewhat confusing. I think you need to frame your specific tests better in terms of your a priori predictions. In other words, what information are you getting from specific color comparisons? What statistical results would support what predictions. Maybe a short table could be useful here or in the results matching the plumage comparisons and results to your predictions.

    L306: Unclear why you don’t compare among the three monochromatic lineages on V. This could strengthen your independent evolution argument if there are differences, could it not? The taxonomic scale is a bit different from your main comparison between SOL and VFS, but it still seems relevant to the independent evolution argument even if its among lineages in the same archipelago.

    L320: I think you may have missed an opportunity to do a more “nuanced” ASR considering you have quantitative info on plumage color (spec data). There are models that can place continually-varying trait data on a tree in an ASR-type framework. That might give you more power and would be a better use of your hard-earned plumage data rather than collapsing the dichromatism data into subjective categories. I understand the three plumage modes are certainly qualitatively different, and your current ASR does give you some relevant info, but using a continuous scale might give more insights and shouldn’t be too onerous. This could be an issue when including the outgroups for which you don’t have spec data though.

    Results:

    L362: Use subscripts to refer to each panel of Figure 3 when presenting corresponding results.

    L363: Include DeltaK and Ln(P)K figures in supplement?

    L386: Not sure you can call this the VFS lineage here as there is no S in this analysis.

    L407: Use subscripts to refer to each panel of Figure 4 when presenting corresponding results. Also you discuss morphometrics first but it is the fourth panel in the figure.

    L408: As mentioned above, you could do a DFA or other clustering test here. Although your PCA certainly looks like there is no morphometric signal of divergence.

    L428: It’s not clear how quantifying color differences among the three modes is relevant to your predictions. Brown and black are obviously different. It seems like the more important test is among either drab or elaborate lineages, are males and females different?

    L439: Same comment as above- what do we gain by comparing within the dichromatic taxa? You may be able to simplify things (including your figures) by only focusing on comparisons among the monochromatic lineages.

    L442: You refer to Figure 4 here and several other times when discussing hue and saturation but this figure only shows brightness. Perhaps an identical supplemental figure displaying hue and saturation variation? Either way, you also need to better justify why Figure 4 focuses only on brightness.

    L444: Not sure “lighter hue” is informative. Hue = color. “Light vs dark” to me connotes brightness variation, not hue.

    L449: Separate figures for hue and saturation are especially necessary to evaluate these results. Also- you focus in the text only on back when it appears from Figure 4b and c that there are differences in the other patches as well. Finally, don’t you also want to compare these values among the dull lineages on V (soror, feminina, and cognata)?

    L460: Specify what you mean by “lighter”. Higher brightness values? And do you have stats for these comparisons? Some of the boxplots in Figure 4 don’t look all that different (e.g., drab V M vs F back brightness).

    L497: Can’t really say this re: VFS since there are no intron data from S.

    L526: Delete “support”.

    L530: Delete comma after “While”.

    L533: Refer specifically to Figures 5c and 5d where appropriate here.

    Discussion:

    L588: Insert a line space before this paragraph.

    Figures:

    L1037: Change “2” to “two”.

    L1048: Insert a space in “boxplotsdepict”.

    Figure 3a: Use a color other than green for the third population emerging from K = 3. I have been associating green with the Samoa population up to this point.

    Figure S4: This figure is really tough to decipher. I know there are tons of comparisons to make, but I think perhaps a series of boxplots might be better (ala Figure S3). Also, there are no hue data here.

    Reviewed by
    Cite this review
    Endorsed by
    Ongoing discussion (0 comments - click to toggle)
All peer review content displayed here is covered by a Creative Commons CC BY 4.0 license.