Species biology and demographic history determine species vulnerability to climate change in tropical island endemic birds

Article activity feed

1. 

   Version published to 10.7554/elife.106369.2 on eLife

   Jun 18, 2026
2. 

   Version published to 10.7554/elife.106369 on eLife

   Jun 18, 2026
3. 

   eLife

   Jun 16, 2026

   eLife Assessment

   Tropical single-island endemic bird populations are particularly vulnerable to climate change. This study investigates genetic evidence of how such species dealt with climate change in the past as a possible predictor of how they will respond in the future, which could provide an important example for the fields of conservation genetics and island biogeography. The authors' integration of genomics and habitat modeling is commendable, but we find that the support for their conclusions is currently inadequate: some model parameter choices do not seem to reflect the biology of the studied species or to be well founded, which can cause misalignment of modeled dynamics with glaciation windows crucial for interpreting the study's results against its claims.

   Read the original source
4. 

   eLife

   Jun 16, 2026

   Reviewer #1 (Public review):

   Summary:

   The authors combine PSMC and habitat modeling to try to connect habitat change during the Last Glacial Period to changes in Ne.

   Strengths:

   Observing how tropical single-island endemic bird species responded to habitat change in the past may help inform conservation interventions for these particularly vulnerable species. The combination of genomics and habitat modeling is a good idea-this sort of interdisciplinary thinking is what is needed to tackle these complex questions. Additionally, the use of PSMC makes it possible to perform this analysis on poorly-studied species with only a single genome available.

   Room for Improvement:

   A paper was cited to support the idea, but why coalescent Ne is a better predictor of extinction risk than current genomic diversity or current Ne isn't explicitly …

   Reviewer #1 (Public review):

   Summary:

   The authors combine PSMC and habitat modeling to try to connect habitat change during the Last Glacial Period to changes in Ne.

   Strengths:

   Observing how tropical single-island endemic bird species responded to habitat change in the past may help inform conservation interventions for these particularly vulnerable species. The combination of genomics and habitat modeling is a good idea-this sort of interdisciplinary thinking is what is needed to tackle these complex questions. Additionally, the use of PSMC makes it possible to perform this analysis on poorly-studied species with only a single genome available.

   Room for Improvement:

   A paper was cited to support the idea, but why coalescent Ne is a better predictor of extinction risk than current genomic diversity or current Ne isn't explicitly explained in this paper.

   Differing PSMC parameters may also impact results: the differences between passerines and non-passerines was one of their main results. They explain why they chose different mutation rates for the two groups, but they do not provide any analysis to show this difference was not driven by the different mutation rates used for the two groups.

   For five of the species tested, PSMC parameter differences led to different results, but the species shown in table S4 are different from what is listed in the manuscript.

   Ecosystems are highly complex; there may also be other variables influencing past demographic change other than those explored here. Results should be interpreted with caution.

   Read the original source
5. 

   eLife

   Jun 16, 2026

   Reviewer #2 (Public review):

   Summary and strengths:

   In this manuscript, Karjee and colleagues used coalescent based effective population size reconstruction (PSMC) from single genomes to understand past population trends in island birds and related this to life history traits and glacial patterns. In this analysis they chose to use a generation time of 2 years for passerines and 1 year for non-passerines. Non-passerine birds include Amazona vittata which only reaches sexual maturity at 3-5 years; Amazona guildingii which reaches sexual maturity at ~5 years; Amblyornis subalaris at 7 years etc. This means that the choice of generation time is very poorly matched to the species biology of many of the focal systems. What this will do is to "squash" the PSMC plot, meaning that population trends will not match with when they actually …

   Reviewer #2 (Public review):

   Summary and strengths:

   In this manuscript, Karjee and colleagues used coalescent based effective population size reconstruction (PSMC) from single genomes to understand past population trends in island birds and related this to life history traits and glacial patterns. In this analysis they chose to use a generation time of 2 years for passerines and 1 year for non-passerines. Non-passerine birds include Amazona vittata which only reaches sexual maturity at 3-5 years; Amazona guildingii which reaches sexual maturity at ~5 years; Amblyornis subalaris at 7 years etc. This means that the choice of generation time is very poorly matched to the species biology of many of the focal systems. What this will do is to "squash" the PSMC plot, meaning that population trends will not match with when they actually occurred. As a result, glaciation windows are not correctly placed. It is my opinion that the results are not interpretable in the current form.

   The authors must adjust the generation time to roughly the median period between average age of sexual maturity and age of death. It should represent the time when an individual has had 50% of their offspring. After which all analyses must be repeated.

   Read the original source
6. 

   eLife

   Jun 16, 2026

   Author response:

   The following is the authors’ response to the original reviews.

   eLife Assessment

   Tropical single-island endemic bird populations are particularly vulnerable to climate change. The authors investigate genetic evidence of how such species dealt with climate changes in the past as a possible predictor for how they will respond to change in the future, which could provide an important example for the fields of conservation genetics and island biogeography. The authors' integration of genomics and habitat modeling is commendable, but we find that the support for their conclusions is incomplete: at times, the results presented appear to contradict each other, the authors do not fully account for key variables, and the limited taxonomic scope may cause problematic biases for the conclusion.

   We thank the editors for …

   Author response:

   The following is the authors’ response to the original reviews.

   eLife Assessment

   Tropical single-island endemic bird populations are particularly vulnerable to climate change. The authors investigate genetic evidence of how such species dealt with climate changes in the past as a possible predictor for how they will respond to change in the future, which could provide an important example for the fields of conservation genetics and island biogeography. The authors' integration of genomics and habitat modeling is commendable, but we find that the support for their conclusions is incomplete: at times, the results presented appear to contradict each other, the authors do not fully account for key variables, and the limited taxonomic scope may cause problematic biases for the conclusion.

   We thank the editors for supporting the premise of this study and highlighting the importance of the study approach. Based on the lacuna identified by the editors and the reviewers, we have modified the manuscript and details of the same are given below. We believe that these revisions have now substantially improved the flow and scope of the manuscript and have addressed the concerns raised by the reviewers.

   Reviewer #1 (Public review):

   Summary:

   The authors combine PSMC and habitat modeling to try to connect habitat change during the Last Glacial Period to changes in Ne.

   Strengths:

   Observing how tropical single-island endemic bird species responded to habitat change in the past may help inform conservation interventions for these particularly vulnerable species. The combination of genomics and habitat modeling is a good idea - this sort of interdisciplinary thinking is what is needed to tackle these complex questions. Additionally, the use of PSMC makes it possible to perform this analysis on poorly-studied species with only a single genome available.

   Room for Improvement:

   Why coalescent Ne is a better predictor of extinction risk than current genomic diversity, or current Ne, isn't explicitly explained. PSMC in particular has many caveats, and some are not acknowledged or adequately addressed by the authors. For example, the authors note that population structure is a confounding factor with PSMC, but that it is not a problem in this instance. They do not provide compelling evidence for why this would be the case, they simply state that the species studied are all single-island endemics. However, single-island endemic species are not necessarily panmictic; this is even less likely to be true for species studied here that inhabit a large geographic area (ie, Australian species). Differing PSMC parameters may also impact results: the differences between passerines and non-passerines were one of their main results, but they do not provide any analysis to show that this difference was not driven by the different mutation rates used for the two groups.

   Parameters for many steps are not described, and choices that are described (such as the PSMC parameters) are not always fully explained. It is unclear why all data was mapped to the autosomes rather than removing reads that map to the sex chromosomes first. Using all the data, the reads belonging to the sex chromosomes could potentially map to other areas of the genome. It does not seem like a mapping quality filter was used, so these potential spurious alignments would not have been removed prior to analysis.

   There are points where the results are described in ways that appear to potentially differ from the supplementary figures. The authors state that even for species where PSMC results differed between models, "trends of Ne increase or decrease from the LIG to LGM were robust across all three PSMC models considered." The figures in the supplement for Pachycephala philippinensis, Rhynochetos jubatus, and Zosterops hypoxanthus appear to potentially contradict this statement, but it is difficult to tell, as the time period observed is not clearly marked on the graphs. How this robustness of trends was determined is not explained, leaving the precision of the analysis unclear.

   Table 1 also includes some information that contradicts what is in the Supplementary Tables, leading to a lack of clarity. Centropus unirufus, Chaetorhynchus papuensis, and Cnemophilus loriae are not included in Supplementary Table 4. Table 1 says Eulacestoma nigropectus, Paradisaea rubra, and Parotia lawesii did not undergo PSMC analysis, but Supplementary Table 4 says PSMC and modeling trends matched for these species. Table 1 says Rhagologus leucostigma underwent both PSMC and climate modeling, but Supplementary Table 4 says "NA" as if it was missing one of these analyses.

   Additionally, some of the results appear to contradict each other. For example, they show that there is no impact of habitat change in larger-bodied species, but also that larger-bodied species saw a decrease in Ne during the LGP. In another example, they state that when a species saw an increase in habitat during the LGP, they also had an increase in Ne. However, they also state that this was not the case for non-passerines.

   Ecosystems are highly complex; there may also be other variables influencing past demographic change other than those explored here. Results should be interpreted with caution.

   We thank the reviewer for their comments, which has helped us in improving the scope of the manuscript while also removing errors in the supporting information. We have improved the section of the manuscript which addressed the drawbacks of PSMC in our revised version. Details and rational for parameter choice are now included in the revised manuscript.

   We performed additional PSMC analyses for a subset of the samples (n = 5), wherein the scaffolds mapping to the sex chromosome were removed only after mapping the reads. We compared the new approach suggested by the reviewer to our original approach and no differences in the PSMC pattern were observed, highlighting the robustness of the results (Supplementary Information Fig. S3).

   Additionally, we have included multiple box-plot and tables in the revised manuscript that helps with interpreting the changes in effective population size. The details of the revisions are presented below in the “Recommendations for the authors” section. We believe that these changes have improved the scope of the manuscript and removed any redundancies and conflicts.

   Reviewer #2 (Public review):

   Summary and strengths:

   In this manuscript, Karjee and colleagues used coalescent-based effective population size reconstruction (PSMC) from single genomes to understand past population trends in island birds and related this to life history traits and glacial patterns. This concept is fairly new, as there are still relatively few multiple PSMC synthesis studies. I also thought that the focus on island endemics was unique and adds value to this paper. I enjoyed seeing a paper focused on South East Asia and think that this could help contribute to our knowledge of the important biodiversity within this region.

   Major weaknesses:

   My biggest concern with this paper is that the analyses are limited to 20-30 species, and significant taxonomic bias is present (there are multiple species of passerine but only 1-2 representatives of other groups). While this is not an issue alone, many of the life history traits or geographical traits are conflated with phylogenetic diversity (e.g., there are no large-bodied passerines). Thus, it is my opinion that the impact of these drivers of past population size is conflated and cannot be disentangled with the current data. The authors themselves state that the core hypothesis surrounding Ne and habitat availability is not supported by their entire dataset (only seen in Passerines). This was not clear enough in the abstract, and conclusions cannot be drawn here as the impact of taxonomy cannot be separated from data richness, traits, etc. The PSMC analysis was done according to the most recent recommendations, and this part of the manuscript is fairly robust. However, in several places, it is incorrectly stated that the PSMC measures or can infer genetic diversity; PSMC only infers past effective population size. It cannot measure genetic diversity in the past. I cannot review the habitat reconstruction modelling as I am a conservation genomics specialist.

   Appraisal:

   I am not convinced about the findings within the paper. I do not think that the results are sufficiently supported at this time, largely due to the conflation of taxonomy with other variables. As this type of comparison is new, I do think that there is a chance for reasonable impact on the field of genomics and island biogeography if the manuscript's constraints are addressed. I do not see scope for impact on conservation at this time and find the conclusions in the abstract regarding conservation relevance to be unfounded.

   We thank the reviewer for highlighting the unique and robust analytical approaches we have taken in this study. We agree with the reviewer that our sample size currently is small. However, we do observe a robust correlation between habitat fluctuation and change in effective population size. Further, the study also highlights the predicament of tropical island endemics, which are currently understudied and future studies are necessary to safeguard the biodiversity. We have highlighted this while also addressing the concerns in the revised version of the manuscript.

   Recommendations for the authors:

   Reviewer #1 (Recommendations for the authors):

   Overall:

   This starts with a great premise - looking at how tropical single-island endemic bird species dealt with climate changes in the past may be a predictor of how they will respond to change in the future. Since these species are at high risk of extinction in the face of climate change, tailored approaches to conservation are a good idea. While the premise is solid, I have some questions and recommendations. At times while reading, I did feel a bit confused, which may be due to the fact that this isn't my exact area of expertise. However, if I'm confused, that means a reader from a general audience is also likely to be confused. Some results appear to be conflicting, some claims about data seem possibly inaccurate, and some major limitations are not acknowledged or fully addressed.

   Below I've noted areas that I feel could benefit from revisions. That being said, I liked the integration of habitat modeling and genomics! These sorts of multifaceted approaches are necessary when it comes to unraveling the complex dynamics involved in ecology and evolution.

   Crucial Issues to Address:

   (1) Line 75: With the lower sea levels and habitat change, you say animals can disperse across barriers of land and sea. When it comes to these single-island endemics, were they always confined to a single island? Is there no possibility of introgression with ancient populations of birds on other islands during these periods?

   We thank the reviewers for identifying the potential artifact in effective population size estimates that may occur due to hybridization/introgression. Most of our species belong to small and oligotypic families as has been addressed in the discussion section already, making them likely to be newly arisen lineages rather than refugial ones. There is scant information available in the literature on where the species in our dataset originated from, and further species-specific studies are required to identify signatures of hybridization/introgression. However, we have included this caveat in the revised version of the manuscript (line numbers: 73-78 and 303–305).

   (2) Lines 149-151 "However, in these species as well, trends of Ne increase or decrease from the LIG to LGM were robust across all three PSMC models considered." Please double-check this claim. Some of your figures in the supplement appear to contradict this. In particular, Pachycephala philippinensis, Rhynochetos jubatus, and Zosterops hypoxanthus appear to differ a bit in the time frame described, but it is difficult to tell-I would recommend adding some shading on the graphs to indicate the time period observed. If there was a way you determined this that is more precise than eyeballing the figures like I did, this should also be explained.

   We thank the reviewer for this comment and have reworded the sentence by cross verifying with the PSMC graphs. In addition, we have calculated the precise values of effective population size at the Last Interglacial (LIG) and Last Glacial Maximum (LGM) for each species using custom scripts and used these to evaluate whether the change in Ne during the Last Glacial Period (LGP) was significantly different for the three PSMC settings used. A table depicting these effective population size changes from LIG to LGM are also included in the revised version of the manuscript (Supplementary table S4; line numbers: 145­-156 and 345-357).

   (3) Lines 280-292: Issues with PSMC that are not acknowledged here are my largest concern. The situation being investigated does not necessarily meet all the assumptions PSMC makes (ie, neutral evolution and panmixia), which should be explained in this section. I'll point out the two issues I think should be acknowledged and addressed: First, selection is a confounding factor with PSMC, which is not mentioned here. While that's likely not an issue due to the size of the genome, this is still something that should be stated and explained. Second, the following statement is what I take the most issue with: "Population structure is thus a confounding factor. However, this is unlikely to be a problem given that all our species are single-island endemics". This needs justification. You state that in the past, islands could be connected (see my first comment regarding line 75), so it seems unlikely that 1) migration between past populations on other islands never happened, and 2) there is no population structure *on* the island.

   We thank the reviewer and have modified the PSMC caveats section of the revised version of the manuscript (line numbers: 289-307).

   (4) Line 310: Mapping all the data to the autosomes seems inappropriate to me. The sex chromosome reads could potentially map to other areas of the genome. Unless this information was accidentally left out of the methods section, it doesn't seem like any mapping quality filter was used, so spurious alignments aren't being removed. To remove sex chromosome data, I would instead align data to the whole genome, remove all reads that map to the sex chromosomes, and then map the remaining reads to the autosomes.

   As mentioned earlier, for a subset of the species (n =5), we directly mapped raw reads files onto the genome and then called SNPs on only autosomal regions using the SAMtools mpileup-bcftools pipeline, after which we performed PSMC as above (Supplementary Information Fig. S3). We did not observe and significant difference between the two approaches. Further, only high-quality mapped reads were used for SNP calling as mentioned in the previous version of the manuscript (line numbers: 338-343; Supplementary Information Fig. S3).

   (4) Table 1 includes some information that contradicts what is in the Supplementary Tables: Centropus unirufus, Chaetorhynchus papuensis and Cnemophilus loriae are not included in Supplementary Table 4. Table 1 says Eulacestoma nigropectus, Paradisaea rubra, and Parotia lawesii did not undergo PSMC analysis, but Supplementary Table 4 says PSMC and modeling trends matched for these species. "Pseudorectes ferrugineus" and "Rhynochetos jubatus" are spelled differently in Supplementary Table 4. Table 1 says Rhagologus leucostigma underwent both PSMC and climate modeling, but Supplementary Table 4 says "NA" as if it was missing one of these analyses.

   We thank the reviewer for identifying the errors and we have corrected for these in the revised version of the manuscript. Please see the detailed changes for these comments outlined below

   Centropus unirufus, Chaetorhynchus papuensis and Cnemophilus loriae are not included in Supplementary Table S4 (Now Supplementary table S2): we have added these species to the revised table S2.

   Table 1 says Eulacestoma nigropectus, Paradisaea rubra, and Parotia lawesii did not undergo PSMC analysis, but Supplementary Table 4 says PSMC and modeling trends matched for these species: The genomes for these samples were obtained from museums and exhibited high error rates. Hence, we excluded these samples from further analysis. However, the supplementary table S2 was not updated, and we have corrected this error in the revised version of the manuscript.

   "Pseudorectes ferrugineus" and "Rhynochetos jubatus" are spelled differently in Supplementary Table 4 (Now table S2): we have corrected the typographical error in the revised manuscript.

   Table 1 says Rhagologus leucostigma underwent both PSMC and climate modeling, but Supplementary Table 4 (Now table S2) says "NA" as if it was missing one of these analyses: This was a typographical error, and we have updated it to “mismatch”.

   Major Issues to Address:

   (1) Lines 97-99: "Information on tropical, single-island endemics' demographic responses to past climate change can inform conservation efforts, owing to the genomic signatures that predispose a species to extinction". This needs more explanation. For example, why couldn't we just look at these genomic signatures instead of recreating demographic responses? I'm not sure I fully understand what you mean here.

   We thank the reviewer for this comment and have modified the introduction to highlight the importance of demographic history in predicting species extinction. Comparison of genomic diversity and demographic history of over 200 mammalian genomes, highlights the importance of demographic history in predicting species endangerment and extinction risk (Wilder et al., 2023) (line numbers: 99-104).

   (2) Line 181-182: Whether or not a species was a passerine was an important predictor of Ne only in combination with the change in habitat from LIG to LGM". This is a major finding, but "respond positively to habitat change" (line 183) is a bit ambiguous. Were they responding to habitat expansion? Habitat contraction? Increase in rainfall? What is the change happening? Not all habitat changes are equal.

   We thank the reviewer for this comment and have modified this section for clarity in the revised results and discussion section of the manuscript. We observed a positive correlation between effective population size and availability of suitable habitat. Further, we observed precipitation of the warmest quarter to be the largest contributing bioclimatic variable for all but one Caribbean species (line numbers: 172-­191; 196-211).

   (3) Line 184-185: "The interaction between habitat change and body mass (β = 10.05, 95% CI: [-0.3, 24.41) suggests that there is no impact of habitat change in larger species." Doesn't this contradict the earlier finding of larger-bodied species seeing a decrease in Ne? Or do you mean the decrease in Ne was not due to habitat change?

   We have edited this section for clarity. With the inclusion of additional species, we observed a significant positive relationship between body size and effective population size (line number: 191-193).

   (4) Lines 206-207: "Our results also reveal that both passerine and non-passerine island endemics have entered the Holocene with low genetic diversity." How does this align with the statement that passerines responded positively to habitat change?

   The observation that passerines respond positively to habitat change is based on a systematic analysis of the last glacial period. However, a close look at the entire species’ demographic history reveals the often the Ne is at the lowest following the LGM, and coinciding with the advent of Holocene, the current interglacial. We have therefore modified the sentence in the revised version of the manuscript (line numbers: 213-214).

   (5) Line 215: If we already know flightless birds and endemics are particularly prone to extinction, what is the benefit of this study? Be clear about how your method can be used in a way that is better than what people are already doing. It would be good to explicitly explain why coalescent Ne is a better predictor of extinction risk than current genomic diversity or current Ne.

   We thank the reviewer for this comment and have modified this section in the revised version of the manuscript (line numbers: 221-224).

   (6) Line 259-261: "Habitat change in the LGP was positively associated with Ne fluctuations (Figure 3, β = 9.45), that is, species which showed an increase in habitat in the LGP also showed a concurrent increase in Ne." Is this true in all instances? I thought you found it had no effect for some, or did I misunderstand?

   We thank the reviewers for pointing this out. Species which showed an increase in habitat in the LGP did not always show a concurrent increase in Ne. Our results instead reflect an overall trend and this is clarified in the revised version of the manuscript (line numbers: 268-269).

   Lines 328-330: Could the different mutation rates used for passerines and non-passerines be driving the differences found between the two groups?

   The difference in the mutation rate is low and using the passerine specific mutation rate for non-passerines only shifts the PSMC graph slightly. As our analysis is considering the change in Ne across the LGP, this shift is minimal and does not affect the overall results.

   How are you connecting the demographic changes to species traits? I'm a bit confused about that, so I think some further explanation would be beneficial.

   We have modified the discussion to highlight the role of species traits in shaping the species response to habitat modification and ultimately the change in effective population size. We have included this in the revised version of the manuscript (line numbers: 437­-439).

   Minor Issues to Address:

   (1) Lines 165-168: "Habitat change was poorly associated with change in Ne for the 20 species for which both PSMC and ENM analyses were possible (Cramer's V = 0.15). However, passerine species only showed a strong association (Cramer's V = 0.96), while non-passerines showed a weak negative association (Cramer's V = -0.15)." This is phrased in a way that is a bit confusing. I'd consider rephrasing for clarity.

   We have modified this section in the revised version of the manuscript (line numbers: 167­-170).

   (2) Line 177: The confidence interval says "16.27, -2.61". I think it's supposed to be -16.27?

   We have corrected the typographical error in the revised version of the manuscript.

   (3) Line 185-187: "Finally, the random intercept for Country (sd (Intercept)) showed a marginal positive influence (β = 0.85, 95% CI: [0.04, 2.24])". What does this mean? This needs further explanation.

   We modified this sentence in the revised version of the manuscript (line number: 189-191).

   (4) Line 204: landbridge is misspelled as "landbride".

   We have fixed the typographical error.

   (5) Line 310: What were your Trimmomatic parameters?

   We have included the parameters used for Trimmomatic in the revised version of the manuscript (line numbers: 324-326).

   (6) Line 311: What were your bwa parameters?

   We used default parameters for bwa alignment and this is included in the revised version of the manuscript (line numbers: 328-329).

   (7) Line 322-324: Why did you choose those specific parameters for PSMC? Splitting up the first time window makes sense (as shown in Hilgers 2025), but why did you choose t=5, r=1, and 84 atomic time intervals? Did you choose these parameters independently, or did you decide to use them because they were used by Nadachowska-Brzyska et al? Either way, that information is important to state.

   The parameter selection followed the suggestions based on both Hilgers et al. 2025 and Nadachowska-Brzyska et al. 2016. The information is included in the revised version of the manuscript (line numbers: 345-350).

   (8) Lines 325-326: What did you use for bootstrapping? If not Psmcfa, why?

   We have used “splitfa” to generate files for bootstrap analysis and have included this information in the revised version of the manuscript (line numbers: 350-351).

   (9) Lines 350-354: Please explain the reasoning behind using the different resolution and worldclim for Amazona guildingii.

   Based on the reviewer’s comment, we have re-run the habitat model with the same resolution for Amazona guildingii and include this in the revised version of the manuscript.

   (10) Line 412-413: "For the response variable i.e., the change in Ne, a Bernoulli distribution with a logit link because it is a binary response variable." I think this sentence might be missing some words.

   We have fixed the typographical error in the revised version of the manuscript (line numbers: 444-445).

   (11) Figure 1 is difficult to read, especially the top left panel. I would consider presenting this differently.

   We have supplemented Figure 1 with boxplots of effective population size values estimated during the Last Interglacial and the Last Glacial Maximum which should aid in clarity.

   Reviewer #2 (Recommendations for the authors):

   The authors state that they intentionally chose to remove several avian species that would be suitable for this analysis, because they were subject to larger studies elsewhere. This seems like an unnecessary constraint, and it is my opinion that the authors need to add this data in. I am not aware of what species were excluded, but I hope this will increase the non-passerine proportion of their dataset to help them robustly address their questions. An alternative solution would be for the authors to only include passerines, but this will come at the expense of statistical power with the current dataset and so would also require an increase in sample size. Overall, I recommend including more non-passerine species with traits similar to your passerine species.

   This was a typographical error from the previous versions of the manuscript arising from the fact that we excluded museum species from our samples. We have modified this sentence in the revised version of the manuscript as well as included one new species (Melanocharis versteri) in our study panel (line number: 311-314).

   It was not clear how or if PSMC bootstrapping was included in the comparisons across species, i.e. how did you include bootstrapping when you turned PSMC into a response variable within your statistical analysis? Failing to account for it would introduce measurement error into the data, and I would suggest that the authors explore how to incorporate this.

   We thank the reviewer for this comment and have calculated the precise values of effective population size during the LIG and the LGM using custom scripts to generate boxplots. These boxplots were used to investigate if effective population size values were significantly different during the LGP for all three PSMC parameter settings. Non-significant results were treated as “no change” in effective population size for further statistical analyses. The bootstrap values were used for this analysis, in addition to circumventing the issue of selection on the genome.

   I would also like to see a greater discussion on what aspects of the PSMC curve were used for comparisons and the limitations therein. These cross-species comparisons are still relatively new, and I think they will add value to this paper.

   In our study, the change in Ne from LIG to LGM is considered. We have elaborated this in the revised version of the manuscript. Addition analysis, depicting the changes in Ne as box plots were also included to help understand the fluctuations in Ne.

   Lines 164-168, which refer to your core hypothesis, are really unclear. What was actually found here? Please rephrase.

   We have rephrased the sentence for clarity in the revised version of the manuscript (line numbers: 169­-172).

   PSMC measures effective population size, not genetic diversity. Please change throughout.

   Based on the reviewer’s comment we have changed this in the revised version of the manuscript.

   I was surprised to see some references to conservation within the abstract of the paper. It is important that this is also included in the discussion so that the authors ensure their logic is accessible to managers. It would also be good to discuss the risks of using PSMC to inform conservation from just one genome, as I see these being quite high.

   We thank the reviewer for this comment and have included both pros and cons of using PSMC in the revised version of the manuscript (line numbers: 229-237).

   As this paper is based on public reference genomes, it is best practice that the original notes or reference genome papers are cited to acknowledge the data holders.

   We thank the reviewer for this comment and have included a supplementary table (Supplementary Table S7) acknowledging all the data holders.

   Read the original source
7. 

   eLife

   Jun 27, 2025

   eLife Assessment

   Tropical single-island endemic bird populations are particularly vulnerable to climate change. The authors investigate genetic evidence of how such species dealt with climate changes in the past as a possible predictor for how they will respond to change in the future, which could provide an important example for the fields of conservation genetics and island biogeography. The authors' integration of genomics and habitat modeling is commendable, but we find that the support for their conclusions is incomplete: at times, the results presented appear to contradict each other, the authors do not fully account for key variables, and the limited taxonomic scope may cause problematic biases for the conclusions.

   Read the original source
8. 

   eLife

   Jun 27, 2025

   Reviewer #1 (Public review):

   Summary:

   The authors combine PSMC and habitat modeling to try to connect habitat change during the Last Glacial Period to changes in Ne.

   Strengths:

   Observing how tropical single-island endemic bird species responded to habitat change in the past may help inform conservation interventions for these particularly vulnerable species. The combination of genomics and habitat modeling is a good idea - this sort of interdisciplinary thinking is what is needed to tackle these complex questions. Additionally, the use of PSMC makes it possible to perform this analysis on poorly-studied species with only a single genome available.

   Room for Improvement:

   Why coalescent Ne is a better predictor of extinction risk than current genomic diversity, or current Ne, isn't explicitly explained. PSMC in particular has many caveats, …

   Reviewer #1 (Public review):

   Summary:

   The authors combine PSMC and habitat modeling to try to connect habitat change during the Last Glacial Period to changes in Ne.

   Strengths:

   Observing how tropical single-island endemic bird species responded to habitat change in the past may help inform conservation interventions for these particularly vulnerable species. The combination of genomics and habitat modeling is a good idea - this sort of interdisciplinary thinking is what is needed to tackle these complex questions. Additionally, the use of PSMC makes it possible to perform this analysis on poorly-studied species with only a single genome available.

   Room for Improvement:

   Why coalescent Ne is a better predictor of extinction risk than current genomic diversity, or current Ne, isn't explicitly explained. PSMC in particular has many caveats, and some are not acknowledged or adequately addressed by the authors. For example, the authors note that population structure is a confounding factor with PSMC, but that it is not a problem in this instance. They do not provide compelling evidence for why this would be the case, they simply state that the species studied are all single-island endemics. However, single-island endemic species are not necessarily panmictic; this is even less likely to be true for species studied here that inhabit a large geographic area (ie, Australian species). Differing PSMC parameters may also impact results: the differences between passerines and non-passerines were one of their main results, but they do not provide any analysis to show that this difference was not driven by the different mutation rates used for the two groups.

   Parameters for many steps are not described, and choices that are described (such as the PSMC parameters) are not always fully explained. It is unclear why all data was mapped to the autosomes rather than removing reads that map to the sex chromosomes first. Using all the data, the reads belonging to the sex chromosomes could potentially map to other areas of the genome. It does not seem like a mapping quality filter was used, so these potential spurious alignments would not have been removed prior to analysis.

   There are points where the results are described in ways that appear to potentially differ from the supplementary figures. The authors state that even for species where PSMC results differed between models, "trends of Ne increase or decrease from the LIG to LGM were robust across all three PSMC models considered." The figures in the supplement for Pachycephala philippinensis, Rhynochetos jubatus, and Zosterops hypoxanthus appear to potentially contradict this statement, but it is difficult to tell, as the time period observed is not clearly marked on the graphs. How this robustness of trends was determined is not explained, leaving the precision of the analysis unclear.

   Table 1 also includes some information that contradicts what is in the Supplementary Tables, leading to a lack of clarity. Centropus unirufus, Chaetorhynchus papuensis, and Cnemophilus loriae are not included in Supplementary Table 4. Table 1 says Eulacestoma nigropectus, Paradisaea rubra, and Parotia lawesii did not undergo PSMC analysis, but Supplementary Table 4 says PSMC and modeling trends matched for these species. Table 1 says Rhagologus leucostigma underwent both PSMC and climate modeling, but Supplementary Table 4 says "NA" as if it was missing one of these analyses.

   Additionally, some of the results appear to contradict each other. For example, they show that there is no impact of habitat change in larger-bodied species, but also that larger-bodied species saw a decrease in Ne during the LGP. In another example, they state that when a species saw an increase in habitat during the LGP, they also had an increase in Ne. However, they also state that this was not the case for non-passerines.

   Ecosystems are highly complex; there may also be other variables influencing past demographic change other than those explored here. Results should be interpreted with caution.

   Read the original source
9. 

   eLife

   Jun 27, 2025

   Reviewer #2 (Public review):

   Summary and strengths:

   In this manuscript, Karjee and colleagues used coalescent-based effective population size reconstruction (PSMC) from single genomes to understand past population trends in island birds and related this to life history traits and glacial patterns. This concept is fairly new, as there are still relatively few multiple PSMC synthesis studies. I also thought that the focus on island endemics was unique and adds value to this paper. I enjoyed seeing a paper focused on South East Asia and think that this could help contribute to our knowledge of the important biodiversity within this region.

   Major weaknesses:

   My biggest concern with this paper is that the analyses are limited to 20-30 species, and significant taxonomic bias is present (there are multiple species of passerine but only 1-2 …

   Reviewer #2 (Public review):

   Summary and strengths:

   In this manuscript, Karjee and colleagues used coalescent-based effective population size reconstruction (PSMC) from single genomes to understand past population trends in island birds and related this to life history traits and glacial patterns. This concept is fairly new, as there are still relatively few multiple PSMC synthesis studies. I also thought that the focus on island endemics was unique and adds value to this paper. I enjoyed seeing a paper focused on South East Asia and think that this could help contribute to our knowledge of the important biodiversity within this region.

   Major weaknesses:

   My biggest concern with this paper is that the analyses are limited to 20-30 species, and significant taxonomic bias is present (there are multiple species of passerine but only 1-2 representatives of other groups). While this is not an issue alone, many of the life history traits or geographical traits are conflated with phylogenetic diversity (e.g., there are no large-bodied passerines). Thus, it is my opinion that the impact of these drivers of past population size is conflated and cannot be disentangled with the current data. The authors themselves state that the core hypothesis surrounding Ne and habitat availability is not supported by their entire dataset (only seen in Passerines). This was not clear enough in the abstract, and conclusions cannot be drawn here as the impact of taxonomy cannot be separated from data richness, traits, etc. The PSMC analysis was done according to the most recent recommendations, and this part of the manuscript is fairly robust. However, in several places, it is incorrectly stated that the PSMC measures or can infer genetic diversity; PSMC only infers past effective population size. It cannot measure genetic diversity in the past. I cannot review the habitat reconstruction modelling as I am a conservation genomics specialist.

   Appraisal:

   I am not convinced about the findings within the paper. I do not think that the results are sufficiently supported at this time, largely due to the conflation of taxonomy with other variables. As this type of comparison is new, I do think that there is a chance for reasonable impact on the field of genomics and island biogeography if the manuscript's constraints are addressed. I do not see scope for impact on conservation at this time and find the conclusions in the abstract regarding conservation relevance to be unfounded.

   Read the original source
10. 

    eLife

    Jun 27, 2025

    Author response:

    We thank the editors and the reviewers for their positive comments regarding our manuscript and the methodological approach we have taken to understand the historical demographic response of endemic island birds to climate change. We acknowledge the issues of uneven sample sizes and plan to include additional species of island endemic birds for which genomic data is now available. As requested by reviewer 1, we will also address the issues related to the PSMC analysis in the revised version of the manuscript.

    Read the original source
11. 

    Version published to 10.7554/elife.106369.1 on eLife

    Jun 27, 2025
12. 

    Version published to 10.1101/2024.11.14.623644 on bioRxiv

    Nov 15, 2024