Building a DNA Reference for Madagascar’s Marine Fishes: Expanding the COI Barcode Library and Establishing the First 12S Dataset for eDNA Monitoring

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript describes a comprehensive reference database for COI and 12S sequences of marine fish from Madagascar. The sampling covered different gear and seasons. The analyses are appropriately, figures and tables are well presented (see minor exceptions below). The underlying dataset is very valuable for future work in the whole Western Indian Ocean and a significant contribution to the scarce public available genetic data for Africa. Currently BOLD is lacking in particular 12S references for the biodiversity rich Western Indian Ocean. This is an excellent contribution to address this issue. There are minor revisions needed, mandatory in particular within the method and result sections.

Abstract

Line 31: To me it is confusing that there is a general remark on “The COI database”. I think the MADFI container or “Malagasy COI database” should be named.

Lines 36-38: Please state clearly if here taxonomic assignment is not possible from the sequence data or if the this is a missing data issue.

Lines 38,39: Maybe it would be better to explain the approach as “Among the five metabarcodes evaluated” . The current wording comes a bit out of the blue and may benefit from an introductory wording such as "The new 1000bp 12S sequences were used to evaluate five commonly used 12S metabarcoding marker …"

From the abstract it is not clear to me how many new data come along with this publication and what was there before. From table 1 I learn that some tissues were collected for other studies. Were these only tissues or sequences deposited with COI only and supplemented by 12S, or were COI + 12S data already published and were merged with the new sequences for a comprehensive regional database? I think it would be better to clearly describe what this study contributes and how it was supplemented from existing MADFI container data.

Introduction

Lines 70-77: Please rephrase as I am not sure if eDNA emerged as tool for Malagasy fish diversity but if this is rather a general trend to involve eDNA in fisheries or MPA monitoring (Jacquemot et al. 2024; Sahu et al. 2025)? Also, I am not sure if eDNA is a superior method but rather complementary to more traditional monitoring, but this is my personal opinion.

Material and Methods

Line 127: This sampling was designed ….

Line 138 ff Table 1: Why is the doi link within the table? The citation would be sufficient. Are these samples stored somewhere? You could add storage information here or refer to BOLD in the legend to find the information there. You could also specify how the dataset can be found.

Line 144: MOTU in brackets.

Lines 145-147: It is difficult to understand what was downloaded and what is new data from this study, please specify.

Line 158: I do not understand “This facilitated the subsequent matching of tissue samples with their morphospecies during DNA barcoding.” Please rephrase or delete.

Line 199: See separate file for Fig 1

Lines 238-254: As the method description is not yet published it is not appropriate to reference the detailed workflow without explanation. Please give the software or what kind of similarities (distance metrics) were used to cluster sequences. To me it is hard to understand what was done how. (Or ask the Ruiz et al. submitted to provide a copy on a preprint server.)

Give somewhere the length of each 12S marker (and arrangement?) used in the inference on taxonomic resolution.

Results

Table2: Please improve layout. It is very difficult to locate the header throughout the table. Despite this is core data for the study, I suggest moving this table to supplemental material as the 19 pages table disrupt the flow of the manuscript. It might be possible to summarize this information within the main manuscript on family level (but if you do so, please do not duplicate information in tables and figures, eg Fig. 3b.) Also, I am not sure if the current formatting matches expectations of the journal.

Line 276: Sp. complex: Y (Yes) indicates ….

Line 290, Fig.3c: Unknown species no= 8, but 7 are mentioned in line 337

Line 292: Abbreviations need full name in figure legend if I understood the journals recommendations (MADFI)

Line 301: Could somewhere the number of BINs be explained? Here and Fig4(c) 502 is mentioned, 488 BIN's is mentioned for the MADFI project (line 319, 328). What are the other 14 BIN's?

Line 310: For better readability I suggest to reduce text within the figure x-axes and move parts to the legend. (eg “Valid WoRMS species”) and in the legend refer to 424 species for panel a,b,d,e,f. etc. This is not a mandatory request.

Line 312: Refer to the panel (a in this case) and not to the position (Top left)

Line 337: see line 290 above.

Line 366: What are “complete fragments”? Either a gene is complete or a fragment. Maybe you should delete or somewhere have a figure explaining your PCR products and the five markers.

Line 393, Fig5: X-axes labeling is missing! (You could use the empty panel to analyse the Leray COI fragment but I have no idea how demanding these analyses are (see comments in the method description 2.5)

Line 404: 12 S has loop regions which can evolve rapidly whereas COI more stringent due to the coding of amino acids with just the 3^rd position allowing for flexibility. This I am not sure if a COI BIN captures the correct evolutionary unit while 12S cannot. However, it is true that BIN is the established measure, but I suggest another wording such as “mismatch” instead of “error”. (Not mandatory)

Line 419: Please do not use red and green, but other color contrast such as green/magenta or red/cyan

Line 422: Could you rephrase “Each percentage represents the optimal similarity threshold for each primer”

 

Discussion

You could highlight that the new data also allow for testing primer binding behavior. However, but I think in general the Discussion is rather extensive as that I miss a lot. Thus, please only add something when something else is deleted.

Line 530: Why BIN is explained here? Please check throughout the manuscript if abbreviations are written out at the position when they appear for the first time.

 

References

15. DOI link does not work. This might be better: https://link.springer.com/article/10.1007/s00227-005-0090-6

 

Jacquemot, L., B. P. V. Hunt, S. Li, A. D. Schulze, C. M. Deeg, B. J. G. Sutherland, A. Tabata, C. Lovejoy & K. M. Miller, 2024. Mapping Biodiversity Coast-to-Coast-to-Coast Across Canada's Three Oceans Using eDNA Metabarcoding. Environmental DNA 6(6):e70028 doi:https://doi.org/10.1002/edn3.70028.

Sahu, A., M. Singh, A. Amin, M. M. Malik, S. N. Qadri, A. Abubakr, S. S. Teja, S. A. Dar & I. Ahmad, 2025. A systematic review on environmental DNA (eDNA) Science: An eco-friendly survey method for conservation and restoration of fragile ecosystems. Ecological Indicators 173:113441 doi:https://doi.org/10.1016/j.ecolind.2025.113441.

 

 

Comments for author File: Comments.pdf

Author Response

Comment 1: [Line 31: To me it is confusing that there is a general remark on “The COI database”. I think the MADFI container or “Malagasy COI database” should be named.]

Response 1: [Thank you for pointing this out. We agree with this comment. We have replaced the expression with “Malagasy COI database” in the manuscript.]

Comment 2: [Lines 36-38: Please state clearly if here taxonomic assignment is not possible from the sequence data or if the this is a missing data issue.]

Response 2: [Thank you for this comment. For clarification, the fact that some species are present only in one of the COI or 12S datasets results from a missing data issue. Depending on the samples, some amplifications were successful with the COI primer pair but failed with 12S, and vice versa. This lack of data is therefore related to the variable efficiency of PCR amplification.]

Comment 3: [Lines 38,39: Maybe it would be better to explain the approach as “Among the five metabarcodes evaluated”. The current wording comes a bit out of the blue and may benefit from an introductory wording such as "The new 1000bp 12S sequences were used to evaluate five commonly used 12S metabarcoding marker …"]

Response 3: [Thank you for pointing this out. We agree with this comment. We have modified the sentence as followed: “For 319 species with complete 12S gene sequences associated with COI BINs were used to evaluate the taxonomic resolution of five 12S metabarcodes.”]

Comment 4: [From the abstract it is not clear to me how many new data come along with this publication and what was there before. From table 1 I learn that some tissues were collected for other studies. Were these only tissues or sequences deposited with COI only and supplemented by 12S, or were COI + 12S data already published and were merged with the new sequences for a comprehensive regional database? I think it would be better to clearly describe what this study contributes and how it was supplemented from existing MADFI container data.]

Response 4: [Thank you for this insightful comment. To clarify the contribution of this study, we would like to specify that the MADFI (Reference DNA barcodes library of Malagasy marine fishes)’s BOLD project currently includes 2,146 COI sequences, of which 953 were newly generated during the present study (as mentioned lines 302-303 in the revised ms). The remaining COI sequences were obtained from previous studies (those mentioned in table 1) and integrated to build a comprehensive and regionally representative reference database.

Regarding the 12S marker, all sequences used in this study are newly generated and have never been published before. These original 12S sequences represent 446 species (mentioned line 414 in the revised ms), and constitute a major advancement toward the development of a robust reference library for eDNA metabarcoding applications in the region.

This study thus provides a significant contribution by combining newly generated and existing COI data with a completely novel 12S dataset, offering, for the first time within the MADFI framework, a dual-marker reference database to support molecular biodiversity assessment and taxonomic identification in marine ecosystems.]

Comment 5: [Lines 70-77: Please rephrase as I am not sure if eDNA emerged as tool for Malagasy fish diversity but if this is rather a general trend to involve eDNA in fisheries or MPA monitoring (Jacquemot et al. 2024; Sahu et al. 2025)? Also, I am not sure if eDNA is a superior method but rather complementary to more traditional monitoring, but this is my personal opinion.]

Response 5: [We thank the reviewer for the suggestion. The sentence has been rephrased to reflect that eDNA is part of a global trend in biodiversity monitoring and is used as a complementary, not superior, tool to traditional methods. See lines 72 to 78 in the revised ms:

“In line with a global trend toward the integration of molecular tools into biodiversity monitoring, environmental DNA (eDNA) has increasingly been used as a complementary approach to traditional survey methods, including in the context of fisheries and marine protected area (MPA) monitoring (e.g. [9,10] ). In Madagascar, eDNA offers promising opportunities to improve our understanding of fish diversity and community structure across coastal habitats by detecting species through genetic material shed into the environment via secretions, excretions, tissues, or carcasses.”]

Comment 6: [Line 127: This sampling was designed ….]

Response 6: [Thank you for pointing this out. We agree with this comment, and we have corrected the text.]

Comment 7: [Line 138 ff Table 1: Why is the doi link within the table? The citation would be sufficient. Are these samples stored somewhere? You could add storage information here or refer to BOLD in the legend to find the information there. You could also specify how the dataset can be found.]

Response 7: [Thank you for pointing this out. We agree with the suggestion and have removed the DOI link from the table, keeping only the bibliographic citation. Regarding the storage location of the samples, we have added in the legend that the projects mentioned in the table are publicly available through the BOLD System v5, where detailed information about sample storage and metadata can be accessed.]

Comment 8: [Line 144: MOTU in brackets.]

Response 8: [Thank you for pointing this out. We agree with this comment, and we have corrected the text.]

Comment 9: [Lines 145-147: It is difficult to understand what was downloaded and what is new data from this study, please specify.]

Response 9: [Thank you for pointing this out. We agree with this comment, and we rephrased the sentence as follows to improve clarity:

Line 156 to 160 “All 12S sequences generated in this study are original and unpublished. They were obtained from two sources of DNA: (i) DNA extracts previously used in earlier studies to generate COI sequences [24,30], and (ii) newly extracted DNA from specimens collected specifically for the present study to expand the COI database. »]

Comment 10: [Line 158: I do not understand “This facilitated the subsequent matching of tissue samples with their morphospecies during DNA barcoding.” Please rephrase or delete.]

Response 10: [Thank you for pointing this out. We agree with this comment, and we have deleted this sentence.]

Comment 11: [Line 199: See separate file for Fig 1]

Response 11: [We did not find any separate file for Figure 1 attached to the review comments, or we may have misunderstood the referee’s request. Could you please clarify or resend the file if necessary?]

Comment 12: [Lines 238-254: As the method description is not yet published it is not appropriate to reference the detailed workflow without explanation. Please give the software or what kind of similarities (distance metrics) were used to cluster sequences. To me it is hard to understand what was done how. (Or ask the Ruiz et al. submitted to provide a copy on a preprint server.)]

Response 12: [Due to the lack of reviewer, this paper submitted one year ago still did not got published even though the promising first revision round gives us hope that this article will be published soon. The paper being currently reviewed, it will not be possible to publish a preprint with in-depth analysis. However, trying to stay concise, we detailed the workflow methodology with further details and methodologies used, notably precising the method using to obtain similarities. See lines 265 to 269 & lines 274 to 278.]

Comment 13: [Give somewhere the length of each 12S marker (and arrangement?) used in the inference on taxonomic resolution.]

Response 13: [Thank you for pointing this out. We agree with this comment, and we have inserted the length of the 12S markers in the revised ms lines 259 to 260.]

Comment 14: [Table2: Please improve layout. It is very difficult to locate the header throughout the table. Despite this is core data for the study, I suggest moving this table to supplemental material as the 19 pages table disrupt the flow of the manuscript. It might be possible to summarize this information within the main manuscript on family level (but if you do so, please do not duplicate information in tables and figures, eg Fig. 3b.) Also, I am not sure if the current formatting matches expectations of the journal.]

Response 14: [Thank you for your valuable comment regarding Table 2. To improve the readability and flow of the manuscript, we have moved Table 2 to the end of the document, just before the references section. This allows us to keep the complete dataset accessible to readers without interrupting the main text.]

Comment 15: [Line 276: Sp. complex: Y (Yes) indicates ….]

Response 15: [Thank you for pointing this out. We agree with this comment, and we have corrected the text.]

Comment 16: [Line 290, Fig.3c: Unknown species no= 8, but 7 are mentioned in line 337]

Response 16: [Thank you for this relevant comment. For clarification, in Figure 3c, the 165 BINs correspond to complex identification cases that could not be directly resolved through sequence comparison in BOLD and required phylogenetic analysis. Among these 165 BINs, phylogenetic reconstruction revealed that 8 BINs correspond to unknown species (identified only at the genus level). Of these 8 BINs, 7 are associated with a single species name in BOLD, while 1 BIN is associated with multiple species names. In line 380 of the revised version, we refer only to the BINs associated with a single species name in BOLD.]

Comment 17: [Line 292: Abbreviations need full name in figure legend if I understood the journals recommendations (MADFI)]

Response 17: [Thank you for pointing this out. We agree with this comment, and we have corrected the text.]

Comment 18: [Line 301: Could somewhere the number of BINs be explained? Here and Fig4(c) 502 is mentioned, 488 BIN's is mentioned for the MADFI project (line 319, 328). What are the other 14 BIN's?]

Response 18: [The MADFI project comprises a total of 502 BINs. Of these, 14 were newly generated in the present study and are not yet registered in BOLD, while the remaining 488 BINs were already available in the BOLD database. We partially rewrote the paragraph (lines 336–342) to improve clarity as follows:

: “Of the 502 BINs identified in this study, 488 correspond to BINs already registered in the BOLD database as part of the MADFI project, and 14 are newly generated BINs not yet present in BOLD. Among the 502 BINs, 437 were identified at the species level, including 133 cryptic species, 63 at the genus level, and two at the family level (Figure 3b and 3c). Among the 488 BINs already registered in BOLD, 26 BINs are newly recorded for the West Indo-Pacific, and 290 BINs for Madagascar (Figure 4c).”]

Comment 19: [Line 310: For better readability I suggest to reduce text within the figure x-axes and move parts to the legend. (eg “Valid WoRMS species”) and in the legend refer to 424 species for panel a,b,d,e,f. etc. This is not a mandatory request.]

Response 19: [Thank you for pointing this out. We agree with the suggestion. For better readability, we have removed the text from the X-axes and moved it to the legend of Figure 4.]

Comment 20: [Line 312: Refer to the panel (a in this case) and not to the position (Top left)]

Response 20: [Thank you for pointing this out. We agree with this comment, and we have replaced positional references with panel letters in each figure.]

Comment 21: [Line 337: see line 290 above.]

Response 21: [Thank you for this relevant comment. As we explained in the response to the comments on line 290 (line 324 of the revised version), , in Figure 3c, the 165 BINs correspond to complex identification cases that could not be directly resolved through sequence comparison in BOLD and required phylogenetic analysis. Among these 165 BINs, phylogenetic reconstruction revealed that 8 BINs correspond to unknown species (identified only at the genus level). Of these 8 BINs, 7 are associated with a single species name in BOLD, while 1 BIN is associated with multiple species names. In line 380 of the revised version, we refer only to the BINs associated with a single species name in BOLD.]

Comment 22: [Line 366: What are “complete fragments”? Either a gene is complete or a fragment. Maybe you should delete or somewhere have a figure explaining your PCR products and the five markers.]

Response 22: [Thank you for the observation. We agree that the term "complete fragments" may be misleading and have clarified the text accordingly. In our 12S barcode library, most individuals yielded near full-length 12S sequences (~1000 bp) using long-range primers. However, for some individuals, likely due to mutations at primer binding sites or DNA degradation, amplification failed with the full-length primers. In these cases, we used alternative primers targeting a shorter internal region of approximately 600 bp to recover usable sequence data. We have updated the manuscript to clarify this point and to avoid ambiguity regarding the term "complete." See lines 409 to 412 of the revised version:

“The 12S barcode library comprises 524 new sequences, of which 89% correspond to near full-length 12S rRNA sequences (~1000 bp), obtained with long-range primers. The remaining 11% are shorter sequences (~600 bp), generated using internal primers when full-length amplification failed, likely due to mutations at primer binding sites or DNA degradation”]

Comment 23: [Line 393, Fig5: X-axes labeling is missing! (You could use the empty panel to analyse the Leray COI fragment but I have no idea how demanding these analyses are (see comments in the method description 2.5)]

Response 23: [Thank you for pointing this out. We have added the missing X-axis labels to Figure 5. Regarding the suggestion to include the Leray COI fragment in our analysis, we fully agree that it is an interesting and relevant idea. However, due to the short timeframe for this revision (one week), we were unable to perform the necessary additional analyses. Moreover, the Leray primers were not specifically designed for fish identification, and their taxonomic resolution is likely lower than that of the other primers included in our study, which were developed with fish as the target taxonomic group. Including this marker would also have required substantial changes to the Introduction and Materials & Methods sections. We nonetheless consider this an important direction for future work.]

Comment 24: [Line 404: 12 S has loop regions which can evolve rapidly whereas COI more stringent due to the coding of amino acids with just the 3^rd position allowing for flexibility. This I am not sure if a COI BIN captures the correct evolutionary unit while 12S cannot. However, it is true that BIN is the established measure, but I suggest another wording such as “mismatch” instead of “error”. (Not mandatory)]

Response 24: [Thank you for pointing this out. We have carefully considered your suggestion to use a term like “mismatch” instead of “error.” However, we have chosen to retain the expression “discrimination error,” as it explicitly refers to cases of incorrect clustering (false positives) or unjustified splitting (false negatives) relative to the BIN reference.]

Comment 25: [Line 419: Please do not use red and green, but other color contrast such as green/magenta or red/cyan]

Response 25: [Thank you for pointing this out. We agree with this comment. We have modified the colors of this figure in accordance with your suggestion.]

Comment 26: [Line 422: Could you rephrase “Each percentage represents the optimal similarity threshold for each primer”]

Response 26: [Thank you for pointing this out. We agree with this comment. We have rephrased this sentence.]

Comment 27: [You could highlight that the new data also allow for testing primer binding behavior. However, but I think in general the Discussion is rather extensive as that I miss a lot. Thus, please only add something when something else is deleted.]

Response 27: [Thank you for this suggestion. We acknowledge the importance of balancing clarity and conciseness. However, after careful consideration, we chose not to add further content or shorten the Discussion. We believe the current version includes the essential elements necessary to support and contextualize our findings.]

Comment 28: [Line 530: Why BIN is explained here? Please check throughout the manuscript if abbreviations are written out at the position when they appear for the first time.]

Response 28: [Thank you for pointing this out. We agree with this comment. The acronym BIN has been removed at this point. After reviewing the entire manuscript, it is properly defined at its first occurrence.]

Comment 29:[

1. DOI link does not work. This might be better: https://link.springer.com/article/10.1007/s00227-005-0090-6 ]

 

Response 29: [Thank you for pointing this out. We agree with this comment. The DOI link has been replaced with the one you proposed.]

Reviewer 2 Report

Comments and Suggestions for Authors

This paper presents important data-driven and methodological research findings, providing valuable references for eDNA-based fish diversity monitoring in Madagascar, a hotspot of marine fish diversity. This paper demonstrates a clear theoretical framework and in general carefully written. It is recommended for publication pending minor revisions. Below are specific suggestions for improvement:

1. Line 31: The phrase "since 2018" should be revised to "from 2018 to 2023".
2. Introduction Section: The current structure contains an excessive number of paragraphs. Consider streamlining and consolidating them for better coherence.
3. Line 167: The citation should be presented in-text as "by Ward et al. [31]", consistent with the formatting of citations [30] and [34] in Line 184.
4. Line 238: The cited paper (submitted) is not recommended for inclusion. Please replace it with more appropriate references.
5. Figure 2 and Legend: The definitions of false-negative and false-positive errors are incorrectly reversed. Please verify and correct this.
6. Table 2: The table is excessively lengthy (spanning 19 pages) and would be better relocated to the supplementary materials.
7. Subtitles 3.1 and 3.3: The current titles are imprecise. Suggest revising them to "Taxonomic Coverage of COI Sequences" and "Taxonomic Coverage of 12S Sequences", respectively.
8. Line 345: The statement "Among 33 Bins, 25 were identified at the genus level and 12 at the species level" contains a numerical discrepancy. Please check the counts.

Author Response

Comment 1: [Line 31: The phrase "since 2018" should be revised to "from 2018 to 2023".]

Response 1: [Thank you for pointing this out. We agree with this comment. The sentence has been corrected and now begins with “from 2018 to 2023.]

Comment 2: [Introduction Section: The current structure contains an excessive number of paragraphs. Consider streamlining and consolidating them for better coherence.]

Response 2: [Thank you for pointing this out. We agree with this comment. To improve coherence, I merged paragraphs 1 and 2 as well as paragraphs 4 and 5 in the Introduction section.]

Comment 3: [Line 167: The citation should be presented in-text as "by Ward et al. [31]", consistent with the formatting of citations [30] and [34] in Line 184.]

 

Response 3: [Thank you for pointing this out. We agree with this comment. These citations have been modified according to the requested format.]

Comment 4: [Line 238: The cited paper (submitted) is not recommended for inclusion. Please replace it with more appropriate references.]

Response 4: [Thank you for the recommendation. Thank you for the recommendation. As the cited manuscript (now submitted) is the first to describe this new method, no published alternative is currently available. However, to address this concern, we have provided additional methodological details in the revised version of the manuscript to ensure clarity and reproducibility.]

Comment 5: [Figure 2 and Legend: The definitions of false-negative and false-positive errors are incorrectly reversed. Please verify and correct this.]

Response 5:  [Thank you for pointing this out. We agree with this comment. Indeed, false positives typically correspond to the splitting of a single taxon, while false negatives correspond to the merging of different taxa.]

Comment 6: [Table 2: The table is excessively lengthy (spanning 19 pages) and would be better relocated to the supplementary materials.]

Response 6: [Thank you for your suggestion. As Table 2 represents the core dataset of our study, we believe it is important to retain it within the main text. However, to improve the manuscript's readability and flow, we have moved it to the end of the document, just before the references.]

Comment 7: [Subtitles 3.1 and 3.3: The current titles are imprecise. Suggest revising them to "Taxonomic Coverage of COI Sequences" and "Taxonomic Coverage of 12S Sequences", respectively.]

Response 7: [Thank you for pointing this out. We agree with this comment. The titles of sections 3.1 and 3.3 have been revised as recommended.]

Comment 8: [Line 345: The statement "Among 33 Bins, 25 were identified at the genus level and 12 at the species level" contains a numerical discrepancy. Please check the counts.]

Response 8: [Thank you for pointing this out. We agree with this comment. After verification, the sentence has been corrected. I noticed a numerical inconsistency: I had written 12 instead of 8.]