Genetic Diversity of the Critically Endangered Big-Headed Turtle (Platysternon megacephalum) Based on Wild and Traded Samples: Implications for Conservation

Round 1

Reviewer 1 Report

REVIEW of “Genetic diversity of the Critically Endangered Big-headed Turtle (Platysternon megacephalum) based on wild and traded samples: implications for conservation”

The submitted study describes the results from an extensive genetic analysis using sequences from the NADH Dehydrogenase 4 mitochondrial gene derived from a very substantial survey of more than 450 samples of the critically endangered big-headed turtle from Vietnam. These were used to assess the levels of genetic diversity of the species in Vietnam and address two main concerns of relevance to conservation programs, as stated by the authors: assess the genetic diversity of the species in Vietnam (confiscated from trade, and those sampled from known localities) to eventually be able to identify and select correct sites into which return seized animals. This desirable outcome of genetic studies are reflected in a note the authors include in the introduction: “Detailed knowledge of phylogeographic patterns of threatened species is essential when developing ex-situ conservation programs in order to maintain the genetic integrity of captive populations and to minimize the risks of genetic pollution to wild populations when the animals are released”. The relevance to conservation of the study is that this, like many other endangered turtle species, are widely extracted unsustainably in the region and, whenever traffiqued animals are confiscated, the genetic identity is necessary to guarantee that the turtles are returned to their correct population of origin.

 Using ND4, the genetic analyses provide a robust support for the presence of at least two subspecies (Platysternon megacephalum megacephalum and Platysternon megacephalum peguense). A third subspecies (Platysternon megacephalum shiui), at least on the basis of the samples from the presumed range for this taxon, remains of doubtful validity because it was found to have low genetic differentiation with respect to P. m. megacephalum and with inconsistent morphologically distinguishable features.

While these results are an important contribution providing an improved understanding of the phylo-geographic patterns for the species in Vietnam, I have two major reservations. One is that the molecular data (the sequences) have not been sufficiently processed to ascertain the degree of genetic variation within each subspecies category. Second, there has been a scant effort to analyze for possible insights on the “geographic” distribution of the resolved genetic variation.

As a respectful suggestion to address my first issue, I have outlined how the authors could collapse redundant ND4 sequences for each taxon into a catalog of unique haplotypes using appropriate nomenclature and then use these to quantify levels of variation. Essentially this scheme is what was employed in one publication referenced in the current study: Kundu et al., 2022 reference [63].  The establishment of a haplotype catalog would also greatly aid future studies using the same molecular marker, as sequences would be comparable if everyone used the same standard haplotype lengths and nomenclature.

The standard haplotypes would also allow estimates of resolved ND4 variants (haplotypes) in each sample site or case, for each subspecies. In its current form, this is something that remains impossible to ascertain from the article.  Furthermore, while one can agree that the confiscated animals from which samples were taken represent a mixture of unknown sources, and so estimates of haplotype frequencies would not necessarily apply to the natural populations of origin, I think it may still provide insight into possible regional differences in the distribution of ND4 haplotypes. I would go as far as suggesting an attempt be made to illustrate the geographic distribution of the identified haplotypes in the map, cladogram, and network. Again, a good example can be seen in Figure 2 of Kundu et al., 2022.

 

Finally, I respectfully suggest that in the Discussion, as a complement to the current excellent review of difficulties facing the conservation of the Big-headed Turtle, the following issues are addressed, many as proposals to be included when designing future significant studies:

·         To gain an appropriate understanding of the geographic distribution of the units of conservation, future efforts are necessary to sample and analyze a representative proportion of natural populations from the major distribution ranges of the subspecies.

·         The authors could discuss whether or not the applied molecular marker (ND4) provides sufficient resolution to identify the population units of conservation relevance for the taxa in question. With the data provided by the authors in the current version of the paper, this is difficult to assess. If high levels of spatial genetic structuring are present in the taxa it may be necessary to combine or replace the mtDNA marker with others that afford greater resolution. Nuclear microsatellites could be one such candidate and the authors cite a key publication with application in Asian turtles (e.g. Tiedemann et al. 2014 reference [40]). The mtDNA control region is another marker that has proven valuable in distinguishing population-level management units in other turtle species.

In summary, I consider this an important study that merits publication but I would prefer that the issues raised in my appraisal are addressed in a careful revision of the paper.

-----------------------------------------------------------------------------------

 

Issues to correct or address in the various sections of the article:

 

Materials and methods

Taxonomic sampling

·         p. 4, parag. 2. Add that GenBank accession number MH822616 was used as reference for Platysternon megacephalum peguense. Note that in Supp Table 1, the entry for this sequence is catalogued as P.m.shiui so that needs correcting.

·         last line says “and one of outgroup species, Chrysemys picta (NC002073) based on their phylogenetic relationships”. Suggest changing to “and one for the closely related taxon Chrysemys picta (NC002073), to be used as an outgroup species.

·         A statement is made that from the samples “…four were taken from individuals collected from within the distribution range of Platysternon megacephalum shiui and 11 from those of P. megacephalum peguense.” Are these referring to the N=15 that came from wild collected samples? This is confusing as written. I suggest changing to

o   “In total, 465 new samples of Platysternon megacephalum were incorporated into the analysis. Of these, 15 were wild collected, four from within the distribution range of Platysternon megacephalum shiui and 11 from that of P. megacephalum peguense.  The rest of the samples (450) were from individuals confiscated from the trade (N=450) (Fig. 1). “

o   These 15 samples could be emphasized, as appropriate, whenever they appear in figures and tables with a symbol (e.g. asterisk) and in the legend indicate these represent samples from known origins. Indeed, if they were used to corroborate the subspecies identification from the sequences, say so.

·         GenBank reports for both DQ256377 and DQ016387 state they come from for P. megacephalum. In Supp. Table 1 they are both identified as P.m.megacephalum, presumably from information available in Luo et al (2019). If that’s the case explain here and add a note to that effect in Supp Table 1.

 

Molecular data

·         The Qiagen kit is spelled DNeasy instead of Dneasy

·         Specify the size of the ND4 fragments you were generating in your protocol. If you carry out the haplotyping I am suggesting below, you can specify the typical lengths produced and the standard lengths they were truncated to.

·         Change “A negative and positive control was used..” to “A negative and positive control were used…”

·         To confirm if the correct size of the fragment was amplified didn’t you use a size standard or ladder? If so, state the brand and catalog number.

·         Second paragraph, you state “The sequences were edited using…” You could add that they were edited and truncated to a standard size, if you prefer over my previous suggestion.

·         Last line. If you had identified the optimal model for nucleotide evolution, why did you calculate “Uncorrected pairwise genetic distances” in PAUP 4.0? State your reasoning for doing so.

·         I am very uncomfortable seeing that each and every single sample is not only analyzed as if they were a different sequence  but also uploaded as individual GenBank accessions. This is needlessly redundant, confusing and may even hinder future population genetic studies. It also makes it impossible to understand the true extent of ND4 sequence variation present. I propose, is something typically performed in population genetics (see Kundu et al 2023 for an example):

1.       Review the sequences and decide on a uniform length that includes a stretch of the sequence that incorporates as much as possible of optimal quality readings across the entire set of readings. Use the identified 5’ and 3’ positions of the two ends as reference for standard trimming points to be applied to all sequences. In case of sequences with shorter lengths due to unreliable readings at either end, substitute these with “Ns”; do not leave as gaps.

2.       Revise the entire sequence datasets and for each taxon (subspecies/clade level, but not at subclade level) make a catalog with the unique sequences or ND4 “haplotypes”. A program such as DNAsp (Generate/Haplotype Data File) can help with this. Ascribe standard names to each haplotype, for each taxon. A nomenclature system could use subspecies initials followed by sequential numbering e.g. PMS001.. PMS00n; PMP001…PMP00n.

3.       Simple haplotype statistics (number or haplotype frequency) for each sampled “case” or natural site can then be presented in a results table, stating that for cases were the samples are from unknown origins and therefore uneven representation of the natural populations, these values may not be applicable to natural populations.

4.       Using this approach, only the haplotypes (indicating the nomenclature for each) would be uploaded to GenBank.

5.       The Supp Table 1 would need to be modified, adding new columns for:

§  the ascribed standard haplotype for each sample.

§  GenBank accession numbers which would be much reduced, limited only to the individual haplotypes

§  (another suggestion) new columns for a) type of origin (trade or natural), b) region (could be the Vietnamese provinces)

6.       A results table would be needed to list absolute frequency of the resolved haplotypes, for each subspecies at each case or site

 

 

Results

·         P.m.peguense is stated as having at least three subclades, if there was a criteria for these explain in Materials and Methods.

·         Neither in Materials and methods nor in Results do the authors explain what criteria they used to identify sequences as P. megacephalum shiui. If no reliable published reference ND4 sequence is associated with P. megacephalum shiui, explain how this subspecies was identified. Is it because all of the samples from a limited area in NE Vietnam known to harbor this species clustered together?

·         Last paragraph, I suggest modifying to:

o   Only a few of the confiscated samples (N=?) clustered with the Platysternon megacephalum shiui reference. Most of them (approximately 96%), are clustered in Subclade 3 of P. megacephalum peguense. As expected, the distribution of the origin for samples identified as P. megacephalum shiui in the cladogram were limited to a small area in northeastern Vietnam. However, the distribution of Subclade 3 (N=??) covers a broad geographic region extending from the northwestern part of the country to the Central Highlands (Fig. 1). Locations for samples in Subclades 1 (N=??) and 2 (N=??) are currently unknown because none had reliable locality information.

·         If you accept the use of haplotype classifications and the resulting haplotype frequencies,

o   Figure 1 could be modified, at least for sample sites where wild collections were made, replacing the current simple symbols with pies reflecting the haplotype frequencies. This depiction could reveal if any geographic haplotype frequency cline is present.

o   Figure 2B would be simplified, using only the haplotypes (unique sequence) or, if there are many haplotypes present, representative haplotypes at nodes in each of the taxons. Subclade labels could still be applied to reflect evolutionary variation within P.m.peguense

Discussion

·         In the first paragraph, Luo et al. (2019) should be converted to reference format used in this journal: [37]

·         Last sentence in p. 8. I suggest changing to “On the other hand, P. m. megacephalum and P. megacephalum peguense exhibit a level of genetic differentiation (5 – 6.7%) equivalent to independent species when compared with analogous estimates in other turtle taxa [60,61].

·         First parag., p. 9  states “It is therefore very challenging to determine the exact geographic provenance of confiscated turtles in Quang Ninh, especially as populations within Subclade 3 are not well differentiated genetically.” Do you mean that the differences in the sequence comparisons is too low as to suggest genetic structure at a population level? If the authors are questioning if the resolving capacity of ND4 sequences is not adequate to identify demographic units of conservation relevance (i.e. populations, stocks, or management units), then this needs to be developed in the Discussion as it is relevant to consider in future studies.

·         Third parag., p. 9 you mention “our georeferenced samples are still very limited and could not be used to locate the natural ranges of Subclade 1 and 2 of P. megacephalum peguense.” Did you try plotting the georeferences for the sites and the subclade they belong to? Perhaps the modifications I am suggesting to the figures can help resolve this issue.

·         What about subclade 3? In Figure 1 there seems to be a broad distribution with a north-south spread. Don’t these samples indicate a spatial cline in genetic variation along the same direction?

 

Figures

·         Figure 1.

o   In text change “Distribution of” to “Origin of”

o   In legend change “The most popular confiscated sites” to “Sites with most confiscations”

·         Fig 2

o   A. Label for DQ016387 is duplicated, I believe the second needs to be eliminated.

o   B. Why is DQ256377 labelled to species name when in Fig 2A it is as subspecies?

o   Legend. Clarify further in legend that the sequence from GenBank accession number MH822616 (P. m. peguense) clusters in Subclade 3 in both figures. Also, mention that a larger, more detailed cladogram for Fig 2A is available in Supp data Figure 1.

 

·         Supp data Figure 1.

o   I suggest modifying the format to incorporate visual coding of subspecies, region (province), and type origin (natural, trade). A beautiful example of this style is in Figure 3 of Kundu et al 2023.

o   To avoid confusion, include subspecies names in the GenBank referenced sequence labels, similar to what is included in Fig 1A and B

o   Add to legend text stating something like this “Branch color codes as in Figures 2A and B”

 

Tables

·         Table 1. Assuming that a geographic orientation of the study is desirable (as stated in the last paragraph of the Introduction, when the authors say “Detailed knowledge of phylogeographic patterns of threatened species is essential…”), I would recommend that the data in this table is ordered by locality. This could be done if a column is added with the province where the sample was taken and then use that column for ordering the samples, not the sampling date.

·         Table 2

o   Why are the “average” distances not written as single values and instead as ranges? Unless there is a specific purpose, I suggest you limit to single values representing the average.

o   Change Sub-calde to Sub-clade

o   I suggest you use smaller sized text so that every line fits into a single row.

o   You could add a column to state how many samples in your dataset clustered in each of the clades

 

Supplementary data, Table 1

·         Sample No. 3 (GenBank accession No. MH822616) is labelled as P.m.shiui whereas in the Genbank record and in the paper’s Figure 1 it clusters with branches of Platysternon megacephalum peguense. Correct the subspecies in the table, as well as the color coding.

·         I know this may not be feasible, but I insist that being able to do a georeferenced analysis of the samples, particularly for P. m. peguense would be very informative. For that, having latitude/longitude information (even if very approximate) for samples would help. In addition, as mentioned elsewhere, adding and classifying by province would be useful

·         What does TCC mean?

In general the English is very adequate and only requires a few corrections which are mentioned in my review.

Author Response

Dear Reviewer,

Please find our responses to our comments in the attachment.

Thank you very much for your suggestions!

Best regards,

 

Minh (on behalf of other authors)

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

Please, see some comments in the file attached.

 

Kind regards

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

Please find our responses to our comments in the attachment.

Thank you very much for your suggestions!

Best regards,

 

Minh (on behalf of other authors)

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear Authors,

Thank you for your commitment to making the corrections submitted by the reviewers