Applying Target Capture Sequencing to Unravel the Anthurium Section Pachyneurium (Araceae), with Emphasis on Brazilian Species

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript "Applying Target Capture Sequencing to Unravel the Anthurium Section Pachyneurium (Araceae), with Emphasis on Brazilian Species" by Mel C. Camelo and colleagues employs Hyb-Seq with the Angiosperms353 probe set to resolve phylogenetic relationships in Anthurium section Pachyneurium, emphasizing Brazilian species diversity. It demonstrates non-monophyly of the section and identifies three major Brazilian clades linked to Miocene geoclimatic events. Strengths include the first genome-wide phylogeny for this group, robust support for biogeographic hypotheses, and comprehensive Brazilian sampling (18 Pachyneurium species and 36 species in the total sample). The manuscript presents a clear, well-structured phylogenomic study highly relevant to plant systematics, particularly for megadiverse Neotropical genera like Anthurium. The experimental design effectively tests the hypothesis of sectional monophyly using appropriate Hyb-Seq methods, with rigorous filtering of loci to ensure data quality despite variable recovery rates.

General Comments

Results appear reproducible, as detailed pipelines and public resources are specified, though raw reads and alignments should be deposited in a public repository beyond "available upon request."

The discussion, in my opinion, is very limited and written in a general manner. The discussion is very limited and written in a general manner. While a few key points are clear, much of the text consists of vague generalities that add volume without substance; it references related works but fails to clarify connections to the results obtained. For example, lines 413–417 ("By providing a well-supported phylogenetic hypothesis... directly addressing the Wallacean shortfall") lack specifics on how these shortfalls are addressed. Consider clarifying these links, dividing the discussion into subsections (e.g., phylogenetic implications, biogeography), and adding how findings inform conservation for Atlantic Forest endemics.

Despite the study's important results, the conclusions significantly underestimate its value. The section is vague, contains lengthy arguments, and lacks specifics tied to concrete results; it should emphasize novelty over prior hints of non-monophyly.

Specific comments

– In my opinion, the section name (presumably) is shown after the species name in the figures presents in not quite successful view. In the current version, firstly, without the word "sect.", it looks like a continuation of the species name, and secondly, it greatly complicates the drawing. I would suggest changing the visual component of the concept, for example, formatting the section names as abbreviations or callouts. At worst, you can add the word "sect.", if that is what is meant, of course, or at least put it in brackets.

– Line 381-383: ‘Species traditionally placed in section Pachyneurium, based on morphological characters, like rosulate habit and involute vernation, are distributed across distantly related, geographically separate clades.’ Here, it's important to specify which specific species are being referred to, as this is currently too general a conclusion. Perhaps it would be helpful to somehow schematically illustrate the incongruence between the molecular and morphological data directly on the phylogenetic tree (figures), or provide a separate scheme.

– Line 417-419. The "unexpected phylogenetic affinity between the Atlantic Forest Pachyneurium clade and A. coriaceum (sect. Polyphyllum)" is unclear, as figures label A. coriaceum as ‘Urospadix’, clarify sectional placement and its implications.

– Please, standardize section names (e.g., ’sect. Pachyneurium’ consistently), consistently; italicize all species or section names consistently; italicize all species names (e.g., line 306: 'A. longipeltatum' or line 365 'Anthurium sect. Pachyneurium' and other).​

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted.

Comment 1
Results appear reproducible, as detailed pipelines and public resources are specified, though raw reads and alignments should be deposited in a public repository beyond "available upon request."

Response:  We agree with this comment. To ensure full transparency and reproducibility, we include in the Data Availability Statement: The raw sequence reads have been deposited in the NCBI Sequence Read Archive (SRA) under accession number PRJNA1413566.

Comment 2
The discussion is very limited and written in a general manner… lacks specifics tied to concrete results… Consider clarifying these links, dividing the discussion into subsections, and adding how findings inform conservation for Atlantic Forest endemics.

Response: We agree and expanded the Discussion to directly link our phylogenetic results to broader systematic, biogeographic, and conservation implications. The Discussion is now explicitly structured into three subsections:
•    4.1 Implications for the Systematics and Taxonomy of Anthurium sect. Pachyneurium
•    4.2 Biogeographic history and diversification of Brazilian lineages
•    4.3 Methodological insights and contributions to biodiversity knowledge
These revisions provide concrete examples drawn from the phylogenetic analyses, explicitly discuss morphological convergence, clarify how the Darwinian and Wallacean shortfalls are addressed, and include a dedicated conservation-oriented discussion focusing on Atlantic Forest and Northeastern Brazilian endemics (Section 4.3).

Comment 3
Despite the study's important results, the conclusions significantly underestimate its value.

Response: We agree. The Conclusion section has been rewritten to better emphasize the novelty and significance of this study. The revised text highlights that our work represents the first densely sampled phylogenomic framework for Anthurium sect. Pachyneurium and constitutes a substantial advance over previous molecular studies that only suggested non-monophyly. The Conclusions now explicitly summarize the main findings and their broader systematic, biogeographic, and conservation implications.

Comment 4
The section name shown after the species name in the figures is visually confusing… consider adding “sect.” or using callouts or abbreviations.

Response: We revised Figures 2 and 3 to improve clarity. Section names are now consistently indicated using the abbreviation “sect.”, and formatting has been adjusted to avoid confusion with species names. The figure legends were also updated accordingly.

Comment 5
Line 381–383: The statement about species being distributed across distant clades is too general; specific species should be mentioned.

Response: We agree and have revised both the Results and Discussion to explicitly name representative species illustrating this pattern. In particular, we now cite A. affine, A. bonplandii, and A. leonii as concrete examples of morphologically similar species recovered in distinct phylogenetic clades. These changes were implemented in Results, Section 3.4, and expanded in Discussion, Section 4.1.

Comment 6
The unexpected affinity between Atlantic Forest Pachyneurium and A. coriaceum is unclear; figures label it as sect. Urospadix.

Response: We have clarified throughout the manuscript that A. coriaceum belongs to Anthurium sect. Urospadix and corrected any inconsistent references. The systematic implications of its strongly supported sister-group relationship with the Atlantic Forest Pachyneurium clade are now explicitly discussed in Section 4.1.

Comment 7
Please standardize section names and italicization.

Response:: We agree and have thoroughly revised the manuscript to ensure consistent usage of “sect.” for sectional names and proper italicization of all genus and species names throughout the text, tables, and figure legends. 

 

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The study is potentially interesting and suitable for publication in Plants. The introduction is well written and satisfactory, with solid and promising premises and a clearly stated research aim. The methods employed are modern and appropriate for the objective, namely reconstructing a phylogeny at the sectional level of the difficult and extensive genus Anthurium. However, the sampling is not fully adequate for the aims of the work, as some sections are very poorly represented.

Consequently, the discussion is somewhat incomplete, since the study does not address the taxonomic implications of the results. In particular, the authors are not able to resolve the polyphyletic section, nor to assess the taxonomic validity of the various sections of Anthurium. Moreover, they do not provide any meaningful comparison between morphology and the obtained phylogenetic results that would allow the identification of monophyletic taxa and evolutionarily meaningful sections. Their explanation of the observed patterns in terms of convergent evolution is also limited. A tree with plotted characters would have been valuable in this respect.

Therefore, the work is good but would benefit from being completed. While I understand that some of these issues cannot be fully addressed at this stage, the authors could nonetheless make an effort to enrich the discussion and examine their results in greater depth. They may not be able to propose a new infrageneric treatment of Anthurium, but they could certainly move in the direction suggested above. A good starting point would also be to improve the overall structure of the manuscript.

Comments for author File: Comments.pdf

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted.

Comment 1
Sampling is not fully adequate for the aims of the work, as some sections are very poorly represented.

Response: We respectfully clarify that the primary objective of this study is not to evaluate the monophyly or delimitation of all infrageneric sections of Anthurium, but rather to reconstruct the phylogenetic relationships and evolutionary history of species traditionally assigned to Anthurium sect. Pachyneurium, with particular emphasis on Brazilian taxa. Accordingly, our sampling strategy was designed to densely represent this morphological complex, while including representatives of other sections only to provide phylogenetic context.
This targeted approach allowed us to robustly identify three well-supported Brazilian clades (Amazonian, Atlantic Forest, and Caatinga/Cerrado), each with distinct evolutionary histories. Limited representation of other sections therefore does not compromise our central conclusions. A comprehensive infrageneric revision of Anthurium will require much broader taxon sampling and is the focus of an ongoing phylogenomic project including approximately 400 species.

Comment 2
The discussion is incomplete and does not sufficiently address the taxonomic implications of the results.

Response: We agree and have substantially expanded the Discussion to more clearly articulate the taxonomic and systematic implications of our results within the context of Anthurium sect. Pachyneurium. In Section 4.1, we now demonstrate that the traditional morphology-based circumscription of this section does not reflect evolutionary relationships and that the diagnostic vegetative traits historically used to define sect. Pachyneurium are homoplastic.
Rather than proposing formal taxonomic changes, which would be premature given the broader complexity of the genus, our study provides a robust phylogenomic framework that identifies evolutionarily cohesive lineages and clarifies the limits of vegetative morphology for infrageneric classification. These results establish a foundation for future systematic revisions based on expanded sampling.

Comment 3
The explanation of convergent evolution is limited; character mapping would have been valuable.

Response: We agree that explicit character-mapping analyses can be informative; however, in the case of Anthurium sect. Pachyneurium, such analyses would not substantially clarify the observed phylogenetic patterns. Some vegetative characters traditionally used to define the section—such as involute prefoliation—are shared with one section (sect. Multivernia), reflecting a complex history of convergence rather than shared ancestry.
Moreover, sect. Multivernia was historically treated as a series within Pachyneurium and only later recognized as a distinct section based on combined molecular and morphological evidence (e.g., orange fruit color, venation pattern, and Andean distribution). Plotting these characters onto a phylogeny in which sect. Pachyneurium is demonstrably non-monophyletic would likely introduce confusion rather than provide explanatory power.
Instead, we strengthened the Discussion (Section 4.1) by explicitly interpreting these traits as homoplastic and emphasizing that morphological similarity in this group does not reliably predict evolutionary relationships. Formal character-mapping analyses will be more informative when conducted within a broader phylogenomic framework encompassing multiple sections, which is currently in preparation.

Comment 4
The authors should consider including morphological information in the phylogenetic figures to better illustrate the taxonomic implications.

Response: We appreciate this suggestion and carefully considered the inclusion of morphological characters in the phylogenetic figures. However, because the present study demonstrates that sect. Pachyneurium is not monophyletic and that its diagnostic characters are homoplastic and partly shared with one section (sect. Multivernia), plotting these traits directly onto the phylogeny would risk obscuring rather than clarifying the evolutionary signal.
To improve interpretability without introducing misleading visual correlations, we instead revised Figures 2 and 3 to include sectional names in parentheses, allowing readers to readily assess the discordance between traditional sectional assignments and phylogenetic relationships. This approach maintains conceptual clarity while reinforcing one of the study’s central conclusions: that morphology-based sectional delimitations fail to capture the evolutionary history of this group.

Comment 5
The overall structure of the manuscript could be improved.

Response: We agree and have revised the manuscript structure accordingly. The Discussion is now organized into three focused subsections: (i) implications for the systematics and taxonomy of sect. Pachyneurium, (ii) biogeographic history and diversification of Brazilian lineages, and (iii) methodological insights and broader contributions to biodiversity knowledge. This structure highlights the central importance of the three Brazilian clades and their distinct evolutionary trajectories, while clearly linking phylogenetic results to biogeographic and conceptual interpretations.

Response to Comments on the Quality of English Language

Response: We check your Language Score in Wiley Editing services and we made a 95% language score. It’s a high score. 

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Applying Target Capture Sequencing to Unravel the Anthurium section Pachyneurium (Araceae), with emphasis on Brazilian species

The manuscript requires quite a few concrete revisions. I am sending comments by sections. Please revise carefully and respond to the questions.

ABSTRACT
Please revise. The methods are described in too much detail, while the main scientific results are not clearly highlighted.

INTRODUCTION
I suggest shortening the introduction, as it is very long and partly repetitive (the homoplastic nature of morphological characters is mentioned several times). Please add the hypotheses you formulated and clearly state your scientific objectives. How will this study contribute to the revision of Anthurium sections?

MATERIALS AND METHODS
The methods are generally appropriate, but they are not optimally justified and are not presented critically enough. Overall, this section lacks precise information on software versions, parameters for all analytical steps, and details on how much data were excluded at each stage.

Taxon sampling and genomic DNA extraction
Please justify why you selected this particular set of species. Explain why only one sample per species was used. The outgroup (Pothos scandens) is not methodologically justified. There is no information on the geographic coverage of individual species (i.e., whether the full distribution range was sampled).

Please also state whether type specimens or reference herbarium specimens were used. How many species exist in sect. Pachyneurium in total, and what proportion of them is covered by your study?

Library preparation, target enrichment and sequencing
Please add information on the average sequencing depth (coverage) and whether negative/positive control samples were used. Please explain more clearly whether libraries were normalized before pooling, as this cannot be determined from the current text. Data on hybridization success rates are also missing.

Capture sequencing data processing
The filtering criteria are not justified – please revise this. Did you test the effect of filtering on tree topology? Please state more clearly how paralogs were treated. Did you use any automated trimming (e.g., Gblocks, trimAl), or only manual editing? How many genes were excluded due to potential non-orthology?

Phylogenetic analysis
Two methods were used (concatenation + ASTRAL), which is methodologically appropriate. However, you did not explain why these two methods were chosen. Please also explain how conflicts between gene trees and the species tree were handled. A description of how gene tree discordance was assessed is missing. In addition, it is not entirely clear why branches with BS < 30% were removed prior to ASTRAL.

Divergence Time Estimation
Only two calibration points were used → weak temporal anchors. A sensitivity analysis to priors is missing. It is not explained why these particular fossils were selected. It is unclear whether the same topology as in the ML tree was used or whether BEAST simultaneously estimated topology. Was the topology fixed in BEAST or re-estimated?

RESULTS
Given the large amount of data, the results are too descriptive, insufficiently structured, and partly unclear. You included biogeographic explanations, ecological interpretations, and historical hypotheses, which do not belong in the Results section but in the Discussion. I recommend better organization of the Results section and clearer figures. Please clearly separate results from discussion. A summary table or schematic overview and a clear hierarchy of results are missing. Statistical analyses of your results are also very weak.

Gene recovery
Please explain Figure 1 more precisely.
Please explain why the success rate was so low (27%). Are certain taxa problematic? Do you think this affects the reliability of the results?
Figures 2–4 should have clear legends, main clades labeled, and time scales with confidence intervals.

Phylogenomic analysis
For readability, please correct typographical errors (“Pachyneuriumleonii”, “andand”). Which relationships are most stable and which taxa are unstable? Are there any polytomies?

DISCUSSION
The discussion contains a large amount of information, which is good, but it is not sufficiently critical. The statements are not always closely tied to the concrete results of this paper, and there is too little comparison with existing literature. A discussion is not merely a list of findings but an interpretation of them.

You often begin sentences with “This supports previous hypotheses that…”, but you do not specify exactly which results support which hypotheses. It would be better to write, for example, “Our ML and …”.

The discussion hardly addresses the limitations of sampling (one specimen per species), weak divergence-time calibration (only two fossils), and possible alternative explanations (e.g., hybridization, incomplete lineage sorting).

Systematics
Please specify which morphological characters are most homoplastic. Are there any characters that correspond with the phylogeny? What are the consequences for taxonomy (proposal of a new classification)?

Biogeography
Please revise carefully! The interpretations are speculative (Miocene dry diagonal, refugia). Why did you not perform a formal biogeographic reconstruction? How do you distinguish between vicariance and dispersal?

Methodological insights
This section is very long and not critical enough. What are the main limitations of Hyb-Seq in this study? How would different sampling change the results? What is the greatest methodological weakness of the paper?

CONCLUSION
This section repeats points from the discussion. It does not focus on what is new, what you have discovered, what future research could address, what the next necessary step is, and how your findings will contribute to science.

English language
Overall, the manuscript contains numerous typographical errors. Sentences are unclear, long, and overly complex. A thorough revision of the English language is mandatory.

References
Please check carefully, as some references are duplicated and others are cited incorrectly (e.g., references 1, 9, 10, and 11 are repeated).

Comments on the Quality of English Language

Overall, the manuscript contains numerous typographical errors. Sentences are unclear, long, and overly complex. A thorough revision of the English language is mandatory.

Author Response

Response to reviewer 3:

ABSTRACT
Please revise. The methods are described in too much detail, while the main scientific results are not clearly highlighted.

Response: We have revised the abstract to reduce methodological details and highlight the main scientific results, following the structured format required by the journal.

INTRODUCTION
I suggest shortening the introduction, as it is very long and partly repetitive (the homoplastic nature of morphological characters is mentioned several times). Please add the hypotheses you formulated and clearly state your scientific objectives. How will this study contribute to the revision of Anthurium sections?

Response: We have revised the Introduction to more clearly articulate the scientific objectives of this study: (1) to generate a high-resolution genomic dataset for Anthurium using a Hyb-Seq approach; (2) to reconstruct the phylogeny of sect. Pachyneurium and explicitly test its monophyly, with particular emphasis on Brazilian species; and (3) to estimate the timing of origin and diversification of the genus and its principal lineages.
Based on previous phylogenetic evidence, we explicitly state the working hypothesis that sect. Pachyneurium is not monophyletic and that Brazilian taxa may form geographically structured clades.
We also clarify that this study provides a phylogenomic framework for identifying evolutionarily coherent lineages within sect. Pachyneurium, particularly among Brazilian taxa. These results contribute to ongoing efforts to reassess the infrageneric classification of Anthurium, which will require broader taxon sampling across all sections.

MATERIALS AND METHODS
The methods are generally appropriate, but they are not optimally justified and are not presented critically enough. Overall, this section lacks precise information on software versions, parameters for all analytical steps, and details on how much data were excluded at each stage.

1-    Overall, this section lacks precise information on software versions, parameters for all analytical steps, and details on how much data were excluded at each stage. 

Response: We respectfully note that software versions and key analytical parameters were already specified throughout the Methods section (e.g., Trimmomatic v0.39 with explicit trimming parameters; HybPiper v2.3.2; hybpiper-nf v1.0.4; SPAdes 3.13.1; MAFFT v7.505; RAxML-NG v1.2.2; ASTRAL-IV v1.22.4.6; BEAST v2.7.7; Tracer v1.7.2; TreeAnnotator v2.7.7).
However, we appreciate the reviewer’s request for greater clarity and have revised the manuscript to further detail filtering steps and to explicitly quantify data retention and exclusion at each stage of the pipeline. 
The revised Methods section now specifies:
(i) the number of loci initially recovered;
(ii) the number of loci excluded due to paralogy or missing data thresholds;
(iii) the alignment filtering criteria and resulting data matrix dimensions; and
(iv) additional clarification of parameter choices where applicable.
We believe these additions substantially improve transparency and reproducibility.

2-    Taxon sampling and genomic DNA extraction
Please justify why you selected this particular set of species. Explain why only one sample per species was used. The outgroup (Pothos scandens) is not methodologically justified. There is no information on the geographic coverage of individual species (i.e., whether the full distribution range was sampled). Please also state whether type specimens or reference herbarium specimens were used. How many species exist in sect. Pachyneurium in total, and what proportion of them is covered by your study? 

Response:  We have addressed each point below.
1. Species selection criteria and justification for using one sample per species:
The selection of species for Anthurium sect. Pachyneurium was guided by two main criteria: (i) to represent the breadth of morphological and geographic variation within the section, and (ii) to focus on Brazilian taxa, which are the primary subject of this study. Sect. Pachyneurium comprises approximately 120 Neotropical species, of which 21 are recorded in Brazil. Our sampling includes 18 species of the section (15% of the total diversity), including 11 Brazilian species, which represents 52 % of the species occurring in the country (including eight endemics). This coverage captures a substantial proportion of the morphological and geographic variation, encompassing the principal centers of diversity for the section (the Brazilian Amazon, Atlantic Forest, Caatinga/Cerrado, the Caribbean/Central America, and the Andean region).
The use of a single sample per species is justified by the primary goal of this study: to test the monophyly of sect. Pachyneurium and the relationships among its major lineages at a sectional level. This study was not designed to investigate population structure or infraspecific variation. We acknowledge that denser population sampling is necessary for future phylogeographic or species-delimitation studies and have noted this as a perspective for future research in the manuscript.
2. Geographic coverage and use of voucher/reference specimens:
While we did not sample across the entire distribution range of each species, all material was collected from the species' typical areas of occurrence, covering the major biogeographic regions mentioned above. Geographic coordinates and voucher information for all accessions are provided in detail in Table S1. Instead, material was obtained from field collections or from living collections at the Rio de Janeiro Botanical Garden (JBRJ) and the Missouri Botanical Garden (MO) we did not use it herbarium sample.  All accessions were vouchered and are deposited at the herbaria RB and MO, ensuring traceability and verifiability.
3. Inclusion of additional ingroup taxa:
To robustly test the monophyly of sect. Pachyneurium within a broader context, we included representatives from 13 other sections of Anthurium as additional ingroup references.
4. Outgroup justification:
The choice of Pothos scandens as the outgroup is based on well-established phylogenetic relationships within the Araceae family, where Pothos (subfamily Pothoideae) is consistently resolved as the sister lineage to Anthurium. This makes P. scandens an appropriate taxon for rooting the genus-level analysis.
5. Inclusion of the section's type species:
The type species of sect. Pachyneurium, Anthurium crenatum, was intentionally included in the sampling to anchor the nomenclatural and morphological definition of the group.
These clarifications will be incorporated into the revised manuscript.

The filtering criteria are not justified – please revise this. Did you test the effect of filtering on tree topology? Please state more clearly how paralogs were treated.

1. Filtering criteria and the effect of filtering on tree topology: see the explanation for the question “Please explain why the success rate was so low (27%)”, where we added a figure to explain both questions
2. Paralogs: Regarding the paralogs, the full explanation is presented below in the response to “How many genes were excluded due to potential non-orthology?”

Did you use any automated trimming (e.g., Gblocks, trimAl), or only manual editing?

Response: Automated trimming methods were indeed used (trimAL, ClipKit and CIAlign). However, the results were discordant and invariably resulted in over-trimming. We also observed several artifacts that could not be treated by the automated trimming methods, such as intron retention events and highly heterogeneous sequence lengths.

This led us to conduct a thorough inspection of individual gene alignments followed by manual editing. 

How many genes were excluded due to potential non-orthology?

In sequence capture datasets, the potential identification of paralogous sequences is a well-recognized challenge, as their inclusion can compromise the accuracy of phylogenetic analyses. In our study, we addressed this issue by utilizing the HybPiper pipeline v2.3.2 (doi:10.3732/apps.1600016), which incorporates specific procedures for detecting and filtering paralogs. Comprehensive details of these procedures are provided in the referenced publication: Identification of paralogous sequences, alleles, or contaminants

In many target enrichment analysis pipelines, correct orthology of enriched sequences is inferred using BLAST searches to the target proteome (e.g., Bi et al., 2012; Bragg et al., 2015), but this method will be inefficient when genomic resources in the target taxa are incomplete. In HybPiper we provide a streamlined method for identification of potential paralogs that can be further analyzed using gene phylogenies. Typically, if HybPiper identifies a single contig that subsumes the range of other contigs, it is retained and the smaller contigs are discarded. However, sequences assembled using SPAdes occasionally result in multiple, long contigs, each representing the entire target sequence. During the extraction of exon sequences, the HybPiper script exonerate_hits.py identifies contigs that span more than 85% of the length of the reference sequence.

 

Johnson MG, Gardner EM, Liu Y, et al. HybPiper: Extracting coding sequence and introns for phylogenetics from high-throughput sequencing reads using target enrichment. Appl Plant Sci. 2016;4(7):apps.1600016. Published 2016 Jul 12. doi:10.3732/apps.1600016

Paralogs were excluded from the analysis, and an overview of the genes containing paralogs is illustrated in the figure below, where colored spots indicate the presence of a discarded paralog.

Figure 1:

Phylogenetic analysis

Two methods were used (concatenation + ASTRAL), which is methodologically appropriate. However, you did not explain why these two methods were chosen. The supermatrix (concatenation) approach and the multi-species coalescent model (e.g., ASTRAL) are widely recognized as the primary methods for constructing species trees in phylogenomic analyses (doi:10.1038/s41576-023-00620-x, doi:10.3390/bioengineering11050480 [section 3.1 and 3.2], doi:10.1038/s41576-020-0233-0). These methodologies are integral to the field, and their application is often considered a standard practice in phylogenomics. A key excerpt from an influential paper highlights the significance of these approaches.

Concatenation versus coalescence
Phylogenomic data matrices can be analysed as a single supermatrix (an approach known as concatenation) or each gene alignment can be analysed separately under the multispecies coalescent framework (an approach known as coalescence). The two approaches sometimes yield different tree topologies, contributing to incongruence. Determining which approach is more appropriate for a phylogenomic data set is difficult. For example, using simulated multi-locus data, concatenation slightly outperformed a fully coalescent-based approach (wherein gene trees and species trees are coestimated), whereas using coalescent independent sites, both approaches performed comparably. However, an extensive evaluation of coalescent-based and concatenation-based approaches when different biological and analytical factors are at play is lacking, hindering our knowledge of best practices.

Steenwyk, J.L., Li, Y., Zhou, X. et al. Incongruence in the phylogenomics era. Nat Rev Genet 24, 834–850 (2023). doi:10.1038/s41576-023-00620-x

As described above, species trees generated by these two methods may produce differing topologies, primarily due to the phenomenon of incomplete lineage sorting (doi:10.1038/s41576-020-0233-0). For consistency, we have presented the results from both methods, which, in our case, were congruent (see below). 

To address the reviewer’s comment, we included the first citation along with a concise explanation of the rationale behind using the two methods in section 3.4, second paragraph.

 

Please also explain how conflicts between gene trees and the species tree were handled. 
A description of how gene tree discordance was assessed is missing. 

There was no observed conflict between the species trees generated by the coalescence (ASTRAL) and concatenated gene trees. We indicated this fact in the text. The tanglegram comparing the tree topologies is shown below:

 

Since the topologies are the same, we did not present this image.

 

In addition, it is not entirely clear why branches with BS < 30% were removed prior to ASTRAL.

The removal of low-support branches follows the recommendation of ASTRAL developers. In their article (doi:10.1007/978-3-319-67979-2_4), the authors address several research questions, including: “Can contracting low support branches improve the accuracy of ASTRAL?” In their findings, they claim that “However, contracting low support branches with 3–33% thresholds dramatically reduces the discordance with the reference tree” and “Thus, consistent with simulations, contracting very low support branches seems to produce the best results, when judged by similarity with the reference trees”. Therefore, we applied low-bootstrap filtering, as recommended in:

·    https://github.com/smirarab/ASTRAL/blob/master/astral-tutorial.md#running-with-unresolved-gene-trees 
·    https://github.com/smirarab/ASTRAL/issues/39 

Divergence time estimation

Response: First, the topology was fixed to the maximum-likelihood (ML) species tree during the BEAST analysis. Our objective was to estimate divergence times based on a well-supported phylogenomic topology rather than to re-estimate relationships under a Bayesian framework. Fixing the topology allowed us to focus exclusively on temporal inference while avoiding additional topological uncertainty unrelated to the dating objectives. This clarification has now been explicitly added to the Methods section.
Regarding fossil calibrations, we acknowledge that only two calibration points were used. However, confidently assignable fossils within Araceae—particularly within Pothoideae and Anthurium—are limited. The selected fossils represent the most reliable and phylogenetically informative calibration points currently available. Specifically:
•    The root calibration (Pothoideae crown) is based on early fossil evidence attributed to Pothoideae.
•    The crown Anthurium calibration is based on †Petrocardium cerrejonense, whose leaf morphology closely resembles extant Anthurium and has been discussed in detail in the cited studies [49,50].
These fossils were selected because they exhibit strong morphological affinity with extant lineages and have been previously interpreted as reliable temporal anchors in Araceae evolutionary studies. We have clarified this justification in the revised manuscript.
We did not conduct a formal sensitivity analysis to alternative prior configurations. However, posterior age estimates closely align with the fossil constraints and display reasonably bounded 95% highest posterior density (HPD) intervals, suggesting that the priors were not overly restrictive. The HPD intervals are now explicitly reported in the Results section to provide transparency regarding temporal uncertainty.
We believe these clarifications substantially improve the methodological transparency and robustness of the divergence time analyses.

 

RESULTS

1-Given the large amount of data, the results are too descriptive, insufficiently structured, and partly unclear. You included biogeographic explanations, ecological interpretations, and historical hypotheses, which do not belong in the Results section but in the Discussion. I recommend better organization of the Results section and clearer figures. Please clearly separate results from discussion. A summary table or schematic overview and a clear hierarchy of results are missing. Statistical analyses of your results are also very weak.

Response: We have reorganized the Results section to present findings in a clearer hierarchical structure, separating strictly empirical outcomes from interpretative statements. Biogeographic, ecological, and historical interpretations have been moved to the Discussion section to ensure a clearer distinction between data presentation and inference.
Additionally, we have streamlined descriptive passages and emphasized key results rather than listing all topological details.

 

2-Gene recovery. Please explain Figure 1 more precisely. Please explain why the success rate was so low (27%). Are certain taxa problematic? Do you think this affects the reliability of the results?

Response: Figure 1 now explicitly illustrates the recovery patterns of the 68 high-confidence loci retained after filtering from the initial set of 336 targeted Angiosperms353 genes. 

Please explain why the success rate was so low (27%). Are certain taxa problematic? Do you think this affects the reliability of the results?

The observed low success rate primarily stems from differences in sequencing coverage among the sampled species, despite standardized DNA extraction and quantification protocols. Variations in sequencing coverage and capture efficiency led to fragmented or missing guided assemblies for several genes. Additionally, the assemblies differed in the extent of reference gene length recovered. The figure below shows the percentage of reference gene length recovered for each gene (x-axis) across all samples (y-axis). Spots closer to black (as shown in the scale on the right) indicate genes that were nearly fully reconstructed, while all-white columns represent genes with no recovery, accounting for most of the low success rate. To address this, two potential experimental improvements are possible: redesigning capture probes to include Anthurium species from the section Pachyneurium, or increasing sequencing coverage. However, the latter would incur higher costs (for which funding is currently unavailable) and may not necessarily improve the overall reconstruction of target genes.

 

 

Furthermore, the bottom of the figure shows that for two samples (#7 = Anthurium lucidum and #8 = A. plowmanii), no sequences were recovered. These samples were therefore excluded from further analyses.

Given the recovery scenario illustrated in the figure, we applied several filtering criteria to maximize the quality of sequence data for downstream phylogenetic analyses. One key filter required a minimum of 24 species to have genes mapping to at least 75% of the reference gene length. This criterion was implemented to avoid cases where only a few species contained reliable sequences for alignment—thus reducing missing data in the supermatrix approach. Such missing data can produce artifacts such as phylogenetic terraces (doi:10.1186/s12862-024-02309-z; excerpt: “A concept related to the presence of multiple optimal solutions is called “phylogenetic terraces” – regions of the tree space where all trees have the same score purely due to certain patterns of missing data.”). These filters were therefore applied to enhance the reliability of our analyses.

 

3- Figures 2–4 should have clear legends, main clades labeled, and time scales with confidence intervals.

Response: Figures 2–4 have been revised to improve clarity and readability.
Specifically, we have:
•    Labeled major clades directly on the phylogenetic trees.
•    Expanded the figure legends to clearly define node support metrics (bootstrap values and local posterior probabilities).
•    Clarified the color coding of geographic lineages.
•    Explicitly indicated the time scale and 95% highest posterior density (HPD) intervals in the time-calibrated tree (Figure 4).
•    Removed interpretative phrasing from the legends to ensure they remain strictly descriptive.

4-Phylogenomic analysis. Which relationships are most stable and which taxa are unstable? Are there any polytomies?

Response: 
No polytomies were found. A limited number of shallow nodes showed moderate support values (<70% bootstrap), reflecting reduced resolution at recent divergence levels. Minor positional variation among a few terminal taxa was observed between ML and ASTRAL analyses; however, these differences did not affect the recovery of the principal clades.
We have now explicitly indicated lower-support nodes in the revised Results and clarified clade labeling in the figure legends.

 

DISCUSSION
The discussion contains a large amount of information, which is good, but it is not sufficiently critical. The statements are not always closely tied to the concrete results of this paper, and there is too little comparison with existing literature. A discussion is not merely a list of findings but an interpretation of them.
You often begin sentences with “This supports previous hypotheses that…”, but you do not specify exactly which results support which hypotheses. It would be better to write, for example, “Our ML and …”.

The discussion hardly addresses the limitations of sampling (one specimen per species), weak divergence-time calibration (only two fossils), and possible alternative explanations (e.g., hybridization, incomplete lineage sorting).

Response:  We have revised the Discussion to strengthen its critical interpretation and to link statements more explicitly to our results. General expressions such as “This supports previous hypotheses…” were replaced by direct references to our analytical findings (e.g., congruence between ML and ASTRAL trees, support values, divergence-time estimates). We also expanded comparisons with previous phylogenetic and taxonomic studies of Anthurium, clarifying agreements and differences.
This study focuses specifically on sect. Pachyneurium, particularly Brazilian species. Representatives from other sections were included only to test the monophyly and phylogenetic placement of the focal section, not to revise the genus as a whole.
We now explicitly acknowledge key limitations: (1) one accession per species, which precludes assessment of intraspecific variation; (2) divergence-time estimation based on two fossil calibrations, resulting in provisional age estimates; and (3) potential effects of incomplete lineage sorting or hybridization, which would require denser sampling and population-level data to evaluate fully.

 

Systematics
Please specify which morphological characters are most homoplastic. Are there any characters that correspond with the phylogeny? What are the consequences for taxonomy (proposal of a new classification)?

Response: We now explicitly state which morphological characters are inferred to be homoplastic. Rosulate habit, involute prefoliation, primary lateral veins free, and absence of a collective vein occur in multiple, distantly related lineages and therefore do not correspond to a single evolutionary lineage.
Some morphological patterns show partial correspondence with geographically structured clades, but no clear vegetative synapomorphy was identified. Given the incomplete taxon sampling (~120 species in the section), we do not propose a formal recircumscription at this stage.

Biogeography
Please revise carefully! The interpretations are speculative (Miocene dry diagonal, refugia). Why did you not perform a formal biogeographic reconstruction? How do you distinguish between vicariance and dispersal?

Response: Formal biogeographic reconstruction using model-based approaches (e.g., DEC or DIVA) was not performed because our sampling, although representative of Brazilian lineages, does not yet encompass the full geographic and taxonomic diversity of sect. Pachyneurium. Such analyses will be more informative once the ongoing broader sampling effort (ca. 400 species) is completed.
Distinguishing between vicariance and dispersal with confidence requires denser taxon sampling and well-constrained divergence-time estimates. Therefore, the historical scenarios discussed here are presented as testable hypotheses based on temporal patterns rather than as explicit ancestral-area reconstructions.

 

Methodological insights
This section is very long and not critical enough. What are the main limitations of Hyb-Seq in this study? How would different sampling change the results? What is the greatest methodological weakness of the paper?

Response: The main limitation of Hyb-Seq in this study was the relatively low recovery rate of Angiosperms353 loci (~27%). This likely reflects sequence divergence between universal probes and Anthurium, as well as DNA quality variation. Although missing data may reduce resolution at some nodes, the congruence between concatenation and coalescent-based analyses suggests that the principal topological patterns are stable.
A second limitation is taxon sampling. Only one accession per species was included, which prevents assessment of intraspecific variation, hybridization, or cryptic diversity. Expanded sampling across the geographic range of the section may refine relationships within clades and alter support values for shallow nodes.
The greatest methodological weakness of the present study is the limited temporal calibration, as divergence-time estimation relied on only two fossil constraints. Consequently, age estimates should be interpreted as provisional.
These limitations are now explicitly acknowledged in the manuscript, and interpretations are presented cautiously.

 

CONCLUSION
This section repeats points from the discussion. It does not focus on what is new, what you have discovered, what future research could address, what the next necessary step is, and how your findings will contribute to science.

Response: We have rewritten the Conclusion to avoid repetition of the Discussion. The revised section now emphasizes the main findings of the study, specifically the non-monophyly of sect. Pachyneurium, the identification of geographically structured evolutionary lineages, and the implications of morphological homoplasy for infrageneric classification in Anthurium.
The Conclusion now also highlights the broader contribution of this study as the first phylogenomic framework for Brazilian representatives of the section and outlines future research priorities, including expanded taxon sampling, formal biogeographic reconstruction, and integrative taxonomic revision.

 

English language

Response: 
The manuscript now was revised by a native: Dr. Simon J. Mayo (Kew Gardens).

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I read the reply and the new version by the authors and, if on the one hand I understand the difficulties in improving the manuscript in the sense I pointed out before, on the other hand I became aware that a much more extensive work is being carried out by the authors themselves. So I wonder about the opportunity to publish such a "half-made" contribution. The authors insist on the relevance of having discovered three distinct lineages within a single section of Anthurium. As the sections themselves are highly polyphyletic, how can we be sure that,based on a geographically and taxonomically limited sampling, the section Pachyneuron itself is adequately represented? I think that the authors actually highlighted simply that, among the Brazilian species of Anthurium, three evolutionary lineages can be recognized, without any definitive taxonomic implication. Probably, I have to admit that I would be more "indulgent" if the authors had not used terms such as "solid", "robust", "unravel" and so on already in the abstract.

On this regard, I would point out that the sentence "future classifications of Anthurium may benefit from prioritizing historically cohesive biogeographic lineages over traditional morphology-based sections" is in my opinion rather insidious, because it seems to suggest formal recognition of infrageneric taxa even when, as in this case, that option turns out to be impractical. The examination of the provided trees seems indeed to suggest to refrain from further subdivision and, on the contrary, to favour the synonymization of others.

By experience, yielding to geography is often due to insufficient sampling and the overlooking of reticulate evolution or incomplete lineage sorting.

On this point, I strongly suggest to choose different tones for the figures as they are difficult to read.

Moreover, the authors seem unable to relate these lineages to any morphological synapomorphies. Maybe this is not possible, because of the rapid radiation of the genus in South America, but no useful information about macromorphology, micromorphology or cytology is provided. Honestly, a greater effort by the author team would be expected, especially by colleagues expert in Brazilian flora.

Therefore, in order not to hinder the publication of the results achieved so far, as is my custom, I accept the manuscript with minor revision, recommending mitigating some "highly sounding" sentences and modifying the graphical tree layouts. However, I leave to the Editor the important decision to publish in an international and high-rated journal such as Plants this limited contribution which will be likely soon superseded.

 

Author Response

Response to Reviewer 2

We thank the reviewer for their insightful and constructive comments, which have been instrumental in refining the manuscript.
1. Regarding sampling representativeness and study scope:
We acknowledge that comprehensive sampling of the entire section Pachyneurium would require a global effort. However, the primary aim of this study was to elucidate the Brazilian diversity within this group. To this end, our sampling includes the majority of Pachyneurium species documented for Brazil, representing a substantial increase in taxonomic coverage compared to prior studies. To accurately reflect the scope and inherent limitations of our inferences, we have made the following adjustments:
•    We have moderated the language throughout the manuscript, replacing categorical assertions with more measured phrasing such as “suggest,” “indicate,” and “recover.”
•    We have explicitly contextualized the findings by referring to “the sampled Brazilian species” and “lineages recovered in our analyses.”
•    We have clarified that the groups identified (now consistently termed “lineages” in the Discussion) do not constitute a formal taxonomic proposal.
2. Concerning assertive language and taxonomic implications:
•    We have conducted a thorough revision to temper overly definitive language. The Abstract, Discussion, and Conclusion now employ a more cautious and academically appropriate tone commensurate with the presented data.
•    The sentence suggesting that “future classifications may benefit from prioritizing historically cohesive biogeographic lineages over traditional morphology-based sections” has been removed. It has been replaced with the statement that “historical biogeography may be an important factor to consider in future classifications,” thereby positioning it as one element within an integrative taxonomic framework.
3. Pertaining to figure legibility and morphological discussion:
•    To enhance clarity, Figures 2, 3, and 4 have been redesigned. We implemented a higher-contrast color scheme and included explanatory text in each figure legend describing the color scheme, which adheres to Color Universal Design (CUD) principles for accessibility.
•    We recognize that a detailed morphological analysis extends beyond the primary phylogenomic focus of this work. Nevertheless, we have strengthened the discussion to emphasize that the traditional morphological characters defining the section are homoplastic likely resulting from ecological convergence in xeric environments and are therefore poor predictors of evolutionary relationships.

We have carefully addressed all points raised by Reviewer 2.

We thank the reviewer for their valuable input, which has improved our work.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

Abstract: Although you state that you have “reduced the methodological details,” you have in fact only slightly softened them rather than truly shortened them. The abstract is still methodologically oriented instead of being results-focused.

Introduction: Although you have clarified the objectives and hypotheses, you have not substantially shortened the Introduction and it still contains repetitive arguments. I recommend further condensation, particularly in the sections addressing homoplastic morphology.

Materials and Methods:
You have only partially addressed this comment. Although the revised Methods section now includes more detailed software version information and quantitative data on filtering, several key methodological issues remain insufficiently justified or documented. In particular, the filtering criteria are not adequately explained, and it is unclear whether filtering decisions were tested for their impact on phylogenetic topology. The criteria for paralog exclusion should be more explicitly justified and consistently presented in the Methods section.

Furthermore, although you explain how paralogs are detected in HybPiper, you do not specify how many loci were excluded in this study due to potential non-orthology. Please provide explicit numbers and clarify how these relate to the final datasets (94 and 68 loci).

Finally, the manuscript lacks a description of the procedure used to assess conflicts between gene trees and the species tree. Simply stating that no conflict was observed is insufficient without explaining the method used to evaluate gene tree discordance.

Results:
You have only partially addressed this comment. The Results section remains largely descriptive and would benefit from a clearer synthesis of the findings (e.g., a summary table or schematic overview) and a more explicit treatment of statistical support.

Although the explanation of Figure 1 and the low gene recovery rate is clearer and generally satisfactory, you do not fully address whether certain taxa were consistently problematic, nor do you explicitly discuss how the low recovery rate may affect the reliability and potential bias of the phylogenetic results. This limitation should be more clearly acknowledged in the manuscript.

Furthermore, although you state that no polytomies were detected and that only minor differences among a few terminal taxa were observed among analyses, it would be useful to explicitly identify which taxa or nodes show instability and which relationships are consistently recovered with high statistical support.

Biogeography: You have only partially addressed this comment. The biogeographic interpretations remain largely speculative and should be more clearly framed as hypotheses rather than as firm conclusions.

Author Response

Response to Reviewer 3

We sincerely thank the reviewer for the detailed and constructive comments, which have helped us substantially improve the manuscript. Below we provide a point-by-point response indicating how each comment has been addressed in the revised version. All line numbers refer to the revised manuscript with tracked changes.

Comment 1 – Abstract:

"Although you state that you have 'reduced the methodological details,' you have in fact only slightly softened them rather than truly shortened them. The abstract is still methodologically oriented instead of being results-focused."

Our response: We revised the Abstract to focus on results rather than methods. The revised version now includes only one brief mention of the methodology ("using target capture sequencing with the Angiosperms353 probe set"), while the remaining text synthesizes the main findings: the non-monophyly of sect. Pachyneurium, the three geographically structured Brazilian lineages, the unexpected sister relationship with sect. Urospadix, the identification of a putatively new species (A. giuliettiae sp. nov. ined.), divergence time estimates, and the implications for taxonomy and conservation. The Abstract now has approximately 195 words and is results-focused as requested (lines 23–40).

 Comment 2 – Introduction:

"Although you have clarified the objectives and hypotheses, you have not substantially shortened the Introduction and it still contains repetitive arguments. I recommend further condensation, particularly in the sections addressing homoplastic morphology."

Our response: We have substantially condensed the Introduction, reducing it from five to four paragraphs. Repetitive arguments about morphological homoplasy have been eliminated; this concept is now mentioned only once in the first paragraph. The paragraph on phylogenomic methods has been significantly shortened to focus on the essential information needed to justify our approach, while maintaining the key references requested (lines 62–67). The objectives and hypotheses remain clearly stated in the final paragraph (lines 69–78).

 Comment 3 – Materials and Methods (filtering criteria):

"The filtering criteria are not adequately explained, and it is unclear whether filtering decisions were tested for their impact on phylogenetic topology."

Our response: We have expanded the explanation of filtering criteria in section 2.3 (lines 144–165). The thresholds (50% and 90%) are now explicitly justified: the 50% threshold was chosen to retain loci with moderate representation across samples while minimizing missing data, and the 90% ultra-stringent filter was applied to obtain a core set of highly complete loci for downstream analyses. More importantly, we tested the impact of filtering decisions by comparing topologies obtained from the 94-locus and 68-locus datasets; both recovered congruent relationships with comparable support values, demonstrating that our conclusions are robust to filtering thresholds (lines 163–165).

Comment 4 – Materials and Methods (paralog exclusion):

"The criteria for paralog exclusion should be more explicitly justified and consistently presented in the Methods section. Furthermore, although you explain how paralogs are detected in HybPiper, you do not specify how many loci were excluded in this study due to potential non-orthology. Please provide explicit numbers and clarify how these relate to the final datasets (94 and 68 loci)."

Our response: We have clarified the paralog exclusion criteria in section 2.3 (lines 152–162). The HybPiper paralog detection pipeline flags loci with multiple long contigs (>85% of reference length), and all such putative paralogs were excluded to ensure orthology. Regarding the number of loci excluded, we now explicitly state that from the initial 336 targeted genes, 242 loci (72.02%) were excluded after applying quality filters. However, because the filtering criteria for paralogy and mapping rate were applied concurrently, and because loci with paralogy issues frequently also exhibited low mapping rates, the exact number of loci excluded specifically due to paralogy cannot be quantified separately from those excluded due to insufficient mapping. We have added a transparent statement acknowledging this limitation (lines 158–162). The relationship with the final datasets is now clearly explained: the 94-locus dataset represents loci retained after filtering (≥50% samples), and the 68-locus dataset represents the core set recovered in ≥90% of samples.

Comment 5 – Materials and Methods (gene tree conflict):

"The manuscript lacks a description of the procedure used to assess conflicts between gene trees and the species tree. Simply stating that no conflict was observed is insufficient without explaining the method used to evaluate gene tree discordance."

Our response: We have added a detailed description of the procedure used to assess gene tree conflict in section 2.4.1 (lines 204–212). Specifically, we examined the distribution of quartet scores computed by ASTRAL-III, which represent the proportion of gene trees supporting each species tree relationship. Additionally, we compared topologies obtained from concatenated (RAxML-NG) and coalescent-based (ASTRAL-IV) approaches. The high congruence between methods (Figs. 2, 3) suggests that conflicting signals do not substantially obscure the primary phylogenetic relationships.

Comment 6 – Results (synthesis and statistical support):

"The Results section remains largely descriptive and would benefit from a clearer synthesis of the findings (e.g., a summary table or schematic overview) and a more explicit treatment of statistical support."

Our response: We have substantially restructured the Results section to provide a clearer synthesis of the findings. A new summary table (Table 2) has been added, presenting the major lineages recovered within sect. Pachyneurium, their representative species, statistical support from both ML bootstrap and ASTRAL local posterior probabilities, and lineage status (high support, moderate support, or unstable). In section 3.4, we now explicitly list which relationships are recovered with high statistical support (BS ≥90%, LPP ≥0.95) and which nodes show moderate support or instability (lines 252–285). This provides a more explicit treatment of statistical support as requested.

Comment 7 – Results (problematic taxa and low recovery bias):

"You do not fully address whether certain taxa were consistently problematic, nor do you explicitly discuss how the low recovery rate may affect the reliability and potential bias of the phylogenetic results. This limitation should be more clearly acknowledged in the manuscript."

Our response: We have addressed this in two ways. First, in section 3.1, we now explicitly identify taxa with consistently lower recovery rates: A. paraguayense, A. ernestii, and A. salvinii showed mapping rates below 0.6 for several loci (lines 232–235). Second, we discuss the potential implications of the low overall recovery rate (27.98%), noting that while this likely reflects evolutionary divergence between the Angiosperms353 probes and Anthurium, the congruence between analyses using different filtering thresholds and between concatenated and coalescent approaches indicates that the recovered loci contain sufficient phylogenetic signal to resolve the major relationships within sect. Pachyneurium (lines 236–242).

Comment 8 – Results (instability vs. consistent relationships):

"It would be useful to explicitly identify which taxa or nodes show instability and which relationships are consistently recovered with high statistical support."

Our response: We have added this information in section 3.4. Relationships consistently recovered with high statistical support (BS ≥90%, LPP ≥0.95) are now explicitly listed, including the three Brazilian lineages, the sister relationship between the Atlantic Forest lineage and A. coriaceum, and the Caribbean/Central American lineage (lines 258–273). Taxa showing instability between concatenated and coalescent analyses, coupled with low support values (BS <70%, LPP <0.80), are identified as A. paraguayense, A. ernestii, and A. salvinii, with a description of their conflicting placements (lines 280–288). We have also indicated these unstable taxa with asterisks in Figures 2 and 3 for clarity.

Comment 9 – Biogeography:

"The biogeographic interpretations remain largely speculative and should be more clearly framed as hypotheses rather than as firm conclusions."

Our response: We have thoroughly revised section 4.2 to frame all biogeographic interpretations as testable hypotheses rather than firm conclusions. The section now begins by explicitly acknowledging sampling limitations: "given that our sampling includes only 15% of the section's total diversity... the following interpretations are presented as testable hypotheses to guide future studies with expanded taxon sampling" (lines 353–356). Throughout the section, we have replaced conclusive language with hypothesis-framing terminology such as "is consistent with," "may represent," "has been hypothesized," "raises the hypothesis that," and "could be consistent with" (lines 357–390). We also explicitly note that formal biogeographic reconstruction was not performed due to incomplete sampling, and that our hypotheses require testing with expanded datasets (lines 391–396).

 

Author Response File: Author Response.docx