Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear All,

The manuscript investigates soil bacterial and fungal communities under four vegetation types - monsoon forest (MF), lowland rainforest (LR), montane rainforest (MR), and artificial forest (AF) - in the Jianfengling section of Hainan Tropical Rainforest National Park. Twenty-seven 20 × 20 m plots were sampled (14 LR, 6 MF, 4 MR, and 3 AF), with five soil subsamples composited per plot from 0-20 cm. The authors combine soil physicochemical characterization with shotgun metagenomic sequencing and use alpha-diversity indices, Bray–Curtis PCoA, Pearson correlations, RDA, and Mantel tests to identify associations between vegetation type, edaphic properties, and microbial communities. The results indicate differences in several soil properties and microbial taxonomic profiles among vegetation types, with TN, TOC, and pH proposed as principal environmental drivers. The topic is relevant to tropical forest microbial ecology, and the inclusion of both natural and artificial forests is potentially valuable. However, substantial revision is required because the current manuscript repeatedly interprets ordination patterns as statistically demonstrated community differentiation, correlations as environmental “driving” effects, and differences in elemental concentrations as nutrient-retention capacity. More importantly, the strongly unbalanced sampling design (14, 6, 4, and only 3 plots) requires greater statistical scrutiny, and beta-diversity differences do not appear to have been formally tested using PERMANOVA or an equivalent permutation-based method. The manuscript also needs to address potentially serious environmental confounding among vegetation types, particularly elevation, topography, soil type, and plantation history, before attributing microbial differences primarily to vegetation.

Strong points

The study examines an ecologically important tropical rainforest system with considerable conservation value. Jianfengling contains multiple vegetation formations within a relatively restricted geographical region, providing a potentially informative setting for examining associations between vegetation and belowground microbial communities. The authors also include an artificial forest, which can provide useful information regarding differences between natural and managed forest ecosystems.

The use of shotgun metagenomic sequencing, rather than conventional 16S/ITS amplicon sequencing, is potentially a substantial strength. The Methods indicate assembly using MEGAHIT, ORF prediction with Prodigal, non-redundant gene construction with CD-HIT, and taxonomic annotation against NR using DIAMOND. This dataset could potentially provide considerably more ecological information than is presently extracted from it.

The study also evaluates several soil properties - SWC, bulk density, pH, TN, TP, TOC, Mn, and Fe - and attempts to integrate these variables with microbial-community patterns. This is appropriate because forest vegetation and microbial communities are jointly structured by multiple edaphic and environmental factors.

Finally, the authors appropriately retain the individual plot as the sampling unit rather than treating the five within-plot sampling points as independent replicates. The compositing strategy is clearly described.

Weaker aspects

The most important weakness is causal interpretation. This is an observational, cross-sectional comparison among naturally occurring vegetation types. Consequently, TN, TOC, and pH can be identified as variables associated with microbial community variation, but they cannot be established as “driving factors” from RDA, Mantel tests, or correlations alone. Vegetation type itself is also potentially confounded with elevation, topography, soil parent material, forest age/history, and other environmental variables.

A second major weakness concerns beta diversity. The authors visually interpret PCoA separation/concentration as evidence that bacterial communities differ substantially among habitats whereas fungal communities do not undergo significant succession. PCoA is an ordination/visualization method, not a statistical significance test. Contemporary microbial-community studies normally accompany Bray–Curtis ordination with PERMANOVA and assess multivariate dispersion.

A third important issue is the strongly unbalanced design: LR n = 14, MF n = 6, MR n = 4, and AF n = 3. This is particularly problematic for AF and affects ANOVA assumptions, multivariate comparisons, dispersion tests, correlations, and environmental fitting.

The manuscript also makes ecological inferences from variables that do not directly measure the claimed processes. Higher TN, TP, Mn, and Fe concentrations do not by themselves demonstrate greater nutrient retention capacity; community composition does not demonstrate ecological function; and the presence or relative abundance of a taxon cannot establish its functional contribution.

Finally, the shotgun metagenomic dataset appears underused. The manuscript is framed almost entirely around taxonomic composition and diversity, although shotgun metagenomics potentially allows analysis of functional genes and metabolic pathways. For a Microbiology-section paper, this represents an important missed opportunity.

So, I have annotated below an attempt to clarify certain ideas, but the authors should examine my suggested wording changes carefully, to be sure that I have not misinterpreted what they wanted to say.

#1. P4 L128-137: The replication is highly unbalanced: LR = 14, MF = 6, MR = 4, AF = 3 plots. Please provide the ecological and statistical rationale for this allocation. In particular, only three independent AF plots provide limited information concerning within-habitat variability and reduce confidence in comparisons involving AF.

#2. Please clarify the spatial independence of the 27 plots. What was the minimum distance between plots? Were plots from each vegetation type distributed among multiple independent forest stands, or were several plots clustered within the same stand?

This distinction is essential because multiple plots within one forest stand may constitute spatial subsamples rather than truly independent vegetation-type replicates.

#3. Please report the coordinates and elevation of each sampling plot, preferably as supplementary data. The study region has pronounced vertical zonation, and the Introduction itself describes Jianfengling as having a complex vegetation gradient. Vegetation type could therefore be confounded with elevation.

#4. This is particularly important because the manuscript repeatedly attributes differences to vegetation type. If MF, LR, MR, and AF occur systematically at different elevations, then temperature, precipitation, soil moisture, parent material, weathering, and other environmental variables may contribute to the observed microbial patterns.

#5. Please provide a substantially more detailed description of the artificial forest. Tree species, plantation age, stand density, management history, fertilization, disturbance, previous land use, and litter management are necessary for interpreting comparisons with natural forests.

#6. The same information should be provided, where possible, for the natural forest types: dominant tree species, canopy characteristics, stand structure, litter properties, and disturbance history. “Vegetation type” is presently treated as a categorical explanatory variable without adequate ecological characterization.

#7. P4 L125-126: The statement that soils are predominantly “red and yellow soils” is insufficient for an international scientific article. Please provide formal soil classification/taxonomy, preferably including the Chinese Soil Taxonomy and an internationally recognizable WRB or USDA Soil Taxonomy equivalent.

#8. Because the objective is to identify environmental factors associated with microbial communities, soil texture should also be reported. Clay content can strongly influence water retention, organic matter stabilization, nutrient availability, microbial habitat structure, and therefore microbial-community composition.

#9. The authors should clarify whether all four vegetation types occur on comparable soil parent materials and soil taxonomic units. If not, soil type itself becomes a potentially major confounder.

#10. P5 L173–186: The authors use one-way ANOVA followed by Duncan's multiple-range test. Given the strongly unequal group sizes and apparently large variance differences in several soil properties, please report formal assessment of ANOVA assumptions, particularly residual normality and homoscedasticity.

#11. Some values in Table 1 deserve verification. For example, TP is reported as 0.10 ± 0.44 g kg-1 for LR and 0.09 ± 0.34 g kg-1 for AF, where the reported dispersion greatly exceeds the mean. Please verify these values and state explicitly whether ± represents SD or SE.

#12.

Likewise, AF bulk density is reported as 1.35 ± 0.96 g cm-3, which indicates extremely high relative variability compared with the other vegetation types. Please verify the value and provide the individual observations where n = 3.

#13. Given these distributions and unequal sample sizes, the authors should consider Welch's ANOVA with an appropriate heteroscedastic post-hoc procedure, transformation, or robust/non-parametric alternatives where conventional ANOVA assumptions are not satisfied.

#14. P5 L177-178: PCoA based on Bray-Curtis dissimilarity is appropriate for visualization, but PCoA itself does not test whether communities differ statistically among vegetation types.

A PERMANOVA should be performed for bacterial and fungal community composition, reporting pseudo-F, R², P-value, number of permutations, and permutation scheme.

#15.PERMANOVA should be accompanied by PERMDISP/betadisper or an equivalent test of multivariate dispersion. Otherwise, a significant PERMANOVA result could arise partly from differences in within-group dispersion rather than centroid location.

This is particularly important with the markedly unequal sample sizes.

#16. The Abstract states that PCoA revealed “relatively pronounced bacterial differentiation” and that fungal clustering indicates vegetation type is not a prominent driver of fungal community succession. Neither conclusion can be established by visual inspection of PCoA alone.

#17. Accordingly, statements such as “bacterial communities are characterized by substantial habitat specificity” should be used only if supported by formal beta-diversity statistics. Habitat specificity itself is also a stronger ecological claim than community separation.

#18. Similarly, the conclusion that vegetation type “does not drive significant [fungal] community succession” is problematic. First, succession is a temporal ecological process, whereas this study sampled once in November 2024. Second, no temporal succession was measured.

Please replace “succession” with community composition/turnover where appropriate.

#19. The manuscript repeatedly describes TN, TOC, and pH as the “primary driving factors.” RDA, Mantel tests, and Pearson correlations demonstrate statistical association, not causation. “Environmental variables most strongly associated with microbial-community variation” would be more defensible.

#20. Please report the RDA permutation tests, including significance of the global model, individual environmental terms, and constrained axes. Merely identifying the longest arrows or strongest apparent relationships in an ordination does not establish statistical significance.

#21. The authors should assess multicollinearity among soil predictors before RDA. TN, TOC, TP, pH, Fe, Mn, SWC, and other soil variables may covary substantially. Please report VIF or an equivalent diagnostic and remove/recombine highly collinear variables.

#22. Where TN and TOC are strongly correlated, interpreting them as two independent “drivers” may be particularly misleading. Variation partitioning or partial RDA could help distinguish their independent and shared explanatory contributions.

#23. P5 L179-182: Please clarify precisely how the Mantel test was performed and reconsider the statement that it was used to “quantify the relative contribution” of different soil factors. A Mantel correlation coefficient is not equivalent to variance explained or relative contribution.

#24. If the authors wish to quantify the contribution of environmental predictors, methods such as adjusted R² from constrained ordination, variation partitioning, or another explicitly defined multivariate approach would be more appropriate.

#25. The manuscript performs numerous Pearson correlations between environmental variables, taxa, and diversity indices. Please state whether multiplicity was controlled. With many simultaneous correlations, unadjusted P < 0.05 substantially increases false-positive risk. FDR correction should be considered.

#26. Before using Pearson correlation, distributions and linearity should also be assessed. With n = 27 overall - and considerably smaller n within individual vegetation groups - individual extreme observations can strongly affect correlation coefficients.

#27. The negative relationships between TN/TOC and ACE, Chao1, and Shannon are interpreted as possible evidence of “terrestrial eutrophication,” increased selective pressure, expansion of dominant taxa, and intensified niche competition. This mechanistic sequence was not measured.

The data demonstrate correlations; they do not establish eutrophication, competition, or competitive exclusion.

#28. The proposed explanation is additionally difficult because TN and TOC did not differ significantly among vegetation types in Table 1. Please reconcile how these variables become the dominant explanatory factors and distinguish among-group differences from continuous across-plot correlations.

#29. The statement that natural forests have a stronger “nutrient retention capacity” because TN, TP, Mn, and Fe concentrations are higher than in AF is not demonstrated. Soil concentrations reflect the net outcome of inputs, outputs, mineral weathering, plant uptake, litter return, leaching, parent material, and soil mass.

#30. Moreover, the Abstract states that natural forests contained higher TN, TP, Mn, and Fe, although TN was not significantly different among vegetation types according to Table 1. This apparent contradiction should be corrected throughout the Abstract, Results, Discussion, Simple Summary, and Conclusions.

#31. If nutrient stocks are to be compared, concentrations alone are insufficient. The authors would need to consider bulk density and sampling depth to estimate element stocks per unit land area. Even then, “retention capacity” would require stronger process-based evidence.

#32. The manuscript states that the common dominant phyla indicate their “broad adaptability” to different soil ecosystems. The presence of Pseudomonadota, Actinomycetota, Acidobacteriota, and Verrucomicrobiota across four forest types is compatible with broad occurrence, but does not experimentally demonstrate adaptability.

#33. Likewise, identifying dominant phyla does not demonstrate their “critical ecological functions.” Taxonomic abundance and ecological function should be distinguished unless functional genes/pathways are explicitly analyzed.

#34. This issue is especially important because the authors performed shotgun metagenomic sequencing, which can potentially provide functional information. Yet the manuscript appears to exploit the sequencing data primarily for taxonomic profiles.

The authors should explain why functional gene/pathway analyses were not included or, preferably, add a carefully controlled functional metagenomic analysis relevant to C, N, and P cycling.

#35. If functional annotation is added, predicted gene abundance should still not be equated automatically with process rate or gene expression. Metagenomic DNA indicates functional potential; it does not establish transcription or biogeochemical flux.

#36. P5 L158 - 171: Please provide comprehensive metagenomic quality statistics for each sample: raw reads, clean reads, bases retained, assembly size, N50, number of contigs, mapping rate, predicted ORFs, non-redundant gene counts, and taxonomic annotation rate.

#37. The quality-filter description also requires clarification. The phrase “ambiguous base ‘N’ < 5 bp” appears potentially inconsistent with a read-removal criterion. Please provide the exact fastp parameters rather than an ambiguous verbal description.

#38.  Please explain how gene abundances were normalized before calculating taxonomic relative abundances. Were gene length and sequencing depth accounted for? This is essential when comparing metagenomic abundance among samples.

#39. Please clarify whether alpha-diversity indices were calculated from species-level taxonomic profiles derived from shotgun reads/genes, and describe rarefaction or other sequencing-depth normalization procedures.

#40. The authors should provide sequencing-depth/coverage information demonstrating that differences in richness (ACE/Chao1) were not driven by unequal sequencing effort.

#41. The interpretation of the Simpson index requires particular attention. There are multiple Simpson formulations (D, 1- D, 1/D) with opposite directional interpretations. The manuscript should state the exact equation used and what a higher value represents.

#42. This is essential because the manuscript simultaneously reports lower/higher Shannon and Simpson values and interprets them as diversity responses. Without defining the Simpson formulation, the biological interpretation cannot be evaluated reliably.

#43. The manuscript states that fungi generally prefer “slightly acidic environments.” The sampled soils span only a relatively narrow acidic range, approximately pH 5.0–5.5. These data do not establish a general fungal pH optimum.

#44. Furthermore, different fungal taxa and functional guilds can respond differently to pH. Please avoid translating a community-level correlation into a universal physiological preference of fungi.

#45. The manuscript should distinguish alpha diversity from beta diversity more carefully. A vegetation type can show no difference in Shannon richness/evenness metrics while still differing substantially in taxonomic composition. Therefore, the bacterial results are not intrinsically contradictory.

#46. Conversely, significant differences in fungal alpha diversity do not imply that vegetation type restructures fungal composition. Alpha and beta diversity describe different ecological dimensions and should be interpreted independently before integration.

#47. because sampling was performed only once, seasonality cannot be separated from vegetation effects, and the results represent microbial communities at the November 2024 sampling date rather than stable year-round forest microbiomes.

#48. The authors should explicitly recognize that the study is observational rather than experimental. “Effects of vegetation type” and “vegetation-driven differentiation” imply stronger causality than the design permits. “Differences associated with vegetation type” is preferable.

#49. Similarly, the manuscript cannot determine whether vegetation altered soil properties and thereby microbial communities, whether pre-existing edaphic differences contributed to vegetation distribution, or whether both processes operate simultaneously.

#50. A conceptual framework acknowledging vegetation ↔ soil ↔ microbial feedbacks, rather than a unidirectional vegetation → soil → microbiome mechanism, would substantially improve the ecological interpretation.

#51. the claim that these findings provide a scientific basis for “effective conservation and restoration strategies” is too strong. No restoration treatment, ecosystem function, management intervention, or conservation outcome was tested.

#52. The practical implication should therefore be reframed as providing baseline information on soil microbial-community variation across Jianfengling vegetation types, potentially useful for future monitoring and restoration research.

53. (Conclusions): Several statements should be moderated. In particular, higher nutrient concentrations do not establish retention capacity; common dominant phyla do not demonstrate high adaptability; PCoA does not demonstrate habitat specificity; cross-sectional sampling cannot establish community succession; and correlations do not identify causal environmental drivers. These issues currently recur almost verbatim in the Abstract and Conclusions.

#54. Page 11, Lines 342-356 (Discussion – vegetation/management effects on microbial community structure)

NOTICE - Suggested insertion (after the discussion that different vegetation types may contribute to microbial-community differentiation): “The association between plant cover and belowground microbial organization is also supported by evidence from contrasting plant-management systems, in which changes in plant composition and rotation can restructure soil microbial communities, emphasizing the close coupling between aboveground vegetation and belowground microbial assemblages (Cezar et al., 2025 - https://doi.org/10.1590/1678-992X-2024-0134).

#55. 

Page 11-12, Lines 356-373 (Discussion – soil physicochemical conditions as correlates of microbial communities)

NOTICE - Suggested insertion: “These relationships are consistent with evidence that changes in soil conditions associated with plant and management interventions can modify microbiological attributes, reinforcing the need to interpret microbial-community variation within the broader soil–plant environmental context rather than as an isolated response to vegetation identity (Galina et al., 2024 - https://doi.org/10.1590/1678-992X-2023-0214).” 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Brief summary.

The manuscript presents a methodologically sound descriptive study investigating the influence of vegetation type (three natural forest types and artificial plantations) on the structure and diversity of soil bacterial and fungal communities in the Jianfengling tropical forest (Hainan Island). The study employs modern methodologies including shotgun metagenomic sequencing, multivariate statistical approaches (PCoA, RDA), and correlation analysis. Relevance of the topic is beyond question: understanding the linkage between aboveground vegetation and belowground microbial pools is critically important for predicting the consequences of climate change and anthropogenic transformation of tropical ecosystems. The authors justifiably point to the lack of comprehensive studies covering four vegetation types simultaneously in this region, including artificial forests.

Manuscript is clear, relevant for the field and presented in a well-structured manner. The review is clear, comprehensive and of relevance to the field cited references are current. The manuscript is scientifically sound and is the experimental design appropriate to test the hypothesis. The data are interpreted appropriately and consistently throughout the manuscript. The conclusions are consistent with the evidence and arguments presented.

 

Specific comments:

 129-130 - Section 2.2 (p. 4).  Replication numbers by vegetation type: LR — n=14, MF — n=6, MR — n=4, AF — n=3. With n=3 for artificial forests (AF), any parametric tests (ANOVA, t-test) involving this group are statistically invalid. The standard error of the mean for AF is underestimated, leading to a risk of false-positive results (Type I error) in pairwise comparisons. Conclusions about differences between AF and natural forests, as stated in the Abstract («Natural forests (mon- soon forests, lowland rainforests, and montane rainforests) had a significantly greater capacity to retain soil nutrients than that of artificial forests») (19-21,  Simple Summary) lack rigorous statistical support given this design.

Sections 2.3.3 and 3.3 (pp. 5, 7) Absence of Rarefaction Analysis (Substantial). ACE and Chao1 are extrapolation-based estimates of species richness and are critically dependent on sequencing depth. In this study, shotgun metagenomic data were assembled into contigs, followed by ORF prediction and construction of a non-redundant gene set, after which relative abundances were calculated as the proportion of annotated genes assigned to each taxon. Under this approach, it is essential to demonstrate that sequencing depth was sufficient to achieve saturation of gene discovery for each sample. The manuscript does not provide:  rarefaction curves (or gene accumulation curves) demonstrating saturation of gene discovery across samples; Good's coverage estimates or similar metrics.

Without this information, it remains unclear whether the observed differences in ACE and Chao1 indices reflect genuine biological variation or artifacts arising from incomplete sampling of genetic diversity in some samples.

 

 

Fig. 5C.  pp. 9–13 -- RDA shows that Mn explains >40% of the variation in bacterial communities, yet this crucial finding is virtually unexplored in the Discussion. Mn is a key micronutrient involved in redox processes, enzymatic reactions (peroxidases, superoxide dismutases), and serves as a cofactor for many enzymes.

Table 1 (p. 6) - Extremely high standard deviation for TOC in MR (±12.29 with mean 18.08) — the spread is 68% of the mean, indicating high heterogeneity within this forest type, which requires commentary in the text.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is quite interesting, and, according to the authors, this is the first time such studies have been conducted in this region, although they themselves refer to the works of Huang et al. and Lin et al. So, is it the first time or not? I’m a bit confused. Metagenomic sequencing, soil analysis, and multivariate statistical studies — all of this indicates a high methodological level. Overall, it’s a worthy study that can be published, but there are a number of minor issues. The history of artificial forests is unclear — what is the history of land use there? Perhaps the data found regarding the genera and their predominance is the result of previous human activity? Introduced with seeds or tree seedlings? Here lies the biggest flaw in the article’s completeness; perhaps it would be worth adding a short section about the specific history of these particular forest stands? This also raises doubts about a single sampling during one season — the soil microbiota may change depending on the season.

Line 82 — A dubious statement. Many bacteria function excellently in rock formations, in the depths of oil wells, in the conditions of the Far North, etc. In environments that are poor, with low nutrient content and limited resources.

Line 107 – The study is of high value; this statement is indeed true — I couldn’t find any other similar works in the literature.

Line 122 – AF (Artificial) forests — what kind of forests? Plains, mountains, monsoon forests? It is still necessary to provide some kind of characterization of them.

Line 126 — Here, it is still necessary to describe the soil characteristics, at least briefly.

Line 132 — What is the basis for choosing a depth of 20 centimeters?

Line 134 — Is it really necessary to remove roots and stones? Could you have lost some of the microbes that form associations with plants or biofilms on stones?

Line 140 — It is necessary to attach a high‑resolution image (Study Area) to the Supplement; the zone boundaries merge in the selected colors.

Line 192 — Regarding this result, it is necessary to draw some conclusions in the conclusions section. I didn’t see any assumptions about why this is the case.

Line 199 — Could this be related to the large number of such bacteria that accumulate Mn and Fe?

Line 219 — The predominant genera are clearly visible on the diagrams, while the others are not so obvious. Perhaps it would be better to introduce a separate table with percentages for other genera and also analyze the differences between them for different types of forest?

Line 245 — Were coniferous plants present in these biotopes? There should have been more fungi. The vegetation, both coniferous and deciduous, affects the number of mushrooms.

Line 303 — Could it be because these are young forests with an unstable community, rather than the result of anthropogenic impact?

The work requires minor corrections before publication. But in my opinion, it is worthy of publication in the chosen journal.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear All,

The authors have made substantial efforts to address the comments raised during the first review. The revised manuscript includes a more balanced sampling design, additional information on plot separation and elevation, improved characterization of the artificial forest, formal soil classification, and more comprehensive statistical analyses. The inclusion of PERMANOVA, PERMDISP, multicollinearity diagnostics, false-discovery-rate correction, and environmental-variable significance testing represents meaningful progress.

The authors have also moderated several previously unsupported ecological interpretations, particularly those concerning nutrient retention, microbial adaptability, causal environmental drivers, and conservation outcomes. These changes have improved the scientific presentation.

Nevertheless, several important methodological and inferential concerns remain unresolved. Most notably, the revised study combines metagenomic data originating from different research teams, does not adequately standardize sequencing effort for alpha-diversity estimation, and continues to face substantial confounding between vegetation type and elevation. Additional inconsistencies in the multivariate analyses and supplementary information require correction.

The manuscript has improved, but the remaining issues affect the reliability and interpretation of its principal findings.

Remaining Major Comments

1. Integration of metagenomic datasets and potential batch effects

The authors' response reveals that the artificial-forest metagenomic data include measurements generated by other research teams, although the samples were reportedly collected during the same period. This is important methodological information that should have been clearly disclosed in the original manuscript.

The authors must specify whether all 35 samples were processed using identical protocols for soil collection, DNA extraction, library preparation, sequencing, quality filtering, assembly, gene prediction, and taxonomic annotation.

If the samples originated from different sequencing batches or analytical workflows, potential batch effects must be assessed. Differences attributed to vegetation type could otherwise reflect technical variation, particularly if batch and vegetation category are confounded.

The authors should provide a sample-level metadata table identifying the origin, sequencing batch, and analytical workflow for each sample. Where appropriate, sensitivity analyses should establish whether the principal community-composition results remain consistent after accounting for technical differences.

If vegetation type and sequencing batch are completely confounded, this limitation cannot be eliminated statistically and must be explicitly acknowledged.

2. Alpha-diversity estimation and sequencing-depth standardization

The authors clarify that bacterial and fungal alpha-diversity indices were calculated from genus-level taxonomic abundance profiles derived from shotgun metagenomic sequencing, without rarefaction or another sequencing-depth standardization procedure.

This remains a significant concern, particularly for richness estimators such as ACE and Chao1. Differences in sequencing depth, taxonomic assignment efficiency, and the number of reads assigned to each microbial group can influence apparent richness independently of biological diversity. Although Good's coverage values have been supplied, high coverage alone does not demonstrate that richness estimates are comparable across samples or that sequencing effort has no effect on the results.

The authors should re-evaluate alpha diversity using a clearly justified approach that accounts for differences in sampling effort and taxonomic detection. They should also report the number of reads assigned to bacteria and fungi separately for each sample, together with the relevant coverage or completeness diagnostics.

The distinction between genus-level richness and species richness must be maintained throughout the manuscript. Conclusions about fungal diversity differences should be reconsidered if the results are not robust to appropriate standardization.

3. Confounding between vegetation type, elevation, and soil characteristics

The revised manuscript now reports the elevation ranges of the four vegetation types, which is a welcome improvement. However, montane rainforest occurs at 875-1130 m, whereas the other vegetation types occupy substantially lower elevations.

Consequently, vegetation identity and elevation remain strongly associated. Differences in microbial communities may reflect temperature, moisture regime, soil development, parent material, or other environmental gradients rather than vegetation type itself. Although elevation has been incorporated into the environmental analyses, its inclusion as a predictor does not, by itself, establish an independent vegetation effect.

The authors should assess the extent of vegetation-elevation confounding and, where the data permit, evaluate whether vegetation-associated community differences persist after accounting for elevation. The interpretation should distinguish comparisons among overlapping elevation ranges from those involving montane rainforest, for which direct environmental comparability is limited.

The absence of soil texture and parent-material information should also be explicitly recognized as a limitation rather than treated as a resolved methodological issue.

4. PERMANOVA, multivariate dispersion, and consistency of reported results

The addition of PERMANOVA and PERMDISP is appropriate. However, the revised manuscript does not consistently report the corresponding statistics.

The Abstract reports a fungal PERMANOVA p-value of 0.1117, whereas the Results report p = 0.1777. These values must be reconciled. Moreover, the authors' response indicates that both PERMANOVA and PERMDISP yielded significant results in support of bacterial community differentiation. A significant PERMDISP result, however, indicates differences in within-group multivariate dispersion; it does not independently support habitat specificity or separation of group centroids.

The authors should report the PERMANOVA pseudo-F, R², p-value, and permutation procedure, together with the PERMDISP statistic and p-value, for both bacterial and fungal communities. The interpretation must distinguish location effects from dispersion effects.

If significant differences in dispersion are present, particularly with unequal group sizes, conclusions concerning vegetation-associated community differentiation must be appropriately qualified.

5. Environmental-variable importance and interpretation of constrained ordination

The addition of VIF screening and permutation-based significance testing is valuable. Nevertheless, the revised manuscript continues to describe environmental variables as the most influential factors without consistently distinguishing explanatory ranking from statistical significance.

For the fungal community, the Results explicitly indicate that none of the individual environmental variables remained significant after BH-FDR correction. Despite this, TOC is subsequently described as dominating fungal community composition.

This conclusion is stronger than the statistical evidence supports. TOC may have the highest observed marginal explanatory value, but this does not establish a statistically supported independent effect. Furthermore, a Monte Carlo permutation test does not itself quantify an environmental variable's relative contribution. The authors should identify the statistic used to measure explanatory power, clarify whether tests were marginal or conditional, and distinguish unadjusted from adjusted R² where applicable.

The significance of the overall RDA model, constrained axes, and individual environmental terms should be reported consistently. The authors should also avoid interpreting correlated environmental predictors as independent ecological drivers.

6. Completeness and internal consistency of the metagenomic supplementary information

The supplementary sequencing-quality table now contains 35 samples, corresponding to the revised sampling design. However, the samples are identified only by numbers, without assignment to vegetation type, sampling plot, or sequencing batch.

This prevents independent verification of the relationship between sample-level sequencing quality and the biological comparisons.

The mapping rates also vary substantially among samples, ranging from approximately 7.8% to 32.8%. The authors should clarify what reference or assembly was used to calculate these values, explain their relatively low magnitude, and assess whether variation in mapping efficiency could affect downstream taxonomic profiles.

The supplementary tables also contain an evident labeling error: Table S2 is identified as fungal relative abundance but lists bacterial phyla, whereas Table S3 is identified as bacterial relative abundance but lists fungal phyla.

These errors must be corrected. The supplementary information should also include sample-level taxonomic assignment statistics and the relevant normalized abundance data or an accessible repository reference.

Remaining Comments

7. Interpretation of natural versus artificial forests

The revised manuscript provides useful information concerning the artificial forest, including the dominance of Pinus caribaea and its management history. However, the natural forests remain insufficiently characterized in terms of stand structure, dominant tree species, litter properties, and disturbance history.

Consequently, differences between natural and artificial forests should not be attributed solely to vegetation origin or management.

The statement that natural forests had higher nutrient concentrations should also distinguish numerical differences from statistically significant differences, particularly for TN and TOC, which did not differ significantly among vegetation types.

8. Ecological interpretation of microbial diversity

The authors have appropriately clarified the direction of the Simpson index, with higher values indicating lower diversity under the formulation employed. However, the Results and Discussion should consistently distinguish richness, evenness, and diversity indices. Opposite trends in Shannon and Simpson indices require careful interpretation and should not be reduced to a general statement that one vegetation type supports greater microbial diversity.

Furthermore, differences in taxonomic profiles cannot independently establish microbial ecological function, nutrient cycling rates, or ecosystem stability.

9. Data availability and reproducibility

The manuscript should provide sufficient information to reproduce the principal analyses, including sample metadata, the taxonomic abundance matrix, environmental measurements, statistical procedures, and the sequencing-data accession numbers.

A statement that the relevant contributions are included in the article and supplementary materials is insufficient if the underlying sequencing reads and sample-level datasets are not accessible.

The authors should deposit the raw metagenomic sequences in an appropriate public repository or provide a clear, justified explanation for any access restrictions.

Minor Comments

• Abstract: Reconcile the fungal PERMANOVA p-value with the Results and avoid describing nonsignificant environmental associations as established explanatory effects.

• Title and terminology: The revised title appropriately replaces causal “driving factors” with “associated environmental variables.” This terminology should be maintained throughout the manuscript.

• Methods: Clarify the abundance-normalization procedure in the manuscript itself, rather than only in the response letter.

• Results: Avoid interpreting the absence of a statistically significant difference as evidence that communities are equivalent.

• Supplementary Tables S2 and S3: Correct the reversed bacterial and fungal headings.

• Language: Further English-language editing is necessary to improve precision, particularly in the statistical and ecological interpretations.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf