Review Reports - Dynamic Interactions of Stand Characteristics and Site on <i>Quercus</i> spp. Volume in China Under Climate Change

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors General comments:

The English is fine. I have highlighted some minor typos in the specific comments.

The article analyses a big dataset, from almost 66000 plots with Quercus spp. from China, from three consecutive periods of NFIs.

The comparison between artificial and natural stands, as well as between different elevation gradients, is interesting.

However, some aspects need clarification.

It is not clear in Figures 3B, 3C, 4, 5 and 6 if the values are from the last NFI or an average of all three. Considering these are permanent plots, at least for the NF, some of them should be found in all three periods. The authors seem to have treated them as different.

You only have data for 15 years. Please be more explicit when addressing the link between your data with climate change.

If I understood correctly, from ~66000 plots, you have aggregated the data to 42-50 cases (total count from Table 1). Please give a short explanation of why you did that in the article.

The authors sometimes present AF and NF separately, sometimes sum them up. As for example figure 5: In 5A I presume they are combined, whereas in 5B they are divided.

A list of the Quercus species used in this study would be nice (even in the Supplementary file).

From Figure 6, for the AF stands, it seems that AC 3 (41-50 years old) for mid-elevation and AC 5 (higher than 71 years old) have a stand volume of 0 or close to it. Are they exploited at this age? It would be nice to have a paragraph regarding the rotation age of Quercus species in China. If artificial and natural stands have different exploitation ages, that would be even better to add.

In the Discussion section, a paragraph comparing the stand volume of Quercus species with other species from China could be useful.

Please also address the different number of cases for different altitudes, from 2 for high altitudes to around 30 cases in low altitudes.

Specific comments:

Keywords: I’m not sure if “National forest inventory” is relevant.

L17-18: The last year of meteorological records is 2021, not 2020/2022.

L56: habitat

L61: variabilis in italic font

L96-99: Xu et al. is not cited properly. Neither 7 nor 16 represents their article.

L115-116: Presenting the formulas and parameters would be preferable. That way, the reader would not have to search for them.

L117: “The per-hectare stand volume of Quercus spp. across different age classes and provinces was then aggregated” - some minor details about the resulting X cases or provinces analysed would add clarity to the manuscript.

L121-130: A map and/or table would improve the readability of that long paragraph.

Figure 1: A scatter plot of the Mean Annual Temperature would be a nice addition to the figure.

Figure 3: They are not changes, maybe "differences".

Figure 3 B and C: Are the values from the last NFI, or an average of the three of them?

Same for Figure 4…

Table 1:

Some columns with the minimum and maximum age would be nice

For the first two periods, some standard deviations of AF are higher than the mean, which could indicate skewness or outliers in your data.

Also, it seems there are AF plots where the stand volume/ha is 0. I understand that seedlings planted do not have a DBH higher than 5 cm. Please explain the reasoning for keeping them in the analysis.

You should address these issues in the Discussion section.

Figure 5: Again, are the values from all three NFIs?

L226: I would delete: “in Climate Change 2022: Impacts, Adaptation and Vulnerability”.

L232: while, not While.

L241: delete: “While”

Figure 7: should be in the Results section.

L 346-348: “Specifically, natural forests exhibited significantly higher stand volume than plantations (p<0.05), with a consistent elevational pattern of high-elevation > mid-elevation > low-elevation stands.”

It’s contradicted by Figure 6 and Figure 7 (which is on the same page). Or are the colour codes wrong?

Author Response

It is not clear in Figures 3B, 3C, 4, 5 and 6 if the values are from the last NFI or an average of all three. Considering these are permanent plots, at least for the NF, some of them should be found in all three periods. The authors seem to have treated them as different.

Response: Many thanks for the comments. The data presented in the figures represent means derived from three National Forest Inventory (NFI) cycles, as specified in the captions of Figures 3B, 3C, 4, 5, and 6. To mitigate pseudo-replication concerns in the permanent plot data, our analysis compared several methodological approaches, with line mixed-effects models (LMM) providing a robust countermeasure against spatiotemporal non-independence. For stand volume, the fixed effects in the model included elevation gradient, stand origin, survey period, and their interaction terms. Random effects were specified using plot ID to account for intrinsic, time-invariant plot characteristics and the correlation structure introduced by repeated measurements. Results from the LMM analysis have been incorporated into the results section, with corresponding interpretation provided in the discussion.

You only have data for 15 years. Please be more explicit when addressing the link between your data with climate change.

Response: Many thanks for this important comment. We acknowledge that the 15-year timeframe of our forest inventory data (2004-2018) presents limitations for detecting long-term ecological trends primarily driven by climate change. Although our analysis revealed significant temporal changes in stand volume and we conducted corresponding examinations, a further limitation lies in the spatial resolution of the temperature reanalysis data, which may not precisely match the scale of individual plots. It is important to note that our plot data cover a national scale, and as a large-scale spatial pattern study, our findings can still offer preliminary evidence and directional insights into the potential influence of climate change on forest volume. We have included a dedicated discussion of these limitations related to data and scale in the manuscript and hope this response adequately addresses the reviewer's concerns.

If I understood correctly, from ~66000 plots, you have aggregated the data to 42-50 cases (total count from Table 1). Please give a short explanation of why you did that in the article.

Response: Thank you for this observation. You are correct that the data were aggregated from individual plots to provincial-level means. This approach was adopted because the original data were structured as official provincial-level statistics, where plot-level measurements had already been summarized by regional forestry authorities for reporting purposes. Consequently, the analysis unit in our study becomes the province, resulting in the number of cases presented in Table 1. We have clarified this data structure and the reason for using provincial-level aggregation in the revised manuscript.

The authors sometimes present AF and NF separately, sometimes sum them up. As for example figure 5: In 5A I presume they are combined, whereas in 5B they are divided.

Response: Many thanks for this keen observation. We did employ different data presentation formats across the figures. In Figure 5A, data from artificial and natural forests were combined to highlight the significant interactive effect between time and elevation on Quercus spp. stand volume, irrespective of forest origin. In Figure 5B, data are presented separately to clearly illustrate the significant interaction between forest origin and age class. This dual presentation strategy allows us to address distinct scientific questions within the same study. We have clarified the rationale for these presentation choices in the caption of Figure 5 and the corresponding methods section in the revised manuscript to prevent any potential confusion.

A list of the Quercus species used in this study would be nice (even in the Supplementary file).

Response: Many thanks for the comment. We have added this table in the supplementary materials and included a direct reference to it in the main text.

From Figure 6, for the AF stands, it seems that AC 3 (41-50 years old) for mid-elevation and AC 5 (higher than 71 years old) have a stand volume of 0 or close to it. Are they exploited at this age? It would be nice to have a paragraph regarding the rotation age of Quercus species in China. If artificial and natural stands have different exploitation ages, that would be even better to add.

Response: We thank the reviewer for this insightful observation. The reviewer is correct in noting the significantly reduced stand volumes for mid-elevation AC3 (41-50 years) and AC5 (>71 years) plantations in Figure 6. This pattern is indeed a direct result of management practices. As we have now clarified in the revised manuscript, China's national forestry standard (LY/T 2908-2017) sets the official rotation age for oak timber plantations at 51 years. The low volume in AC3 (41-50 years) occurs because these stands are approaching this legal harvesting age and are often subject to preparatory cutting. Notably, the stand volume in AC4-5 (>51 years) does not approach zero completely, which reflects China's current forest policy emphasis on ecological conservation. In plantation management, there is an active exploration of close-to-nature silviculture practices that gradually replace clear-cutting operations, allowing some older plantations to be retained or managed under more natural regimes. This regulated harvesting cycle, which creates the characteristic "stepwise decline" in plantation volume, stands in stark contrast to natural forests that are often protected under conservation policies and can develop over much longer periods, in some cases exceeding 150 years to reach quantitative maturity.

In the Discussion section, a paragraph comparing the stand volume of Quercus species with other species from China could be useful.

Response: Many thanks for this suggestion. After careful consideration, we have decided to retain the current focus of the Discussion on the dynamics within Quercus spp. The primary reason is that introducing comparisons with other tree species, each with unique growth patterns and influenced by different factors, could complicate the narrative and distract from the key mechanisms we aimed to highlight for Quercus spp. We believe that a focused discussion best serves the objectives of this paper.

Please also address the different number of cases for different altitudes, from 2 for high altitudes to around 30 cases in low altitudes.

Response: Many thanks for the comment. This disparity primarily stems from the inherent limitations of the National Forest Inventory's (NFI) systematic sampling design and the natural distribution pattern of Quercus spp. forests in China. Unlike the extensive oak forests found at mid and low elevations, natural Quercus stands at high elevations are inherently scarce and fragmented, which inevitably leads to a smaller number of sample plots. Furthermore, regarding artificial Quercus forests at high elevations, a comprehensive review of the literature and field observations confirm that they remain uncommon in practice. Any existing stands are typically in newly established or early growth stages. In this revision, we have chosen to represent these nascent stands with a value of zero in the dataset. We consider this approach scientifically more transparent than complete data omission, as it accurately reflects the current status of such stands, thereby avoiding potential misinterpretations that could arise from their exclusion.

Specific comments:

Keywords: I’m not sure if “National forest inventory” is relevant.

Response: Many thanks for this comment. We have revised the keywords accordingly.

L17-18: The last year of meteorological records is 2021, not 2020/2022.

Response: We apologize for the oversight. The necessary correction has been made.

L56: habitat

Response: Sorry for the mistake. We have revised it.

L61: variabilis in italic font

Response: Sorry for the mistake. We have revised it.

L96-99: Xu et al. is not cited properly. Neither 7 nor 16 represents their article.

Response: We apologize for this misunderstanding. The citation markers in the text were indeed correct; however, an automated parsing error in Zotero reversed the family names and given names of the Chinese authors in the reference list. We have now manually corrected all affected entries to ensure accuracy.

L115-116: Presenting the formulas and parameters would be preferable. That way, the reader would not have to search for them.

Response: We thank the reviewer for this suggestion. As revised in our methodology, this study utilizes officially published provincial-level forestry statistics in which plot-level data had already been aggregated into mean stand volume values by provincial authorities. Therefore, the unit of analysis is the province, and the number of cases reflects this level of aggregation. Regarding the formulas and parameters, the compilation of provincial data involved multiple localized standards and equations, which cannot be comprehensively listed due to space constraints and their regional specificity. We have ensured that the data processing and aggregation procedures are clearly described.

L117: “The per-hectare stand volume of Quercus spp. across different age classes and provinces was then aggregated” - some minor details about the resulting X cases or provinces analysed would add clarity to the manuscript.

Response: Many thanks for the comment. We have revised the manuscript to clarify that the aggregation process integrated data across 30 provinces, five age classes, and two stand origins, incorporating repeated measurements and resulting in a comprehensive dataset of 745 cases for analysis. This clarification has been added to the Methods section.

L121-130: A map and/or table would improve the readability of that long paragraph.

Response: Many thanks for the comment. We have added a new table (Table 1) summarizing the classification of provinces into three elevations, which significantly improves the clarity and readability of the methodology.

Figure 1: A scatter plot of the Mean Annual Temperature would be a nice addition to the figure.

Response: Many thanks for this suggestion. We have revised Figure 1 to include a scatter plot of Mean Annual Temperature, which enhances the visualization of climate gradients across the study area.

Figure 3: They are not changes, maybe "differences".

Response: Many thanks for this comment. We have revised it.

Figure 3 B and C: Are the values from the last NFI, or an average of the three of them?Same for Figure 4…

Response: Many thanks for the comment. The values in Figures 3B, 3C, and 4 represent the average stand volume calculated across all three NFI cycles. Following your previous comment, we have revised these figures and, crucially, have now explicitly stated this data source in the captions of Figures 3B, 3C, 4, 5, and 6 to prevent any potential confusion.

Table 1:

Some columns with the minimum and maximum age would be nice

Response: We thank the reviewer for this suggestion. We believe there might be a misunderstanding regarding Table 1, which provides a statistical description of stand volume across temporal, elevational, and origin gradients. As this table does not encompass age class or stand age as analytical dimensions, the addition of minimum and maximum age columns falls beyond its current scope..

For the first two periods, some standard deviations of AF are higher than the mean, which could indicate skewness or outliers in your data.

Response: Many thanks for this insightful observation. The reviewer is correct that the high standard deviations in artificial forest (AF) data during the initial periods could indicate skewness, which may indeed be influenced by the presence of zero values. To address this issue more comprehensively, we have introduced two additional modeling approaches in our revision: Linear Mixed-effects models (LMM) and Hurdle models. These are now analyzed alongside the Multi-factor ANOVA results. Specifically, the Hurdle model effectively accounts for and minimizes the potential bias caused by zero values in the dataset, providing a more robust interpretation of the stand volume patterns in artificial forests.

Also, it seems there are AF plots where the stand volume/ha is 0. I understand that seedlings planted do not have a DBH higher than 5 cm. Please explain the reasoning for keeping them in the analysis.

Response: Many thanks for the comment. The inclusion of plots with zero stand volume was a deliberate decision based on our data structure, ecological considerations, and valuable input from reviewers regarding data consistency, as clarified in the revised Methods. First, our analysis uses officially aggregated provincial-level NFI statistics that inherently include all original plot records to ensure representativeness; removing zero-volume plots would compromise data integrity. Second, excluding these plots would misrepresent field conditions—zero values accurately reflect real ecological outcomes such as failed plantation establishment or young trees below the 5 cm DBH threshold in high-elevation areas. Finally, to appropriately handle this zero-inflated distribution, we incorporated hurdle models, which separately model stand presence-absence and positive volume values, as noted in our previous response.

You should address these issues in the Discussion section.
Figure 5: Again, are the values from all three NFIs?

Response: Many thanks for the comment. The values in Figures 5 represent the average stand volume calculated across all three NFI cycles. Following your previous comment, we have revised these figures and, crucially, have now explicitly stated this data source in the captions of Figures 3B, 3C, 4, 5, and 6 to prevent any potential confusion.

L226: I would delete: “in Climate Change 2022: Impacts, Adaptation and Vulnerability”.

Response: Many thans for this comment. We have deleted it.

L232: while, not While.

Response: Sorry for the mistake. We have revised it.

L241: delete: “While”

Response: Many thanks for this comment. We have deleted it.

Figure 7: should be in the Results section.

Response: Many thanks for the comment. The three-way interaction between time, elevation, and origin on stand volume was statistically non-significant. Additionally, reviewers rightly pointed out that the short-term temporal dynamics could not adequately capture this level of variation. Weighing these considerations, we have decided to remove the corresponding interaction plot. The Discussion section has also been revised accordingly to reflect this change.

L346-348: “Specifically, natural forests exhibited significantly higher stand volume than plantations (p<0.05), with a consistent elevational pattern of high-elevation > mid-elevation > low-elevation stands.” It’s contradicted by Figure 6 and Figure 7 (which is on the same page). Or are the colour codes wrong?

Response: We thank the reviewer for pointing out this inconsistency. We apologize for the error in the original statement regarding the elevational pattern. The claim of a "consistent elevational pattern of high-elevation > mid-elevation > low-elevation" was indeed a misjudgment based solely on the initial ANOVA results. As the reviewer correctly observed, this pattern is not consistently supported by the data presented in Figures 6 and 7. In response, we have thoroughly re-analyzed the data using both Linear Mixed-Effects Models (LMMs) and Hurdle Models. These more robust statistical approaches, which better account for the hierarchical structure of our nationwide data and the zero-inflated distribution of stand volume, demonstrate non-significant direct differences across elevation zones. However, they reveal significant interactive effects on stand volume among elevation, origin, and age classes. This revised analysis reveals that the relationship is more complex than a simple elevational gradient, being driven primarily by these multi-factor interactions rather than by elevation alone. The text has been corrected and the entire manuscript has been revised accordingly to ensure all conclusions align with the results from these improved analytical methods. We believe the revised analysis and text now provide a more accurate and nuanced interpretation of the elevational trends in Quercus stand volume.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript has an interesting topic, but still needs some improvements, especially to strengthen and clarify the authors' statements and justify the methods used.

INTRODUCTION

Line 35: Please, directly relate the impact of global warming to trees/forests (e.g., phenological shifts, growth response, etc.).

L.48: Haven't there been regional studies on this? The limited ones are at the national level?

L.58: It would be better to specify which Quercus species are dominant and the focus of this paper to clarify the ecological context. The use of "Quercus spp." is too broad (more than 30 species in China).

L.76: This hypothesis assumes a significant trend at high elevations, but the previous paragraphs lack a strong empirical basis; the hypothesis of an increasing trend across all regions overlaps with the first hypothesis. Consider using the following hypotheses:

H1: Stand volume of Quercus spp. varies significantly with elevation under climate warming. H2: Interactive effects of forest origin and age modify this elevational response.

MATERIALS AND METHODS

L. 86: Using NCEP 2.5°, the data are too coarse for spatial analysis of NFI plots (~0.06 ha). The interpolation or downscaling method used by the authors should be explained to make the climate data relevant to the forest scale.

L. 101: The NFI only covers 2004–2018, but the results are linked to climate trends from 1948–2021 (L. 91) without analytical explanation.

L. 124: The classification of provinces based on “core zone elevation” is not representative of the actual plot data. The authors should consistently use the actual elevation of NFI plots (mean elevation per plot) rather than administrative categories for ecological validity.

RESULTS

L. 155: Where does this 0.01℃ value come from?

L. 172: Although it states “...no significant differences (p>0.05) across the three survey periods.” However, it then states an increase from 51.5 to 65.5 m³/ha (a 27% increase). If it is not significant, the variance is likely large, so it is best to present the SD, F, df, and p-value.

L. 197: The statement "initial increase-decrease-rise" is ambiguous without significant figures or p-values between age classes. It is best to present ANOVAs per elevation and age with F values and post hoc grouping.

DISCUSSION

L. 220: This statement should be supported with data on the inter-seasonal trend difference from Fig. 2 in the results section to provide a solid discussion.

L. 239–241: If the results are consistent with the "Long study in Hunan," it is best to provide a correlation or R² value between elevation and volume to support this claim.

L. 254–258: Please add citations.

L. 324: The first hypothesis was not proven (p>0.05), but it is still stated that "essential patterns were captured."

Abstract and conclusion should be adjusted according to the results of the improvements. The conclusion will be more concise and focused on answering the research objectives.

Author Response

This manuscript has an interesting topic, but still needs some improvements, especially to strengthen and clarify the authors' statements and justify the methods used.

INTRODUCTION

Line 35: Please, directly relate the impact of global warming to trees/forests (e.g., phenological shifts, growth response, etc.).

Response: Many thanks for the comment. We have revised this sentence

L.48: Haven't there been regional studies on this? The limited ones are at the national level?

Response: Thank you for this observation, which allows us to clarify a key point. The studies we cited by Liu et al. [7] and Fu et al. [9] are indeed regional-scale investigations, focusing on northern China and Northeast China, respectively. We used these examples to illustrate the existing body of regional research. The "limited ones at the national level" we referred to is the context for identifying the specific research gap our study aims to fill: a nationwide analysis of stand volume responses across elevation gradients, which remains unreported.

Response: Thank you for the comment. We have adopted the suggestion. The specific Quercus species of focus are now detailed in the Methods section, with a full list provided in the supplementary materials (Table S6).

H1: Stand volume of Quercus spp. varies significantly with elevation under climate warming. H2: Interactive effects of forest origin and age modify this elevational response.

Response: We thank the reviewer for this constructive feedback. We have adopted the suggestion and, considering our substantial revisions, further optimized the hypotheses as shown in the manuscript. We hope the updated versions now meet the reviewer's expectations.

MATERIALS AND METHODS

86: Using NCEP 2.5°, the data are too coarse for spatial analysis of NFI plots (~0.06 ha). The interpolation or downscaling method used by the authors should be explained to make the climate data relevant to the forest scale.

Response: We sincerely thank the reviewer for raising this important point. The issue noted is indeed valid—there is a scale mismatch between the resolution of the NCEP climate data and the forest plot data. For this reason, our analysis did not perform direct correlation analyses between these datasets, but instead focused on revealing broad-scale, directional trends. We fully agree that substantial potential remains for exploring deeper connections among the indicators. We look forward to future studies where plot- or region-specific climate data can be obtained to further validate and substantiate the conclusions drawn in this research.

101: The NFI only covers 2004–2018, but the results are linked to climate trends from 1948–2021 (L. 91) without analytical explanation.

Response: We fully agree with the reviewer regarding this limitation. As addressed in our previous response, we did not establish correlation tests in our analysis. Furthermore, we have added dedicated descriptions in the Discussion section regarding the limitations related to data scale and temporal span, in order to clarify the macro-scale exploratory positioning and core rationale of this study.

124: The classification of provinces based on “core zone elevation” is not representative of the actual plot data. The authors should consistently use the actual elevation of NFI plots (mean elevation per plot) rather than administrative categories for ecological validity.

Response: Many thanks for the comment. We thank the reviewer for this thoughtful comment. We have clarified in the Methods section that our analysis is based on officially compiled provincial-level forestry statistics (as stated on lines 105-111). Since the original NFI plot data had already been aggregated by authorities into mean stand volume values at the provincial level, our study unit is defined accordingly. This administrative-level approach aligns with China's official NFI reporting system and is appropriate for the national-scale analysis conducted here. A full description of the data integration process, which resulted in 745 cases across provinces, age classes, and stand origins, has been added (lines 126-130).

RESULTS

155: Where does this 0.01℃ value come from?

Response : We apologize for this error in the figure. The value of 0.01°C was indeed incorrect and resulted from a miscalculation in the code, as the reviewer rightly pointed out. We have now corrected it to 0.13°C—the actual warming rate per decade calculated from the temperature data. This correction has been made in the revised manuscript. Thank you again for your careful review.

172: Although it states “...no significant differences (p>0.05) across the three survey periods.” However, it then states an increase from 51.5 to 65.5 m³/ha (a 27% increase). If it is not significant, the variance is likely large, so it is best to present the SD, F, df, and p-value.

Response: We apologize for this misunderstanding. The multi-factor ANOVA method we originally employed had certain limitations, particularly when handling datasets containing zero values. To address this, we have now introduced both Linear Mixed Effects models and Hurdle Models for comprehensive analysis. The results indeed differ considerably, and we have updated the manuscript accordingly. The full statistical parameters of the multi-factor ANOVA are provided in the supplementary materials, while those for the Linear Mixed Effects and Hurdle Models are fully reported in the main text.

197: The statement "initial increase-decrease-rise" is ambiguous without significant figures or p-values between age classes. It is best to present ANOVAs per elevation and age with F valuesand post hoc grouping.

Response: We thank the reviewer for this comment. The phrase "initial increase-decrease-rise" was intended as a visual description of the interaction effect. The detailed statistical results for this interaction are provided in Table S5. Furthermore, in our revisions we have applied Linear Mixed Effects models to validate the significance of this interaction. The results consistently confirm the interaction effect, aligning with the findings from the multi-factor ANOVA.

DISCUSSION

220: This statement should be supported with data on the inter-seasonal trend difference from Fig. 2 in the results section to provide a solid discussion.

Response: We thank the reviewer for this suggestion. As now detailed in the Discussion section (lines XX–YY), we have explicitly linked our interpretation to the specific inter-seasonal trend differences shown in Fig. 2 and quantified in Table S2. This allows us to more substantively argue that the pronounced temperature variability in spring, autumn, and winter—compared to summer—may have distinct ecological implications for Quercus species, particularly through impacts on key phenological events.

239–241: If the results are consistent with the "Long study in Hunan," it is best to provide a correlation or R² value between elevation and volume to support this claim.

Response: Many thanks for this suggestion. In response, we have revised the relevant section in the manuscript. As our analysis has been updated with more robust statistical methods (Linear Mixed-Effects and Hurdle Models), the statistically significant effect of elevation on stand volume can no longer be definitively supported. Therefore, we have modified the core conclusion throughout the text and no longer assert a direct alignment with the findings of the "Long study in Hunan" in this specific context. The discussion has been reframed to more accurately reflect the nuanced relationships identified by the new models.

254–258: Please add citations.

Response: Many thanks for this comment. In response to the valid point regarding the need for citations to support the statement, we have reconsidered the passage. Upon reflection, we agree that the description lacked objective references and carried a subjective tone. Therefore, we have chosen to remove this sentence entirely from the manuscript.

324: The first hypothesis was not proven (p>0.05), but it is still stated that "essential patterns were captured."

Response: Many thanks for this comment. We apologize for the misunderstanding caused by the original phrasing. The sentence in question has been removed, and the conclusions in this section have been fully reorganized and rewritten based on the results of our updated statistical analysis.

Abstract and conclusion should be adjusted according to the results of the improvements. The conclusion will be more concise and focused on answering the research objectives.

Response: We thank the reviewer for this constructive feedback. We have revised both the Abstract and the Conclusion sections according to the improved results and analyses presented in the manuscript. The Conclusion has been refined to be more concise and is now sharply focused on directly addressing the research objectives.

Reviewer 3 Report

Comments and Suggestions for Authors

Reviewer Comments

The study analyzes Quercus stand volume using three National Forest Inventory cycles (2004–2018) alongside long-term temperature data, and reports that national stand volume does not show a significant temporal trend, natural stands exceed plantations, and interactions among elevation, stand origin, and age are important. The topic is relevant and the national scope is valuable; however, several methodological choices and gaps in reporting currently limit the strength and generalizability of the conclusions.

The most consequential issue is how elevation is handled. Elevation classes are assigned at the province level rather than using plot-level altitudes, which risks aggregation bias and modifiable areal unit problems. Many plots within a single province can occupy very different elevations; treating them as one class conflates true topographic effects with administrative boundaries. Closely related is the dependence and imbalance in the sample: some strata, especially high-elevation cells, appear to have very small sample sizes while others are large, and repeated NFI measurements likely introduce temporal and spatial dependence that standard one-way and multi-factor ANOVA do not address. A hierarchical/mixed-effects framework with random effects for province and plot (and, ideally, spatial correlation structures) would allow uneven sample sizes, repeated measures, and cross-level interactions to be modeled appropriately, and would yield more credible effect sizes with uncertainty.

Climate covariates and scaling also need attention. The analysis relies on coarse-resolution reanalysis temperature and omits precipitation and drought metrics, even though water balance strongly conditions oak growth. At minimum, precipitation, an index such as SPEI/PDSI, or growing-season water balance should be added, and sensitivity to alternative climate datasets or lapse-rate adjustments should be tested to reduce scale mismatch. The treatment of zero volumes is inconsistent—excluded in one-way ANOVA but included in multi-factor analyses—which can bias inference. Please adopt a single, principled approach: if zeros are structural, consider hurdle or zero-inflated models; if they reflect measurement noise, keep them consistently and report residual diagnostics. The temporal analysis also deserves a stronger design. With only three inventory epochs, ANOVA on period factors is underpowered and ignores within-epoch variability; modeling time as continuous (inventory mid-year) in a mixed framework, with possible plot-level random slopes and interactions with elevation and origin, would provide more informative trend estimates and confidence intervals.

Causal language should be tempered throughout. The manuscript attributes patterns to warming and logging bans, but the models do not include policy variables, management intensity, site index, soils, or competition. Unless those covariates are introduced, conclusions should be framed as associations and, if policy effects are central, a difference-in-differences or interrupted time-series design should be considered. Uncertainty reporting is also thin: beyond mean differences and multiple-comparison letters, readers need standardized effect sizes (e.g., partial η²), confidence intervals, variance partitioning (e.g., intraclass correlations), and prediction intervals where appropriate. Reproducibility would benefit from full model formulas, factor codings, and statistical settings, plus analysis code and an anonymized or aggregated dataset that allows others to replicate the main tables and figures.

Figures and maps should be strengthened with professional cartography and clearer spatial context. Please include geospatial visualizations of plot density (or confidentiality-respecting kernels), elevation distributions, and the spatial footprint of each stratum. A few editorial points will also help: verify the reported warming rate (0.01 °C per decade seems implausibly small—check units and slope specification); define terminology consistently (age class, origin, elevation class) and all acronyms at first use; provide normality and homoscedasticity checks for model residuals; assess multicollinearity among elevation, age, and origin (e.g., VIFs); ensure key policy claims cite authoritative sources; and standardize units (m³ ha⁻¹) and sample sizes in tables, avoiding mingling descriptive and inferential content without clear labeling. A light language edit would improve clarity and consistency.

Overall, the dataset and question are strong, but the current stratification, statistical framework, climate covariates, and uncertainty treatment limit the robustness of the conclusions. Re-analyzing with plot-level elevation, a mixed-effects design, richer climate metrics, clearer uncertainty reporting, and upgraded cartography would substantially improve rigor and impact.

Author Response

Response: Many thanks for the comment. The province-level elevation classification was necessitated by the nature of our dataset, which consists of officially aggregated provincial statistics as detailed in the Methods section regarding data sourcing and composition. While we recognize that this approach may not capture within-province topographic heterogeneity and could introduce the modifiable areal unit problem, it was the most appropriate method to match the administrative scale of our data and to address our research focus on national-scale patterns. In response to this comment, we have now explicitly discussed this methodological consideration and its potential implications as a limitation in the revised discussion section.

Closely related is the dependence and imbalance in the sample: some strata, especially high-elevation cells, appear to have very small sample sizes while others are large, and repeated NFI measurements likely introduce temporal and spatial dependence that standard one-way and multi-factor ANOVA do not address. A hierarchical/mixed-effects framework with random effects for province and plot (and, ideally, spatial correlation structures) would allow uneven sample sizes, repeated measures, and cross-level interactions to be modeled appropriately, and would yield more credible effect sizes with uncertainty.

Response: We sincerely thank the reviewer for this insightful suggestion regarding sample dependence and imbalance. In direct response to this comment, we have incorporated Linear Mixed-Effects Models (LMMs) into our revised analysis. This framework explicitly includes random effects for province and plot to account for the hierarchical data structure. It effectively handles the uneven sample sizes across strata (particularly the small sample sizes in high-elevation cells), incorporates the temporal and spatial dependence introduced by repeated NFI measurements, and appropriately models cross-level interactions. We agree that this approach provides more credible effect estimates with robust uncertainty quantification, and we have updated the results and discussion accordingly.

Response: We thank the reviewer for highlighting the need for additional climate covariates like precipitation and drought metrics. We agree that including water balance indicators (e.g., SPEI/PDSI) or growing-season water balance would enhance our analysis, particularly given their physiological relevance to oak growth. However, procuring, processing, and robustly integrating these additional datasets—coupled with performing essential sensitivity tests (e.g., evaluating alternative climate products or lapse-rate corrections to mitigate scale discrepancies)—would demand considerable additional time and research effort. A further fundamental constraint is that currently available climate products still face challenges in achieving precise spatial alignment with our plot-level observations. We look forward to incorporating more diverse and spatially refined climate variables in future studies as a meaningful direction for further exploration.

The treatment of zero volumes is inconsistent—excluded in one-way ANOVA but included in multi-factor analyses—which can bias inference. Please adopt a single, principled approach: if zeros are structural, consider hurdle or zero-inflated models; if they reflect measurement noise, keep them consistently and report residual diagnostics.

Response: We thank the reviewer for raising this critical point regarding inconsistent treatment of zero stand volume values. We fully agree that maintaining methodological consistency is essential for unbiased inference. In response, we have adopted a unified and principled approach throughout our revised analysis. Recognizing that zero values predominantly represent meaningful ecological states (e.g., failed plantation establishment or young trees below the measurement threshold) rather than measurement noise, we have retained all zero-volume plots across all analyses. To appropriately model this zero-inflated data structure, we have incorporated hurdle models, which simultaneously analyze the presence-absence probability of stands and the positive volume values. This approach has been consistently applied in conjunction with linear mixed-effects models, ensuring robust and ecologically interpretable results. We have updated the manuscript accordingly.

The temporal analysis also deserves a stronger design. With only three inventory epochs, ANOVA on period factors is underpowered and ignores within-epoch variability; modeling time as continuous (inventory mid-year) in a mixed framework, with possible plot-level random slopes and interactions with elevation and origin, would provide more informative trend estimates and confidence intervals.

Response: We thank the reviewer for this valuable suggestion to strengthen the temporal analysis design. We fully agree that treating time as a continuous variable provides more statistical power and ecological insight than analyzing three discrete inventory periods. In response, we have revised our analytical approach by incorporating time as a continuous variable (using inventory mid-year) within our linear mixed-effects modeling framework. This enhanced model includes plot-level random slopes for time and examines its interactions with elevation and stand origin. As the reviewer rightly anticipated, this approach now provides more informative estimates of temporal trends and more precise confidence intervals, while also accounting for within-epoch variability. We have updated all relevant results and discussion sections accordingly.

Response: We thank the reviewer for this important suggestion. We fully agree that causal language should be used with greater caution throughout the manuscript. In response, we have systematically revised the text to replace all causal assertions with correlational descriptions (e.g., changing "attributed to" to "associated with"). We acknowledge that our current analysis cannot establish strict causal relationships, as the models do not incorporate covariates such as policy variables, management intensity, site index, soils, or competition. If policy effects are to be rigorously evaluated in future studies, methods such as difference-in-differences or interrupted time-series designs would be more appropriate. We have explicitly addressed this limitation in the Discussion section and have thoroughly modified the relevant wording.

Uncertainty reporting is also thin: beyond mean differences and multiple-comparison letters, readers need standardized effect sizes (e.g., partial η²), confidence intervals, variance partitioning (e.g., intraclass correlations), and prediction intervals where appropriate. Reproducibility would benefit from full model formulas, factor codings, and statistical settings, plus analysis code and an anonymized or aggregated dataset that allows others to replicate the main tables and figures.

Response: Many thanks for these specific and valuable suggestions regarding uncertainty reporting and reproducibility. In response, we have now included five supplementary tables in the appendix (Table S7-S11) that provide complete model formulas, statistical parameters (including standardized effect sizes, confidence intervals, and variance partitioning metrics), and detailed uncertainty measures. Regarding data accessibility, due to the specific nature of the official national forest inventory data used in this study, full public dissemination is not permitted by the data policy. However, the anonymized aggregated dataset and analysis code are available from the corresponding author upon reasonable request to ensure verifiability and reproducibility.

Response: We thank the reviewer for these constructive suggestions for improving our figures and maps. In response, we have enhanced the majority of the figures using professional cartographic standards to improve visual clarity and spatial context. However, due to the province-aggregated nature of our dataset, we were unable to incorporate plot-level visualizations such as plot density kernels or detailed elevation gradients, as these would require access to the raw, non-aggregated plot coordinates and measurements which are not available in our current data structure. We have instead focused on maximizing the cartographic quality and informational value of the available data in the revised figures.

A few editorial points will also help: verify the reported warming rate (0.01 °C per decade seems implausibly small—check units and slope specification); define terminology consistently (age class, origin, elevation class) and all acronyms at first use; provide normality and homoscedasticity checks for model residuals; assess multicollinearity among elevation, age, and origin (e.g., VIFs); ensure key policy claims cite authoritative sources; and standardize units (m³ ha⁻¹) and sample sizes in tables, avoiding mingling descriptive and inferential content without clear labeling. A light language edit would improve clarity and consistency.

Response: Many thanks for these thoughtful editorial suggestions. We have carefully addressed each point in the revised manuscript, which included verifying and correcting the warming rate, ensuring consistent terminology and acronym usage throughout, supplementing relevant model parameters, standardizing citations for policy statements, unifying measurement units and table formatting, and performing comprehensive language polishing to enhance clarity and consistency.

Response: We sincerely thank the reviewer for this professional assessment. While our dataset has inherent limitations in plot-level elevation and climate metrics, we have made every effort to enhance the analytical robustness by: (1) enriching our modeling framework to address data constraints, (2) explicitly quantifying and reporting uncertainties, (3) systematically handling zero-inflated distributions, and (4) completely restructuring the manuscript narrative. All figures and tables have been professionally optimized for clarity and visual impact. We hope these comprehensive revisions meet the reviewer's expectations.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript has been revised in accordance with the suggestions, and its acceptance for publication is recommended.