Spatiotemporal Dynamics and Trade-Off Analysis of Ecosystem Services in the Caijiachuan Watershed of the Loess Plateau

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

This manuscript investigates ecosystem service dynamics in a representative watershed of China's Loess Plateau using the InVEST model framework, which addresses research topics that align well with Agronomy's scope covering agroecology, farming systems, soil health and plant nutrition, agricultural climatology, and ecosystem services of cropping systems. The study examines ecological restoration in an agricultural landscape context, addressing water management, soil conservation, and carbon sequestration—all core themes within the journal's agricultural and agroecological focus. However, while the research addresses relevant questions for agricultural sustainability and agroecosystem management, it exhibits several critical methodological limitations that require substantial revision before publication consideration.

The most significant concern relates to the fundamentally inadequate sampling design employed throughout the study. The authors utilized only 36 vegetation plots across 10 land use types in a 39 km² watershed, resulting in highly uneven representation with as few as 3 plots for some forest types and up to 14 for others. This sampling intensity is demonstrably insufficient for drawing robust conclusions about ecosystem service provision across such a heterogeneous landscape, particularly given the authors' claim of analyzing "over 3000 patches." The stratified layout based on area and spatial distribution lacks statistical rigor, with no randomization protocol described, no power analysis conducted to determine adequate sample sizes, and no consideration of spatial autocorrelation effects that are critical in agricultural landscape studies. Most problematically, field data collection occurred exclusively in 2024, yet the study attempts to model ecosystem services for 2002, 2012, and 2024, creating a fundamental temporal mismatch that severely undermines the validity of historical assessments in this supposed longitudinal study.

The technical framework presents additional serious limitations that compromise the study's scientific soundness. The carbon storage module claims to assess four carbon pools but provides insufficient detail on belowground biomass and dead organic matter quantification methods, which are crucial for agricultural soils where carbon dynamics differ significantly from natural systems. The soil retention calculations using the Universal Soil Loss Equation and EPIC model applications lack proper validation against local agricultural conditions, despite the Loess Plateau's unique soil characteristics being well-documented in agricultural literature. The water yield assessment through the InVEST module requires careful parameterization of plant available water content and root restricting layer depth, parameters that are particularly critical in agricultural watersheds but are inadequately addressed in the current methodology.

Statistical analysis represents another major weakness, with Pearson correlation coefficients employed without testing assumptions of normality and linearity—basic requirements for valid correlation analysis. The authors ignore potential spatial dependencies in ecosystem service values that could severely bias correlation estimates, particularly problematic in agricultural landscapes where management practices create spatial patterns. No correction for multiple testing is applied when examining pairwise service relationships, inflating the risk of Type I errors. The correlation coefficients are presented without significance testing or confidence intervals, with the practical significance of observed trade-offs lacking adequate discussion relevant to agricultural management decisions.

Data quality issues further compound these methodological concerns. The manual delineation of "over 3000 small forest patches" lacks any accuracy assessment or validation protocols, which is particularly concerning given that accurate land use classification is fundamental to ecosystem service modeling in agricultural landscapes. Reliance on data from a single meteorological station for a 39 km² watershed ignores spatial climate variability that is well-known to affect agricultural productivity and ecosystem services. The complete absence of model validation through cross-validation against observed data or sensitivity analysis to assess parameter robustness represents a fundamental gap that would not meet standards for agricultural systems modeling.

The results presentation, while demonstrating clear temporal trends in ecosystem service provision, suffers from critical interpretation flaws that undermine confidence in the findings. The discussion attributes trade-offs to "combined effects of natural geography and ecological processes" without sufficient mechanistic detail relevant to agricultural management, exhibits limited engagement with recent agroecological literature on ecosystem service trade-offs, and provides insufficient translation of findings into actionable recommendations for agricultural land management in the Loess Plateau region. The authors fail to adequately discuss how their findings relate to the Grain-for-Green Program's impacts on agricultural productivity and farmer livelihoods, which should be central to this analysis.

The manuscript's alignment with Agronomy's focus requires strengthening through enhanced discussion of implications for agricultural systems, soil health, and sustainable farming practices in the Loess Plateau. The study would benefit from explicit consideration of how ecosystem service trade-offs affect agricultural decision-making, crop production systems, and soil conservation practices that are central to the journal's mission. The current treatment inadequately addresses the agricultural context despite studying a watershed affected by major agricultural land use changes.

Comments on the Quality of English Language

While the English language quality does not prevent understanding of the scientific content, it falls short of the polished presentation expected for an international sustainability journal. The issues are sufficiently numerous and consistent to warrant moderate revision, focusing specifically on language clarity and grammatical accuracy before resubmission.

Author Response

We sincerely thank the editor and reviewers for your valuable time and effort in evaluating our manuscript. We are deeply grateful for your thoughtful, constructive, and detailed comments, which have greatly helped us identify and address key issues in our research—particularly in areas such as sampling design, model assumptions, and the implications for agricultural management. Your suggestions have been instrumental in improving the scientific rigor, clarity, and practical significance of our study.

We have carefully revised the manuscript in response to each of your comments. All changes made are clearly marked in yellow highlight in the revised manuscript to facilitate your review. We truly appreciate your guidance, and we hope that the revised version meets your expectations. Below, we provide a point-by-point response to all the comments raised.

Question：The most significant concern relates to the fundamentally inadequate sampling design employed throughout the study. The authors utilized only 36 vegetation plots across 10 land use types in a 39 km² watershed, resulting in highly uneven representation with as few as 3 plots for some forest types and up to 14 for others. This sampling intensity is demonstrably insufficient for drawing robust conclusions about ecosystem service provision across such a heterogeneous landscape, particularly given the authors' claim of analyzing "over 3000 patches." The stratified layout based on area and spatial distribution lacks statistical rigor, with no randomization protocol described, no power analysis conducted to determine adequate sample sizes, and no consideration of spatial autocorrelation effects that are critical in agricultural landscape studies. Most problematically, field data collection occurred exclusively in 2024, yet the study attempts to model ecosystem services for 2002, 2012, and 2024, creating a fundamental temporal mismatch that severely undermines the validity of historical assessments in this supposed longitudinal study.

Response: We sincerely thank the reviewer for this important comment. To ensure spatial representativeness along the main gradients of topography and vegetation, we employed a stratified sampling strategy, where sample plot allocation was based on the area proportion and spatial distribution of different land use types. Given the complex terrain, highly fragmented land cover, and field accessibility constraints in the Caijiachuan watershed, a completely randomized sampling design was not feasible. Instead, our sampling intensity followed the standards adopted in similar-scale studies conducted in the Loess Plateau region, ensuring comparability and methodological consistency.

We acknowledge that mixed forests and orchards were represented by only three plots each due to their limited spatial extent, whereas dominant monoculture forests were sampled proportionally with a higher number of plots. Regarding the “over 3000 patches” mentioned in the manuscript, we clarify that these refer not to sample plots, but to manually delineated micro-patches derived from long-term field surveys by our research team. These patches include detailed vegetation structure and composition information (e.g., shrub and herbaceous layers), and were later used as the basis for land use classification in our modeling process.

As for the temporal mismatch raised by the reviewer, we deeply regret the confusion caused by our earlier description. While field data collection for this study occurred in 2024, we also incorporated historical records from the Caijiachuan National Field Observation Station, a long-established national-level ecological monitoring site. Since its establishment, the station has systematically recorded vegetation, herbaceous species composition, and daily rainfall data, using consistent protocols across time. Thus, the 2002 and 2012 datasets used in this study were not simulated retroactively using 2024 data, but obtained directly from this authoritative database, aligned with the same data collection standards as our 2024 survey.

Regarding the concerns on statistical rigor—particularly the lack of a formal power analysis, randomization, and treatment of spatial autocorrelation—we appreciate this critical observation. We acknowledge that we did not explicitly report a power analysis, and that spatial autocorrelation effects were not addressed in the initial version. In the revised manuscript, we have now included a spatial autocorrelation test (Moran's I) to assess potential biases in sampling distribution and variable clustering. We have also provided a justification for the adopted sample sizes based on coverage of dominant land use types and precedents from comparable studies, and clarified the rationale for our stratified—but not randomized—sampling layout.

Question：The technical framework presents additional serious limitations that compromise the study's scientific soundness. The carbon storage module claims to assess four carbon pools but provides insufficient detail on belowground biomass and dead organic matter quantification methods, which are crucial for agricultural soils where carbon dynamics differ significantly from natural systems. The soil retention calculations using the Universal Soil Loss Equation and EPIC model applications lack proper validation against local agricultural conditions, despite the Loess Plateau's unique soil characteristics being well-documented in agricultural literature. The water yield assessment through the InVEST module requires careful parameterization of plant available water content and root restricting layer depth, parameters that are particularly critical in agricultural watersheds but are inadequately addressed in the current methodology.

Response: We sincerely thank the reviewer for this constructive and insightful comment. In response, we have made substantial revisions to clarify and enhance the technical details of our modeling framework:

Carbon Storage Module: We have now provided explicit descriptions of the data sources and calculation methods for belowground biomass and dead organic matter in the revised Section 2.2 (Vegetation Survey and Sample Collection). These values were derived using standard root-to-shoot ratios based on vegetation type, supported by field measurements and published literature specific to the Loess Plateau region.

Soil Retention Module: We have supplemented the description of the Universal Soil Loss Equation (USLE) parameters and local calibration strategy in Section 2.4.3. This includes clarification of how rainfall erosivity (R), soil erodibility (K), and cover-management (C) factors were parameterized using field survey data and regional empirical studies. Moreover, the LS factor was calculated using a 1 m resolution DEM, and P factors were determined through both literature review and local ground truthing.

Water Yield Module: We have now detailed the parameterization process of plant available water content (PAWC) and root restricting layer depth in Section 2.4.2. These parameters were derived from site-specific soil profiles, cross-referenced with the national soil database and validated through in-situ soil sampling.

Regarding the reviewer’s concern about the lack of validation under local agricultural conditions, we fully acknowledge that this is a critical limitation. Although the Caijiachuan watershed is not dominated by intensive agricultural practices, some orchard and dryland farming areas do exist. In this study, due to limited historical records of plot-scale soil erosion or runoff measurements in these agricultural areas, we were not able to conduct direct validation at the field level. However, to partially address this issue, we referenced regional calibration values from peer-reviewed studies conducted under similar loess soil and slope conditions. These studies served as indirect validation benchmarks for our parameter selection. We have now made this limitation and our mitigation strategy explicit in Section 4 (Discussion), and added suggestions for future field-based validation work.

Question：Statistical analysis represents another major weakness, with Pearson correlation coefficients employed without testing assumptions of normality and linearity—basic requirements for valid correlation analysis. The authors ignore potential spatial dependencies in ecosystem service values that could severely bias correlation estimates, particularly problematic in agricultural landscapes where management practices create spatial patterns. No correction for multiple testing is applied when examining pairwise service relationships, inflating the risk of Type I errors. The correlation coefficients are presented without significance testing or confidence intervals, with the practical significance of observed trade-offs lacking adequate discussion relevant to agricultural management decisions.

Response: We sincerely thank the reviewer for this critical comment. We acknowledge that the original version of the manuscript lacked a sufficiently detailed description of the statistical procedures, which may have led to misunderstandings regarding the analytical rigor.

In response to the reviewer’s concerns, we have substantially revised our statistical methodology. Specifically, we have replaced Pearson correlation with Spearman’s rank correlation, as the distributions of our ecosystem service indicators are non-normal and potentially non-linear. This adjustment ensures the robustness of correlation estimates.

To address the issue of multiple testing, we applied the False Discovery Rate (FDR) correction to all p-values. In addition, we computed 95% confidence intervals using bootstrapping (n = 1000), to enhance the reliability of our inferences.

We have also conducted Moran’s I spatial autocorrelation tests to assess spatial dependency in the data. The implications of spatial clustering were further considered in the revised Discussion section, especially in relation to land-use patterns in agricultural landscapes.

Finally, we expanded our interpretation of correlation results in the Discussion to emphasize the practical significance of observed trade-offs and synergies for agricultural management and ecological planning.

All these revisions are now explicitly described in Section 2.5 (Statistical Analysis), with updated results presented in Figures 8–10. We believe that these improvements significantly enhance the methodological transparency and scientific robustness of the study.

Question：Data quality issues further compound these methodological concerns. The manual delineation of "over 3000 small forest patches" lacks any accuracy assessment or validation protocols, which is particularly concerning given that accurate land use classification is fundamental to ecosystem service modeling in agricultural landscapes. Reliance on data from a single meteorological station for a 39 km² watershed ignores spatial climate variability that is well-known to affect agricultural productivity and ecosystem services. The complete absence of model validation through cross-validation against observed data or sensitivity analysis to assess parameter robustness represents a fundamental gap that would not meet standards for agricultural systems modeling.

Response: We sincerely thank the reviewer for this insightful comment. We acknowledge that our original description may have led to confusion regarding data collection and validation procedures.

First, we would like to clarify that the delineation of over 3000 small forest patches was not based on remote interpretation alone, but rather conducted through intensive field surveys, involving systematic ground-truthing and manual sketching during on-foot investigations. We have clarified this process and corrected our wording in Section 2.3 (Data Sources) to reflect the actual protocol.

Second, regarding the concern about meteorological data, we apologize for the oversimplified description in the previous version. In fact, our watershed is part of the Caijiachuan National Ecosystem Observation and Research Station, which maintains 18 long-term, networked monitoring sites, including climate, hydrology, and vegetation observation points. These data were integrated into our modeling workflow, and we have updated Section 2.3 to clearly describe this.

Finally, while we did not perform a full sensitivity analysis, we conducted parameter consistency checks across the time series to ensure the robustness of model inputs and outputs. We believe these clarifications and methodological enhancements better support the validity of our ecosystem service assessments.

Question：The results presentation, while demonstrating clear temporal trends in ecosystem service provision, suffers from critical interpretation flaws that undermine confidence in the findings. The discussion attributes trade-offs to "combined effects of natural geography and ecological processes" without sufficient mechanistic detail relevant to agricultural management, exhibits limited engagement with recent agroecological literature on ecosystem service trade-offs, and provides insufficient translation of findings into actionable recommendations for agricultural land management in the Loess Plateau region. The authors fail to adequately discuss how their findings relate to the Grain-for-Green Program's impacts on agricultural productivity and farmer livelihoods, which should be central to this analysis.

Response: Thank you very much for this valuable and constructive comment. We fully agree that the implications of ecosystem service (ES) trade-offs for agricultural systems and rural livelihoods—particularly in the context of the Grain-for-Green Program (GFGP)—are essential to this study.

To address this, we have added a dedicated subsection entitled “4.5 Ecosystem Service Trade-offs, Agricultural Adaptation, and Rural Livelihoods” to our revised manuscript. In this section, we explore how observed trade-offs between carbon storage and water yield influence agricultural decision-making, soil management, and crop production. We discuss how reduced water yield in afforested areas may pose challenges for irrigation-dependent farming, and we examine how increased carbon sequestration may affect long-term soil fertility and productivity.

Furthermore, we have engaged more deeply with recent agroecological and land system science literature, which provide insights into the socio-ecological outcomes of the GFGP and its implications for smallholder farmers' adaptive strategies.

This revision strengthens the policy relevance of our work and more explicitly connects our findings with agricultural sustainability and rural development goals in the Loess Plateau region.

Question：The manuscript's alignment with Agronomy's focus requires strengthening through enhanced discussion of implications for agricultural systems, soil health, and sustainable farming practices in the Loess Plateau. The study would benefit from explicit consideration of how ecosystem service trade-offs affect agricultural decision-making, crop production systems, and soil conservation practices that are central to the journal's mission. The current treatment inadequately addresses the agricultural context despite studying a watershed affected by major agricultural land use changes.

Response:We sincerely thank the reviewer for this important suggestion. We recognize the need to strengthen the agricultural context and policy relevance of our study to better align with Agronomy’s scope.

In the revised manuscript, we have significantly expanded our discussion of the implications of ecosystem service (ES) dynamics for agricultural systems, soil health, and sustainable farming practices on the Loess Plateau. Specifically, in the newly added Section 4.5, we explicitly address how trade-offs—particularly between carbon storage and water yield—can influence irrigation needs, land use decisions, and agroecological resilience in hilly agricultural regions.

We also elaborate on the soil conservation benefits and potential soil moisture limitations associated with afforestation under the Grain-for-Green Program, highlighting implications for long-term agricultural productivity. These revisions help connect our biophysical findings to farm-level management decisions and landscape-scale planning, which we believe directly supports the journal’s mission and readership.

Furthermore, we have incorporated recent literature on agroecosystem trade-offs, adaptive agricultural practices, and the impacts of ecological restoration programs on rural livelihoods to better situate our findings within the broader context of agricultural sustainability.

 

We sincerely hope that the revisions we have made meet the expectations of the reviewers and the editorial team, and that the improved manuscript now aligns with the standards of agronomy. We truly appreciate the opportunity to revise our work and are grateful for your insightful feedback, which has significantly enhanced the quality of this study.

We respectfully look forward to your positive response and would be honored to see our manuscript considered for publication.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper evaluates the spatiotemporal dynamics of various ecosystem services in the Caijiachuan watershed, China, and is considered a useful resource for ecosystem restoration research. However, the following revisions and improvements are necessary:

Lines 100–103: Please clearly state the source of the meteorological data provided. For example, specify the location of the meteorological station and the observation period.

Line 118 and Lines 164–167: The vegetation survey was conducted once in 2024, and the vegetation and soil data were used as input for the carbon storage and soil retention modules of the InVEST model. However, in the results section, carbon storage and soil retention outcomes are presented not only for 2024 but also for 2002 and 2012. Therefore, please clarify whether the 2024 vegetation survey data were also used to estimate the results for 2002 and 2012.

Section 2.4 InVEST Model Assessment: Among the four ecosystem services analyzed in this study—carbon storage, water yield, soil retention, and habitat quality—only the first three are described. Please include a description of the habitat quality as well.

The InVEST model plays a central role in this paper. Therefore, a more detailed explanation of the model should be added in the Materials and Methods section.

Line 255: Revise the sentence "After extracting the four ecosystem services at the pixel level" to:
"After extracting the four ecosystem services—soil retention, water yield, carbon storage, and habitat quality—at the pixel level".

The titles of the figures in the results section should be written more clearly and in accordance with academic formatting standards. Additionally, the resolution of all figures needs to be improved, especially Figure 8, which is too low in quality to be legible.

Author Response

We sincerely thank the editor and reviewers for your valuable time and effort in evaluating our manuscript. We are deeply grateful for your thoughtful, constructive, and detailed comments, which have greatly helped us identify and address key issues in our research—particularly in areas such as sampling design, model assumptions, and the implications for agricultural management. Your suggestions have been instrumental in improving the scientific rigor, clarity, and practical significance of our study.

We have carefully revised the manuscript in response to each of your comments. All changes made are clearly marked in yellow highlight in the revised manuscript to facilitate your review. We truly appreciate your guidance, and we hope that the revised version meets your expectations. Below, we provide a point-by-point response to all the comments raised.

Comment 1 (Lines 100–103): Please clearly state the source of the meteorological data provided. For example, specify the location of the meteorological station and the observation period.

Response: Thank you for pointing this out. We have revised Lines 100–103 to clarify that the meteorological data used in this study were obtained from the Caijiachuan National Ecosystem Observation and Research Station, which has provided continuous and standardized meteorological observations since its establishment. Data for the years 2002, 2012, and 2024 were extracted to match the ecosystem service assessment periods. This clarification has been added to Section 2.3 (Data Sources).

Comment 2 (Line 118 and Lines 164–167): The vegetation survey was conducted once in 2024, yet results are reported for 2002 and 2012. Please clarify whether 2024 data were also used for historical estimates.

Response: We appreciate the reviewer’s attention to this temporal mismatch. We have clarified in the revised manuscript (Section 2.2 and 2.3) that although the vegetation survey for this specific study was conducted in 2024, we also used archived vegetation and soil data for 2002 and 2012. These data were obtained from long-term monitoring records at the Caijiachuan National Observation Station and include historical records of vegetation structure, species composition, and soil properties collected using consistent protocols. Therefore, model inputs for all three years (2002, 2012, and 2024) were year-specific and not retroactively inferred from the 2024 dataset.

Comment 3 (Section 2.4 - InVEST Model Assessment): Among the four ecosystem services analyzed, only three are described. Please include a description of the habitat quality assessment.

Response: Thank you for highlighting this omission. We have now added a subsection in Section 2.4 to describe the Habitat Quality Module of the InVEST model. This includes the selection of land use/land cover maps, threat factors (e.g., roads and agricultural expansion), sensitivity scores, and habitat suitability values used in the assessment. The description also clarifies how the model integrates anthropogenic disturbance and land use dynamics to evaluate spatial habitat quality.

Comment 4: The InVEST model plays a central role in this paper. Therefore, a more detailed explanation of the model should be added in the Materials and Methods section.

Response: We fully agree. The revised manuscript now includes a more comprehensive overview of the InVEST model in Section 2.4. This includes a general introduction to the model’s modular structure, assumptions, and limitations, as well as an explanation of how input parameters were selected and validated in the context of the Caijiachuan watershed. Additional citations to key InVEST documentation and recent applications in similar agroecological contexts have also been added.

Comment 5 (Line 255): Revise the sentence "After extracting the four ecosystem services at the pixel level" to: "After extracting the four ecosystem services—soil retention, water yield, carbon storage, and habitat quality—at the pixel level".

Response: Thank you for the suggestion. We have revised the sentence as requested for clarity and completeness.

Comment 6: The titles of the figures in the results section should be written more clearly and in accordance with academic formatting standards. Additionally, the resolution of all figures needs to be improved, especially Figure 8, which is too low in quality to be legible.

Response: Thank you for the suggestion. We have revised all figure titles for clarity and academic consistency, and updated all figures—especially Figure 8—to high-resolution versions to ensure readability.

 

We sincerely hope that the revisions we have made meet the expectations of the reviewers and the editorial team, and that the improved manuscript now aligns with the standards of agronomy. We truly appreciate the opportunity to revise our work and are grateful for your insightful feedback, which has significantly enhanced the quality of this study.

We respectfully look forward to your positive response and would be honored to see our manuscript considered for publication.

 

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The revised manuscript exhibits significant enhancements over the previous version and addresses numerous fundamental methodological concerns that were raised in the initial review, thereby bringing the work much closer to the publication standards of Agronomy. The authors have made big improvements in many areas that make their research much more scientifically sound and useful in real life. However, there are a few small things that need to be fixed before the manuscript is fully ready for publication. In the attached file, I've written down these small details.

Comments for author File: Comments.pdf

Author Response

Dear Editor and Reviewers,

We sincerely thank you for your thoughtful and constructive comments on our manuscript. We greatly appreciate the time and effort you have devoted to reviewing our work. Your insights have significantly improved the clarity, depth, and scientific rigor of the study.

Below, we provide a detailed point-by-point response to each of the reviewers’ comments. All revisions have been incorporated into the revised manuscript and are highlighted in red for clarity. Reviewer comments are shown in bold, and our responses follow directly below each point.

Question1

Correlations of this magnitude (R² < 0.01) indicate virtually no meaningful relationship in practical terms. The authors should acknowledge that these correlations, while potentially statistically significant due to large sample sizes (n > 5000 pixels), represent negligible practical associations that would not inform ecosystem management strategies.

Answer1

We thank the reviewer for this important observation. As suggested, we have explicitly acknowledged in Section 3.5 that correlations with extremely low effect sizes (e.g., R² < 0.01) are statistically significant primarily due to the large sample size (n > 5000), but have limited practical relevance for informing ecosystem management strategies. We now emphasize that these weak associations—such as between carbon storage and soil retention—should be interpreted with caution and are unlikely to reflect ecologically meaningful interactions.

We appreciate this helpful suggestion and will incorporate such critical distinctions more systematically in future work.

Question2

The authors must discuss what this change means for ecosystem functioning. Does this shift from weak negative to near-zero correlation indicate: Successful ecological restoration reducing conflicts between services? Maturation of forest stands allowing habitat and carbon benefits to coexist?Methodological artifacts in habitat quality assessment over time?

Answer2

We appreciate the reviewer’s insightful question. In the revised Results section, we now interpret the weakening correlation between carbon storage and habitat quality (from r = –0.25 in 2002 to r = –0.03 in 2024) as an ecological signal of reduced trade-offs. Specifically, we suggest that this shift likely reflects the positive effects of long-term vegetation restoration and forest maturation, which promoted greater structural complexity and functional complementarity between carbon accumulation and habitat provisioning.

To address the possibility of methodological artifacts, The model parameters were kept consistent within each individual year’s analysis (2002, 2012, and 2024, respectively), but differ somewhat across years to reflect temporal changes in ecological conditions. This approach balances the need for internal consistency within each time slice and ecological realism across time, although it may introduce some variability in cross-year comparisons. Nevertheless, the core modeling framework remained unchanged, supporting the reliability of observed temporal trends

Question3

 The authors must present or interpret these intervals.For management applications, knowing whether a correlation of -0.43 has confidence bounds of [-0.39, -0.47] versus [-0.30, -0.56] substantially affects the certainty of trade-off predictions.

Answer3

We appreciate the reviewer’s insightful comment. In response, we have clearly presented 95% bootstrap confidence intervals (CIs) for all correlation coefficients and incorporated their interpretation in the revised manuscript. As requested, we now emphasize that the width of the CIs reflects the uncertainty of the estimates—narrow intervals indicate high precision and reliability, while wider intervals suggest lower certainty and require more cautious interpretation.

For example, in 2002, the negative correlation between carbon storage and water yield (ρ = –0.43, 95% CI: –0.46 to –0.42) is associated with a narrow interval, indicating a stable and robust trade-off, suitable for guiding restoration planning. In contrast, the correlation between carbon storage and habitat quality in 2024 (ρ = –0.03, 95% CI: –0.07 to 0.01) shows a wide interval overlapping zero, reflecting high uncertainty and limited management value.

All such results, including interval widths and implications, are now clearly reported and interpreted in Section 3.6 of the revised manuscript.

Question4

The manuscript lacks comparison with effect sizes from similar ecosystem service studies. Cohen's conventional benchmarks (small: 0.1, medium: 0.3, large: 0.5 for correlations) or ecological restoration literature standards would help readers evaluate whether observed relationships represent meaningful trade-offs worth addressing in restoration planning.

The authors must, explicitly, acknowledge when large sample sizes may produce statistically significant but practically meaningless correlations, particularly for the weak positive relationships.

Answer4

We thank the reviewer for highlighting the importance of interpreting effect sizes beyond mere statistical significance. We have now added a dedicated subsection (4.6) discussing this issue. In particular, we acknowledge that while our study applied rigorous statistical controls—including FDR correction and bootstrap confidence intervals—the large sample size (n > 5000 pixels per year) means that very weak correlations (e.g., ρ = 0.04) can be statistically significant without ecological relevance.

We explicitly reference Cohen’s conventional benchmarks (small: ρ = 0.10, medium: ρ = 0.30, large: ρ = 0.50) and note that most weak positive correlations in our results fall below the threshold of a small effect size, indicating limited practical importance. By contrast, stronger and consistent negative correlations (such as between carbon storage and water yield, ρ ≈ –0.42) approach a medium effect and are likely to represent meaningful trade-offs that warrant attention in restoration planning.

We also emphasize that future work should integrate effect size interpretation and cross-study comparisons to better discern ecologically meaningful patterns from those arising due to large sample sizes and high-resolution spatial data.窗体底端

Question5

Additional information about the effect size in ecological studies would be helpful in interpreting R2 values of about 0.28 for the carbon-water trade-off.

In ecological restoration contexts, an R² of 0.28 indicates that carbon storage explains only 28% of the variance in water yield, leaving 72% unexplained. The authors should discuss whether this effect size is: Sufficient to guide management decisions about forest plantation density Comparable to other ecosystem service trade-offs reported in restoration literature

Strong enough to warrant policy interventions balancing carbon sequestration with water provisioning

Answer5

We thank the reviewer for this important comment. In response, we have revised Section 5.2 to further discuss the practical implications of the observed effect size (R² ≈ 0.28) for the carbon-water trade-off. We clarify that while this represents a moderate and consistent correlation, it also indicates that 72% of the variation in water yield remains unexplained, likely due to factors such as soil properties, slope, and climate variability.

Accordingly, we emphasize that relying solely on this trade-off is insufficient to directly guide management decisions such as plantation density. We have supplemented the discussion by recommending integration of field-based hydrological data and mechanistic modeling to develop more robust and site-specific management strategies.The revised text in Section 5.2 reflects these points in detail.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

I sincerely appreciate the authors’ thoughtful and comprehensive revisions. The revised manuscript demonstrates significant improvements in clarity, structure organization, and methodological transparency. Most of the comments from the first review have been well addressed, particularly regarding data source clarification, detailed InVEST module descriptions, and the explanation of how vegetation data from 2024 were used to support past-year estimates. The newly added subsection on habitat quality (Section 2.4.4) and enhancements in Section 3.6 have notably strengthened the manuscript.

However, before final acceptance, I recommend addressing the following minor formatting and editorial issues to meet the publication standards and improve readability.

1. Figure and Table Captions: All figure captions should follow proper academic format, starting with "Figure X." and written as complete sentences ending with a period. For example: "Figure 3. Changes in carbon storage from 2002 to 2024." Several captions (Figures 3–8) currently begin with lowercase letters and/or lack terminal punctuation. Please revise for consistency.

 

2. Spacing and Punctuation: Add a space between text and parentheses where needed. For example, "In 2002(Figure 3a)" should be "In 2002 (Figure 3a)".

3. Use consistent scientific notation for units. For instance: t·hm⁻² or Mg C ha⁻¹, with correct spacing. Avoid mixing styles such as t/hm² and t·hm⁻² within the same section.

4. Section numbers and headings should be properly spaced. For example: "2.4.4Habitat Quality Module" should be "2.4.4. Habitat Quality Module".

5. Reference Consistency:  Avoid duplicated references (e.g., Wischmeier & Smith, 1978 appears more than once). Ensure consistent formatting of author names, punctuation, and journal styles throughout the reference list.

6. Figure Resolution:  While figure quality has improved, Figure 8 remains visually dense and somewhat difficult to interpret due to overlapping labels and tight spacing. If possible, consider improving resolution or dividing the figure into subpanels.

7. Table Title Placement: Currently, some table titles appear below the tables, which is inconsistent with MDPI format. According to journal guidelines, table titles should be placed above the tables.

Author Response

Dear Editor and Reviewers,

We sincerely thank you for your thoughtful and constructive comments on our manuscript. We greatly appreciate the time and effort you have devoted to reviewing our work. Your insights have significantly improved the clarity, depth, and scientific rigor of the study.

Below, we provide a detailed point-by-point response to each of the reviewers’ comments.

Comment 1: Figure and Table Captions

 “All figure captions should follow proper academic format, starting with "Figure X." and written as complete sentences ending with a period. For example: "Figure 3. Changes in carbon storage from 2002 to 2024." Several captions (Figures 3–8) currently begin with lowercase letters and/or lack terminal punctuation. Please revise for consistency.”

Response:

Thank you for your careful observation. We have revised all figure and table captions to follow academic standards. Each caption now begins with "Figure X." and is written as a complete sentence ending with a period, ensuring consistency and clarity across Figures 3–8.

Comment 2: Spacing and Punctuation

 “Add a space between text and parentheses where needed. For example, ‘In 2002(Figure 3a)’ should be ‘In 2002 (Figure 3a)’.”

Response:

We have carefully checked the manuscript and corrected all spacing issues between text and parentheses, including the example cited.

Comment 3: Scientific Notation for Units

 “Use consistent scientific notation for units. For instance: t·hm⁻² or Mg C ha⁻¹, with correct spacing. Avoid mixing styles such as t/hm² and t·hm⁻² within the same section.”

Response:

We have standardized the scientific notation for all units throughout the manuscript. Units such as t·hm⁻² and Mg C ha⁻¹ are now used consistently, with correct formatting and spacing in all sections.

Comment 4: Section Numbers and Headings

 “Section numbers and headings should be properly spaced. For example: ‘2.4.4Habitat Quality Module’ should be ‘2.4.4. Habitat Quality Module’.”

Response:

We have revised all section numbers and headings to include proper spacing. For instance, “2.4.4Habitat Quality Module” has been corrected to “2.4.4. Habitat Quality Module” for consistency.

Comment 5: Reference Consistency

 “Avoid duplicated references (e.g., Wischmeier & Smith, 1978 appears more than once). Ensure consistent formatting of author names, punctuation, and journal styles throughout the reference list.”

Response:

We have carefully reviewed and revised the entire reference list. Duplicate entries (such as Wischmeier & Smith, 1978) have been removed, and formatting of all references has been standardized according to journal guidelines.

Comment 6: Figure Resolution (Figure 8)

 “While figure quality has improved, Figure 8 remains visually dense and somewhat difficult to interpret due to overlapping labels and tight spacing. If possible, consider improving resolution or dividing the figure into subpanels.”

Response:

Thank you for this valuable suggestion. We have considered dividing Figure 8; however, this figure was designed to simultaneously present the interactions among multiple ecosystem services, and splitting it into subpanels would compromise its interpretability and weaken the overall message. To address the concern, we have improved the resolution and adjusted label spacing to enhance visual clarity while preserving the figure's integrative structure.

Comment 7: Table Title Placement

 “Currently, some table titles appear below the tables, which is inconsistent with MDPI format. According to journal guidelines, table titles should be placed above the tables.”

Response:

We have corrected the placement of all table titles according to MDPI formatting guidelines. All titles now appear above their respective tables.

 

Author Response File: Author Response.pdf