Review Reports - Vegetation Coverage Evolution Mechanism and Driving Factors in Dongting Lake Basin (China), 2000 to 2020

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Clarify data & preprocessing. Specify sources, years, spatial resolution, and resampling for climate (T/P/evap), population density, and GDP; state exactly how each factor was rasterized/categorized and matched to the 500 m NDVI grid.
Geodetector details. Explain the dependent variable (mean NDVI vs. trend/slope), the binning method and number of classes for each continuous driver, and whether you ran significance/permutation tests for q-values.
Seasonality: add interpretation. You report faster winter greening and seasonal “mutations”; add 2–3 sentences on plausible drivers (e.g., warmer winters, phenology shifts, cropping calendars, flood regime) with 1–2 citations.
Hydrology context. Dongting’s wetlands are flood-controlled; briefly discuss how lake/river regulation and inundation patterns could contribute to NDVI variability/decline in the northern plains.
Human drivers: nuance. Clarify how higher population/GDP aligns with greening (e.g., targeting of restoration programs, farmland-to-forest conversions) while acknowledging urbanization pressures; add one supporting reference or statistic if available.
Cross-check narrative consistency. The abstract highlights the DEM×population interaction; mirror this in Conclusions and ensure the factor ranking is consistent throughout.
Minor presentation fixes. Clean up subsection numbering, smooth a few awkward sentences, and standardize units (NDVI change as “per year,” not “%” unless clearly stated as relative).
Limitations & validation. Note NDVI saturation in dense forests and inundation effects in wetlands; if possible, add a brief cross-check with any available land-cover/forest-gain dataset (even qualitative).
Figures. Ensure all figure axes, colorbars, and thresholds are labeled; keep the intensity-analysis panels focused on the most policy-relevant transitions.

Author Response

Comment (1): Clarify data & preprocessing. Specify sources, years, spatial resolution, and resampling for climate (T/P/evap), population density, and GDP; state exactly how each factor was rasterized/categorized and matched to the 500 m NDVI grid.

Response 1: Thanks very much for you comment. We agree with this comment. Therefore in the revised manuscript, the source of climate, population density, and GDP was added. As pointed in 2.2 data source and processing (paragraph 3, page 4), the climate data was obtained from national meteorological science data center (http://data.cma.cn) and was interpolated using ANUSPLIN to get raster climate data with spatial resolution of 500 m. the population density and GDP data was acquired from the Chinese Statistic Yearbook at county-level.

Comment (2): Geodetector details. Explain the dependent variable (mean NDVI vs. trend/slope), the binning method and number of classes for each continuous driver, and whether you ran significance/permutation tests for q-values.

Response 2: Thanks very much for pointing this out. We agree with your comment. Therefore we added the details of Geodetector in the revised paper. As pointed in 2.3.5 Geodetector (page 7), sample points were generated across the study area using ArcGIS 10.6, and the NDVI value at 2020 were extracted to the sample points. In order to meet the requirement of Geodetector, the continuous independent variables were discretized into categories using the natural breaks. After removing outliers, the Geodetector model was conducted on the remaining 3280 samples.

In the factor detector model, a p-value was used to determine the statistical significance of the relationship between explanatory factors and dependent variable. The related content was added in the revised manuscript.

Comment (3): Seasonality: add interpretation. You report faster winter greening and seasonal “mutations”; add 2–3 sentences on plausible drivers (e.g., warmer winters, phenology shifts, cropping calendars, flood regime) with 1–2 citations.

Response 3: Thanks very much for your suggestion. We added statement on plausible drivers in the result section marked in red (paragraph 2, page 9).

Comment (4): Hydrology context. Dongting’s wetlands are flood-controlled; briefly discuss how lake/river regulation and inundation patterns could contribute to NDVI variability/decline in the northern plains.

Response 4: Thanks very much for your comments. We completely agree with what you said. The Dongting Lake is the second largest freshwater lake in China, playing a key role in water regulation and biodiversity conversion. The water level significantly influenced the temporal dynamics of vegetation coverage. Therefore we added discussion about how lake regulation contribute to NDVI variability in the northern plains. The revision are shown in red in Discussion (paragraph 2, page 16).

Comment (5): Human drivers: nuance. Clarify how higher population/GDP aligns with greening (e.g., targeting of restoration programs, farmland-to-forest conversions) while acknowledging urbanization pressures; add one supporting reference or statistic if available.

Response 5: Thanks very much for your comments. Human activities have both positive and negative impact on vegetation succession. On one hand, ecological protection strategies such as Grain to Green policy have converted farm-land with slopes greater than 25 degrees into forest land, which lead to an increase on vegetation coverage. On the other hand, rapid urbanization due to high population density lead to conversion from vegetation covered land to construction land, which caused a reduction on vegetation coverage. In the revised manuscript, related content was added in page 17 in red.

Comment (6): Cross-check narrative consistency. The abstract highlights the DEM×population interaction; mirror this in Conclusions and ensure the factor ranking is consistent throughout.

Response 6: Thanks very much for your comment. According to the results of interactive detector in Geodetector model, the interaction between DEM and population density enhanced the explanatory power for the vegetation coverage variation. We also read through the entire text and checked the statements about the Geodetector’s results. The factor ranking in Conclusions (paragraph 6, page 18) is consistent with abstract.

Comment (7): Minor presentation fixes. Clean up subsection numbering, smooth a few awkward sentences, and standardize units (NDVI change as “per year,” not “%” unless clearly stated as relative).

Response 7: Thanks very much for your comments. The subsection numbering has been checked throughout the whole paper. Meanwhile, the English statements were corrected by the English editing institute. The units in the whole paper were corrected. The revision are shown in red.

Comment (8): Limitations & validation. Note NDVI saturation in dense forests and inundation effects in wetlands; if possible, add a brief cross-check with any available land-cover/forest-gain dataset (even qualitative).

Response 8: Thanks very much for you comment. We agree with your suggestion, therefore we modified the Discussion section, the limitations of our research have been stated in Discussion (page 17 and page 18). We added a validation section in Methods (page 8). The TimeSync plus method was used to choose validation samples from high resolution time series Landsat images and Google Earth images. A total of 500 samples were selected to validate the vegetation change dynamics. Related contents was added in 2.3.6 Validation and accuracy assessment.

Comment (9): Figures. Ensure all figure axes, colorbars, and thresholds are labeled; keep the intensity-analysis panels focused on the most policy-relevant transitions.

Response 9: Thanks very much for your comments. Figures in the whole paper have been checked carefully. The axes, colorbars, and thresholds are all labelled. Considering the importance of vegetation coverage on carbon storage, soil conservation, and flood regulation, it is important to improve the vegetation condition. Therefore, in the transition intensity analysis, we mainly focused on transition among the medium, medium-high and high NDVI areas. In the revised paper, the intensity analysis panels remained the figures about medium, medium-high, and high NDVI transition (Figure 10 in page 14), as well as the figure statement (paragraph 3, page 13). The revised part was marked in red.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The reviewed paper studies the evolution of vegetation in a lake in China over a period of time using various approaches.

It presents many mathematical developments, some of which are unnecessary as they are obvious, and therefore it is suggested that some equations be eliminated.

It is a well-designed work with a strong mathematical focus, which in this case is applied to vegetation but could be used for any other parameter.

To better summarize the review, a series of questions are answered:

What is the main question addressed by the research? It involves justifying the evolution of vegetation cover in a region and over a period of time. The reason for this period must be justified.

Do you consider the topic original or relevant to the field? Do you? The study of vegetation variation is interesting if it can be substantiated that it has evolved and the reason for choosing that area.

Does it address a specific gap in the field? Please also explain why this is/is not the case. The study fills in the vegetation variability that appears to have been unknown.

What does it contribute to the subject area compared to other published literature? A region in China has been used, but it is understood that the application can be applied to similar areas anywhere in the world.

What specific improvements should the authors consider regarding the methodology? The methodology is not very accessible and is somewhat complicated due to the large amount of mathematical apparatus. It is recommended to develop a straightforward application program and avoid formulation.

Are the conclusions consistent with the evidence and arguments presented? And do they address the main question posed? Please also explain why this is/is not the case. The conclusions are based on the results and the initial hypotheses.

Are the references appropriate? The references are appropriate, but their presentation should be unified.

Any additional comments on the tables and figures?

Line 112. Indicate the climate type in the Geiger-Köppen classification.

Line 222: Eq. 1 and 2 are mentioned. If so, when (1) is mentioned, it should be indicated that it is Eq. 1 and all the same.

There is an excessive number of equations. Delete unnecessary ones.

Figure 2c). Indicate the elevation units on the abscissa axis.

Author Response

Comment (1): What is the main question addressed by the research? It involves justifying the evolution of vegetation cover in a region and over a period of time. The reason for this period must be justified.

Response 1: Thank you very much for this comment. The main questions were: (1) what changes occurred in Dongting Lake basin from 2000 to 2020? (2) What is the key factors affecting the spatial variation of vegetation coverage. We added related content in revised manuscript (paragraph 2, page 3). We also provided supplementary information related to the reasons for choosing 2000 to 2020 as study period for this study. The devastating flood that occurred in 1998 made the government realize the importance of environmental protection. Starting in 1999, a series of environmental protection policies, such as the Natural Forest Protection Project and the Grain to Green, brought significant changes to the vegetation coverage in the basin. At the same time, human-related engineering measures (the construction of the Three Gorges Dam) also altered the local water and heat cycles, indirectly affecting vegetation growth. Therefore, the vegetation coverage changed a lot during 2000 to 2020 (paragraph 2, page 3), we chose this period to detect the change pattern of vegetation coverage.

Comment (2): Do you consider the topic original or relevant to the field? Do you? The study of vegetation variation is interesting if it can be substantiated that it has evolved and the reason for choosing that area.

Response 2: Thanks very much for your comments. Previous studies have already conducted to analyze the vegetation cover change in Yangtze River, Dongting Lake Basin. So the topic is relevant to the field. Most of research was about wetland vegetation cover change or vegetation in east Dongting Lake. However, the whole Dongting Lake basin not only plays an important role in maintaining ecological security but is also an important grain producing area in the middle and lower reaches of the Yangtze River. Therefore, we chose the Dongting Lake base as the study area. Previous studies have mostly focused on analyzing the patterns of decrease or increase in vegetation cover, with relatively few analysis on the mutual conversion between different levels of vegetation coverage. In this paper, the intensity analysis method was used to analyze the transitions in and out of different levels of vegetation coverage. In the revised manuscript, the related content was added to the Introduction (page 3) of the manuscript shown in red.

Comment (3): Does it address a specific gap in the field? Please also explain why this is/is not the case. The study fills in the vegetation variability that appears to have been unknown.

Response 3: Thanks very much for your comments. Previous studies was mainly focusing on the wetland vegetation in Dongting Lake, and briefly analyzed the driving mechanism of vegetation coverage, but further studies on the conversion mechanisms between different vegetation coverage levels and the key driver of spatial and temporal evolution are still insufficient. In this paper, we used Sen’s Slope to analyze the inter-annual variation of vegetation coverage, and further explored the category intensity and transition intensity of different NDVI levels. The gains and losses of different NDVI levels between 2000-2010 and 2010-2020 were detailed analyzed, and the conversion mechanisms between different NDVI levels were demonstrated using the transition intensity (paragraph 1, page 3).

Comment (4): What does it contribute to the subject area compared to other published literature? A region in China has been used, but it is understood that the application can be applied to similar areas anywhere in the world.

Response 4: Thanks very much for your comments. In this study, we attempt to use intensity analysis to find out which period is active in vegetation coverage change during 2000 to 2020, and also to investigate which level of NDVI changed actively or dormant during the study period. For the vegetation coverage change, previous studies focused on the change trend of vegetation coverage, less consideration have been taken into the transformation mechanism among different NDVI levels. We also added some statement about limitation and future research in the Discussion part (paragraph 4, page 17).

Comment (5): What specific improvements should the authors consider regarding the methodology? The methodology is not very accessible and is somewhat complicated due to the large amount of mathematical apparatus. It is recommended to develop a straightforward application program and avoid formulation.

Response 5: Thanks very much for your valuable suggestions. We made some improvements to the method to make it easier to understand (page 5). The large amount of mathematical apparatus in intensity analysis are intended to show how each type of intensity is calculated. Aldwaik and Pontius have already developed an Excel based application program. Once the needed data is input, the intensity for category level and transition level can be calculated. In the future, we will make effort to develop a straightforward application program.

Comment (6): Are the conclusions consistent with the evidence and arguments presented? And do they address the main question posed? Please also explain why this is/is not the case. The conclusions are based on the results and the initial hypotheses.

Response 6: Thanks very much for your comments. We checked through the whole paper, and make sure that the conclusions are consistent with the evidence and arguments presented in the paper. Through the conclusion, the main questions were addressed properly. Considering the importance of the Dongting Lake basin in flood regulation, biodiversity conservation, and carbon storage, it is important to understand the spatial and temporal dynamics of vegetation coverage and its driving mechanism. The reason to choose Dongting Lake basin as case study was added in the revised manuscript (paragraph 4, page 2; paragraph 2, page 3).

Comment (7): Are the references appropriate? The references are appropriate, but their presentation should be unified.

Response 7: Thanks very much for your comments. All the references in the paper are valid citations and are listed in the references. We carefully checked the references, and make sure all the presentation have been unified.

Comment (8) Line 222: Eq. 1 and 2 are mentioned. If so, when (1) is mentioned, it should be indicated that it is Eq. 1 and all the same.

Response 8: Thanks very much for you careful review. We carefully checked the equation number mentioned in the statement, and modified inappropriate equation number in the original manuscript (paragraph 2 and paragraph 3, page 6).

Comment (9) There is an excessive number of equations. Delete unnecessary ones.

Response 9: Thanks very much for your comments. As you suggested, the number of equations has been reduced. The unnecessary ones have been deleted (page 5).

Comment (10) Figure 2c). Indicate the elevation units on the abscissa axis.

Response 10: Thank you very much for your careful review. We added the elevation units on the abscissa axis in Figure 2c in the revised manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

After I reviewed this manuscript, I found that although the topic is relevant and fits well within the journal’s scope, the methodological weaknesses seriously compromise its scientific reliability. The exclusive use of monthly MODIS NDVI composites smooths out short-term vegetation changes and introduces bias through the Maximum Value Composite approach, while the 500 m resolution is too coarse to detect localized land-use variation. The absence of any validation protocol (no field verification, no high-resolution imagery comparison) renders the reported NDVI trends uncertain. Key issues such as NDVI saturation, mixed pixels, and atmospheric effects were not addressed, further weakening the conclusions.

From a statistical standpoint, the methods selected (Sen’s slope, Mann–Kendall, and intensity analysis) are appropriate in theory but poorly implemented. The analysis neglects spatial autocorrelation and multicollinearity, employs arbitrary NDVI thresholds, and fails to provide confidence intervals or uncertainty measures. The application of the geodetector model is especially problematic, as it treats correlated variables without diagnostics and interprets associations as causal. The lack of significance testing, model validation, and correction for multiple comparisons undermines the credibility of the findings. More robust alternatives, such as spatial autoregressive models, variance partitioning, or BFAST, would be required to produce dependable results.

In the discussion, I observed frequent confusion between correlation and causation, unsupported claims about climate and human drivers, and an absence of quantitative comparison with other regional studies. The paper also omits a limitations section and fails to link NDVI trends to actual land-use or policy data. To make this work publishable, the authors need to reconstruct the analytical framework: integrate validation and spatial models, explicitly quantify uncertainty, and use attribution methods that separate climatic from anthropogenic effects. Until these revisions are made, the study’s conclusions remain methodologically weak and scientifically unreliable.

Comments on the Quality of English Language

The English is generally clear but requires moderate editing to reach publication standards. The main problems involve long and complex sentences that reduce readability, occasional grammatical errors such as subject-verb disagreement, inconsistent use of terms like “vegetation cover” and “vegetation coverage,” and uneven hyphenation in key expressions. Some words and phrases are awkward or ambiguous, and articles and prepositions are inconsistently applied. Abbreviations are not always defined when first used, and mathematical notation is sometimes blended with text in a way that interrupts flow.

Author Response

Comments (1):The exclusive use of monthly MODIS NDVI composites smooths out short-term vegetation changes and introduces bias through the Maximum Value Composite approach, while the 500 m resolution is too coarse to detect localized land-use variation. The absence of any validation protocol (no field verification, no high-resolution imagery comparison) renders the reported NDVI trends uncertain. Key issues such as NDVI saturation, mixed pixels, and atmospheric effects were not addressed.

Response 1: Thanks very much for your comments. Given that remote sensing imagery can only capture land cover at specific time, and vegetation index is calculated from such imagery, most current vegetation index products have a temporal resolution of 16 days. Based on a review of relevant literature, the most widely used method for deriving monthly NDVI data from current vegetation index products is the Maximum Value Composite (MVC) method. As you pointed out, while the MVC approach may smooth out short-term vegetation fluctuations, it effectively mitigates atmospheric and cloud contamination. Therefore, in this study, the MVC method was used to generate annual NDVI data from monthly NDVI composites (paragraph 2, page 4).

Regarding spatial resolution, the 500 meter resolution was selected to accommodate the extensive scope of the study area, as it adequately meets the research requirements. However, in future study, we will consider utilizing higher spatial resolution data, such as 250 meter products or Landsat imagery with 30 meter resolution, to obtain more detailed vegetation indices. We added a discussion about the high resolution vegetation index in page 18.

To validate vegetation coverage changes, sample points were selected from high resolution Landsat imagery and Google Earth imagery using the TimeSync plus software. The overall accuracy achieved was 83.4%, indicating that the detected vegetation changes meet acceptable precision standards. The related content was added in revised manuscript 2.3.6 Validation and accuracy assessment (page 8) in red.

Furthermore, concerning NDVI saturation issues. Since the Dongting Lake basin is a major grain-producing region in the middle and lower reaches of Yangtze River, and NDVI saturation primarily occurs in densely forested areas, NDVI remains effective for representing surface vegetation coverage in this context. Nevertheless, your suggestion is highly valuable, and we plan to explore other vegetation indices, such as EVI or TSAVI, in subsequent research to enhance the robustness of our analysis (page 18).

Comments (2): From a statistical standpoint, the methods selected (Sen’s slope, Mann–Kendall, and intensity analysis) are appropriate in theory but poorly implemented. The analysis neglects spatial autocorrelation and multicollinearity, employs arbitrary NDVI thresholds, and fails to provide confidence intervals or uncertainty measures. The application of the geodetector model is especially problematic, as it treats correlated variables without diagnostics and interprets associations as causal. The lack of significance testing, model validation, and correction for multiple comparisons undermines the credibility of the findings. More robust alternatives, such as spatial autoregressive models, variance partitioning, or BFAST, would be required to produce dependable results.

Response 2: Thank you very much for pointing this out. We checked our data and ensured that the NDVI was classified by natural breaks classification, which the class boundaries are not determined by arbitrary, pre-defined rules, but are found within the data’s own natural groupings. We modified the statement in page 7. About the geodetector model, we reviewed some related papers, and found that the Geodetector model is based on spatial variance rather than correlation, multicollinearity between factors is not a problem for factor detector. Meanwhile, the factor detector determined the contribution degree of each factor through the q value and assessed the statistical significance of each factor using p-value. In this study, the p-value for all factors were less than 0.05, indicating that the influence of each factor on vegetation coverage change was statistically significant. In the revised paper, related content was added in page 7 marked in red.

Comments (3): In the discussion, I observed frequent confusion between correlation and causation, unsupported claims about climate and human drivers, and an absence of quantitative comparison with other regional studies. The paper also omits a limitations section and fails to link NDVI trends to actual land-use or policy data. To make this work publishable, the authors need to reconstruct the analytical framework: integrate validation and spatial models, explicitly quantify uncertainty, and use attribution methods that separate climatic from anthropogenic effects.

Response 3: Thank you very much for pointing out this. The Geodetector was used to identify the main drivers influencing the spatial variation of vegetation coverage. The explanatory power and p-value were applied to determine the strength and significance of each individual factor’s impact on vegetation coverage. The driving factors affecting vegetation distribution have been discussed in the Discussion section. As you suggested, we added a “Limitation” subsection to the Discussion, where we address the shortcomings the NDVI data used and future works (4.3 research limitations in page 17).

Comments (4): Comments on the Quality of English Language

Response 4: Thank you very much for your review. This article has been polished by the professional English editing agency Editage. The "vegetation cover" and "vegetation coverage" has been uniformly modified to "vegetation coverage". Mathematical notation is consistent with statement.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Revise the Abstract to highlight the novelty and implications for sustainability science, rather than focusing solely on technical results.
Strengthen the Discussion & Conclusions by:
- Comparing more directly with similar NDVI studies in China and globally.
- Explicitly stating how findings guide ecological restoration, land-use policy, or climate resilience.

Comments on the Quality of English Language

Generally understandable, but some sentences are long, repetitive, or awkwardly phrased. A thorough English proofreading would improve readability and professionalism.
Some sections (methods, equations) have formatting issues.

Author Response

Comments 1:Revised the Abstract to highlight the novelty and implications for sustainability science, rather than focusing solely on technical results.

Response 1: Thank you so much for you suggestion. We modified part of Abstract and added some implications for sustainability science. The revised part was marked in red.

Comment 2: Strength the Discussions & conclusions by comparing more directly with similar NDVI studies in China and globally. Explicitly stating how findings guide ecological restoration, land use policy, or climate resilience.

Response 2: Thank you for your comments. As you suggested, we strengthened the Discussion part and stated the guidance for ecological protection. The revised part can be found in page 20 (line 615-629) and page 21 (703-709).

Comment 3: Comments on the quality of English Language.

Response 3: Thank you for pointing this out. In order to improve the English expression, we sent our manuscript to the English Editing company to polish the English Language. If there are still some problems, we will choose the Author Services to improve the English Language.

Reviewer 2 Report

Comments and Suggestions for Authors

With the changes introduced the work can be accepted in the present form

Author Response

Comment 1: With the changes introduced the work can be accepted in the present form

Response 1: Thank you so much for your careful review.

Reviewer 3 Report

Comments and Suggestions for Authors

I consider the improvements made to the document are not sufficient to recommend its publication in Sustainability without a major revision.

The manuscript examines vegetation changes in the Dongting Lake Basin over two decades, using MODIS NDVI data and analytical methods such as intensity analysis and geodetector models. It offers meaningful insights into how vegetation relates to climate and human activity in an ecologically important region. However, several methodological and interpretive flaws weaken its conclusions. The study often treats spatial correlations as causal relationships, especially when describing elevation or topography as “drivers” of vegetation change. Because the geodetector analysis relies on NDVI data from a single year, it cannot explain changes over time. The authors should reinterpret these findings as factors describing spatial distribution rather than causes of temporal variation.

The treatment of climatic factors also raises questions. Reported results suggest a weak influence of temperature and precipitation on vegetation, contradicting established evidence that climate strongly shapes vegetation in humid areas. This likely results from the static nature of the data used rather than ecological reality. Using only 2020 climate data ignores temporal fluctuations essential for understanding long-term dynamics. The study should therefore separate factors explaining spatial variation (such as topography) from those driving temporal change, including climate variability and policy effects. Incorporating time-series correlations between NDVI trends and meteorological records would clarify these relationships.

Several technical and structural issues further limit the study’s reliability. Relying solely on NDVI without validating it against other indices, such as EVI, overlooks known saturation problems in dense forests. The classification scheme for NDVI values is arbitrary and lacks ecological justification, while the limited temporal resolution (three time points) fails to capture intermediate or nonlinear changes. The validation method is poorly explained and unsuitable for continuous NDVI data, using too few samples to represent such a large region. The discussion of the Three Gorges Dam’s impact is superficial despite its known influence, and statistical procedures lack rigor, with no corrections for multiple testing or clear details on data processing parameters.

Despite these shortcomings, the study has potential. It uses an extensive dataset, applies recognized analytical methods, and addresses a region of clear ecological and policy relevance. To reach publishable quality, the authors must reframe their causal interpretations, improve validation procedures, test methodological sensitivity, and expand temporal analyses. Strengthening the discussion through clearer comparisons with existing studies and integrating hydrological and policy data would make the work more robust and informative. With these major revisions, the research could provide a valuable contribution to understanding vegetation dynamics in the Dongting Lake Basin.

Comments on the Quality of English Language

Author Response

Comment 1:The study often treats spatial correlations as causal relationships, especially when describing elevation or topography as “drivers” of vegetation change. Because the geodetector analysis relies on NDVI data from a single year, it cannot explain changes over time. The authors should reinterpret these findings as factors describing spatial distribution rather than causes of temporal variation.

Response: Thank you for pointing this out. We agree with your comment. Therefore we used the Geodetector model to explore the factors influencing the spatial differentiation of vegetation coverage. As you suggested, we have read the manuscript and revised the descriptions to depict the relationship between topographical factors and vegetation coverage distribution. Meanwhile, to investigate the response of vegetation coverage change to climate change, the partial correlation analysis was applied to examine the effects of temperature and precipitation on vegetation dynamics. The revised part was in red in page 17. (3.5 Relationship between climate change and NDVI change).

Comment 2: The treatment of climatic factors also raises questions. Reported results suggest a weak influence of temperature and precipitation on vegetation, contradicting established evidence that climate strongly shapes vegetation in humid areas. This likely results from the static nature of the data used rather than ecological reality. Using only 2020 climate data ignores temporal fluctuations essential for understanding long-term dynamics. The study should therefore separate factors explaining spatial variation (such as topography) from those driving temporal change, including climate variability and policy effects. Incorporating time-series correlations between NDVI trends and meteorological records would clarify these relationships.

Response: Thank you for pointing this out. We agree with your comments. Climate change is the main factor affecting vegetation coverage change. Therefore, in order to examine the relative influence strength of temperature and precipitation to vegetation dynamics, we applied partial correlation analysis, using the time series of NDVI as dependent variable, and the annual average temperature and annual precipitation data from 2000 to 2020 as independent variable, to investigate the effect of temperature and precipitation on NDVI temporal change. The description of the method was added at 2.3.6 Partial correlation analysis (page 8), and the result of partial correlation analysis was added at 3.5 Relationship between climate change and NDVI change (page 17).

Comment 3: Several technical and structural issues further limit the study’s reliability. Relying solely on NDVI without validating it against other indices, such as EVI, overlooks known saturation problems in dense forests. The classification scheme for NDVI values is arbitrary and lacks ecological justification, while the limited temporal resolution (three time points) fails to capture intermediate or nonlinear changes. The validation method is poorly explained and unsuitable for continuous NDVI data, using too few samples to represent such a large region. The discussion of the Three Gorges Dam’s impact is superficial despite its known influence.

Response: Thank you for pointing this out. In this paper, we only used NDVI dataset to detect the vegetation coverage change in DLB, because the NDVI is an effective and broadly used indices for environment monitoring. Although NDVI has some disadvantage in dense forests, we reviewed many paper using NDVI to analyze vegetation coverage in Dongtinghu area, such as “Spatiotemporal dynamics of vegetation coverage and its response to hydrological regime in the Dongting Lake, China” and “Spatiotemporal variation of vegetation and its driving mechanisms in Dongting Lake Basin”, which proved that NDVI is suitable for the vegetation coverage analysis in DLB. However, in the future work, we will try to employ the EVI or SAVI to improve the result. We added discussion on vegetation indices in Page 20, line 8.

For the classification of NDVI values, consultations of relevant studies have determined that the natural breaks method was used for NDVI classification. This is because the natural breaks classification can minimize the variance within classes and maximize the variance between classes. Some previous research already used natural breaks classification in vegetation coverage research, such as “Spatio-temporal dynamics of vegetation cover and its relationship with climate factors in the Yellow River Basin, China” and “Quantifying influences of natural factors on vegetation NDVI changes based on geographical detector in Sichuan, western China”.

For the validation samples, we added the samples from 500 to 1500, and the NDVI change detection was validated by the samples, the accuracy rate is fit for the research. The revised part is in page 9 (line 328- 335). As you suggested, the discussion of the Three Gorges Dam’s impact was deleted in the article.

Comment 4: Comments on the Quality of English Language.

Response: Thank you for pointing this out. In order to improve the English expression, we sent our manuscript to the English Editing company to polish the English Language. If there are still some problems, we will choose the Author Services to improve the English Language.

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

This third version maintain a clear and well-supported analysis of vegetation-cover dynamics in the Dongting Lake Basin between 2000 and 2020, using MODIS-NDVI data, intensity analysis, geodetector, and partial correlation. The methodology is coherent, and the findings offer useful insight into spatial and temporal vegetation changes and the factors influencing them.

The study’s strongest elements include its broad methodological approach, the validation based on 1,500 samples with an accuracy of 85.2%, the multilevel examination of temporal intervals and NDVI transitions, the identification of DEM as the dominant factor and temperature as the most relevant climatic variable, and the detection of a change point in 2010 through the Mann–Kendall test.

Several issues still need attention. The manuscript acknowledges but does not resolve NDVI saturation in dense forests, the absence of quantified political drivers, and the fact that the intensity analysis does not consider spatial distribution. These gaps should be addressed for publication.

Specific recommendations:

Additional validation: Use EVI or other indices less prone to saturation in areas with high canopy cover.
Policy quantification: Introduce measurable proxies of policy implementation (e.g., reforested area, investment levels) as variables for the geodetector analysis.
Improved spatial analysis: Incorporate measures of spatial autocorrelation such as Moran’s I to complement the intensity analysis.
Causal mechanisms: Although the results show correlations, the theoretical explanation of how driving factors influence NDVI change needs further development.
Practical implications: Broaden the conclusions (lines 703–709) with concrete management recommendations informed by the findings.

Recommended improvements include complementing NDVI with EVI or another index less susceptible to saturation, incorporating measurable proxies for policy implementation into the geodetector analysis, and adding a spatial-autocorrelation measure such as Moran’s I to balance the limits of intensity analysis. The discussion of causal mechanisms should be expanded, since the current results show correlations but do not fully explain how the driving factors influence vegetation change. The conclusions would also benefit from more concrete management recommendations based on the study’s findings.

Comments on the Quality of English Language

The English needs refinement to express the research more clearly. Examples:

Line 14: “is crucial in improving” should be “is crucial for improving” or “is crucial to improve.”
Lines 29–30: “stronger explanatory power for NDVI differentiation” is vague; “stronger influence on NDVI spatial variation” is clearer.
Line 78: “as they develop strategies” lacks logical connection; “when they develop strategies” is more precise.
Line 532: “factors affecting the spatial distribution” is inconsistent with the earlier use of “factors influencing.”
Lines 46–47: “multiple variables related to topographical factors, climate change, and human activities” is ambiguous because it is unclear whether “related to” applies to all elements.
Line 106: “Our sstudy objective” contains a typographical error.
Lines 223–227: sentences are excessively long and overloaded with subordinate clauses, which affects comprehension.
Line 478: “Considering the importance of vegetation coverage on carbon storage” should be “for carbon storage.”
Section 3.1 shifts between simple present (“shows,” line 388) and simple past (“showed,” line 420) when referring to the same analysis.
Lines 519–521 show an unwarranted switch from past to present in the description of results.
Line 173: “ndvi S is the dynamic trend” lacks the article “the.”
Line 290: “the explanatory power of each factor for the spatial variability” would be clearer as “on the spatial variability.”
Line 625: “the partial correlation results showed a stronger climatic signal than in some other subtropical regions” is awkwardly phrased and needs restructuring.
Line 120: “vegetation coverage in this area luxuriant” should be “vegetation coverage in this area is luxuriant” or “luxuriant vegetation coverage.”
Lines 372–380 contain a dense, repetitive passage that hinders readability.
Line 612: “As we know” is too colloquial for academic writing; “It is well established that” is more appropriate.
The manuscript alternates between “vegetation coverage” and “vegetation cover” without justification.
Terms such as “NDVI dynamics,” “NDVI change,” and “NDVI variation” are used interchangeably without defining their distinctions.

Author Response

Comment (1): Recommended improvements include complementing NDVI with EVI or another index less susceptible to saturation, incorporating measurable proxies for policy implementation into the geodetector analysis, and adding a spatial-autocorrelation measure such as Moran’s I to balance the limits of intensity analysis. The discussion of causal mechanisms should be expanded, since the current results show correlations but do not fully explain how the driving factors influence vegetation change. The conclusions would also benefit from more concrete management recommendations based on the study’s findings.

Response: Thank you very much for your careful review. For the saturation problem of NDVI, we reviewed some research, and found that in Dongting Lake Basin, the NDVI can be used to analyze the vegetation coverage dynamics, however, in further study, we will try to use the enhanced vegetation index to improve the accuracy of vegetation coverage change detection.

As you suggested, we add an indicator of reforested area into the geodetector model to reflect the influence of policy implementation aspect in vegetation coverage distribution. The revised part was marked in red. (Page 4, line 154 and line 165; page 16, line 548-551; and Figure 12 and 13 in page 17).

In order to balance the limits of intensity analysis on spatial distribution patterns, a spatial autocorrelation analysis method was added in the manuscript, as shown in page 7 (2.3.5 Spatial autocorrelation analysis). Correspondingly, the result part of spatial autocorrelation analysis was modified in page 16 (3.4 Spatial distribution characteristic in DLB). The revised parts were marked in red.

The discussion of causal mechanisms was expanded in page 20 (lines 646-663), marked in red.

As you suggested, the conclusions part also made some revision, which can be found in page 22 (lines 737 -745).

Comment (2): The English needs refinement to express the research more clearly.

Response: Thank you so much for pointing out the English problem, and we checked the manuscript carefully and corrected all the wrong place pointed above. The “vegetation coverage” and “vegetation cover” was unified to “vegetation coverage”. The “NDVI dynamics,” “NDVI change,” and “NDVI variation” were unified to “NDVI variation”.