Impacts of Landscape Mosaic Patterns on Habitat Quality Using OLS and GWR Models in Taihang Mountains of Hebei Province, China

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

The manuscript presents a highly relevant and original contribution to the assessment of habitat quality (HQ) using landscape mosaic (LM) patterns, OLS, and GWR models in the Taihang Mountains. The integration of PCA, SOM, and ED models to classify LM types, followed by the application of the InVEST-HQ framework, demonstrates methodological sophistication and innovation. However, some points should be strengthened to improve the paper’s technical rigor, clarity, and practical relevance. Here there are some suggestions.

1. Methodological clarification (Section 2, pages 4–6)

The methodological framework adopted in this manuscript is both innovative and ambitious, integrating a diverse set of analytical tools including PCA, SOM, ED, OLS, and GWR. While the breadth of methods is commendable, the current presentation lacks a clear, systematic explanation of how each method contributes to the overall research objectives and how they interrelate. This creates a challenge for readers, particularly those from interdisciplinary backgrounds, in fully understanding the workflow and analytical logic.

Specifically, Section 2.3 (pages 5–6) would benefit from a detailed justification for the combined use of OLS and GWR models. While OLS provides a global estimate of relationships across the study area, GWR offers local estimates that account for spatial heterogeneity. However, the manuscript does not explain why both were necessary, nor does it provide a theoretical rationale for their sequential or comparative application. The authors should clarify whether the intention was to validate the OLS results using GWR, to contrast global and local patterns, or to identify specific areas where spatial nonstationarity is pronounced. This clarification is essential to demonstrate the scientific rigor of the approach and to avoid the impression of methodological redundancy.

In addition, the section would greatly benefit from the inclusion of a schematic diagram or flowchart that visually summarizes the entire analytical pipeline — from data collection and preprocessing (including LULC and LM classification using PCA, SOM, and ED) to HQ modeling with the InVEST tool and statistical analysis with OLS and GWR. Such a figure could either update Figure 1 or be presented as a new figure, and it would significantly improve reader comprehension by providing an intuitive overview of a complex workflow.

Moreover, the authors should briefly describe the key assumptions and limitations of each method — for example, the spatial stationarity assumption in OLS, the bandwidth selection in GWR, and the dependence on input data quality in the InVEST model. A short discussion of how these assumptions were addressed or mitigated (e.g., through cross-validation or sensitivity analysis) would enhance the transparency and reproducibility of the study.

Finally, it is recommended that the authors explain the sequence in which the methods were applied and how the outputs of one informed the inputs of another. This will help demonstrate the methodological coherence of the study and make it more accessible to readers from disciplines such as landscape ecology, spatial statistics, and conservation planning.

2. Comparison between LM and LULC approaches (Section 3.2, pages 7–9; Discussion, Section 4, pages 12–13)

While the manuscript emphasizes the conceptual advantages of landscape mosaic (LM) classification over traditional land use and land cover (LULC) metrics, the empirical comparison between these approaches remains insufficiently developed. In Section 3.2 (pages 7–9), the authors present HQ results derived from LM classification but do not explicitly contrast them with those obtained using LULC categories. To strengthen the scientific contribution, it is recommended that the authors provide quantitative comparisons of model performance (e.g., R², AIC, BIC, RMSE) or present spatial overlays showing differences in HQ predictions under both classification systems. Including such a comparative analysis would empirically validate the claimed superiority of LM over LULC and demonstrate the added value of the proposed approach.

In Section 4 (pages 12–13, lines 425–450), the discussion could further explore the practical implications of switching from LULC to LM classification in conservation planning, including potential challenges such as increased data requirements or analytical complexity. Providing at least one illustrative example or case comparison within the study region — or referencing relevant international examples — would enrich the discussion and broaden the appeal to a global readership.

3. Discussion of limitations (Section 4, page 13; Conclusion, Section 5, page 15)

The manuscript currently addresses study limitations in a brief and somewhat general manner. To improve the scientific transparency and utility for future research, a more critical discussion is needed in Section 4 (page 13, lines 455–470). For example, the authors should elaborate on the limitations related to the spatial resolution and temporal coverage of input datasets (e.g., satellite imagery and socioeconomic variables), the potential for classification errors in LM mapping, and the sensitivity of the InVEST HQ model to input parameter uncertainty. Furthermore, the limitations associated with GWR, such as bandwidth selection and sensitivity to outliers, should be acknowledged.

In Section 5 (page 15), the authors are encouraged to reflect on the generalizability of their findings to other mountainous regions and to outline specific directions for future research, such as testing the approach in diverse ecological or socio-political contexts. This critical reflection would enhance the scientific rigor and position the paper as a reference point for subsequent studies.

4. Practical applications (Section 5, pages 15–16)

While the manuscript appropriately highlights the theoretical contributions of the LM-based HQ assessment, the practical recommendations in Section 5 (pages 15–16) remain rather broad. The authors should elaborate on concrete management or policy actions that could be informed by their findings. For example, they could suggest how LM-derived HQ maps could support the design of ecological corridors, the prioritization of conservation areas, or the evaluation of land-use policies. It would also strengthen the manuscript to mention how the methodology could be adapted to support biodiversity conservation in other temperate mountain ranges or rapidly urbanizing regions. Citing examples from comparable international case studies (e.g., the European Alps, the Appalachian Mountains) would contextualize the work within global sustainability discussions and make it more actionable for policymakers and practitioners.

Suggested references (for inclusion across sections)

Turner, M. G. (2005). Landscape ecology: What is the state of the science? Annual Review of Ecology, Evolution, and Systematics, 36, 319–344. https://doi.org/10.1146/annurev.ecolsys.36.102003.152614

Cushman, S. A., & McGarigal, K. (2021). Landscape metrics: Pitfalls and potential solutions. Landscape Ecology, 36, 237–252. https://doi.org/10.1007/s10980-020-01190-2

Comments on the Quality of English Language

Language and clarity (Sections 2, 3, and 4; pages 4–13)

Although the English language is generally acceptable, improvements are recommended to enhance clarity, precision, and readability, particularly in Sections 2, 3, and 4 (pages 4–13). Specific attention should be paid to technical terminology and complex sentence structures, notably in Section 2 (pages 4–6, lines 210–250), Section 3 (pages 7–11, lines 340–370), and Section 4 (pages 12–13, lines 420–450). It is advised to engage a native English-speaking editor with expertise in environmental sciences to ensure the technical expressions are accurately conveyed and the manuscript reads smoothly for an international audience.

Author Response

 [Comments 1]

The manuscript presents a highly relevant and original contribution to the assessment of habitat quality (HQ) using landscape mosaic (LM) patterns, OLS, and GWR models in the Taihang Mountains. The integration of PCA, SOM, and ED models to classify LM types, followed by the application of the InVEST-HQ framework, demonstrates methodological sophistication and innovation. However, some points should be strengthened to improve the paper’s technical rigor, clarity, and practical relevance. Here there are some suggestions.

[Response 1]

Thank you for your comments. Based on the feedback from the editor and the reviewers, we have made careful revisions. We have thoroughly revised the entire manuscript to address the shortcomings of previous research, aiming to improve the paper’s technical rigor, clarity, and practical relevance. Below are our point-by-point responses to all comments.

 

 [Comments 2]

1. Methodological clarification (Section 2, pages 4–6)

The methodological framework adopted in this manuscript is both innovative and ambitious, integrating a diverse set of analytical tools including PCA, SOM, ED, OLS, and GWR. While the breadth of methods is commendable, the current presentation lacks a clear, systematic explanation of how each method contributes to the overall research objectives and how they interrelate. This creates a challenge for readers, particularly those from interdisciplinary backgrounds, in fully understanding the workflow and analytical logic.

[Response 2]

Thank you for your suggestion. We have supplemented the theoretical framework (Figure 2, page 6), providing a systematic allocation of all research methods employed in this paper, thereby enhancing the clarity of the research workflow and analytical logic. (Line 200)

 

 [Comments 3]

Specifically, Section 2.3 (pages 5–6) would benefit from a detailed justification for the combined use of OLS and GWR models. While OLS provides a global estimate of relationships across the study area, GWR offers local estimates that account for spatial heterogeneity. However, the manuscript does not explain why both were necessary, nor does it provide a theoretical rationale for their sequential or comparative application. The authors should clarify whether the intention was to validate the OLS results using GWR, to contrast global and local patterns, or to identify specific areas where spatial nonstationarity is pronounced. This clarification is essential to demonstrate the scientific rigor of the approach and to avoid the impression of methodological redundancy.

[Response 3]

Thank you for your suggestion. In Section 2.4.4, we summarize the reasons for using OLS and GWR models: "We utilize OLS and GWR models to quantify the correlation between landscape patterns of LM and HQ indices. The primary aim is to understand the impact of LM landscape patterns on habitat quality from both a mathematical statistical and spatial differentiation perspective, while also comparing the computational results between global and local models." (Line 336-340)

 

 [Comments 4]

In addition, the section would greatly benefit from the inclusion of a schematic diagram or flowchart that visually summarizes the entire analytical pipeline — from data collection and preprocessing (including LULC and LM classification using PCA, SOM, and ED) to HQ modeling with the InVEST tool and statistical analysis with OLS and GWR. Such a figure could either update Figure 1 or be presented as a new figure, and it would significantly improve reader comprehension by providing an intuitive overview of a complex workflow.

[Response 4]

Thank you for your suggestion. A complete theoretical framework can greatly facilitate the reading of the paper. In this revision, we will modify the diagram representing the LM classification process to a comprehensive theoretical framework. (Line 200)

 

 [Comments 5]

Moreover, the authors should briefly describe the key assumptions and limitations of each method — for example, the spatial stationarity assumption in OLS, the bandwidth selection in GWR, and the dependence on input data quality in the InVEST model. A short discussion of how these assumptions were addressed or mitigated (e.g., through cross-validation or sensitivity analysis) would enhance the transparency and reproducibility of the study.

[Response 5]

Thank you for your suggestions. We have described the results of the stability calculations for the OLS model in Section 3.3 (p < 0.01, VIF < 7.50) (Line 492). Regarding the scientific rigor of the GWR and InVEST model analyses, we have primarily referenced several representative studies (Liu et al., 2019; Zhu et al., 2020) and have drawn on their research approaches to conduct the work presented in this paper (Line 336).

 

 [Comments 6]

Finally, it is recommended that the authors explain the sequence in which the methods were applied and how the outputs of one informed the inputs of another. This will help demonstrate the methodological coherence of the study and make it more accessible to readers from disciplines such as landscape ecology, spatial statistics, and conservation planning.

[Response 6]

Thank you for your suggestions. In Section 2.4, we summarize the progressive relationships between the various research methods and the models used:" We employed four sets of research methods to quantify the relationship between LM patterns and HQ indices in the study area, including LM classification, InVEST-HQ evaluation, LM pattern assessment, and their impact on InVEST-HQ evaluation results. PCA, SOM, and ED models were utilized for conducting LM classification, and the results of LM classification will serve as crucial foundational data to be imported into the InVEST-HQ model to obtain spatial quantification results of habitat quality in the study area. Meanwhile, the LM classification results will be input into FRAGSTATS software to calculate typical landscape pattern indices. Finally, OLS and GWR models will be employed to assess the impact of LM pattern indices on habitat quality results from the perspectives of mathematical statistics and spatial differentiation." (Line 219-228)

 

 [Comments 7]

2. Comparison between LM and LULC approaches (Section 3.2, pages 7–9; Discussion, Section 4, pages 12–13)

While the manuscript emphasizes the conceptual advantages of landscape mosaic (LM) classification over traditional land use and land cover (LULC) metrics, the empirical comparison between these approaches remains insufficiently developed. In Section 3.2 (pages 7–9), the authors present HQ results derived from LM classification but do not explicitly contrast them with those obtained using LULC categories. To strengthen the scientific contribution, it is recommended that the authors provide quantitative comparisons of model performance (e.g., R², AIC, BIC, RMSE) or present spatial overlays showing differences in HQ predictions under both classification systems. Including such a comparative analysis would empirically validate the claimed superiority of LM over LULC and demonstrate the added value of the proposed approach.

[Response 7]

We have added the HQ evaluation results based on the LULC classification system and calculated the spatial distribution of HQ differences under the LM and LULC classification systems. The results largely respond to the previous research content. The transitional landscape mosaics make the HQ model evaluation results more complex, with impacts primarily reflected in two aspects: 1) The presence of transitional landscape mosaics such as AG, FAG, and GA means that areas with concentrated forest and grassland do not necessarily indicate a higher habitat quality index; 2) The existence of transitional landscape mosaics such as UA, RA, and UAG suggests that areas with concentrated urban built-up regions may have a higher habitat quality index value compared to areas within the city.  (Line 457-479)

 

 [Comments 8]

In Section 4 (pages 12–13, lines 425–450), the discussion could further explore the practical implications of switching from LULC to LM classification in conservation planning, including potential challenges such as increased data requirements or analytical complexity. Providing at least one illustrative example or case comparison within the study region — or referencing relevant international examples — would enrich the discussion and broaden the appeal to a global readership.

[Response 8]

Thank you for your suggestions. In this revised manuscript, we have added two representative case studies in the discussion section. The first case is the rainforest biodiversity in Cameroon, highlighted in the seminal 1997 article published in Science. The second case pertains to the study of forest treelines in the European mountains. (Line 590-593)

 

 [Comments 9]

3. Discussion of limitations (Section 4, page 13; Conclusion, Section 5, page 15)

The manuscript currently addresses study limitations in a brief and somewhat general manner. To improve the scientific transparency and utility for future research, a more critical discussion is needed in Section 4 (page 13, lines 455–470). For example, the authors should elaborate on the limitations related to the spatial resolution and temporal coverage of input datasets (e.g., satellite imagery and socioeconomic variables), the potential for classification errors in LM mapping, and the sensitivity of the InVEST HQ model to input parameter uncertainty. Furthermore, the limitations associated with GWR, such as bandwidth selection and sensitivity to outliers, should be acknowledged.

[Response 9]

Thank you for your suggestions. In Section 4, we have added a focus on the limitations of the study: "We conducted LM classification using 12 factors. The resolution of the relevant factor data, the uniformity of time coverage, and the sensitivity of the LM classification results present scientific issues that warrant further exploration in the future." (Line 573-579) and "The fundamental parameters have a complex impact on the evaluation results of the InVEST-HQ model." (Line 601-602)

 

 [Comments 10]

In Section 5 (page 15), the authors are encouraged to reflect on the generalizability of their findings to other mountainous regions and to outline specific directions for future research, such as testing the approach in diverse ecological or socio-political contexts. This critical reflection would enhance the scientific rigor and position the paper as a reference point for subsequent studies.

[Response 10]

In Section 5, we have added a discussion on the future directions of this research: "This study, taking mountainous areas as an example, explores the impact of composite landscape mosaics on biodiversity. The applicability of this method to other mountainous regions, its feasibility for supporting watershed studies, urban agglomerations, and the differences in various social governance contexts can all serve as future avenues for this research: "This paper takes mountainous areas as an example to explore the impact of composite landscape mosaics on biodiversity. This approach can serve as a future direction for this research, considering its applicability to other mountainous regions, its feasibility for studies on supporting watersheds and urban agglomerations, and the differences in various social governance contexts. " (Line 665-669)

 

 [Comments 11]

4. Practical applications (Section 5, pages 15–16)

While the manuscript appropriately highlights the theoretical contributions of the LM-based HQ assessment, the practical recommendations in Section 5 (pages 15–16) remain rather broad. The authors should elaborate on concrete management or policy actions that could be informed by their findings. For example, they could suggest how LM-derived HQ maps could support the design of ecological corridors, the prioritization of conservation areas, or the evaluation of land-use policies. It would also strengthen the manuscript to mention how the methodology could be adapted to support biodiversity conservation in other temperate mountain ranges or rapidly urbanizing regions. Citing examples from comparable international case studies (e.g., the European Alps, the Appalachian Mountains) would contextualize the work within global sustainability discussions and make it more actionable for policymakers and practitioners.

[Response 11]

Thank you for your suggestions. We have further refined the recommendations in Section 4.3: "This can be achieved by leveraging urban arterial roads, rivers, and other infrastructures to develop large-scale green initiatives, particularly by linking TLM such as forest-grassland, forest-cropland, and grassland-cropland, treating them as critical corridor structures within the ecological network and enhancing their connectivity." (Line 623-626). Additionally, we have included the treeline ecotone of European mountain regions and the Andes mountains in the tropics in Section 4.2.

 

 [Comments 12]

Comments on the Quality of English Language

Language and clarity (Sections 2, 3, and 4; pages 4–13)

Although the English language is generally acceptable, improvements are recommended to enhance clarity, precision, and readability, particularly in Sections 2, 3, and 4 (pages 4–13). Specific attention should be paid to technical terminology and complex sentence structures, notably in Section 2 (pages 4–6, lines 210–250), Section 3 (pages 7–11, lines 340–370), and Section 4 (pages 12–13, lines 420–450). It is advised to engage a native English-speaking editor with expertise in environmental sciences to ensure the technical expressions are accurately conveyed and the manuscript reads smoothly for an international audience.

[Response 12]

Thank you for your suggestions. During the revision of this manuscript, we made efforts to simplify the language structure and enhance the expression of sentences to better meet the needs of an international readership. If this does not meet your acceptance, we hope to have the opportunity to further optimize the language of the paper. (Line 160-669)

Reviewer 2 Report

Comments and Suggestions for Authors

I found the topic of your research extremely interesting, especially your use of landscape mosaic patterns and InVEST to assess decadal changes in HQ. My suggested changes are noted as comments in the manuscript text (uploaded with this review). My general comments are as follows.

Introduction:

Provide more details about InVEST since you use this model to determine HQ.

I believe you should clarify what you mean by landscape elements, especially patches. The landscape mosaic is made of elements such as the matrix and patches within the matrix. It is also referred to as the patch mosaic model (PPM). Introduce landscape metrics as a way to measure landscape heterogeneity. Suggest you cite the Bioscience article by Forman and Godron (1981) and Turner (1989). References are provided in the comments.

Line 97, you note few studies have applied landscape mosaic classifications to HQ assessments. I disagree and suggest you refer to a body of literature from the late 1990s and early 2000s.  See for example, Riedler et al. (2015) and Riedler and Lang (2018), Schindler et al. (2015), Schumaker (1996) and Lausch et al. (2015). Just to name a few. You may also search on landscape structure an ecosystem services, Rieb and Bennett (2020). References are provided.

Line 112, I disagree with your statement that most existing research lacks methodologies for coordinating and implementing LM classification across multiple periods. 

Landscapes have been mapped and characterized using remotely sensed image data (both airborne and satellite based) to classify LULC, combined with terrain characteristics and socio-economic factors for decades (see for example, social vulnerability assessments of landscapes by Cutter et al. 2003). Landscape structure and composition has been computed using landscape metrics for decades. See McGarigal and Marks (1994) Fragstats Manual. Current methods use Python and R code to compute landscape metrics introduced in Fragstats, such as Bosch (2019) and Hesselbarth et al. (2019). Focal and zonal statistics are available in many GIS software (e.g., Esri ArcGIS Pro and Q-GIS). There are plenty of classification algorithms for landscape classification from supervised nearest neighbor and maximum likelihood to object-based image analysis/segmentation to machine learning and deep learning).

"Across multiple periods" simply means analyzing information derived from a time-series of imagery or maps. I don't see this as a gap that needs to be filled. I am interested in this manuscript because SOM is an additional method to characterize landscapes and assess ecosystem services.

Line 118, you say numerous studies indicate LULC changes significantly and impact HQ but you only cite one reference. Suggest you add Suggest you add a couple more citations since you say there are numerous studies. I suggest Lausch and Herzog (2002) and Paudel and Yuan (2012).

Materials and Methods

Lines 189-191, you mention investigating the impact of four LM pattern indices, but you don't say why you selected the, what they mean or cite McGarigal and Marks (1994). You need to say you used Fragstats to compute these landscape metrics much earlier.

Table 1. Please provide some details about the LULC data. What was the date and source of the satellite imagery used to create the LULC? Was there an accuracy assessment performed?

Lines 215-218, Suggest you move details about the LULC to your Data Sources section. What is the overall accuracy of your LULC? This is important since it is a major component of LM.

Lines 222-226, Please clarify your conversion of 30-m LULC to a point grid, and then using a kernel density to create a continuous surface (at 1000-km resolution?). Why did you need to do this since LULC was already in raster format and you can't have a continuous surface of categorial data (thematic classes denoted by numerical values).

Section 2.4.2, It seems like your emphasis on LM being a better indicator of HQ would require you to compute HQ with LM and also with LULC to see if there is a significant difference. Am I correct that you only used LM as input to InVEST to compute HQ? That is fine, but I don't think you can claim LM is superior to LULC since you did not test this.

Table 3 lists the LM classes as Threat factors, but the reader does not know what these class abbreviations mean. Throughout your manuscript you refer to the LM abbreviations, but the reader is forced to scroll down to Appendix A and read a relatively lengthy description of the LM class. At the very least you need to add a short name that the reader can associate with the abbreviation in order to understand everything you go on to say about LMs in your Methods, Results and Discussion.  You also could define each LM abbreviation the first time it is mentioned using the short name. The reader can learn the abbreviations this way and then also have Appendix A to refer to.

Also, where did the minimum distances and weights come from?

Line 270, Equation 1, Shouldn't this be Dxj instead of Dxy?

Table 4, Why is UC (Urban Continuous) listed as zero threat to all of the natural LMs?

Line 283, you say you build on existing research to employ the four landscape metrics, but you cite two recent (2010 and 2021) references from China. There is a large body of literature about patch, class and landscape metrics that dates back to the 1990s.

Line 287, You divided your study area into a 3 x 3km grid. Is this coarse resolution adequate for your study area? In the Methods (Line 239) and Discussion (Line 486) you say you utilized a 300 x 300 m grid cell to conduct the PCA and land cover classification.

Line 309, Equation 3, Shouldn't Xik be Xij? Also, the variables are in upper case in Equation 3 but lower case in your definition of variables.

Results

Line 317, Define pilar and transitional landscape mosaics.

Lines 321-322, you say all ten types of TLM include elements of farmland, so this suggests farmland is the predominant landscape element, but this is explicit in your LULC data as shown in your map and graph in Figure 4b. 

Table 5, you need to define the variables listed as X1 to X12. I had to go back to your workflow to see they are the 7 LULC codes, 2 elevation variables and nighttime light, population and GDP. At least you can define them in the legend or as a footnote to your table.

Section 3.1.2, it is extremely difficult to follow your narrative on the area change of LMs because you only refer to the abbreviations.

Line 352, Based on Figure 3, it looks like there was rapid urbanization in both decades.

Lines 353 to 359, The percent changes in LM are interesting but you only proved a couple of examples. Suggest you add a table in the Appendix that shows all of the LM changes in area and percent for each change period, 2000-2010, 2010-2020 and 2000 to 2020.

Lines 367-368, You could describe the changes more generally and refer to the LM abbreviations. For example, "...include transitions from AP to AR, UA to RA, and AR to UC" can be changed to, "...include LM transitions from agricultural to rural (e.g., AP to AR), urban peripheral to bare soil (UA to BA), and rural to urban (AR to UC)". If you did this throughout, your manuscript would be much more understandable. Also, there is no RA. Do you mean BA?

Figure 5, I really like Sankey diagrams to show landscape changes. However, shouldn't they be presented in the same order? AP should be on the top in 2010 and 2020 to match 2000. Also, I don't see any links that show stability or no change (e.g., AP to AP).

Section 3.2, Let your reader know what the levels of HQ classes are before you start to describe the results. You can say something like, "HQ values derived from the InVEST model were categorized into HQ levels ranging from I (extremely low) to V (extremely high).

Line 424, which landscape metrics (or indices) match the factors you list? Are complex patch edges measured by fractal mean (FRAC_MN) and split, continuity between patches by contagion (CONTAG) and diversity by the Shannon Diversity Index (SHDI)? If so, please add this to your text.

Table 6 is a bit confusing. What is Constant? I thought these were the values of the landscape metrics, but the title says they are standardized regression coefficients of the OLS model for different years. It would be interesting to also see the metric values and how they change over time.

Figure 8d, the mapped SHDI results are opposite from what I expected. Why are the agricultural and urban areas more diverse than the forested areas, and over time, the southern urban areas become more diverse? If this is due to a greater diversity in LMs and not an increase in biodiversity, I think you need to be clearer in your text.  Line 450, you say diverse LM types are more likely to signify higher-quality biological habitat in the plains and foothills, but these types are agricultural, rural and urban that are not typically associated with high-quality habitat for plants and animals. 

Discussion

Lines 459-460, define the abbreviations LPT, MLT, LUs, TPM and LPCT.

Line 467, should be replaced.

Line 469, should this be, "...validate classification results and enhanced the accuracy..."?

Line 500, you say the results showed significant differences in habitat suitability parameters (does this mean HQ?) compared to LULC-based assessments, but you didn't input LULC to InVEST to compare HQ.

Line 533-535, again you say the expansive plains need a greater diversity of LM types to enhance the survival of local organisms, but adding agricultural and urban LMs would not necessarily improve biodiversity and HQ.

Appendix A, please add short names for these LM abbreviations and delineate the descriptions better so the reader can more easily associate the correct description with the abbreviation.

Line 641, References, Number 25, should be Godron instead of Gordon.

 

Comments for author File: Comments.pdf

Author Response

 [Comments 1]

I found the topic of your research extremely interesting, especially your use of landscape mosaic patterns and InVEST to assess decadal changes in HQ. My suggested changes are noted as comments in the manuscript text (uploaded with this review). My general comments are as follows.

[Response 1]

Thank you for recognizing the research theme of this paper. As global ecological and environmental changes intensify, our effort aims to propose a method for expressing landscape heterogeneity that is easily promotable. This method not only builds upon the perspectives of numerous historical academic achievements, such as the Matrix of Landscape Types, but also introduces relatively novel data analysis models, including the Self-Organizing Map (SOM). This approach provides new insights for habitat quality assessment research based on the InVEST model. In response to your comments, we have made careful revisions. Below are our detailed responses to all comments.

 

Introduction:

 [Comments 2]

Provide more details about InVEST since you use this model to determine HQ.

[Response 2]

In section 2.4.2, we provide a detailed description of the InVEST model: "The InVEST model, collaboratively developed by Stanford University, the World Wildlife Fund, and the Nature Conservancy, consists of a suite of spatial models aimed at analyzing and predicting the provision of ecosystem services and habitat availability. By utilizing land-use maps in conjunction with relevant biophysical, economic, and institutional data for a specific region, the model quantitatively evaluates habitat quality from a biodiversity perspective." (Line 277-281)

 

 [Comments 3]

I believe you should clarify what you mean by landscape elements, especially patches. The landscape mosaic is made of elements such as the matrix and patches within the matrix. It is also referred to as the patch mosaic model (PPM). Introduce landscape metrics as a way to measure landscape heterogeneity. Suggest you cite the Bioscience article by Forman and Godron (1981) and Turner (1989). References are provided in the comments.

[Response 3]

In the introduction, we supplemented an explanation regarding the composition of elements in landscape mosaics: "From a landscape ecology perspective, the migration paths and habitat preferences of organisms are often more susceptible to the influences of the landscape mosaic (LM), which encompasses various elemental attributes, including the matrix and the patches within it." In the context of biodiversity conservation, both patches and corridors within the landscape matrix (LM) are particularly significant and critical. The interplay of climate, topography, and plant resources generates a complex mosaic. Meanwhile, we have also included references to the suggested literature (Forman & Gordon, 1996; Turner, 1989). (65-70)

 

 [Comments 4]

Line 97, you note few studies have applied landscape mosaic classifications to HQ assessments. I disagree and suggest you refer to a body of literature from the late 1990s and early 2000s.  See for example, Riedler et al. (2015) and Riedler and Lang (2018), Schindler et al. (2015), Schumaker (1996) and Lausch et al. (2015). Just to name a few. You may also search on landscape structure an ecosystem services, Rieb and Bennett (2020). References are provided.

[Response 4]

In the revised manuscript, we have modified the sentence to: " Overall, in the past few decades, particularly from the late 1990s to the early 21st century, there has been a growing interest in the classification of landscape mosaics, their spatiotemporal evolution, and their impact on ecosystem service functions. Nevertheless, compared to the research outcomes based on LULC classification systems, the academic achievements related to ecosystem services based on LM remain relatively scarce. This highlights a significant contrast for argumentation between the two fields. Additionally, an indirect consequence is that many findings from field investigations remain disconnected from the results of HQ model assessments. Furthermore, we have included the recommended references, such as Riedler et al. (2015), Schumaker (1996), Schindler et al. (2015), Riedler and Lang (2018), and Rieb and Bennett (2020). (Line 95-103)

 

 [Comments 5]

Line 112, I disagree with your statement that most existing research lacks methodologies for coordinating and implementing LM classification across multiple periods. 

[Response 5]

I respectfully disagree with your assertion that the majority of existing research lacks methodologies for coordinating and implementing LM classification across multiple periods. It is essential to recognize that various studies have proposed robust frameworks and methodologies that effectively address this issue. (Line 117-120)

 

 [Comments 6]

Landscapes have been mapped and characterized using remotely sensed image data (both airborne and satellite based) to classify LULC, combined with terrain characteristics and socio-economic factors for decades (see for example, social vulnerability assessments of landscapes by Cutter et al. 2003). Landscape structure and composition has been computed using landscape metrics for decades. See McGarigal and Marks (1994) Fragstats Manual. Current methods use Python and R code to compute landscape metrics introduced in Fragstats, such as Bosch (2019) and Hesselbarth et al. (2019). Focal and zonal statistics are available in many GIS software (e.g., Esri ArcGIS Pro and Q-GIS). There are plenty of classification algorithms for landscape classification from supervised nearest neighbor and maximum likelihood to object-based image analysis/segmentation to machine learning and deep learning).

[Response 6]

Thank you for your systematic introduction to remote sensing image data and the analysis of landscape pattern indices. We have thoroughly reviewed your viewpoints and the references you provided, which we believe are crucial theoretical foundations for our subsequent research. Regarding the existing research methods you mentioned for landscape classification, we will incorporate our considerations of the inadequacies in this field into the limitations of our study, specifically: "Various classification methods, such as landscape features, ecological units, and landscape units, share similarities with the classification of landscape mosaics. Furthermore, there are numerous classification algorithms that can be further explored, including supervised nearest neighbor, machine learning, and deep learning." (Line 569-573)

 

 [Comments 7]

"Across multiple periods" simply means analyzing information derived from a time-series of imagery or maps. I don't see this as a gap that needs to be filled. I am interested in this manuscript because SOM is an additional method to characterize landscapes and assess ecosystem services.

[Response 7]

Your perspective is very enlightening. We have revised "To address this gap" to "In order to facilitate the comprehensive classification of multiple factors and various time periods." (Line 120)

 

 [Comments 8]

Line 118, you say numerous studies indicate LULC changes significantly and impact HQ but you only cite one reference. Suggest you add Suggest you add a couple more citations since you say there are numerous studies. I suggest Lausch and Herzog (2002) and Paudel and Yuan (2012).

[Response 8]

Thank you for your suggestion. We have incorporated the two referenced works you proposed into the corresponding sections of the manuscript. (Line 124)

 

 [Comments 9]

Materials and Methods

Lines 189-191, you mention investigating the impact of four LM pattern indices, but you don't say why you selected the, what they mean or cite McGarigal and Marks (1994). You need to say you used Fragstats to compute these landscape metrics much earlier.

[Response 9]

FRAGSTATS is a spatial pattern analysis program designed to quantify landscape structures. It effectively measures the spatial heterogeneity of landscapes, represented in either categorical maps or continuous surfaces. This approach has become increasingly common for quantifying landscape pattern indices. (Line 301-304)" and "We employed four land metrics with low correlation at the landscape level to represent the landscape mosaic pattern: the Mean Fractal Dimension Index (FRAC_MN), the Splitting Index (SPLIT), Contagion (CONTAG), and Shannon’s Diversity Index (SHDI). These four indices correspond to four dimensions, including shape, aggregation-subdivision, aggregation-interspersion, and diversity."  (Line 308-313)

 

 

 [Comments 10]

Table 1. Please provide some details about the LULC data. What was the date and source of the satellite imagery used to create the LULC? Was there an accuracy assessment performed?

[Response 10]

In Section 2.3, we have made textual additions focusing on the sources of LULC data. Specifically, "LULC is the most important foundational data in this study. We have obtained data for the years 2000, 2010, and 2020. The data utilized is a national-scale, multi-period land use/land cover remote sensing detection database established by the Chinese Academy of Sciences, based on the National Resources and Environment Database and primarily leveraging US Landsat remote sensing imagery." The LULC data is characterized by high accuracy and robust applicability, featuring a spatial resolution of 30 meters and an overall accuracy exceeding 90% (Liu et al., 2014; Wu & Fan, 2024). This data has been validated through research across various domains, including urban expansion, land prediction, and ecosystem service assessment. (Line 205-213)

 

 [Comments 11]

Lines 215-218, Suggest you move details about the LULC to your Data Sources section. What is the overall accuracy of your LULC? This is important since it is a major component of LM.

[Response 11]

Thank you for your suggestions; they highlight a logical oversight in our manuscript. In this revision, we have relocated the information regarding LULC-related data to Section 2.3. Section 2.4.1 now solely presents the processing procedures for the LULC data. (Line 205-213)

 

 [Comments 12]

Lines 222-226, Please clarify your conversion of 30-m LULC to a point grid, and then using a kernel density to create a continuous surface (at 1000-km resolution?). Why did you need to do this since LULC was already in raster format and you can't have a continuous surface of categorial data (thematic classes denoted by numerical values).

[Response 12]

In the classification process of landscape mosaics, we primarily referenced the research methodology of Vizzari et al. (2018). Our initial consideration was to fully cite the methodological logic of their study, but we made adjustments based on it (such as replacing the ISODATA model with the SOM model and adding the ED model to validate classification results). Following your suggestion, there are indeed more convenient pathways available in the ESRI ArcGIS analysis module. Based on this, we supplemented Section 4.1 with the research limitations: "Additionally, the processing of LULC data in this paper is relatively cumbersome; leveraging the powerful computational capabilities of ESRI ArcGIS may allow for significant simplification of this process."  (Line 567-569). Alternatively, we hope to provide some additional time to further develop a simplified classification study of landscape mosaics and engage in comparative discussions with the current research findings.

 

 [Comments 13]

Section 2.4.2, It seems like your emphasis on LM being a better indicator of HQ would require you to compute HQ with LM and also with LULC to see if there is a significant difference. Am I correct that you only used LM as input to InVEST to compute HQ? That is fine, but I don't think you can claim LM is superior to LULC since you did not test this.

[Response 13]

We have supplemented the appendix with the InVEST-HQ assessment results based on LULC (Line 485), and in Section 3.2, we present the spatial distribution of the HQ index derived from the LM assessment minus the HQ index derived from the LULC assessment, to illustrate the differences in evaluation results based on the two classification systems (Line 458-480). The results largely address the content of the previous research. The transitional landscape mosaics complicate the HQ model assessment results, primarily impacting two aspects: 1) The emergence of transitional landscape mosaics such as AG, FAG, and GA means that areas with concentrated forest and grassland do not necessarily indicate higher habitat quality index values; 2) The presence of transitional landscape mosaics such as UA, RA, and UAG suggests that areas with concentrated urban built-up regions may also possess higher habitat quality index values compared to areas within the urban fabric. (Line 476-480)

 

 [Comments 14]

Table 3 lists the LM classes as Threat factors, but the reader does not know what these class abbreviations mean. Throughout your manuscript you refer to the LM abbreviations, but the reader is forced to scroll down to Appendix A and read a relatively lengthy description of the LM class. At the very least you need to add a short name that the reader can associate with the abbreviation in order to understand everything you go on to say about LMs in your Methods, Results and Discussion.  You also could define each LM abbreviation the first time it is mentioned using the short name. The reader can learn the abbreviations this way and then also have Appendix A to refer to.

[Response 14]

We have supplemented the full name of LM in Table 3-4 and Appendix A.1 (Line 294-297, 684-686). Table 3-4 is the first paragraph where the classification results of LM appear, which facilitates readers in establishing the relationship between the full name and the abbreviation of LM, thereby enhancing reading fluency.

 

 [Comments 15]

Also, where did the minimum distances and weights come from?

[Response 15]

The values for the maximum distance and weights of various threat factors are set based on findings from studies similar to the field of this research. The revised manuscript has also incorporated two additional references (Bian et al., 2022; Wang et al., 2021). (Line 288). These references either possess significant academic influence or are highly aligned with the primary types of TLM involved in this study.

 

 [Comments 16]

Line 270, Equation 1, Shouldn't this be Dxj instead of Dxy?

[Response 16]

This is a review of our proofreading oversights. The manuscript has now been revised to address this issue. (Line 290-293)。

 

 [Comments 17]

Table 4, Why is UC (Urban Continuous) listed as zero threat to all of the natural LMs?

[Response 17]

In our study, the UC sensitivity parameter is set based on the InVEST User's Guide and the results of similar research methods. It can be understood that: "Compared to suburban or natural areas, the organisms living in urban core environments generally possess stronger adaptability; therefore, their sensitivity to the external LM environment is 0." In this revised manuscript, we have added a citation for the InVEST User's Guide. (Line 288)

 

 [Comments 18]

Line 283, you say you build on existing research to employ the four landscape metrics, but you cite two recent (2010 and 2021) references from China. There is a large body of literature about patch, class and landscape metrics that dates back to the 1990s.

[Response 18]

In accordance with your suggestions, we have included two earlier references (Tinker et al., 1998; Lu & Li, 2003). (Line 308)。

 

 [Comments 19]

Line 287, You divided your study area into a 3 x 3km grid. Is this coarse resolution adequate for your study area? In the Methods (Line 239) and Discussion (Line 486) you say you utilized a 300 x 300 m grid cell to conduct the PCA and land cover classification.

[Response 19]

The statement has been adjusted to section 2.4.4 and modified to: "Before conducting OLS and GWR analyses, the analysis divided the study area image into a 3 km × 3 km grid using the Create Fishnet tool in ESRI ArcGIS 10.5. Subsequently, variables such as the landscape mosaic pattern index and habitat quality required for the study were aggregated into the grid." (Line 349-352) In this study, we divided the study area into multiple grid units twice: ① For LM classification, the study area was divided into grid units of 300 × 300 m; ② For OLS and GWR analyses, the study area was divided into grid units of 3 × 3 km. These divisions were made for different research purposes, in conjunction with references such as Vizzari et al. (2018).

 

 [Comments 20]

Line 309, Equation 3, Shouldn't Xik be Xij? Also, the variables are in upper case in Equation 3 but lower case in your definition of variables.

[Response 20]

Thank you for your suggestions. We acknowledge the oversight in our review process, and the manuscript has been revised to address this issue. (Line 338-340)。

 

Results

 [Comments 21]

Line 317, Define pilar and transitional landscape mosaics.

[Response 21]

Thank you for your suggestion. We have included the definitions of PLM and TLM in the revised manuscript: "Regarding the definitions of PLM and TLM, we refer to the research methodology of Vizzari et al., which defines a LM as a PLM when the area proportion of a particular LULC type is ≥75%, and as a TLM when the combined area proportion of two or more LULC types is ≥75%." (Line 358-361)

 

 [Comments 22]

Lines 321-322, you say all ten types of TLM include elements of farmland, so this suggests farmland is the predominant landscape element, but this is explicit in your LULC data as shown in your map and graph in Figure 4b. 

[Response 22]

Thank you for your suggestion. In the revised manuscript, we changed this sentence to: "Notably, with the exception of GF, all ten types of TLM incorporate elements of farmland landscapes, which suggests that there is a close interrelationship between farmland and most landscape spaces such as forest, grassland, water, urban districts, and rural settlement." (Line 364-367)

 

 [Comments 23]

Table 5, you need to define the variables listed as X1 to X12. I had to go back to your workflow to see they are the 7 LULC codes, 2 elevation variables and nighttime light, population and GDP. At least you can define them in the legend or as a footnote to your table.

[Response 23]

Thank you for your suggestion. We have added annotations to Table 5 in the manuscript to provide readers with a more intuitive understanding of the twelve classification variables of landscape mosaics. (Line 380-383)

 

 [Comments 24]

Section 3.1.2, it is extremely difficult to follow your narrative on the area change of LMs because you only refer to the abbreviations.

[Response 24]

Thank you for your suggestions. This manuscript has undergone significant revisions in Section 3.1.2, primarily aimed at enhancing the reader's understanding. For example, the original statement has been modified to: "Figure 4-a illustrates the area proportions and change rates of land cover classification results in the study area from 2000 to 2020. It is evident that the PLM, primarily composed of farmland (AP and AH), forests (FM), and grasslands (GH), dominates the study area, collectively accounting for approximately 50% of the total area." The agricultural land in plain areas (AP) comprises approximately 18% of the total area, while the forests in mountainous regions (FM) account for roughly 13%. Additionally, the grasslands and agricultural lands in hillside areas (GH and AH) each represent about 11%. In contrast, TLM constitutes less than 30% of the total area, indicating that a single element predominantly characterizes most landscape mosaic types in the study area. Within the TLM category, the area proportions of the transitional zones between agricultural land and grassland, rural settlements (AG and AR), and between forests and grassland (GF) are more pronounced than those of other TLM types, each maintaining a stable area of around 6%, with a total area ranging between 2000 and 3000 km². Analyzing the temporal evolution, the area and proportion of landscape patches in the study area experienced significant changes throughout the research period, which is a result of the rapid urbanization development in Hebei Province, China. Notably, the overall change in landscape patches from 2000 to 2010 was slightly greater than that from 2010 to 2020. From 2000 to 2010, the total area of Arable land-scape in plains (AP) decreased from 7157.88 km² to 6455.61 km², reflecting a decline of 9.81%, marking the most substantial change in landscape mosaics during the study period. Conversely, the total area of Barren-Arable land transitional area (BA) increased from 54.00 km² to 155.88 km², representing a remarkable increase of 188.67%, the most significant landscape mosaic change event during the study period. Additionally, the dynamic changes in the transitional spaces between cultivated land, urban areas, and rural settlements (UC, UA, and RA) are significant. " (Line 388-422)

 

 [Comments 25]

Line 352, Based on Figure 3, it looks like there was rapid urbanization in both decades.

[Response 25]

As you mentioned, the LULC and LM in the study area underwent significant changes over the two decades of research, with the degree of change being slightly more pronounced from 2000 to 2010. Based on this, we revised the statement in the specified location of the manuscript to: "Analyzing the temporal evolution, the area and proportion of landscape patches in the study area experienced significant changes throughout the research period, which is a result of the rapid urbanization development in Hebei Province, China." (Line 400-403)

 

 [Comments 26]

Lines 353 to 359, The percent changes in LM are interesting but you only proved a couple of examples. Suggest you add a table in the Appendix that shows all of the LM changes in area and percent for each change period, 2000-2010, 2010-2020 and 2000 to 2020.

[Response 26]

Your suggestions are important. In the revised manuscript, we have included a more comprehensive statistical table of the areas of spatial transformations of LMs in Appendix A.5-6. (Line 700-704)

 

 [Comments 27]

Lines 367-368, You could describe the changes more generally and refer to the LM abbreviations. For example, "...include transitions from AP to AR, UA to RA, and AR to UC" can be changed to, "...include LM transitions from agricultural to rural (e.g., AP to AR), urban peripheral to bare soil (UA to BA), and rural to urban (AR to UC)". If you did this throughout, your manuscript would be much more understandable. Also, there is no RA. Do you mean BA?

[Response 27]

Thank you for your suggestions and examples. We have focused on revising Section 3.1.2 to enhance the clarity for readers. As the reasons for the modifications in this section are the same as those mentioned in Response 23, we will not reiterate them here.

 

 [Comments 28]

Figure 5, I really like Sankey diagrams to show landscape changes. However, shouldn't they be presented in the same order? AP should be on the top in 2010 and 2020 to match 2000. Also, I don't see any links that show stability or no change (e.g., AP to AP).

[Response 28]

According to your suggestions, we have adjusted the order of the landscape mosaic bodies in Figure 5 to align with the other figures and tables. Additionally, we have chosen not to display stable or unchanged links for two reasons: (1) The dynamic changes of the 19 types of landscape mosaic bodies are quite complex, and we aim to simplify the complexity of the diagrams to better highlight the spatial transformations occurring within the research area, specifically the landscape mosaics (LM); (2) The unchanged areas occupy a significant proportion of the total area, and including them in the Sankey diagram would result in an excessively large aspect ratio for the figures. (Line 427)

 

 [Comments 29]

Section 3.2, Let your reader know what the levels of HQ classes are before you start to describe the results. You can say something like, "HQ values derived from the InVEST model were categorized into HQ levels ranging from I (extremely low) to V (extremely high).

[Response 29]

Thank you for your suggestion. We have added the following statement at the beginning of Section 3.2: "In order to present the assessment results based on InVEST-HQ more clearly, HQ values were categorized into HQ levels ranging from I (extremely low) to V (extremely high)." (Line 429-431)

 

 [Comments 30]

Line 424, which landscape metrics (or indices) match the factors you list? Are complex patch edges measured by fractal mean (FRAC_MN) and split, continuity between patches by contagion (CONTAG) and diversity by the Shannon Diversity Index (SHDI)? If so, please add this to your text.

[Response 30]

According to your suggestion, we will supplement the landscape pattern metrics after the types of landscape mosaics being measured. The statement has been modified to: "Factors such as complex patch edges (FRAC_MN), continuity between patches (SPLIT, CONTAG), and diversity of patch types (SHDI) all influence the HQ index." (Line 505-507)

 

 [Comments 31]

Table 6 is a bit confusing. What is Constant? I thought these were the values of the landscape metrics, but the title says they are standardized regression coefficients of the OLS model for different years. It would be interesting to also see the metric values and how they change over time.

[Response 31]

The OLS constant refers to the calculated function result obtained when assessing the correlation between the habitat quality index of the study area and the individual landscape mosaic pattern index. In this study, its influence is less significant than that of the landscape mosaic pattern index value. Therefore, the OLS constant term has been removed in the revised manuscript. (Line 510)

 

 [Comments 32]

Figure 8d, the mapped SHDI results are opposite from what I expected. Why are the agricultural and urban areas more diverse than the forested areas, and over time, the southern urban areas become more diverse? If this is due to a greater diversity in LMs and not an increase in biodiversity, I think you need to be clearer in your text.  Line 450, you say diverse LM types are more likely to signify higher-quality biological habitat in the plains and foothills, but these types are agricultural, rural and urban that are not typically associated with high-quality habitat for plants and animals. 

[Response 32]

Upon reviewing the proofs, we found discrepancies during the figure generation process. As you stated, the positive influence of diverse landscape mosaic patterns on habitat quality is primarily distributed in the hillside areas (Line 537-538). The original statement in line 450 has been modified to: "Specifically, diverse LM types are more likely to signify higher-quality biological habitats in the hillsides." (Line 532-533)

 

Discussion

 [Comments 33]

Lines 459-460, define the abbreviations LPT, MLT, LUs, TPM and LPCT.

[Response 33]

Thank you for your suggestions. The introduction of the paper has provided definitions for LPI, MLT, LUs, and TPM (Line109-110). Since these abbreviations have been introduced previously, we recommend retaining their abbreviated forms in the discussion section.

 

 [Comments 34]

Line 467, should be replaced.

[Response 34]

The ISODATA model should be replaced by the SOM model. (Line 549)

                                            

 [Comments 35]

Line 469, should this be, "...validate classification results and enhanced the accuracy..."?

[Response 35]

ED was employed to validate classification results and enhanced the accuracy of the LM classification framework. (Line 551)

 

 [Comments 36]

Line 500, you say the results showed significant differences in habitat suitability parameters (does this mean HQ?) compared to LULC-based assessments, but you didn't input LULC to InVEST to compare HQ.

[Response 36]

In this revised manuscript, we have supplemented the InVEST-HQ assessment results based on the LULC classification system and conducted a comparative analysis with the assessments based on the LM classification system. (Line 458-480)

 

 [Comments 37]

Line 533-535, again you say the expansive plains need a greater diversity of LM types to enhance the survival of local organisms, but adding agricultural and urban LMs would not necessarily improve biodiversity and HQ.

[Response 37]

In this manuscript revision, we have supplemented strategies to enhance the diversity of LM types in the expansive plains. Specifically, we state: "In the expansive plains, the study area necessitates a greater diversity of LM types to enhance the survival opportunities of local organisms, such as the establishment of agricultural shelterbelts and ecological green belts along irrigation channels." (Line 624-628)

 

 [Comments 38]

Appendix A, please add short names for these LM abbreviations and delineate the descriptions better so the reader can more easily associate the correct description with the abbreviation.

[Response 38]

According to your suggestion, we supplemented the LM name in Appendix A. (Line 685-686)

 

 [Comments 39]

Line 641, References, Number 25, should be Godron instead of Gordon.

 [Response 39]

We sincerely apologize for this oversight in our review process. Following your suggestion, we have made the necessary modifications to the authors' names in the references. (Line 762)

Reviewer 3 Report

Comments and Suggestions for Authors

The authors conducted a large volume of research on the assessment and analysis of biodiversity in the mountainous region of China (Taihang Mountains of Hebei Province) using the PCA-SOM-ED composite model.

The article should be accepted for publication after minor revisions. I have the following questions:

The introduction section should be shortened a little. In its current form it looks redundant and unstructured.

It is necessary to characterize the main types of plant communities. This is not the case now. And also to provide the ability to predict their successional changes.

To what extent do the results differ in hilly and flat areas?

Author Response

 [Comments 1]

The authors conducted a large volume of research on the assessment and analysis of biodiversity in the mountainous region of China (Taihang Mountains of Hebei Province) using the PCA-SOM-ED composite model.

 [Response 1]

Thank you for your recognition of this research. Our core innovation lies in the attempt to integrate PCA, SOM, and ED models to more efficiently address issues such as landscape mosaic classification, InVEST model evaluation, and assessments using OLS and GWR models.

 

The article should be accepted for publication after minor revisions. I have the following questions:

 [Comments 2]

The introduction section should be shortened a little. In its current form it looks redundant and unstructured.

 [Response 2]

Thank you for your suggestions. In this manuscript, we have removed redundant text from the introduction section. However, it should be noted that while addressing other reviewers' comments, we have added some theoretical content, which has resulted in no significant reduction in the overall length of the introduction. We hope for your acceptance of these changes; if not, we kindly request some time to continue optimizing the manuscript. (Line 41-159)

 

 [Comments 3]

It is necessary to characterize the main types of plant communities. This is not the case now. And also to provide the ability to predict their successional changes.

 [Response 3]

Thank you for your suggestions. In this revision of the manuscript, we have focused on strengthening Section 3.1.2 by supplementing the types of plants involved with various LMs. However, due to the foundational data of LULC used in this study, which can only identify macro-type plants, it is challenging to discern plant community structures through the current analysis. In response to this limitation, we have added the following to the discussion of our research shortcomings: "At the same time, we conducted LM classification using primary LULC types such as forests and grasslands. In contrast, a secondary LULC classification system can significantly enhance the expression of local plant mosaic characteristics in the study area during LM classification."Line 577-578)

 

 [Comments 4]

To what extent do the results differ in hilly and flat areas?

 [Response 4]

Our research found that within the Taihang Mountain region of Hebei Province, there are notable differences in the classification of landscape mosaics, habitat quality indices, and the relationships between hilly and flat areas. The core conclusion is that forest and grassland-related transitional landscape spaces (GA, FA, FAG), primarily distributed in hillside areas, have a profound impact on the biodiversity of the study area due to factors such as the dynamics of LM evolution and their correlation with habitat quality indices. This warrants increased attention from relevant government agencies or professionals. (Line 613-640)

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the thoughtful and constructive revisions made to the manuscript. The improvements to the methodological explanation—particularly the justification for the combined use of OLS and GWR—have strengthened the clarity and rigor of the study. Section 2 now provides a more coherent and accessible description of the analytical framework, and the logical flow of the research process is much improved.

The integration of PCA, SOM, ED, and the InVEST Habitat Quality model remains a novel and commendable approach. While the inclusion of a schematic diagram or flowchart would have further enhanced the reader’s understanding, the revised textual structure compensates adequately. The conclusions are well supported by the empirical results, and the engagement with relevant literature is appropriate and sufficient for the scope of the study.

The manuscript makes a valuable contribution to landscape ecology and spatial modeling in sustainability science. I believe it is ready for publication in its current form and commend the authors for their careful and responsive revisions.

Congratulations on this excellent work.

Reviewer 2 Report

Comments and Suggestions for Authors

Many thanks to the authors who carefully considered my comments and made excellent revisions to address my concerns.

The descriptions of threat factors and LM types in Tables 3 and 4 are a great addition and will help the readers understand all of the results and discussion that follow.

There may have been a misunderstanding on my part because I thought you were making the statement that most existing research lacks methodologies for assessing LM over multiple periods. I maintain there is existing research with these methodologies, so perhaps, we are in agreement.

[Comments 5]

Line 112, I disagree with your statement that most existing research lacks methodologies for coordinating and implementing LM classification across multiple periods.

[Response 5]

I respectfully disagree with your assertion that the majority of existing research lacks methodologies for coordinating and implementing LM classification across multiple periods. It is essential to recognize that various studies have proposed robust frameworks and methodologies that effectively address this issue. (Line 117-120)

Lines 360 and 361, you have the threshold for both TLM and PLM as > or = 75%. It seems like one should be greater than or equal and the other less than, or one should be greater than and the other less than or equal. This is your statement:

"Regarding the definitions of PLM and TLM, we refer to the research methodology of Vizzari et al. [37,72], which defines a LM as a PLM when the area proportion of a particular LULC type is ≥75%, and as a TLM when the combined area proportion of two or more LULC types is ≥75%."

Overall, you did a great job of responding to my comments.