Predicting Rural Ecological Space Boundaries in the Urban Fringe Area Based on Bayesian Network: A Case Study in Nanjing, China
Round 1
Reviewer 1 Report
Comments
This study uses Bayesian network to predict rural ecological space boundaries in the urban fringe area by training the model to learn the process of ecological space evolution. Authors proposed an innovative research perspective and effective research path reveals the evolution trends and mechanism of rural ecological space, which provides a valuable reference for rural ecological space protection in the future. As a whole, the research method and technical path of this study are achievable, but considering the case study is an administrative village, on a small scale, the index of rural ecological space prediction model selected seems to be a little sample and the classification of variables could hardly reflects landscape heterogeneity in the village, which leads the results predicted by BN model have some unreasonable parts. Therefore, a major revision is recommended for this manuscript and the details are as follows:
Introduction
1. This study is carried out on a small scale, within an administrative village. Nevertheless, ECR is used to be delineated in on a large scale, which leads to the neglect of microhabitat in rural ecological space as well as the inaccuracy of ecological space boundaries. Admittedly, the ecological space protected by ECR is the most important part in regional scale, but it is difficult to adapt to refined governance of rural space in new era that becomes the main driving force of this study. Consequently, it is considered that authors add relevant discussion in this section to reveal the problems brought by ECR facing multi-scale ecological space protection which is probably close connection with ecological space boundaries.
Materials and Methods
1. The manuscript is lack of the explanation of selecting Paifang Village as the case study in Chapter 2.1. Combining with Paifang Village facing the local ecological problems probably is the main reason why choose this village that should be explained systematically.
2. In Chapter 2.2, the study shows the index of rural ecological space prediction model (Table 1) and these indicators have a key impact on the predicted results of ecological space. However, these indicators seem to be a little simple so that they could hardly describe rural landscape characters in details. Therefore, it could be considered that authors enrich the index to some extent, such as some spatial form index.
3. In Chapter 2.3, the element variable values were divided in 2 or 3 classes and from Figure 6 in Chapter 2.4, discrete training sample data, cannot well to reflect landscape heterogeneity of Paifang village so that BN model probably cannot be better trained to learn the deep-seated characteristics of evolution process. Especially for the land use change factor, if it is unknown for the change between different land types (e.g., from construction to woodland), the model trained is adverse to identify objective trends of land use evolution.
4. It is confused that ecological sensitivity (2010, 2020) and importance of ecosystem service (2010,2020) of Paifang Village, shown in Figure 6 and Figure 8, reveal the value of two large reservoir at a medium or low level. There is no doubt that large water bodies have a strong water regulation capacity while they have higher ecological sensitivity than other small ecological space. Hence, the methods and results of these two measurements should be expounded in the research.
5. Ecological elements of different scale might bring different impacts. It is more reasonable to consider the weighted by elements area during measuring the space distance.
Results
1. From the predictive result it is astonishing that water has a large proportion of shrinking ecological space concentrating in large reservoirs. While this result show an assumed trends to some extent, it would be scarcely possible to happen in the future that means the input data or the correlative parameters need to be checked and reconsidered.
2. There is no ecological space in west of village that is contrary to realistic situation. Small shrubs and ponds also play a no negligible role in maintaining habitat diversity which should be considered during input data processing.
Discussion
1. In this section authors proposed some strategies for ecological space protection based on the results predicted in Chapter 3.1 and discussed the driving factors affecting ecological spatial change in Chapter 3.2. However, the explanation for change of ecological space is somewhat insufficient and insight less. It is necessary to strengthen to analysis the possible reason why the change of ecological space appears in the specific area by comparison with other area has similar or different landscape characteristics.
2. This study focus on rural ecological space boundaries and it is one of the core innovation point of this research. Therefore, it is necessary to add relevant discussion and analysis in this section to explore the change of ecological space boundaries as well as its influence brought in depth.
Conclusions
1. In this section authors gave one of conclusions is that the protection effect of ECR on the rural ecological space is remarkable but some parts of water bodies protected by ECR are occupied by non-ecological space. Obviously, this is incongruous to research results and it should be reconsidered.
Comments for author File: 
Comments.pdf
Author Response
Response to Reviewer 1 Comments
Dear Editors and Reviewers,
Thank you for all the reviewers’ comments concerning our manuscript entitled “Predicting rural ecological space boundaries in the urban fringe area based on Bayesian network: a case study in Nanjing, China” (ID: land-1962772). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made correction which we hope meet with approval. The responds to the reviewers’ comments are as following:
Point 1: This study is carried out on a small scale, within an administrative village. Nevertheless, ECR is used to be delineated in on a large scale, which leads to the neglect of microhabitat in rural ecological space as well as the inaccuracy of ecological space boundaries. Admittedly, the ecological space protected by ECR is the most important part in regional scale, but it is difficult to adapt to refined governance of rural space in new era that becomes the main driving force of this study. Consequently, it is considered that authors add relevant discussion in this section to reveal the problems brought by ECR facing multi-scale ecological space protection which is probably close connection with ecological space boundaries.
Response 1: Thank you for the suggestion, it is helpful to clarify the contribution of our study. The relevant content you mentioned has been supplemented and modified in lines 92-97. To reflect the importance of the study of ecological space boundary, we emphasize the insufficiency of ECR for multi-scale ecological space and the limitation of ECR only for protection of core areas.
Point 2: The manuscript is lack of the explanation of selecting Paifang Village as the case study in Chapter 2.1. Combining with Paifang Village facing the local ecological problems probably is the main reason why choose this village that should be explained systematically.
Response 2: Thanks for your suggestion. The reason for choosing Paifang Village as the case study were re-explained in line177-180. Paifang Village is a typical suburban village. Its local ecological problems are mainly due to the frequent flow of urban and rural elements, and the rapid development of rural tourism based on good landscape scenery, which threatens the ecological space.
Point 3: In Chapter 2.2, the study shows the index of rural ecological space prediction model (Table 1) and these indicators have a key impact on the predicted results of ecological space. However, these indicators seem to be a little simple so that they could hardly describe rural landscape characters in details. Therefore, it could be considered that authors enrich the index to some extent, such as some spatial form index.
Response 3: Your comments have helped us a lot, thank you. The table 1 were not showing the indexes of Ecological sensitivity and Importance of ecosystem service value. In this round of modification, we supplemented them in Chapter 2.3 to show the construction of the indicators in more detail. We believe that the Bayesian network, as a simple machine learning method, has a better ability to deal with factors of ecological space, while the spatial pattern index cannot establish a direct causal relationship with other factors in the construction of the Bayesian network structure.
Point 4: In Chapter 2.3, the element variable values were divided in 2 or 3 classes and from Figure 6 in Chapter 2.4, discrete training sample data, cannot well to reflect landscape heterogeneity of Paifang village so that BN model probably cannot be better trained to learn the deep-seated characteristics of evolution process. Especially for the land use change factor, if it is unknown for the change between different land types (e.g., from construction to woodland), the model trained is adverse to identify objective trends of land use evolution.
Response 4: Thank you for the suggestion. We have ambiguity in the naming of factors. So, it was not able to accurately express the type and meaning of factors. The name of factors had revised to Land use expansion to reflect the possible expansion trend of a certain type of land use. The details refer to line 250-251, Table 1. The classes of the element variable values depend on the scale of the research object. They are usually divided into 3-5 classes in large-scale research. According to the characteristics of different variables, the number of classes are different. In this study, we had compared the different numbers of classes, and found small difference between them. The reason for this result may be that the scale of the Paifang Village is small. In the method of Bayesian network Model, the relationship between the influencing factors was showed by the form of conditional probability. Instead of conversion between land use types, the predictions are made by the change of the factors based on conditional probability, which is different from Markov Model.
Point 5: It is confused that ecological sensitivity (2010, 2020) and importance of ecosystem service (2010, 2020) of Paifang Village, shown in Figure 6 and Figure 8, reveal the value of two large reservoir at a medium or low level. There is no doubt that large water bodies have a strong water regulation capacity while they have higher ecological sensitivity than other small ecological space. Hence, the methods and results of these two measurements should be expounded in the research.
Response 5: Your comment is valuable for improving our study, thank you so much. The ecological sensitivity and the importance of ecosystem service value are re-evaluated. We found the data of the importance of ecosystem service value was a certain deviation. We revised and updated Figure 6, Figure 8, and the content in Chapter 2.3. Based on the results of the revision, the value of two large reservoir at a medium level of ecological quality. We believe there are two reasons. One is two reservoirs are both artificially constructed, and surrounding human activities are frequent. Another is the area of two reservoirs is much smaller than the ecological spaces of woodland.
Point 5: Ecological elements of different scale might bring different impacts. It is more reasonable to consider the weighted by elements area during measuring the space distance.
Response 5: Thanks for your suggestion. In the Bayesian network Model, the space distance could reflect the effect of distance on the changes in ecological space, and the Ecological elements of different scale had been considered in ecological sensitivity index. The CPT obtained by BN parameter learning reflects the degree of influence of ecological elements on ecological space changes, which is equivalent to the weight relationship.
Point 6: From the predictive result it is astonishing that water has a large proportion of shrinking ecological space concentrating in large reservoirs. While this result show an assumed trends to some extent, it would be scarcely possible to happen in the future that means the input data or the correlative parameters need to be checked and reconsidered.
Response 6: The data of predictive result has been recalculated according to the revised parameters. The new result is shown on Figure 10,11,14. The ecological spaces of the water area are still shrinking seriously. We had analyzed the reason. On the one hand, combined the results of ecological sensitivity and the importance of ecosystem service value with site investigation, human activities are frequent around the reservoirs, there is indeed a risk of space shrinkage with high ecological value. On the other hand, the predicted results do not reflect the physical ecological space, but rather the decline of their own ecological functions, or the risk of ecological destruction.
Point 7: There is no ecological space in west of village that is contrary to realistic situation. Small shrubs and ponds also play a no negligible role in maintaining habitat diversity which should be considered during input data processing.
Response 7: We have fully considered your suggestion. We have conducted investigation on the Paifang Village. The main land use types on the west side of the village are agricultural land and construction land. There are almost no large ecological spaces such as forest land or shrubs, only has little grassland. The ecological space is characterized by fragmentation. According to the prediction results, it can be seen that there are several small ecological spaces on the west side. Through comparative analysis, these small ecological spaces have evolved from shrubs and ponds. As you metioned, shrubs and ponds play a non-negligible role in the expansion of ecological space.
Point 8: In this section authors proposed some strategies for ecological space protection based on the results predicted in Chapter 3.1 and discussed the driving factors affecting ecological spatial change in Chapter 3.2. However, the explanation for change of ecological space is somewhat insufficient and insight less. It is necessary to strengthen to analysis the possible reason why the change of ecological space appears in the specific area by comparison with other area has similar or different landscape characteristics.
Response 8: Thank for your suggestion. We have analyzed the possible reason for the change of ecological space appears and added the discussion on Chapter 4.1. In this part, the driving forces of ecological spatial changes of Paifang Village has been compared with Guli Village, which is also located in Jiangning District, and Longtan Village in Qixia District of Nanjing. Additionally,the analysis of the causes of ecological space changes was strengthened.
Point 9: This study focus on rural ecological space boundaries and it is one of the core innovation point of this research. Therefore, it is necessary to add relevant discussion and analysis in this section to explore the change of ecological space boundaries as well as its influence brought in depth.
Response 9: Thanks to your suggestion. we have highlighted changes in the ecological space boundaries and their implications in Chapter 4.3 and supplemented the discussion.
Point 10：In this section authors gave one of conclusions is that the protection effect of ECR on the rural ecological space is remarkable but some parts of water bodies protected by ECR are occupied by non-ecological space. Obviously, this is incongruous to research results and it should be reconsidered.
Response 10: Your suggestion is very constructive. Based on the investigation results, we have compared the satellite image with the ECR, and found the ECR did not completely cover the entire area of the Paifang Reservoir, while the Yanhu Reservoir was completely outside the ECR. Furtherly, we have analyzed the predicted ecological space by superimposing the ECR, and found the area of the ecological space shrinkage of Paifang Reservoir is not covered by ECR. The Yanhu Reservoir is not within the protection range of ECR, and the area of ecological space shrinkage is larger. The results of analysis indicated that ECR has a good protective effect on ecological space. We had optimized Figure 14, and the above conclusions can be clearly seen.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper is very well structured. The theoretical framework is broad and excellently supports both the methodological and empirical parts. Perhaps, from this point of view, in view of their scientific importance within the work, one could go more into the ecosystem services that have been considered and offer more specifics in this regard.
The reading of the text is fluent and the English is functional to the subject matter. A minor point that I can highlight is, at times, the repetition of certain linguistic formulas (e.g.: On the one hand...) that could be limited to make the reading more enjoyable.
Author Response
Response to Reviewer 2 Comments
Dear Editors and Reviewers,
Thank you for all the reviewers’ comments concerning our manuscript entitled “Predicting rural ecological space boundaries in the urban fringe area based on Bayesian network: a case study in Nanjing, China” (ID: land-1962772). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made correction which we hope meet with approval. The responds to the reviewers’ comments are as following:
Point 1: The paper is very well structured. The theoretical framework is broad and excellently supports both the methodological and empirical parts. Perhaps, from this point of view, in view of their scientific importance within the work, one could go more into the ecosystem services that have been considered and offer more specifics in this regard.
Response 1: Thanks for your agreement to our paper.
Point 2: The reading of the text is fluent and the English is functional to the subject matter. A minor point that I can highlight is, at times, the repetition of certain linguistic formulas (e.g.: On the one hand...) that could be limited to make the reading more enjoyable.
Response 2: Your comments are of great help to us. We have further optimized the textual expression of the paper.
Author Response File: Author Response.docx
Reviewer 3 Report
Predicting rural ecological space boundaries in the urban fringe area based on Bayesian network: a case study in Nanjing, China
General
The study selected Paifang village in China as the study area and built a Bayesian network model to predict the ecological space boundary for 2030. The study then identified the driving factors and their mechanisms affecting the changes in the rural ecological space in an urban fringe area and put forward targeted suggestions for its protection. The indicators used were appropriately justified.
Specific comments
In lines 33-35 the authors described a village as an ecological unit with basic functions of material circulation and energy flow. Such a definition also applies to urban villages. However, as the authors state the villages to face severe risks of ecological environment damage and ecological function degradation from with activities such as construction, expansion, and development of tourism. So there is need to define very well what villages look like in China.
In lines 66/67 the authors stated that orchards and tea fields have dual functions of ecology and production but failed to outline the functions. The reader was left hanging.
Greater part of the first paragraph of Section 3.2: Sensitivity and Diagnostic Analyses is methods instead of results. In this section results should be presented without repeating the methods used.
No literature citation in the first two sections of the discussion.
Comments for author File: Comments.pdf
Author Response
Response to Reviewer 3 Comments
Dear Editors and Reviewers,
Thank you for all the reviewers’ comments concerning our manuscript entitled “Predicting rural ecological space boundaries in the urban fringe area based on Bayesian network: a case study in Nanjing, China” (ID: land-1962772). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made correction which we hope meet with approval. The responds to the reviewers’ comments are as following:
Point 1: The study selected Paifang village in China as the study area and built a Bayesian network model to predict the ecological space boundary for 2030. The study then identified the driving factors and their mechanisms affecting the changes in the rural ecological space in an urban fringe area and put forward targeted suggestions for its protection. The indicators used were appropriately justified.
Response 1: Thank you for your review of the paper, which is meaningful for our further research.
Point 2: In lines 33-35 the authors described a village as an ecological unit with basic functions of material circulation and energy flow. Such a definition also applies to urban villages. However, as the authors state the villages to face severe risks of ecological environment damage and ecological function degradation from with activities such as construction, expansion, and development of tourism. So there is need to define very well what villages look like in China.
Response 2: Thank you for pointing out the problems in this article. We have optimized the description of the definition of China's countryside in the article, pointing out that in the context of rapid urbanization in China, the countryside is not only an integral part of the regional ecological network, but also needs to provide ecological function services to the city. At the same time, with the development of rural economy, it faces the risk of shrinking ecological space. For details, please refer to lines 39-40.
Point 3: In lines 66/67 the authors stated that orchards and tea fields have dual functions of ecology and production but failed to outline the functions. The reader was left hanging.
Response 3: Thank you for the suggestion. we have made the supplements in lines 70-72.
Point 4: Greater part of the first paragraph of Section 3.2: Sensitivity and Diagnostic Analyses is methods instead of results. In this section results should be presented without repeating the methods used.
Response 4: Thanks for your valuable advice. The method part of Diagnostic analysis and Sensitivity analysis was moved to Chapter 2.6 in Methods.
Point 5: No literature citation in the first two sections of the discussion.
Response 5: Thank you for the suggestion. We have added literature citations in the first two sections of the discussion.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
This revised manuscript paid further attention in data processing and added particular calculations of the evaluations involved, which makes the results reasonable and persuasive. It could be seen that authors strengthened discussion part especially in analysis of ecological space boundaries change and gave significant conservation strategies with logical and relevant demonstrations.
