Reinforcement Learning for Stand Structure Optimization of Pinus yunnanensis Secondary Forests in Southwest China

Round 1

Reviewer 1 Report (Previous Reviewer 1)

I am satisfied by the included revisions in the resubmitted version of the manuscript. 

Congratulations

Author Response

Dear Reviewer,

Thank you very much for your encouraging feedback on the revised version of our manuscript. We're thrilled to hear that you're satisfied with the included revisions. Your valuable insights and suggestions have been instrumental in enhancing the quality of our work.

We greatly appreciate your time and effort in reviewing our manuscript. Your positive evaluation and congratulations mean a lot to us. We are committed to continually improving our research and look forward to any further suggestions or feedback you might have in the future.

Thank you once again for your support and encouragement.

Best regards,

Reviewer 2 Report (Previous Reviewer 3)

Authors improved the manuscript sufficiently.

English is good. A final revision by authors should fix minor problems.

Author Response

Dear Reviewer,

Thank you for your valuable feedback and positive assessment of the improvements made to our manuscript. We greatly appreciate your acknowledgment of the enhanced quality and language proficiency.

We acknowledge your suggestion regarding a final revision to address any remaining minor issues in the manuscript. We will diligently conduct a thorough review to ensure that all minor concerns are appropriately addressed and resolved. Your guidance has been immensely helpful in refining our work, and we are committed to delivering a final version that meets the highest standards.

Once again, thank you for your time, valuable feedback, and continued support throughout this process.

Best regards,

Reviewer 3 Report (Previous Reviewer 4)

The article has been generally corrected according to my previous comments. I still think it's too long. The Conclusion chapter requires improvement. It should use complete affirmative sentences. Additionally, abbreviations should not be applied.

Author Response

Dear Reviewer,

Thank you for your comprehensive assessment and valuable suggestions regarding our manuscript.

Comment1: The article has been generally corrected according to my previous comments. I still think it's too long.

Response 1: To address the issue of excessive length, we've streamlined the content by removing redundant explanations and extraneous details.

Comment2: The Conclusion chapter requires improvement. It should use complete affirmative sentences. Additionally, abbreviations should not be applied.

Response 2: In revising the Conclusion section, we've restructured it to incorporate complete affirmative sentences, ensuring clearer and more concise content. We've also refrained from using abbreviations and provided explanations for initial abbreviations used in this section. This approach aims to enhance reader comprehension and improve overall readability.

We genuinely appreciate your time, invaluable insights, and continuous support throughout this review process.

Warm regards,

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The paper presents a reinforcement learning model for forst stand strcture optimization, by considring Pinus yunnanensis secondary forests in southwest China as case study. Overall, the paper is well presented and well wrote, the results were presented clearly and the intrepration supported by clear results. The findings of the study are interesting and offering a new path of research in the context of stand structure optimization.

I have some remarks the well be helpful for you:

1- Is that the RL first time apply in the cotext of forest optimation? Some references well be helpful to you ; 

Malo, P., Tahvonen, O., Suominen, A., Back, P., & Viitasaari, L. (2021). Reinforcement learning in optimizing forest management. Canadian Journal of Forest Research, 51(10), 1393-1409.

Tahvonen, O., Suominen, A., Malo, P., Viitasaari, L., & Parkatti, V. P. (2022). Optimizing high-dimensional stochastic forestry via reinforcement learning. Journal of Economic Dynamics and Control, 145, 104553.

 2- Since the developed model is based on reinforcement learning, why the authors not discussed the other developed machine learning and deep learning used for forest stand structure optimization? Would be better to develop a clear background ether in introduction and discussion sections for proving the performance of the developed model. 

3- Time is an essential parameter in modeling, ignoring this paramater decreasing the quality of the work. So, would be better to calculate the consumption time for each model and discussed well the obtained results. 

4- Evaluation of the model using only VOF not clear, since it's used for linear programming? why not other paramters ? 

Best of luck

Author Response

Response to Reviewer 1 Comments

 Dear Reviewer,

Thank you for your meticulous review and valuable insights on our manuscript. We greatly appreciate your feedback and have diligently revised the paper based on your suggestions. We firmly believe that these enhancements have strengthened the paper, making it more comprehensive and persuasive. We eagerly await your further guidance and review.

Comment1- Is that the RL first time apply in the cotext of forest optimation? Some references well be helpful to you;

Response1: Paper one explores the application of reinforcement learning in solving high-dimensional stochastic optimization problems, considering various factors in forest management to propose effective decision strategies. It emphasizes the impact of natural disasters on forest management and demonstrates reinforcement learning's ability to handle harvesting choices from different initial states, enhancing forest species diversity.

Paper two extends research on forest rotation models, presenting an optimal harvesting model for mixed-age and size-structured tree species. Findings indicate that one- or two-dimensional models may compromise economic output, while common models might overestimate profitability, introduce timing biases in harvesting, and mitigate stochasticity effects. Reinforcement learning is acknowledged as an effective approach to address these complex models in resource economics.

Through an analysis of these papers and related studies, we note the existing use of reinforcement learning in forest management. However, research gaps persist in the context of multi-objective optimization of forest stand structures. Therefore, our application of reinforcement learning in this domain, marking its inaugural use in forest stand structure optimization studies, has been appropriately articulated and modified in the manuscript.

Additionally, we have included a comparative discussion between these references and our study within the Introduction sections. This serves to highlight the innovative aspects of our research. Specifically, we have juxtaposed our findings with those outlined in the referenced literature, emphasizing the uniqueness and novelty of our approach.

Comment2- Since the developed model is based on reinforcement learning, why the authors not discussed the other developed machine learning and deep learning used for forest stand structure optimization? Would be better to develop a clear background ether in introduction and discussion sections for proving the performance of the developed model.

Response2: We acknowledge the significance of other machine learning and deep learning approaches in optimizing forest structures. Correspondingly, we have made revisions in the introduction and discussion sections to thoroughly explore the applications of existing machine learning and deep learning methods in forest management. These additions aim to highlight the unique advantages of our chosen RL-based model in addressing the dynamic multi-objective optimization challenges within forest structural management.

Comment3- Since the developed model is based on reinforcement learning, why the authors not discussed the other developed machine learning and deep learning used for forest stand structure optimization? Would be better to develop a clear background ether in introduction and discussion sections for proving the performance of the developed model.

Response3: We recorded the number of iterations during model runs, which partially reflects the time required for the model to converge. With an increase in the number of iterations, the time needed for the model to converge varied. While these records don't offer a complete time measurement, they provide some indications of the model's runtime. We will emphasize this in the results.

Comment4- Evaluation of the model using only VOF not clear, since it's used for linear programming? why not other paramters?

Response4: Thank you for your valuable comment regarding the evaluation of our model using VOF. We acknowledge the importance of using a comprehensive metric for model evaluation. The choice of VOF as our evaluation metric is based on several reasons that align with the objectives of our study and the nature of multi-objective forest optimization.

Firstly, VOF, as an assessment metric, has demonstrated its suitability in capturing diverse aspects of forest spatial structures, aligning perfectly with our research's primary focus on optimizing multi-objective forest stand structures. It has been widely endorsed and utilized in previous studies within the field, reflecting its reliability and comparability.

The decision to solely employ VOF for evaluation was deliberate and tied closely to its comprehensive nature in evaluating stand structures, aligning with the optimization model employed in our study. Other parameters might lack the holistic assessment ability offered by VOF in capturing the intricate dynamics of forest stand structures.

While acknowledging the importance of considering multiple evaluation parameters, our study specifically focused on VOF due to its extensive utilization in similar studies and its ability to comprehensively encapsulate various aspects of forest stand structures. This choice aimed to streamline the evaluation process, aligning closely with the optimization objectives.

We have emphasized in the revised manuscript the reasons behind choosing VOF and its relevance to our study objectives. This discussion underscores the appropriateness of VOF in our study context, addressing the multi-dimensional aspects of forest stand structures efficiently.

We extend our heartfelt appreciation to you for your thorough review and invaluable recommendations, which have played a pivotal role in refining our manuscript. The revised version, incorporating your insightful suggestions, has been resubmitted for your kind evaluation. We eagerly anticipate any further feedback you may have and remain sincerely grateful for your time and dedication to enhancing the quality of our work.

Warm regards, 

All Authors

Reviewer 2 Report

This manuscript uses some operations research methods to optimize the structure of the stands belonging to a secondary forest in China. The paper's aims and objectives fit into this journal's editorial policy. After reading the manuscript, unfortunately, I cannot recommend publication in its current form. My comments are the following:

General Comments: 

A reader cannot understand the manuscript since, in the first Sections, some concepts, sentences, and methods appear non-previously explained. For example, the Abstract must be rewritten since a reader does not know the schemes, and he/she does not understand the results included here—the same for parts of the Introduction Section (e.g., l. 126- l.143). Besides, a potential reader cannot replicate the methodologies followed here because the authors have not included some explanations about their advantages to choosing them, and/or the descriptions are very scarce (e.g., VQM, “multiplication and division strategy,” etc.).

Major comments:

• The principal significance of this research outside China should be stressed. Otherwise, I would advise submitting it to a Chinese Journal. Besides, some key references are grey literature [e.g., 10-13], and they are not accessible to a reader who is not fluent in Chinese.

• The authors should revise the way they use bibliographic references. Some sentences need them (or they can be interpreted as a value judgement). In other cases, some references included are not pertinent. For example, not all the references included in l. 78-84 are related to the optimization of FSS. Another example: in the references included in l. 398, some of them do not use Monte Carlo methods. The same for l. 417. In short, these practices should not be included in any scientific publication.

• The authors should clearly explain why some algorithms are used, and not others. The same is true for this multi-criteria technique (MOP).

• QVM: This technique has not been appropriately defined. A reader does not know why Eq 13 is used. Why this normalization system? Why have all the indices the same weight?

• Some justification about the sentence included in l. 333-336 is needed. 

• Discussion Section: the first sentence is a value judgement. 

Minor comments:

• Please define “ecological engineering endeavors”and “mingling”.

• Please move the Algorithm 1 to an Annex or a Supplementary Material.

• Please put Table 5 in the Discussion Section. 

Author Response

Response to Reviewer 2 Comments

 Dear Reviewer,

We extend our sincere gratitude to the reviewer for their valuable insights and constructive feedback on our manuscript. We highly appreciate the detailed evaluation, which helps us improve the quality and comprehensibility of our work. We aim to address each concern comprehensively, as outlined below:

General Comments:

Comment1 - A reader cannot understand the manuscript since, in the first Sections, some concepts, sentences, and methods appear non-previously explained. For example, the Abstract must be rewritten since a reader does not know the schemes, and he/she does not understand the results included here—the same for parts of the Introduction Section (e.g., l. 126- l.143). Besides, a potential reader cannot replicate the methodologies followed here because the authors have not included some explanations about their advantages to choosing them, and/or the descriptions are very scarce (e.g., QVM, “multiplication and division strategy,” etc.).

Response1: We acknowledge the valid point raised regarding the need for enhanced clarity throughout the manuscript. In response, we plan to extensively revise the Abstract and Introduction sections, particularly concerning the explanation of terms such as 'QVM' and 'multiplication and division strategy.' Our objective is to present the schemes, methods, and results in a more accessible manner, ensuring that readers can grasp the concepts seamlessly.

Major comments:

Comment2 • The principal significance of this research outside China should be stressed. Otherwise, I would advise submitting it to a Chinese Journal. Besides, some key references are grey literature [e.g., 10-13], and they are not accessible to a reader who is not fluent in Chinese.

Response2: We will augment the manuscript by underlining the broader significance and implications of our findings beyond the geographical scope in the Introduction and Discussion sections. Furthermore, we will revise the references, prioritizing those accessible to an international readership while ensuring their direct relevance to our study.

Comment3 • The authors should revise the way they use bibliographic references. Some sentences need them (or they can be interpreted as a value judgement). In other cases, some references included are not pertinent. For example, not all the references included in l. 78-84 are related to the optimization of FSS. Another example: in the references included in l. 398, some of them do not use Monte Carlo methods. The same for l. 417. In short, these practices should not be included in any scientific publication.

Response3: To improve the paper's quality, we will meticulously review the usage of bibliographic references throughout the manuscript. Ensuring relevance and accuracy in supporting our claims is our priority. Additionally, we recognize the need to provide clear justifications for the chosen algorithms, enabling readers to understand the rationale behind our selections.

Comment4 • The authors should clearly explain why some algorithms are used, and not others. The same is true for this multi-criteria technique (MOP).

Response4: In selecting the algorithms for this study, we carefully considered their applicability, robustness, and performance in addressing the specific challenges of forest stand structure optimization. The chosen algorithms, such as Monte Carlo and particle swarm optimization (PSO), were selected due to their historical relevance in forestry studies, ease of implementation, and their representation of different optimization paradigms. Regarding the multi-criteria technique (MOP), our focus was on algorithms that specifically catered to stand-level or tree-level optimization scenarios, where the objective was to optimize multiple interrelated factors simultaneously. However, we acknowledge the reviewer's point and will provide a more detailed explanation in the revised manuscript to offer a clearer rationale for the selection of these algorithms over others.

Comment5 • QVM: This technique has not been appropriately defined. A reader does not know why Eq 13 is used. Why this normalization system? Why have all the indices the same weight?

Response5: Your feedback on the QVM technique is invaluable. We will provide a comprehensive definition, along with a detailed explanation of the normalization system and the rationale behind uniform weight allocation to indices. Furthermore, sentences lacking adequate explanation, particularly in the Materials and Methods Section, will be revised and expanded for greater clarity and context.

Comment6 • Some justification about the sentence included in l. 333-336 is needed.

Response6: Your constructive feedback on the VMM is greatly appreciated, and we will supplement this section in the revised manuscript.

Comment7 • Discussion Section: the first sentence is a value judgement.

Response7: In the revised manuscript, we have rephrased the first sentence of this discussion section to make it more objective.

Minor comments:

Comment8 • Please define “ecological engineering endeavors”and “mingling”.

Response8: We have rephrased the sentence mentioning 'ecological engineering endeavors' for easier comprehension by the readers. We provided further elaboration on 'mingling' and referenced relevant literature when first introducing this term.

Comment9 • Please move the Algorithm 1 to an Annex or a Supplementary Material.

Response9: Regarding Algorithm 1, we intend to relocate it to either an Annex or Supplementary Material, as suggested.

Comment10 • Please put Table 5 in the Discussion Section.

Response10: Table 5 will be placed in the Discussion Section, as per the reviewer's recommendation.

We extend our heartfelt appreciation to you for your thorough review and invaluable recommendations, which have played a pivotal role in refining our manuscript. The revised version, incorporating your insightful suggestions, has been resubmitted for your kind evaluation. We eagerly anticipate any further feedback you may have and remain sincerely grateful for your time and dedication to enhancing the quality of our work.

Warm regards, 

All Authors

Reviewer 3 Report

The topic is very specific and interesting for foresters (like me!) but very likely quite tedious for other audience. Although it is difficult to write in plain language in many occasions I think there are some concepts and other types of information that could be explained helping non-experts to understand the study.

I understand that the topic demands long methodological descriptions but the manuscript is really long. Please try to reduce it, move non-essential information to supplementary material. It will only improve readers' interest.

Although English is mostly well written there are some exaggerations making the text almost poetic but sometimes incorrect.

I strongly suggest that authors connect their study to a more general forestry context as it is too study site centered. Can the results be applied to other forest types, ecosystems, continents? Any changes are required for that? Etc.

L12-17 – The citation of specific numerical results is tedious. I suggest the citation  the main trends via % of increase/improvement to give a clearer overall understanding of the differences between methods. The specific values should be cited in the main table.

 L42-43 – explain what the meaning of a skewed Forest Stand Structure is. Remember that the text should be understandable to all.

L47 – what does “unsound structures” stand for?

L63 – Citation for MOFSS.

L94 – Cite in full the meaning of RL.

L120-167– Authors got a bit lost here. There are results, discussions and conclusions. Only the pertinent text to introduction should be kept here (e.g. L128-130 are results etc).

L170 – What are those forest inventories? Why do they matter? Remember that the general audience do not necessarily know what is a FI and what is that for?

L180 – Prominent is not the correct word here. Are those species the most abundant? Higher density, basal area, tallest?

Figure 1 caption – “Unveiling” Study Sites? It sounds poetic but it is inappropriate. Please, just describe fully the figure’s content.

L197 – Write in full the meaning of DBH.

L197 – Explain how tree height (TH) and especially crown width (CW) were measured. CW is not a simple parameter and to have a meaningful assessment it might require several measurement per tree.

L198 - relative coordinates in relation to what?

Table 1 – Essential details in relation to what? Location, terrain and structure?

Table 1 – tree species heading also refers to the number of individuals from some species. Check please.

Figure 2 – Its caption has a title which is fine. However, it should also contain information related to its content. What are those cells, those dots for instance.

L451 and following coding – please move it to the appendix.

Although English is mostly well written there are some exaggerations making the text almost poetic but sometimes incorrect.

Author Response

Response to Reviewer 3 Comments

Thank you for your thorough review and insightful comments on our article. We appreciate your positive assessment of the methodological approach and your constructive suggestions for improvement. We've carefully noted each of your points and will comprehensively address them in our revised manuscript.

Comment1: The topic is very specific and interesting for foresters (like me!) but very likely quite tedious for other audience. Although it is difficult to write in plain language in many occasions I think there are some concepts and other types of information that could be explained helping non-experts to understand the study.

Response1: Thank you for your feedback. We'll conduct a thorough review of the manuscript and strive to write in a more accessible language. For complex terminology and concepts, we'll provide explanations and clarifications to ensure that non-expert readers can easily comprehend our research.

Comment2: I understand that the topic demands long methodological descriptions but the manuscript is really long. Please try to reduce it, move non-essential information to supplementary material. It will only improve readers' interest.

Response2: We appreciate your feedback regarding the length of the manuscript. We will carefully assess the content and aim to streamline the text by relocating non-essential details to the supplementary materials. This adjustment will enhance the overall readability and maintain readers' interest.

Comment3: Although English is mostly well written there are some exaggerations making the text almost poetic but sometimes incorrect.

Response3: Thank you for highlighting this concern. We'll thoroughly review the text to ensure accuracy and clarity, avoiding any unintentional poetic or exaggerated language that might impact the scientific accuracy of our manuscript. Your input is valuable, and we'll strive to maintain a clear and precise tone throughout the document.

Comment4: I strongly suggest that authors connect their study to a more general forestry context as it is too study site centered. Can the results be applied to other forest types, ecosystems, continents? Any changes are required for that? Etc.

Response4: Your point is well taken. We'll work on broadening the discussion to emphasize the potential applicability of our results beyond the specific study sites. We'll explore how the findings might translate to different forest types, ecosystems, or even continents, and we'll highlight any necessary adaptations or considerations for such broader applications. This extension should help readers better understand the wider relevance and potential implications of our study in various forestry contexts.

Comment5: L12-17 – The citation of specific numerical results is tedious. I suggest the citation the main trends via % of increase/improvement to give a clearer overall understanding of the differences between methods. The specific values should be cited in the main table.

Response5: We'll revise the text to focus more on conveying the main trends and differences between methods using percentage increases or improvements. The specific numerical values will be included in the main table for reference, ensuring a clearer understanding of the differences while avoiding an overly detailed citation of specific results in the text.

Comment6: L42-43 – explain what the meaning of a skewed Forest Stand Structure is. Remember that the text should be understandable to all.

Response6: Absolutely, the term "skewed Forest Stand Structure" refers to a situation where the distribution of trees' sizes or ages within a forest stand is significantly imbalanced or disproportionate. It means that the forest stand doesn't have a uniform or balanced distribution of trees across different size or age classes. This imbalance might indicate an uneven growth pattern or irregularities in the forest's development. We have replaced the term 'skewed Forest Stand Structure' with 'uneven forest stand structure' in the relevant sections of the manuscript. We believe this modification will facilitate comprehension for non-specialist readers, enabling them to grasp the intended meaning. Additionally, we have reviewed the presentation of technical terminology throughout the document to ensure effective understanding by non-specialist readers.

Comment7: L47 – what does “unsound structures” stand for?

Response7: We intended to convey the idea of an unbalanced or unhealthy forest stand structure. To ensure better clarity, we've replaced 'unsound structures' with 'unhealthy structures' in the manuscript. We believe this adjustment will help readers grasp the intended meaning more effectively.

Comment8: L63 – Citation for MOFSS.

Response8: MOFSS stands for Multi-Objective Optimization of Forest Stand Structure, which refers to the optimization of various objectives related to forest stand structure. In our manuscript, we explore the application of different optimization techniques to address the complexities of optimizing forest stand structures considering multiple objectives simultaneously. Moreover, we believe that judiciously using abbreviations is beneficial for reducing manuscript length and enhancing the reading experience. We have adequately explained all abbreviations in the manuscript to ensure a smooth and accurate reading for our audience.

Comment9: L94 – Cite in full the meaning of RL.

Response9: RL stands for Reinforcement Learning. We have provided comprehensive explanations for all abbreviations in the manuscript to ensure a smooth and accurate reading experience for the audience.

Comment10: L120-167– Authors got a bit lost here. There are results, discussions and conclusions. Only the pertinent text to introduction should be kept here (e.g. L128-130 are results etc).

Response10: We greatly appreciate your insight. In the revised manuscript, we have rectified this issue by reorganizing the sections related to results, discussions, and conclusions, placing them appropriately within the text. Additionally, we have revised the introduction to ensure its accuracy and coherence.

Comment11: L170 – What are those forest inventories? Why do they matter? Remember that the general audience do not necessarily know what is a FI and what is that for?

Response11: Forest inventories (FIs) encompass comprehensive assessments of forests, documenting various attributes like tree species, density, biomass, and structural characteristics. They serve as fundamental databases aiding in forest management, resource planning, and ecological studies. In the revised manuscript, considering a wider audience that may not be familiar with these terms, we've replaced 'forest inventories' with 'forest data collection' to describe the source of our data.

Comment12: L180 – Prominent is not the correct word here. Are those species the most abundant? Higher density, basal area, tallest?

Response12: We initially intended to convey the presence of various tree species in our study area. To avoid potential ambiguity for readers, we have replaced 'Prominent tree species' with 'Main tree species in the study areas'.

Comment13: Figure 1 caption – “Unveiling” Study Sites? It sounds poetic but it is inappropriate. Please, just describe fully the figure’s content.

Response13: Here's an alternative caption for Figure 1: "Description of Study Sites".

Comment14: L197 – Write in full the meaning of DBH.

Response14: DBH stands for Diameter at Breast Height, representing the diameter of a tree measured at 1.3 meters above the ground. A detailed explanation has been incorporated into the revised manuscript at the relevant section.

Comment15: L197 – Explain how tree height (TH) and especially crown width (CW) were measured. CW is not a simple parameter and to have a meaningful assessment it might require several measurement per tree.

Response15: We utilized specialized altimeters and distance measurer for measurements. Tree Height (TH) was measured as the vertical distance from the tree base to its highest point, while Crown Width (CW) was determined as the average of the longest horizontal distances of the crown in the four cardinal directions. Detailed explanations have been incorporated into the revised manuscript.

Comment16: L198 - relative coordinates in relation to what?

Response16: The coordinate data we measured and recorded are relative coordinates to the center of the sample plots, and the corresponding explanatory notes have been added to the revised manuscript.

Comment17: Table 1 – Essential details in relation to what? Location, terrain and structure?

Response17: The revised title for Table 1 would be: "Key Information about Sample Plot Attributes" to ensure a clear understanding of the table content by the readers.

Comment18: Table 1 – tree species heading also refers to the number of individuals from some species. Check please.

Response18: We utilized the tree composition percentage method to depict the tree species composition within each plot. The numerical values preceding each tree species indicate the proportion of that species for every 10 trees within the plot. We will provide an explanation for this in the table footnote.

Comment19: Figure 2 – Its caption has a title which is fine. However, it should also contain information related to its content. What are those cells, those dots for instance.

Response19: The figure illustrates the spatial distribution of tree species within the study area. The cells represent the effective range around each tree within the study area, while the dots represent the location of trees. We will provide an explanation for this in the figure footnote.

Comment20: L451 and following coding – please move it to the appendix.

Response20: We have relocated this section to the appendix of the manuscript to maintain a clear and focused narrative in the main text.

We extend our heartfelt appreciation to you for your thorough review and invaluable recommendations, which have played a pivotal role in refining our manuscript. The revised version, incorporating your insightful suggestions, has been resubmitted for your kind evaluation. We eagerly anticipate any further feedback you may have and remain sincerely grateful for your time and dedication to enhancing the quality of our work.

Reviewer 4 Report

The article is quite interesting. It contains all the necessary parts. I have no major comments regarding the methodological approach. However, I think it is a bit too long. For this reason, I suggest the Authors review it carefully and especially eliminate those fragments of the text that are repeated.

Detailed comments:

1. There are no e-mails in the Authors' affiliations (except the correspondence one).

2. Keywords should not coincide with the title of the article. Please make appropriate corrections.

3. The article should be written impersonally, avoiding, for example, the words "we" or "our".

4. The text from lines 119-167 can be successfully deleted.

5. Figure 11 is also a repetition of previous results, so there is no need to present it in the Discussion.

6. There should not be located any table inside Conclusions.

7. Lines 677-695. This is more of a Discussion part  than  Conclusions.

Author Response

Response to Reviewer 4 Comments

Dear Reviewer,

Thank you for your thorough review and insightful comments on our article. We appreciate your positive assessment of the methodological approach and your constructive suggestions for improvement. We've carefully noted each of your points and will comprehensively address them in our revised manuscript.

Comment1. There are no e-mails in the Authors' affiliations (except the correspondence one).

Response1: We will thoroughly review and streamline the article, eliminating redundant sections to ensure a more concise and reader-friendly text. The authors' institution information will be updated to include email addresses, thereby enhancing the authors' contact details.

Comment2. Keywords should not coincide with the title of the article. Please make appropriate corrections.

Response2: Keywords will be revised to avoid repetition with the article title.

Comment3. The article should be written impersonally, avoiding, for example, the words "we" or "our".

Response3: To maintain a more objective writing style, we will refrain from using terms like "we" or "our."

Comment4. The text from lines 119-167 can be successfully deleted.

Response4: The suggested content will be reworked and incorporated into the Conclusion section to ensure consistency and conciseness in the article.

Comment5. Figure 11 is also a repetition of previous results, so there is no need to present it in the Discussion.

Response5: Redundant Figure 11 from the Discussion section has been relocated to the appendix to minimize repetition.

Comment6. There should not be located any table inside Conclusions.

Response6: The Conclusion section has been reformatted and no longer includes any tables.

Comment7. Lines 677-695. This is more of a Discussion part than Conclusions.

Response7: Content from the Outlook section (Lines 677-695) has been relocated to the Discussion section to ensure a logical flow and content coherence in the Conclusion.

We extend our heartfelt appreciation to you for your thorough review and invaluable recommendations, which have played a pivotal role in refining our manuscript. The revised version, incorporating your insightful suggestions, has been resubmitted for your kind evaluation. We eagerly anticipate any further feedback you may have and remain sincerely grateful for your time and dedication to enhancing the quality of our work.

Warm regards, 

All Authors

Round 2

Reviewer 1 Report

The authors not explored the literature, especialy explortaion of machine learning applictaion for forst stand structure optimization, rather in introduction and in discussion. In addition, judging the suitability of VOF metric for model evaluation by citing two reference that used it only not agrue to ignore other metric parameters. Further, the model consumption time not explored in the manuscript, exploring itertaion number and VOF paramter not reflect needed time for each model executing.  

Overall, would be better to take your time to adjust and save the paper. 

Best of luck

Author Response

Response to Reviewer 1 Comments

We extend our heartfelt gratitude for your thorough review and invaluable insights provided for our manuscript. Your feedback has been immensely valuable, shaping the revisions we've meticulously implemented throughout the paper. These refinements significantly bolster the paper, enhancing its comprehensiveness and persuasive impact. We eagerly anticipate your continued guidance and insights upon reviewing the updated version.

Comment1- The authors not explored the literature, especialy explortaion of machine learning applictaion for forst stand structure optimization, rather in introduction and in discussion.

Response1:

We greatly appreciate the insightful comment regarding the exploration of machine learning applications within forest stand structure optimization. In response, we've extensively revised the introduction and discussion sections to enhance the literature review. Our focus now involves an extensive examination of various machine learning techniques applied in forestry, particularly in single-objective studies such as fire prediction, yield forecasting, and LiDAR-based forest structure identification.

While acknowledging the effectiveness of deep learning in handling extensive datasets and pattern recognition, its application in forest management faces limitations, particularly in environmental interaction and reward function design, restricting its primary usage to forest inventory and planning. This limitation impedes its ability to comprehensively address complex multi-objective optimization problems.

Our comprehensive literature review revealed a significant gap in the application of these methodologies when addressing multi-objective optimization in forest stand structures. This highlights a crucial inadequacy in these techniques when dealing with complex multi-objective scenarios. To address this gap, our study utilizes reinforcement learning methods, focusing specifically on the multi-objective optimization of secondary forests of Pinus yunnanensis. Diverging from existing research, our focus lies in addressing the challenges posed by multiple objectives—an area that remains underexplored in current literature. This effort aims to carve a new path in forest structure optimization, intending to bridge the research gap in machine learning approaches to multi-objective optimization in forest stand structures.

Comment2- In addition, judging the suitability of VOF metric for model evaluation by citing two reference that used it only not agrue to ignore other metric parameters.

Response2:

Regarding the choice of solely relying on VOF without considering other parameters, we acknowledge the limitations of basing the evaluation of algorithms solely on the reference papers that utilize VOFs. Through further literature review, we discovered that in the domain of multi-objective forest structure optimization, similarly to our study, many scholars integrate multiple evaluation parameters corresponding to various sub-objectives to derive a comprehensive evaluation index (such as objective function values, fitness values, and comprehensive homogeneity index). They use this composite evaluation index along with runtime to gauge the efficacy of optimization algorithms. While we admit the limitations of relying solely on a composite evaluation index to measure the effectiveness of optimization algorithms, in the realm of multi-objective forest structure optimization and within the context of our research, it may indeed represent the most fitting approach. Moving forward, in subsequent research, we aim to focus on exploring other effective algorithm evaluation parameters within this domain.

Comment3- Further, the model consumption time not explored in the manuscript, exploring itertaion number and VOF paramter not reflect needed time for each model executing.

Response3:

Thank you for highlighting the relationship between model execution time, iteration count, and parameter evaluation. We acknowledge the limitations of using iteration count alone as a measure of algorithmic efficiency since the actual run times might vary for the same iteration count. The newly added table comprehensively illustrates the diverse run times required by different optimization approaches across various sample plots to achieve maximum VOFs. The introduction of this table aims to distinctly present the practical computational efficiency variations among different optimization methods. We believe this addition significantly supplements the evaluation of algorithmic time consumption and provides a more comprehensive view of the model's performance in real-time scenarios.

We extend our sincere gratitude for your meticulous review and invaluable recommendations, which have significantly contributed to refining our manuscript. The revised version, now incorporating your insightful suggestions, has been resubmitted for your esteemed evaluation. Your ongoing guidance and feedback are highly valued, and we remain deeply appreciative of your dedicated efforts in enhancing the quality and depth of our work.