A Remote Sensing-Based Inventory of West Africa Tropical Forest Patches: A Basis for Enhancing Their Conservation and Sustainable Use
Round 1
Reviewer 1 Report
Journal: Land (ISSN 2073-445X)
Manuscript ID: land-1968097
Title: A remote sensing-based inventory of West Africa tropical forest patches: a basis for enhancing their conservation and sustainable use
Overall Comments and Suggestions for Authors
Dear authors,
Regarding West Africa tropical forest, this manuscript would provide some information based on a remote sensing-based inventory. The manuscript was written in detail. From my point of view, the Introduction and Materials and Methods were described in detail. I could not find any critical flaws for further process. For the manuscript to be improved, here I enclosed several suggestions to the authors.
I hope that authors are able to find out any advantages from peer-review’s comments. My specific comments were provided in detail as follows.
Kind regards,
Reviewer
Point 1.
Authors may want to follow the guidelines of this target journal in further process. It may require the different citation and bibliography format from the current version. It would be an ordering number for the citation in the text and the corresponding order accordingly in bibliography. Additionally, authors may need to use the superscript for the units or signs, e.g., km2 and R2, throughout the manuscript.
Point 2.
The manuscript could be a reference because it analyzed the forest patch across the countries. I think authors may be able to put a bit more weights on this trial targeted on West Africa forests in Results and Discussion or Conclusion.
Point 3.
I count on the other reviewers with regards to the evaluation of data source quality and quantity for the target area. At least to me, it may be worth examining the results. Still, why don’t authors write more references in which the methodological approaches were similar? What about mentioning any other similar research or concepts to support this manuscript? I consider it would be helpful to make this convincing along with Congalton and Green 2019.
Point 4.
Check the equal sign in P-value. You may want to drop off the equal sign (=) because it is actually based on the normal standard distribution, which is a type of continuous probability distribution. Also, as I mentioned earlier, consider modifying R2 with superscript.
Moreover, in methods and results section, I could not find out why authors did perform the regression analysis with grouped (or merged) sample points instead of just using the raw data. If you just apply the raw data in regression, it would let the model fit closer to the real trend. It could help to decrease uncertainty and reduce the residuals. Authors may be able to consider this approach to offer a better modeling result. Consider adding the number of samples if authors follow my suggestion.
Lastly, regarding the Figure 5 and regression analysis, in addition to the p-value of F statistic, it may require providing the significance of each coefficient in the figure or caption at least. Also, if possible, S.E. of the coefficients would be also informative.
Comments for author File: 
Comments.pdf
Author Response
Point 1.
Authors may want to follow the guidelines of this target journal in further process. It may require the different citation and bibliography format from the current version. It would be an ordering number for the citation in the text and the corresponding order accordingly in bibliography. Additionally, authors may need to use the superscript for the units or signs, e.g., km2 and R2, throughout the manuscript.
As per the reviewers suggestion, we have updated the citation format and bibliography to suit the current journal.
In addition, we have addressed the km2 and R2 issue throughout the manuscript.
Point 2.
The manuscript could be a reference because it analyzed the forest patch across the countries. I think authors may be able to put a bit more weights on this trial targeted on West Africa forests in Results and Discussion or Conclusion.
As per the reviewers suggestion, we have added more emphasis on the novelty of the manuscript throughout the three sections, these revision are highlighted throughout the text.
Point 3.
I count on the other reviewers with regards to the evaluation of data source quality and quantity for the target area. At least to me, it may be worth examining the results. Still, why don’t authors write more references in which the methodological approaches were similar? What about mentioning any other similar research or concepts to support this manuscript? I consider it would be helpful to make this convincing along with Congalton and Green 2019.
We thank the review for their very thoughtful comment, and have proceeded to address it in details, specifically, by referencing several articles with similar methodological approaches, namely: Hansen et al. (2020).
Point 4.
Check the equal sign in P-value. You may want to drop off the equal sign (=) because it is actually based on the normal standard distribution, which is a type of continuous probability distribution. Also, as I mentioned earlier, consider modifying R2 with superscript.
As per the reviewer recommendation, we have dropped off the equal signs. In addition, the superscript has been changed throughout.
Moreover, in methods and results section, I could not find out why authors did perform the regression analysis with grouped (or merged) sample points instead of just using the raw data. If you just apply the raw data in regression, it would let the model fit closer to the real trend. It could help to decrease uncertainty and reduce the residuals. Authors may be able to consider this approach to offer a better modeling result. Consider adding the number of samples if authors follow my suggestion.
We thank the reviewer for their insight comment and proceed to address it below. In line 272-276, we now justify our use of aggregation.
“To process the regression analysis, we aggregated forest fragments into size classes to reduce the amount of noise in the analysis. Specifically, we aggregated the fragments into 1 km2 size classes, such that if a patch were 1.5 km2 it would be grouped into size class 1-2; this resulted in individual forest patches being aggregated into 9, 1 km2 area/size classes of 1-2 to 9-10 km2 comprising the predictor variables. Subsequently, the average value of the variable of interest was taken for each forest fragment size class and comprised the target variable
Lastly, regarding the Figure 5 and regression analysis, in addition to the p-value of F statistic, it may require providing the significance of each coefficient in the figure or caption at least. Also, if possible, S.E. of the coefficients would be also informative.
We thank the reviewer for the pertinent request. However, in a methodologically related paper, namely, Hansen et 2020 (https://www.science.org/doi/10.1126/sciadv.aax8574#sec-4), the present their results (Figure 2) including only the p-value and R2 value. Since their paper is presenting a similar dataset, and is published in a well-known journal, we have opted to retain the current succinct format and judge that it provides sufficient information.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear Authors,
The paper" "A remote sensing-based inventory of West Africa tropical forest patches: a basis for enhancing their conservation and sustainable use" has got potential for good pyblication, but it needs some modificatins:
1. Introduction - is too short, more actual publications should be add here/
2. Study area - it will be better to put here map with localization of study area
3. Discussion - better as seperatted chapter
Author Response
- Introduction - is too short, more actual publications should be add here/
As per the reviewer’s suggestion, we have now added several references to the introduction. For example, the introduction line 74 now includes the sentence: “Global forest change datasets have enabled unprecedented ability to monitoring global changes in forest cover (Hansen et al. 2013),”
- Study area - it will be better to put here map with localization of study area
As per the reviewer’s recommendation, we have now included a map figure with the study area; specifically, the map figure which was Figure 3 was placed under the heading “study area” and renamed.
- Discussion - better as seperatted chapter
We would like to thank the review for the suggestion; however, upon consultation, the reviewers have opted for maintaining the current structure, as it enables readers to better relate the results. For example, in line 352, we state and result and immediate discuss its implications, namely, “Most of the mapped forest patches are small, secondary forest patches, and these small patches are found to experience the greatest overall loss. This result is in line with previous studies showing that smaller forest fragments experience the greatest degree of forest loss (Hansen et al. 2020).”.
Reviewer 3 Report
Based on multi-source remote sensing data sets, this paper has evaluated the changes of forest patches of different scales and grades in West Africa for the past 45 years, which is of great significance to the ecological protection and sustainable use of forests.
Line 87: Without section1.2, the 1.1 title is not necessary.
Line 210-212: A brief introduction to the Global Ecosystem Dynamics Investigation (GEDI) dataset can be presented here.
Line 276-284: ‘R2’ and ‘km2’ should be ‘R2’ and ‘km2’, please check the full content of your manuscript.
Line 289: It is better to show the four quadrants with image (or map) and mark the locations of the selected 20% forest patches.
Line 410-435: The regression analysis results here are described as "smaller patches experience on average less forest loss than larger ones". However, in regression analysis, please consider whether it would be more reasonable to set y as area loss ratio (%).
Author Response
Line 87: Without section1.2, the 1.1 title is not necessary.
This point has been modified as per the reviewer recommendation.
Line 210-212: A brief introduction to the Global Ecosystem Dynamics Investigation (GEDI) dataset can be presented here.
As per the reviewer recommendation, this sentence now reads: “Finally, we apply a 5 m tree height mask using the most recent high-resolution Global Ecosystem Dynamics Investigation (GEDI) dataset which represents global tree heights based on a fusion of spaceborne-lidar data (Potapov et al. 2021).”
Line 276-284: ‘R2’ and ‘km2’ should be ‘R2’ and ‘km2’, please check the full content of your manuscript.
As per the reviewer recommendation, all instances have been replaced with ‘R2’ and ‘km2’
Line 289: It is better to show the four quadrants with image (or map) and mark the locations of the selected 20% forest patches.
We acknowledge the reviewer comment as being insightful and constructive, however, due to the already large number of figures in the manuscript, and the type of figure which is being presented (Figure. 2) is a workflow diagram, we have decided to maintain the figure as a diagram, rather than a map.
Line 410-435: The regression analysis results here are described as "smaller patches experience on average less forest loss than larger ones". However, in regression analysis, please consider whether it would be more reasonable to set y as area loss ratio (%)
We thank the reviewer for their very pertinent observation; however, upon consideration, the authors have decided that the y axis should be remain as actual loss area in km2 , since this is a key variable which is of most interest in this study.
