Review Reports - Possibilities and Limitations of a Geospatial Approach to Refine Habitat Mapping for Greater Gliders (<i>Petauroides</i> spp.)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents an important investigation into using geospatial analysis for identifying potential hollow-bearing trees to refine habitat mapping for greater gliders. While the research question is highly relevant to conservation efforts, there are several areas requiring substantial revision to improve the scientific rigor, clarity, and impact of the work.

Major Concerns

Introduction (Lines 32-82)

Lines 32-40: The introduction begins with a general statement about old-growth forests declining globally, but lacks specific data on the rate of decline or regional variations that would contextualize the problem. Consider adding quantitative information about forest loss in Australia specifically.

Lines 41-42: The statement "Thus, a lack of hollow-bearing trees has negatively impacted many Australian hollow-dependent species" is too general. Specify which species are most affected and provide evidence of population declines linked directly to hollow loss.

Lines 43-56: This section on geospatial analysis methods is underdeveloped. The review of previous studies using LiDAR for tree identification is cursory and misses several important recent advances. The authors should expand this review to include more recent studies (post-2020) on advanced LiDAR applications in forest ecology.

Lines 57-63: The research gap statement is vague. The authors state "there is limited research investigating the use of geospatial analyses to identify hollow-bearing trees in order to refine habitat mapping for hollow-dependent species in Australia," but do not specify what exact gaps exist in current knowledge or methodologies. A more precise articulation of knowledge gaps is needed.

Lines 64-77: The background information on greater gliders is informative but lacks integration with the methodological approach. How do the specific habitat requirements of greater gliders (e.g., hollow size preferences) inform or constrain the geospatial approach?

Lines 78-82: The research aims are presented late in the introduction and lack specificity. The link between aim 2 (correlating tree height and diameter) and the overall goal of mapping hollow-bearing trees is not clearly established. A stronger conceptual framework linking all three aims to the overarching research question would strengthen this section.

Methods (Lines 86-141)

Lines 87-100: The data collection section lacks critical details about the LiDAR data specifications. Information about point density, accuracy, and flight parameters of the 2009 LiDAR acquisition is essential for evaluating the appropriateness of the data for tree detection.

Lines 101-111: The field survey methodology requires more detail. How were transect locations selected? Was there a stratification approach based on ecosystem types or expected tree density? The extrapolation of tree density from belt transects to 1 km² cells needs stronger justification.

Lines 112-124: The geospatial analysis section lacks sufficient technical detail. The description of the lidR package implementation is too general. What specific algorithms were used for ground point classification, normalization, and individual tree detection? What parameters were applied for the variable window size?

Lines 120-124: Figure 2 is referenced but appears to be incorrectly numbered in the manuscript as it shows the same content as Figure 1. This creates significant confusion for the reader.

Lines 125-141: The statistical analysis section is inadequate. The categorization of transects into "high" and "low" density areas is not clearly defined—what thresholds were used to make this distinction? The selection of statistical tests is not sufficiently justified, and there is no mention of assumption testing for the parametric tests employed.

Results and Discussion (Lines 142-254)

Lines 143-169: The presentation of results begins abruptly without a clear structure. The findings regarding LiDAR's ability to determine tree heights are reported without presenting the actual RMSE values, which are critical for evaluating accuracy.

Lines 170-188: The discussion of LiDAR limitations is valuable but would be strengthened by quantitative comparisons with other studies. How does the achieved accuracy of ~5m compare to other similar studies using comparable or more advanced technologies?

Lines 189-201: The results regarding the correlation between tree height and diameter are presented without sufficient statistical detail. The R² value is provided without appropriate context for interpretation, and the significance level (p<0.003) is unconventionally reported.

Lines 202-217: The findings on hollow presence and tree diameter correlation are important but lack nuance. The analysis does not account for tree species or age, which are known to influence hollow formation. The comparison between private properties and state forests is interesting but presented without statistical testing.

Lines 218-234: The core hypothesis test results are presented unclearly. The statement that "63% were correctly differentiated" needs clarification—does this refer to classification accuracy? The lack of statistical significance is attributed to sample size or scale issues, but other potential methodological limitations are not adequately considered.

Lines 235-254: The implications section is too general and lacks specific recommendations for improving the methodology. The suggestions for refining the approach are valuable but need more development and connection to the specific limitations identified in the study.

Specific Issues

Literature Integration: Throughout the manuscript, there is insufficient integration with relevant literature. Many statements lack appropriate citations, and where citations are provided, they are often not the most current or relevant.
Figure Issues: Figure 1 and Figure 2 appear to be duplicated or mislabeled. Figure 3 is referenced but not clearly described. Figure 4 contains statistical test results but lacks clear labels and explanations.
Statistical Reporting: The reporting of statistical results is inconsistent and often incomplete. P-values are reported with varying precision, and effect sizes are not consistently reported.
Methodological Clarity: The transition from field data collection to LiDAR analysis is not clearly described, making it difficult to understand how the two datasets were aligned for comparison.
Temporal Disconnect: The 14-year gap between LiDAR data (2009) and field surveys (2023) is acknowledged but not adequately addressed in the analysis. How were potential changes in forest structure during this period accounted for?

Detailed Line-by-Line Comments

Lines 32-33: "Old growth forests are declining globally, predominantly due to land clearing and wildfires" – This statement needs more specificity and current references.

Lines 37-40: The process of hollow formation is described but lacks reference to time scales for different tree species relevant to the study area.

Lines 49-51: The statement about LiDAR use in Australian landscapes is too general and needs specific examples with appropriate citations.

Lines 52-56: The review of previous studies using geospatial analysis to identify hollow-bearing trees is superficial and needs expansion.

Lines 59-63: The hypothesis statement lacks precision in defining what constitutes "high-densities" of potential hollow-bearing trees.

Lines 64-77: The information on greater gliders should be better integrated with the methodological approach and research aims.

Lines 78-82: The research aims need clearer articulation and stronger connection to the identified knowledge gaps.

Lines 87-91: The data collection section lacks details on the temporal disconnection between LiDAR data (2009) and field surveys (2023).

Lines 93-95: The statement about LiDAR generating a "three-dimensional visual point cloud" is technically imprecise.

Lines 101-105: The rationale for site selection is inadequately explained. What criteria determined the selection of the four sites?

Lines 106-111: The methodology for recording DBH and hollow presence needs more detail on detection methods and observer reliability.

Lines 113-119: The comparison of field measurements to LiDAR-derived tree heights needs more explanation of the matching process.

Lines 120-124: The description of individual tree detection from LiDAR is too general and lacks technical details on algorithms and parameters.

Lines 125-131: The categorization of transects into high and low density areas is not clearly defined or justified.

Lines 132-137: The statistical methods are described without sufficient detail on model selection, assumption testing, or validation approaches.

Lines 143-146: The main finding regarding LiDAR accuracy is presented without quantitative details.

Lines 147-158: The description of the analysis process repeats information from the methods section without adding new insights.

Lines 170-188: The discussion of limitations due to the age of LiDAR data is valuable but needs more specific suggestions for improvement.

Lines 189-201: The correlation between tree height and diameter is discussed without adequate consideration of confounding factors like tree species or growth conditions.

Lines 202-217: The comparison between private properties and state forests introduces interesting observations but lacks rigorous analysis.

Lines 218-234: The central hypothesis test results lack clarity in presentation and interpretation.

Lines 235-254: The implications section needs more specific recommendations for improving the methodology based on the study's findings.

Author Response

For research article “Possibilities and limitations of a geospatial approach to refine habitat mapping for greater gliders (Petauroides spp.).”

We thank the editor and reviewers for their thorough consideration of our manuscript and have thoroughly reviewed and edited the manuscript accordingly. We provide an outline of our changes and responses to the review below (author responses in green text).

Reviewer 1 Comments to Author:

Thank you for taking the time to thoroughly review our manuscript and for your suggestions to improve our manuscript. We have made major revisions to all sections providing more detail and information that we believe has strengthened and improved the clarity of the paper.

Major Concerns

Introduction (Lines 32-82)

Lines 32-40: The introduction begins with a general statement about old-growth forests declining globally, but lacks specific data on the rate of decline or regional variations that would contextualize the problem. Consider adding quantitative information about forest loss in Australia specifically.

This information has been added to section 1.0, lines 42 - 44.

Lines 41-42: The statement "Thus, a lack of hollow-bearing trees has negatively impacted many Australian hollow-dependent species" is too general. Specify which species are most affected and provide evidence of population declines linked directly to hollow loss.

This information has been modified in section 1.0, lines 46 - 50.

Lines 43-56: This section on geospatial analysis methods is underdeveloped. The review of previous studies using LiDAR for tree identification is cursory and misses several important recent advances. The authors should expand this review to include more recent studies (post-2020) on advanced LiDAR applications in forest ecology.

We agree and have added more context and information about geospatial analysis and remote sensing and its uses in section 1.0 from lines 79 – 118. Throughout this paper, we wanted to highlight the need for a cost-effective and transferrable methods to identify habitat trees, via the use of freely available LiDAR and unsupervised machine learning methods and have modified the text to reflect this.

Lines 57-63: The research gap statement is vague. The authors state "there is limited research investigating the use of geospatial analyses to identify hollow-bearing trees in order to refine habitat mapping for hollow-dependent species in Australia," but do not specify what exact gaps exist in current knowledge or methodologies. A more precise articulation of knowledge gaps is needed.

We have added to the text to point out a lack of management tools to identify hollow-bearing trees and the need for them to be identified to conserve hollow-dependent species. We explain the need for a simple, accessible and transferrable geospatial approach to identify potential habitat trees in section 1.0, lines 132-145.

Lines 64-77: The background information on greater gliders is informative but lacks integration with the methodological approach. How do the specific habitat requirements of greater gliders (e.g., hollow size preferences) inform or constrain the geospatial approach?

We have edited the text in section 1.0, lines 150 – 153 to improve information that large hollows, suitable for greater gliders are found in trees >30 cm DBH. For our study we tested the assumptions that we could use 2009 LiDAR to identify large trees, which were likely to have a larger DBH, which were more likely to be hollow-bearing (lines 171 – 178).

Lines 78-82: The research aims are presented late in the introduction and lack specificity. The link between aim 2 (correlating tree height and diameter) and the overall goal of mapping hollow-bearing trees is not clearly established. A stronger conceptual framework linking all three aims to the overarching research question would strengthen this section.

We agree that this was unclear and have added in an overarching aim; to use a simple, transferrable geospatial approach to refine greater glider habitat mapping. We then tested three assumptions (section 1.0, lines 174 – 178) prior to assessing the hypothesis that 2009 LiDAR could be used to identify high-densities of potential hollow-bearing trees on the Fraser Coast.

Methods (Lines 86-141)

Lines 87-100: The data collection section lacks critical details about the LiDAR data specifications. Information about point density, accuracy, and flight parameters of the 2009 LiDAR acquisition is essential for evaluating the appropriateness of the data for tree detection.

This information was added into lines 300 – 303. “The LiDAR point cloud data has a vertical accuracy of 0.15 m and an average of 2 points per square metre were acquired using aerial laser scanning from a fixed-wing aircraft, collecting all returns, within our study sites surveyed between July and September of 2009.”

Lines 101-111: The field survey methodology requires more detail. How were transect locations selected? Was there a stratification approach based on ecosystem types or expected tree density? The extrapolation of tree density from belt transects to 1 km² cells needs stronger justification.

Thank you for pointing this out, we have added information into section 2.2, lines 282 – 291 explain how transects were selected. “After a preliminary analysis of tree height to identify expected density of large trees (>50 cm DBH) across the sites, nineteen 2000m² (100 x 20 meters) belt transects were conducted across the four sites.” Using the lidR package for LiDAR data processing, 1 km² was the default tile processing size.

Lines 112-124: The geospatial analysis section lacks sufficient technical detail. The description of the lidR package implementation is too general. What specific algorithms were used for ground point classification, normalization, and individual tree detection? What parameters were applied for the variable window size?

We agree, this whole section (2.3) was restructured to add in detail for lines 305 – 324.

Lines 120-124: Figure 2 is referenced but appears to be incorrectly numbered in the manuscript as it shows the same content as Figure 1. This creates significant confusion for the reader.

Thank you for pointing this out. A mistake was made when reformatting the manuscript for submission, using the formatting guide. This has been resolved in section 2.3 line 338 – 339.

Lines 125-141: The statistical analysis section is inadequate. The categorization of transects into "high" and "low" density areas is not clearly defined—what thresholds were used to make this distinction? The selection of statistical tests is not sufficiently justified, and there is no mention of assumption testing for the parametric tests employed.

We have added in text to explain tree height threshold calculations in section 2.3 lines 325 – 329. We also added in information to clarify the categorization of densities of large trees in section 2.4, lines 356 – 361.

Results and Discussion (Lines 142-254)

Lines 143-169: The presentation of results begins abruptly without a clear structure. The findings regarding LiDAR's ability to determine tree heights are reported without presenting the actual RMSE values, which are critical for evaluating accuracy.

The results and discussion have now been separated to make it much clearer for the reader. The results are reported in section 3.1 from lines 379 – 388 “the RMSE between tree heights derived using LiDAR from 2009 and on-ground surveyed trees was 5.75 m for trees >30 cm DBH, whereas, for trees >50 cm DBH, the RMSE was 4.69 m.”

Lines 170-188: The discussion of LiDAR limitations is valuable but would be strengthened by quantitative comparisons with other studies. How does the achieved accuracy of ~5m compare to other similar studies using comparable or more advanced technologies?

We agree, this should have been expanded on and is now explained in section 4.0 lines 628 - 653.

Lines 189-201: The results regarding the correlation between tree height and diameter are presented without sufficient statistical detail. The R² value is provided without appropriate context for interpretation, and the significance level (p<0.003) is unconventionally reported.

We’re unsure what is meant by the p value being unconventionally reported. Section 3.2, line 408 refers to Table 1 in the appendix.

Lines 202-217: The findings on hollow presence and tree diameter correlation are important but lack nuance. The analysis does not account for tree species or age, which are known to influence hollow formation. The comparison between private properties and state forests is interesting but presented without statistical testing.

The hypothesis of our study was that LiDAR from 2009 could identify high-densities of potentially hollow-bearing trees. We were unable to determine tree age and species from LiDAR alone, therefore it was left out of this analysis. However, we were able to identify regional ecosystem type using LiDAR, which was shown to influence the correlation between tree height and diameter. The comparison in hollow presence between private properties and state managed forest was merely an observation to point out that private properties can be better managed than state forest in some cases, and that private properties are more likely to have larger trees due to the lack of logging. We believe these are relevant to have in the discussion since they are observations that are of importance to conservation of these landscapes in the context of our study.

Lines 218-234: The core hypothesis test results are presented unclearly. The statement that "63% were correctly differentiated" needs clarification—does this refer to classification accuracy? The lack of statistical significance is attributed to sample size or scale issues, but other potential methodological limitations are not adequately considered.

The density mapping analyses have been edited and are now reported in section 3.4, lines 456 – 463. The impact of LiDAR from 2009 on mapping accuracy was edited for clarity and is discussed in section 4.0, through lines 636 – 653, however this was also a topic discussed throughout the paper. The recommendations have been expanded on to address the limitations from this study, and are discussed in section 4.0, lines 709 – 715.

Lines 235-254: The implications section is too general and lacks specific recommendations for improving the methodology. The suggestions for refining the approach are valuable but need more development and connection to the specific limitations identified in the study.

We agree, this section has been expanded in section 4.0, from lines 705 – 719.

Specific Issues

Literature Integration: Throughout the manuscript, there is insufficient integration with relevant literature. Many statements lack appropriate citations, and where citations are provided, they are often not the most current or relevant.

We agree and have added in relevant and updated citations and references to relevant literature.

Figure Issues: Figure 1 and Figure 2 appear to be duplicated or mislabeled. Figure 3 is referenced but not clearly described. Figure 4 contains statistical test results but lacks clear labels and explanations.

Thank you for pointing this out. Figure captions have now been fixed.

Statistical Reporting: The reporting of statistical results is inconsistent and often incomplete. P-values are reported with varying precision, and effect sizes are not consistently reported.

The statistical results have now been reported in a consistent format to avoid confusion for the reader.

Methodological Clarity: The transition from field data collection to LiDAR analysis is not clearly described, making it difficult to understand how the two datasets were aligned for comparison.

This has been revised with an overarching statement in section 2.1 from lines 237 – 243. Section 2.2 also states “A subset of approximately five trees at each site was also measured for tree height using a Nikon Laser Forestry Pro II rangefinder to measure height for trees with a clear view of the base and canopy (n = 91) to assess the accuracy of LiDAR tree height detection.” And “We compared field measurements to tree heights extracted from 2009 LiDAR for each 1 km² cell which contained a belt transect” before explaining how LiDAR data was extrapolated.

Temporal Disconnect: The 14-year gap between LiDAR data (2009) and field surveys (2023) is acknowledged but not adequately addressed in the analysis. How were potential changes in forest structure during this period accounted for?

The tree height thresholds which were created to determine a large tree, have been used to account for changes in forest structure. This was calculated as the average of the top 10% of tree heights minus the standard deviation of the population. Large trees detected from 2009 LiDAR would then be able to be surveyed in the field as there is little changes in tree height for old growth forests (section 2.3, line 325 - 329).

Detailed Line-by-Line Comments

Lines 32-33: "Old growth forests are declining globally, predominantly due to land clearing and wildfires" – This statement needs more specificity and current references.

Up to date references have been added in for this statement (section 1.0, lines 44 – 46).

Lines 37-40: The process of hollow formation is described but lacks reference to time scales for different tree species relevant to the study area.

This information has been added in section 1.0 lines 53 - 55.

Lines 49-51: The statement about LiDAR use in Australian landscapes is too general and needs specific examples with appropriate citations.

We have added relevant and updated citations to section 1.0, line 123 – 125.

Lines 52-56: The review of previous studies using geospatial analysis to identify hollow-bearing trees is superficial and needs expansion.

We have added context to this section 1.0, lines 127 – 136.

Lines 59-63: The hypothesis statement lacks precision in defining what constitutes "high-densities" of potential hollow-bearing trees.

The LiDAR mapping was only within current high-quality mapping for greater gliders, which only accounts for the certain regional ecosystems which feature the gliders preferred tree species. Each site was then categorised as either a high- or low-density of potentially hollow-bearing trees, within greater glider habitat mapping, using natural jenks during geospatial analysis processing. Information was added in for section 2.4, lines 359 – 363.

Lines 64-77: The information on greater gliders should be better integrated with the methodological approach and research aims.

Information was added in section 1.0, lines 167 – 175 about the current greater glider habitat mapping for the Fraser Coast, the limitations of that mapping and the methods we propose to refine the mapping (by testing our hypothesis).

Lines 78-82: The research aims need clearer articulation and stronger connection to the identified knowledge gaps.

We have added in information to section 1.0, lines 176 – 183, articulating that the assumptions have been tested prior to the hypothesis test in order to address if the hypothesis could be accurately tested.

Lines 87-91: The data collection section lacks details on the temporal disconnection between LiDAR data (2009) and field surveys (2023).

Text has been added in to clarify why the LiDAR used was from 2009 in section 2.3, lines 301 – 305.

Lines 93-95: The statement about LiDAR generating a "three-dimensional visual point cloud" is technically imprecise.

We agree, this text has been removed and the information about what LiDAR is and how it is collected has been moved to the introduction, section 1.0, lines 84 – 98.

Lines 101-105: The rationale for site selection is inadequately explained. What criteria determined the selection of the four sites?

Information was added to clarify the site selection in field surveys section 2.2, lines 260 – 267.

Lines 106-111: The methodology for recording DBH and hollow presence needs more detail on detection methods and observer reliability.

Information has been added in for this section to clarify how hollow presence and DBH were surveyed, section 2.2, lines 290 - 294.

Lines 113-119: The comparison of field measurements to LiDAR-derived tree heights needs more explanation of the matching process.

Text was added to explain how field measurements were compared to LiDAR-derived tree height measurements in section 2.4, lines 338 - 340.

Lines 120-124: The description of individual tree detection from LiDAR is too general and lacks technical details on algorithms and parameters.

Information was added to add context to the individual tree detection processes in section 2.3, lines 314 – 333.

Lines 125-131: The categorization of transects into high and low density areas is not clearly defined or justified.

Information has been added to clarify how density categories were extrapolated in section 2.4, lines 364 – 371.

Lines 132-137: The statistical methods are described without sufficient detail on model selection, assumption testing, or validation approaches.

Lines 143-146: The main finding regarding LiDAR accuracy is presented without quantitative details.

This has been changed to include these details in section 3.1, lines 391 – 392.

Lines 147-158: The description of the analysis process repeats information from the methods section without adding new insights.

We agree, and this has now been removed from the results section (3.1).

Lines 170-188: The discussion of limitations due to the age of LiDAR data is valuable but needs more specific suggestions for improvement.

We have added information about the discussion of limitations due to the LiDAR age in section 4.0, lines 626 – 645. However, we believe that as the aim was to identify a simple geospatial approach, using freely available LiDAR, the discussion should be focused on how effective older LiDAR can still be, and that this methodological approach is accessible and cost-effective, without requiring new LiDAR and deep machine learning techniques.

Lines 189-201: The correlation between tree height and diameter is discussed without adequate consideration of confounding factors like tree species or growth conditions.

We assessed regional ecosystem type and the effect this has on the correlation of tree height and diameter. We were unable to identify tree species from LiDAR, and regional ecosystem types gives also includes growth conditions. However, we do mention that future studies should consider the impact of smaller scale site factors, and the impacts of logging and fire (section 4.0, lines 719 – 721).

Lines 202-217: The comparison between private properties and state forests introduces interesting observations but lacks rigorous analysis.

We agree that these observations lack rigorous analysis, however we still believe that they are relevant to have in the discussion since they are observations that are of importance to conservation of these landscapes in the context of our study.

Lines 218-234: The central hypothesis test results lack clarity in presentation and interpretation.

Information has been added to this section to aid clarity for the reader in section 4.0 lines 593 – 596.

Lines 235-254: The implications section needs more specific recommendations for improving the methodology based on the study's findings.

The implications for conservation and recommendations for future works sections has been reworked to clearly convey the next steps of this work. Information has been added to lines 715 – 729 now give exact recommendations for future work.

Reviewer 2 Report

Comments and Suggestions for Authors

1- Abstract:

In line 16; you must explain this abbreviation “DBH”
Why didn’t you mention enough details about the temporal and spatial data, as it was not clearly stated that the LiDAR data used is old (2009) while the field surveys were conducted in 2023. This time gap is very important, and it should have been mentioned in the abstract because it affects the accuracy of the results.
You must mention the sample size (the number of sites surveyed), as these details will give the reader a better understanding of the power of the study and the generalizability of the results.
Why you didn’t indicate the statistical significance of the results as statistical values (such as p-values or RMSE) that reflect the accuracy of the results are not indicated, even briefly.

2-Introduction:

In the introduction, you should highlight the potential challenges of this study such as:

the difficulty of identifying hollow-bearing trees using LiDAR data alone,
how the accuracy of the results is affected by the time interval between field surveys in 2023 and LiDAR data in 2009,
and the requirement for increasingly sophisticated methods, such as machine learning, to interpret the data more effectively.

Why did the introduction not adequately discuss the applications of the research study, as it focused only on the academic aspect, but did not explain how the research results can be applied to forest management policies or environmental protection initiatives.
The transition to the objectives was a bit abrupt, as there should have been a transitional paragraph summarizing why the current research was needed right before the objectives were presented.\

3-Methods

In line 91; Why you didn’t address the sources of uncertainty in the data, such as:

Discussing limitations of LiDAR data, such as the effect of vegetation density on the accuracy of trees detection.
Analyzing how the quality of earlier data (2009) impacts the current findings.

In section 2.1 ; One of the most important weaknesses of the research is the use of old LiDAR data from 2009, despite field surveys being conducted in 2023. This has a very strong impact on the credibility of the results.
In section 2.2; You must carefully justify the sample selection, why only four sites were chosen, why only 19 belts were selected, and how they were distributed to ensure there is no bias in the findings.

4-Results

There is not enough explanation of inaccuracy for LiDAR data, these factors are likely to be environmental factors (such as vegetation density) or technical issues (such as the accuracy of unsupervised classification).
In section 3.2 ; The relationship between diameter and the presence of hollow was reported to be statistically strong, but the ecological reason for this was not discussed.
The impact of the time gap between LiDAR data and field surveys (from 2009 to 2023) has not been enough discussed. It is possible to analyze whether some trees were removed or died during this period, which may lead to errors in data comparison. Or a technique to calculate the percentage of change in tree cover during that period using additional satellite images could be suggested.

4-Conclusion

After reviewing the MS, there is NO separate "Conclusion" section.

It is a serious scientific error: What does the absence of a conclusion section mean?

It leaves the reader without a clear and focused conclusion.
It diminishes the clarity of practical recommendations and prospects.

Author Response

For research article “Possibilities and limitations of a geospatial approach to refine habitat mapping for greater gliders (Petauroides spp.).”

Reviewer 2 Comments to Author:

- Abstract:

In line 16; you must explain this abbreviation “DBH”

This information has been added in the Abstract, line 19.

Why didn’t you mention enough details about the temporal and spatial data, as it was not clearly stated that the LiDAR data used is old (2009) while the field surveys were conducted in 2023. This time gap is very important, and it should have been mentioned in the abstract because it affects the accuracy of the results.

We have added context in the Abstract, lines 17 – 20.

You must mention the sample size (the number of sites surveyed), as these details will give the reader a better understanding of the power of the study and the generalizability of the results.

We agree, this information has been added in the Abstract, lines 21 – 23.

Why you didn’t indicate the statistical significance of the results as statistical values (such as p-values or RMSE) that reflect the accuracy of the results are not indicated, even briefly.

We agree, these details should have been in the Abstract and have now been added in for lines 27 – 30.

2-Introduction:

In the introduction, you should highlight the potential challenges of this study such as:

the difficulty of identifying hollow-bearing trees using LiDAR data alone,

We have expanded the introduction, section 1.0, to include a large review of research. This shows that potential hollow-bearing trees can be detected using LiDAR (lines 127 – 137), however, here we aim to see if LiDAR from 2009 can be used with a simple geospatial approach to identify potential hollow-bearing trees.

how the accuracy of the results is affected by the time interval between field surveys in 2023 and LiDAR data in 2009

We have added more context and an explanation for the use of 2009 LiDAR in section 4.0 lines 624 – 626, and justified in section 2.3, lines 306 – 311.

and the requirement for increasingly sophisticated methods, such as machine learning, to interpret the data more effectively.

We agree that this information was necessary and have made major revisions to the introduction where we have included more context about remote sensing and machine learning (section 1.0, lines 100 – 119).

Why did the introduction not adequately discuss the applications of the research study, as it focused only on the academic aspect, but did not explain how the research results can be applied to forest management policies or environmental protection initiatives.

Information has been added in the introduction and the justification of study, to explain that this research is needed as a transferrable and accessible approach for land managers and policymakers to make informed decisions more easily (section 1.0, lines 140 – 151).

The transition to the objectives was a bit abrupt, as there should have been a transitional paragraph summarizing why the current research was needed right before the objectives were presented.

We have edited this section to include more background information of why the research was needed, that the current habitat mapping for Queensland is too broad and should be refined (section 1.0, lines 168 – 184).

3-Methods

In line 91; Why you didn’t address the sources of uncertainty in the data, such as:

Discussing limitations of LiDAR data, such as the effect of vegetation density on the accuracy of trees detection.

We have added this inform in section 2.3, lines 328 – 335.

Analyzing how the quality of earlier data (2009) impacts the current findings.
In section 2.1 ; One of the most important weaknesses of the research is the use of old LiDAR data from 2009, despite field surveys being conducted in 2023. This has a very strong impact on the credibility of the results.

We have decided to place this information in section 4.0 to be discussed in depth. Information was added to this section about the impacts of LiDAR from 2009 from lines 627 – 646. We also note that aim of the study was to investigate if we could use old LiDAR to refine habitat mapping as current LiDAR is often expensive (section 1.0, lines 306 – 310).

In section 2.2; You must carefully justify the sample selection, why only four sites were chosen, why only 19 belts were selected, and how they were distributed to ensure there is no bias in the findings.

We agree that this information was missing, and it has now been added to justify the site selection in section 2.2, lines 261 – 268 and lines 288 – 291.

4-Results

There is not enough explanation of inaccuracy for LiDAR data, these factors are likely to be environmental factors (such as vegetation density) or technical issues (such as the accuracy of unsupervised classification).

We have added to and deepened this discussion in section 4.0, lines 630 – 665.

In section 3.2 ; The relationship between diameter and the presence of hollow was reported to be statistically strong, but the ecological reason for this was not discussed.

We have added information to address this in section 4.0, lines 668 – 681.

The impact of the time gap between LiDAR data and field surveys (from 2009 to 2023) has not been enough discussed. It is possible to analyze whether some trees were removed or died during this period, which may lead to errors in data comparison. Or a technique to calculate the percentage of change in tree cover during that period using additional satellite images could be suggested.

We have added in information about the threshold calculation and why it was necessary, to determine what would be considered a large tree for each ecosystem to account for change in the forest over time (section 2.3 lines 336 – 342). We have also added in recommendations for future work to incorporate logging and fire information, prior to site selection and the calculation of thresholds (section 4.0, lines 616 – 621).

4-Conclusion

After reviewing the MS, there is NO separate "Conclusion" section.

It is a serious scientific error: What does the absence of a conclusion section mean?

It leaves the reader without a clear and focused conclusion.
It diminishes the clarity of practical recommendations and prospects.

We strongly agree, and major revisions have been made to separate the results and discussion section to allow for a deeper discussion about the findings of our research, the context of our investigation in the wider research, and future recommendation. We have added a conclusion to finalise the views of our paper (section 5.0).

Reviewer 3 Report

Comments and Suggestions for Authors

This is a contribution to the ecology of well-known marsupial species. Specifically it deals with assessment of habitat quality of the Southern Great Glider.

The strength of the paper lays in its maticulous habitat analysis using remote-sensing method (LiDAR) and correlating tree hight with tree diamater and this in turn with hollow presence.

There are however few weak points:

Recenty the species P.volans has been split in three separate species: P.minor in the north-eastern Quensland, P. armillatus (the Central Great Glider) in the southern Queensland, and P.volans in SE Australia. So, there are three species, but the authors put somewhat confusing information in line 64-66. In the study are certainly there was only one glider species, viz. P.volans.
The study plots distribution in Australia is not clealry shown in Fig. 1. The plots are also not properly described. Glider's habitat prefernce depends on the availability of holes. This is however not the only habitat requsite. Humidity and species composition of tree stand aslo matter (preferred tree species: E.tereticornis, E.intermedia, E.cebra...), but we know nothing about the tree composition, neither humidity of any of the study plot.
Both in 'Introduction' and 'Discussion' there is a complete lack of any information on the abundance of the glider in the study plots; not only there are no population density estimations, but even general information on the abundance is totally lacking. Such information may well-support the LiDAR results.

Author Response

For research article “Possibilities and limitations of a geospatial approach to refine habitat mapping for greater gliders (Petauroides spp.).”

Reviewer 3 Comments to Author:

This is a contribution to the ecology of well-known marsupial species. Specifically it deals with assessment of habitat quality of the Southern Great Glider.

The strength of the paper lays in its maticulous habitat analysis using remote-sensing method (LiDAR) and correlating tree hight with tree diamater and this in turn with hollow presence.

There are however few weak points:

Recenty the species P.volans has been split in three separate species: P.minor in the north-eastern Quensland, P. armillatus (the Central Great Glider) in the southern Queensland, and P.volans in SE Australia. So, there are three species, but the authors put somewhat confusing information in line 64-66. In the study are certainly there was only one glider species, viz. P.volans.

Thank you for identifying this, the information has now been updated in section 1.0, lines 152 – 156.

The study plots distribution in Australia is not clealry shown in Fig. 1. The plots are also not properly described. Glider's habitat prefernce depends on the availability of holes. This is however not the only habitat requsite. Humidity and species composition of tree stand aslo matter (preferred tree species: E.tereticornis, E.intermedia, E.cebra...), but we know nothing about the tree composition, neither humidity of any of the study plot.

We have added in more information to give context about the site matrix, conditions, and preferred den and tree species in section 2.2, lines 265 – 287.

Both in 'Introduction' and 'Discussion' there is a complete lack of any information on the abundance of the glider in the study plots; not only there are no population density estimations, but even general information on the abundance is totally lacking. Such information may well-support the LiDAR results.

Thank you for pointing this out, we have not included this information because it has been included as a part of the current high-quality habitat mapping for greater gliders in Queensland. As the aim of the investigation was to refine the current habitat mapping by identifying potential hollow-bearing trees using LiDAR, we have accounted for the preferred den and feeding tree species utilised by greater gliders.