Associations between Road Density, Urban Forest Landscapes, and Structural-Taxonomic Attributes in Northeastern China: Decoupling and Implications
Round 1
Reviewer 1 Report
The manuscript looks to be improved essentially if comparing with the previous version. Some minor comments:
Short information on how the paper is organized, as the last paragraph of Introduction, could guide better the reader though the text. I.e. what is the structure of the paper? I find such information used in some MDPI journals same important as, e.g. the abstract, to decide whether it is worth further to read the paper.
Not sure whether the processing framework should be introduced in Fig 1. In principle, this part of illustration improves the introduction of methodology very much, however, it could be improved, especially as it regards the style and some formatting issues. Also, I still consider that such general statements, like “remote sensing technology” (too general), “grid method” (did anybody announce the fishnet function as a method), “partition statistics method of ArcGIS” (same comment as before), etc. You manipulate the commercial software names very much associating them with methodological approaches. Could one do the same using other GIS package than ArcGIS – I believe, that yes. Please, consider improving the flowchart to contain just the information of specific approaches used in your study and reduce long texts in the boxes. I would say, that the overall processing framework could be introduced immediately after the introduction of study area.
Could not understand “… road identification was manually interpreted” (line 136), “… after introduction into ArcGIS…” (line 137), “were performed with ArcGIS map”, “Object-based accuracy assessment” (line 194), “regression analysis (maybe the results?) of road density with all landscape metrics are showed …” (line 281 and 805), “landscape regulations” (line 577) and many more... – much because of your terminology used.
Not sure about the journal policies, but do not the abbreviations be introduced in the text, even though they are already introduced in Abstract (e.g. ISA).
“A pre-experiment of this paper has shown that … “ (line 156 and afterwards) does not look convincing nor scientific. Maybe the sampling could be shortly discussed in Discussion, if you consider this to be needed).
Suggestion to revise the style of text, focusing on technical aspects.
Author Response
Thank you very much for your positive suggestion a revision of our manuscript. We have given a response point by point in Attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Thank you for your revised version and the extensive processing and presentation of the arguments.
I totally agree to your revised version. However in some parts you need to improve the presentation of results. In particular it is necessary to present the data inspection in more detail. Please add several relevant diagrams to your appendix. It is necessary to show the raw data distribution and the transformed data distribution. I do not believe your sentence: "We found several parameter was not normal distribution and log transformation made the data bevome normal distribution". Please add qq-plots both for the raw data and the transformed data. Using the QQline and the exact analysis of the distribution of the transformed data, the possibilities for regression application can be discussed. When you are sure about your statement of the log-transformed data please add in addition some tests for normal distribution.
Furthermore i suggest to present a test for multicollinearity when selecting your final dataset for regression analysis.
As already mentioned in my first review I also call for a better description of regression diagnostics. Please check existing r-libraries. Here you will find the standard procedure to investigate your regression analysis. Please make sure to check all model premise: homoscedasticity/heteroscedasticity/Independence of the residuals/Residuals vs. estimated values//Residuals vs. estimated values/Normal distribution of residuals/Incomplete and non-linear model. Please also present the mean error for estimating the dependent variable. All these measurements are important to judge about the quality of your results. Only on this basis can a final decision be made as to whether the regression analysis is applicable and whether the diagnostic results justify publication. I don't want to decide about black box results.
Of course I would like to praise this work and also welcome the approach. But I need more detailed information for a final judgement.
Author Response
Thank you very much for your positive suggestion a revision of our manuscript. We have given a response point by point in attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The paper was improved in terms of data inspection and regression diagnostics.
