How Do Urban Parks Provide Bird Habitats and Birdwatching Service? Evidence from Beijing, China

Round 1

Reviewer 1 Report

The manuscript investigates variables that promote the quality of the parks for birds and attract birdwatchers in one of the largest cities in the world with very high population density. The manuscript is well written and provides a pleasant reading. The only part that needs to be clearer are the methods, as I point out below. These clarifications are very important for the understanding of the results and discussion. Figures and tables also need to be improved.

Introduction

The introduction is well written and encourages the reader to continue in the text.

L84-87: These questions hindered the understanding of the central questions in the manuscript. I recommend authors to keep only the questions that will be evaluated in the manuscript. It is also important to include a central objective of the manuscript, rewrite specific questions and include predictions.

Methods

L137: strange sentence. Rewrite.

L174: explain better how dendrogram trait was done. For example, species × traits matrix was submitted to a hierarchical grouping procedure using UPGMA?

All method steps are very confusing. Although wording is clear, it is difficult to understand the link between the analyzes - why each analysis was done? When I get to the statistical analysis section, I don't understand where the response and predictor variables came from. There is a lack of connection between what was done and why it was done. For example: why was the dendrogram made? What kind of results does it generate to be used in the analysis? Where the bird census and birdwatching record were used? In table 2, why does the number of parks change so much? Why do you have analyzed that used Pearson's correlation and other stepwise regressions? Perhaps due to the large number of predictor variables?

Is it not clear what is a suitable habitat for birds? The way it was calculated is very confusing.

Make it clear what are the predictor variables and responses variables used in the manuscript. In the discussion I have the feeling that I did not understand what the variables were tested. For example, in some analyzes it seems that bird suitable habitat is the predictor variable and in others it is the response variable. This is confusing if it is not well explained so that the reader does not get lost throughout the manuscript.

Were the independent variables in Table S2 correlated? It is important to test collinearity between predictor variables in order not to influence the analyzes.

It is not clear how Anova was conducted. What are the response variables and what are the predictors? What does pairwise comparisons mean in an ANOVA as described in figure 4?

I was able to understand a little more of the analysis in the results. However, the analyzes need to be clear in the methods, which is the session designed to explain how the study was done.

It remains to clarify important points for the reader to be able to replicate his study. Without these clarifications, it is difficult to assess whether statistical analyzes are the most appropriate.

All figures and tables legends must be rewritten to improve the description of their meaning.

Table 1: The types of habitat that dominate each sampling point are not explained in the methods. I don't know how they were used in the manuscript.

The quality of figure 4 is very low. Need to improve. I think only figures 4c and 4d are necessary to understand the result. Include the regression line in Figure 4d.

Table 3 needs to be reformatted completely. Remove “Dependent variable” from the second column. Remove R2, it already appears in the text. Include in the legend the meaning of each independent variable - difficult to link with the variables in Table S2, the names are different...

Discussion

L265-275: The parks mentioned in the discussion are not common for researchers from other places than China or Beijing. So, it would be important to point out where these parks are on the map - perhaps in Figure 4.

L291-293: The analyzes consider only variables inside the parks. But over the course of the discussion, variables such as closer to the city center, roads, bus stops and nearby metro stations, seem to have been evaluated. I really get confused if these variables were or not used in the analyzes. The analyzes need to be clearer.

L300-301: “Another explanation is that birds recorded by birdwatchers in these four parks were from other nearby places” – explain a little better the reason for this statement. Why could this have happened?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors proposed to examine the complex dynamics among parks, birds and
84 birdwatchers. The manuscript was well written, the methodology was well described and results and discussions clearly stated. 

Author Response

Thanks a lot for your comments. We are very much encouraged.

Reviewer 3 Report

The manuscript on “How do urban parks provide bird habitat and birdwatching service: Evidence from Beijing, China,” is an interesting study into the use of remote sensing data to assess bird habitat distribution and suitability in urban parks. Although this is an interesting study, I have few major concerns. I recommend major revisions should this manuscript to be considered for publication in remote sensing.

1. First of all the authors have used commonly applied remote sensing image classification and spatial analysis techniques without any further innovations or modifications to them. Thus, I think that this study lacks originality/novelty, especially in terms of remote sensing techniques, which I believe is the primary focus of MDPI Remote sensing journal. I think this study doesn’t provide any significant contribution to the remote sensing field and/or its readership. Thus, would be more suitable for a journal with more ecology focus. I suggest the authors to be more innovative with their remote sensing data analysis.

2. The introduction is too general.

1. It doesn’t include the research gap or how the authors are planning to use innovative methods to address the research gap.
2. What do you mean by “whole picture?”

3. The methodology is not very clear.

1. There are no specific details about the vegetation types (Composition, structure etc.) of the parks
2. No details on remote sensing data, e.g., the source of DEM, details of satellite data etc.,
3. You have field survey from 2017, bird watching data 2009-2013, and satellite data from 2015, 2016. None of the data acquisitions overlap with each other. Hasn’t there been any impact of vegetation structural, composition and/or spatial distribution changes over these years?
4. How do you justify the gaps in different data acquisitions and it impacts your result?
5. No detailed description of the remote sensing data analysis/image classification in the methodology part. What is the originality of the method you used?
6. I think vegetation structural details (at least for the canopy layer) should be included as it is one of the most critical aspects that influences bird habitats
7. Table 2 and statistical analysis: It is not clear to me why you had chosen different sample sizes for different analyses. You should clearly mention the rationale behind the different sample sizes in further analysis steps.

4. In my opinion, the results/findings are not that significant. The discussion is too general.
   1. Lines 307-323: Too general statements and in my opinion, doesn’t add anything new to what we already know in this arena.
   2. Pls do a thorough lit review on your topic and you will find hundreds of published work that already have seen the same findings throughout the world. I suggest the authors look for more significant (and more specific) research issues that can be answered with your data and some innovative analyses.
   3. Consider revising your discussion to be more comprehensive.
   4. What are the specific future implications of your study and its findings?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

How do urban parks provide bird habitat and birdwatching service: Evidence from Beijing, China by Zhengkai Zhang and Ganlin Huang

General:

Data from a bird census survey (at 159 sample in seven out of 102 parks in Beijing obtained during the 2017 breeding season from March to July) points are subjected to statistical analysis to obtain information on the spatial patterns of parks suitable for bird habitats (and birdwatching) and the factors affecting suitable habitats. Methods are briefly introduced and results are discussed. This paper is an interesting contribution fitting the journal. Some points are raised to provide sufficient background for an interested reader who is not specializing in ecology.

Specific:

1. Introduction:

L69 etc: “Vegetation remote sensing has been used in various fields, such as vegetation cover classification 69 [37-39], land surface temperature pattern analysis [40-44], species habitat characteristics [45-46] or identification [47-51], and ecosystem services assessment [52-53]. Predicting the habitat of the species and estimating the quality of the habitat is necessary to urban planning, green space management, and biodiversity protection. However, it was seldomly used in urban area.”

For this overview the authors may not be aware of a more general embedding of this subject (see for example Cai et al 2019: Causality of Biodiversity Loss: Climate, Vegetation, and Urbanization in China and America. Sensors 19)

2. Materials and methods

Modelling: …We first identified the breeding birds according to the bird checklist in Beijing [66]. Then we drew a trait dendrogram according to body length, body weight, nest location and nest type in RTM program [75-76] and divided breeding birds into sixteen functional guilds (Figure 3). Finally, we generated a presence-absence data set at each observation site for the functional guilds with more than 14 samples, which was required by the MaxEnt model as the minimum sample size. Plus ANOVA…

Here, several methods are being introduced, for which a not too well informed reader may get a brief but concise description. May be, a set of related mini-examples can be helpful (even as Appendix).

3. Results:

L207, L188: AUC ? better

L222: Our results showed that the bird suitable habitat area was significantly associated with the number of birdwatching records (r=0.828, p<0.01, n=22) (Figure 4d).

Is that surprising?

4. Discussion:

L253: Results from the one-way ANOVA analysis revealed significant differences in park size between parks with and without birdwatching record (Table 4, Figure 5). First, parks with birdwatching record were significantly larger than their counterparts. After the effect of park size was considered, results showed that parks with birdwatching record were older, closer to city center, and more accessible than those have no reported birdwatching.

Check my suggestions in italic.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 5 Report

Line 129: Indicates a 25m radius of a water body, is this not problematic if LANDSAT has a 30m pixel?

Line 134: Figure 2. Separate the legend Sample park and another park from the rest as it is now is more confusing.

Line 140-155: In section 2.2.2. It is not clear why to use the Pleiades and then use LANSAT, which has a worse spatial resolution. It is also not well explained because it uses only seven sampling parks and the relationship with the other 102.

Lines 173-178: Clarify where and how the 14 samples. And the incorporation of the variables from Table S3.

Line 291-293: The statement on accessibility, despite the statistical study confirming it is debatable. The birds are linked to the parks' larger size (the authors indicate this several times as in Lines 312-314) and to older forests. Roads and subway buses have developed in recent decades, so the evolution of habitats is older. Ornithologists go where they are more likely to find birds, not always because there are more communications, but rather as the park is larger, it has more birds and species. The higher density of metro-buses and more roads also has to do with the buffer. The larger parks have more buffer surface around than the smaller ones. They need more connections, even the roads expressed in density (I suppose that linear kilometers per unit area). More kilometers have been built around the largest parks because they represent a more significant obstacle that breaks the city's connections. These communication routes cannot cross it, so the solution is to increase the communication routes around to link opposite neighborhoods. However, this would require other studies.

Bird watcher data oversample the areas where it is more likely to locate species, the more exotic and rare, so care must be taken with data collected this way (the authors make some reference in Lines 303-305. A factor related to this can also explain situations such as Lines 316-323, which is statistically correct. However, has the possibility of certain rare or exotic species that attract more observers been verified or eliminated in that park? If it is easy to add, then incorporate it into this study. If it is not, consider the possibility of eliminating these factors with a future study.

Line 329-330: Finally! From the beginning of reading the work, I wondered about the subject of Pleiades 0.5 m and LANDSAT 30 m resolution, and now it is explained. Couldn't they have put this at the beginning in the materials-method chapter? Still, why use the Pleiades?

Line 332: Confused by the signs in parentheses.

Lines 338-341: The limitations of images in remote sensing concerning the 3D structure can be solved in part by incorporating LIDAR data (I do not know if they have them, in those areas, it may be interesting to complement in future studies).    

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Authors have addressed most of my comments and the manuscript has improved significantly. 

About my question no. 6 and your answer to it. I think it would be better if you add your own statement as an assumption. Something like, "land cover and vegetation structure in parks did not experience significant change over recent years. Therefore we assume that the time gap between different data acquisitions had no significant impact on the results of the study"

Author Response

Thanks for your comments. We added a statement in L155-157 to clarify the time gap between difference datasets used in this study.

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.