Citizen Science Data to Measure Human Use of Green Areas and Forests in European Cities
Round 1
Reviewer 1 Report
Interesting paper, ambitious & creative data gathering, analysis, graphic presentations, results and discussion are not too surprising, kind of what we would expect, but still good to have it empirically confirmed and to see how various nations compare to each other. Allows the higher use nations to brag about their accomplishments and lower use nations to perhaps be prodded into increasing green space/forests, or access to or awareness of them. Two slight grammar revisions to suggest: line 149 - - change 'forests' to 'forest' and line 355, delete the 'a' before bad access. This study is a form of 'unobtrusive research' methodology - - that could be mentioned somewhere. Another thought I have that probably should be mentioned in limitations is: the issue of representativeness - - I would think that people who participate in the iNaturalist app- - are people who are already 'tuned in' to nature and the environment - - - likely more so than the average citizen. I agree, that given the magnitude of observations included here, some of that might be averaged out, but probably not all of it. And, this same 'bias' is held constant across all countries in this study, so the comparative findings are probably not distorted much. But, it probably still worth a mention as a limitation. Something to be considered for future comparative studies.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Dear Authors,
the conclusions can be improved.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
This study aims to measure human use of green areas and forests using citizen science observations through the iNaturalist project in European cities. The research topic is interesting; however, the paper in the current form needs substantial revisions before publication is recommended for the Forests. Here are some recommendations for improving the quality of this paper:
1- Please clarify: (1) what quality categories of iNat data (Needs ID, Research Grade, etc.) were used in your study and (2) what types of species records (plants, animals, etc.) were included in your study? As far as I understood, you used the whole iNat’s database (existing for your study area) for this research, but please explain this more clearly.
2- My experience in using the iNat data says that sometimes (when the number of contributions is relatively low), the majority of observations from a green space may come from several enthusiastic naturalists. In this case, the number of observations may not necessarily mean that the green areas were visited by many people because, for example, it is highly accessible. Therefore, I am interested to know if you consider this issue in your study (by, for example aggregating all the contributions of an observer—user ID— in an area and consider all his/her observations as one observation)? Maybe at least you can discuss this issue in the discussion section of your study (if you find my comment valid).
3- I think there are some more potentially valid predictors that can be considered in this study; for example, some elements and characteristics of a green area, such as the existing facilities, activities, sceneries, biodiversity richness, or distance to residential or shopping areas etc. may explain the use of the green space too. Please at least discuss about other relevant potential predictors that were not considered in this study in the discussion section and provide some recommendations for future works.
4- I recommend you to provide the equations for the adopted predictors too (you currently provided them only in verbal form).
5-While different statistical learning methods can be used for the means of explanatory analysis, please explain more why did you use Poisson generalized mixed effect models in this study?
6-Line 176: Why did you conduct the correlation analysis among the selected predictors? If it is related to the negative issue of multicollinearity, you have discussed this in line 193.
7-Line 179: “ The variable "city population" was highly correlated to the number of observations”. I think this is not the correlation analysis among the selected predictors but among a predictor and target value. Please correct me if I am wrong.
8-Line 191: “Predictors with a skewed distribution were log-transformed, and all predictors were standardised prior to modelling.”. Please explain more about the reasons for these.
9-Figure 3: You only provided the abbreviation for the country names without providing the full forms of them. Please kindly provide the full forms of them too.
10- Section 4.3. Citizen Science —spatial—data quality has different dimensions: completeness, positional quality, thematic quality (if the species is identified correctly), etc. I think the most relevant issue to your work is the positional quality of an observation (if it belongs to the green space or is tagged incorrectly in the green space) (If you think the other aspects of the quality of crowdsourced data in Inat also may affect your study please also discuss about that). Please only discusses about the elements of quality that impacted your research in this section.
While most iNat users nowadays use GPS for tagging the location of the observations, still relatively few positional errors may occur in the observations (due to various issues: manual geotagging, GPS errors, etc.). You may refer to this paper for further information:
Vahidi, H., Klinkenberg, B., & Yan, W. (2018). Trust as a proxy indicator for intrinsic quality of Volunteered Geographic Information in biodiversity monitoring programs. GIScience & Remote Sensing, 55(4), 502-538.
This study also developed a fuzzy trust model for assuring the quality of crowdsourced records (positional and thematic quality) in citizen science biodiversity projects (particularly the iNaturalist). So, you can discuss that if we aim to omit such errors from the iNat’s crowdsourced data, it is possible to use such quality assurance methods.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Thank you very much for submitting the revised version of your paper. The quality of the paper has been improved, however, there are still some important issues (particularly from the previous round of review) that should be addressed more clearly before accepting the paper for publication. Hope the following comments can help to solve the remaining issues:
1-Line 133: In your new version, you mentioned that you used only the iNat's Needs ID data. Are you sure that you only used needs id data or it is a type? Just wondering why didn't you used other categories of iNat data (such as research grade) in your research?
2-Line 389-L390:"Indeed,one could argue that observations occurring in one green area or forest can derive from one single enthusiastic user of the mobile app, with a consequent bias.". What kind of bias do you mean?
3-L390 to L392: "The magnitude of the observations, the large geographic context as well as the possible consistency of this bias over the entire data set reduces the importance of this issue."
-the large geographic context? : What do you mean by this? Do you mean large geographic extent?
-"possible consistency of this bias over the entire data" Please clarify what do you mean by this?
4- L393-394: "much of our analysis and discussion focused on urban areas with a minimum number of observations": What do you mean by "minimum number of observations" (maybe you referred to this using another term earlier in your article--but please use a consistent term)?
5-L394-398: "Nevertheless,the 394identity of the single users should be considered in future studies to account for the bias 395brought by participants with high frequency of observationsand this possible bias could 396be overcome by integrating citizen science data fromavailablegeotagged social media 397data.". I think one can also use the user ID of iNat to somehow solve this issue. Please see my comment in the previous round if you are interested in this.
6- L398-400: "Finally, it is important to recall that data on spatial location could have positional errors but this seems to be less likely and important when focusing on well-defined patches, as green spaces, and not on specific species location [81].". The reference No. 81 you used, not discussing about "but this seems to be less likely and important when focusing on well-defined patches, as green spaces, and not on specific species location"
-So, I suggest you to refer to this paper in this way if you find it interesting:
Finally, it is important to recall that data on spatial location could have positional errors [81] but this seems to be less likely and important when focusing on well-defined patches, as green spaces, and not on specific species location.
---More specifically as far as I mentioned this paper is exclusively is about the locational errors in crowdsourced biodiversity observations (such as iNat data---which is the issue of your research) so I suggest you be more clear in the first part of your sentence:
Finally, it is important to recall that the crowdsourced biodiversity observations could have positional errors [81]............
7-L331 "higher number of species, or portion of public urban parks": What do you mean by portion of public urban parks?
8-332-334: "These types of data are usually not available for the majority of the urban areas considered with the spatial detail required in this study" . I suggest you to modify your sentence like this (please feel free to object or improve it): "Some of these data may not available for some of the urban areas....."
9- I personally believe your findings at worst (when there is a limited number of observations in a city) shows the use of "iNat users" from green spaces and forests and at the best (when the observations are high enough in an area or city by various observers) it "could" help to measure human (citizens from different categories) use of green areas and forests in European cities ( as one may claim that when the number of observations from various people increases, it improves the representativeness of the data). So, still, I think it is needed to take care when over-generalizing the results of this study. If you agree on this please briefly and clearly discuss this important issue in your discussion section.
Author Response
Please see the attachment
Author Response File: Author Response.docx