Potential Range Map Dataset of Indian Birds

Round 1

Reviewer 1 Report

The authors have collected a huge data set of bird observations from different internet-based databases and have modelled the distribution of 968 bird species in India. I have checked the most important steps in preparing the data and performing the model. Here are my comments:

Data set and preparation

The authors have used data from open-access citizen science platforms (eBird and GBIF). Generally citizen science data sets come with a number of downsides. However, the authors have done good job by mentioning the downsides in the text and by taking them into account during the data preparation:

- double count were removed;

- spatially weighted data cleaning was applied to the data set;

- very rare species with few observations and occurring in limited geographical area were also removed.

Modelling method

The authors have used MaxEnt, which in my view is appropriate as this method is widely used for such kinds of modelling taks. The method helps creating a good model in cases were reliable absence points are not available.

Model validation

The authors have applied appropriate validation analysis and have provided the statistics in the supplementary materials.

What I miss is probably a summary map of India with a raster showing the accumulated number of species per raster. If the journal’s policy allows putting such summary map in the text summary of the data set, I would highly recommend doing it.

Eventhough, the model performance statistics look great I would recommend to include a paragraph where the authors describe the limitations of their modelled maps. Modelling birds’ distribution at continental scale using citizen science data with a “presence-only data approach” and only climatic variables generally results in coarse-scale resolution maps. I think the high AUC values are proving just that. Therefore, I would suggest to the authors to list shortly the occasions in which their maps are very useful and the once which are less suitable (for instance local scale planing – estimation of the potential impact on the bird community by a local infrastructure project).

I added few comments/correction directly in the text (in the PDF-file).

I am satisfied with the quality of the paper and would like to thank the authors for their effort!

I would suggest to publish the paper after taking my comments into account.

Comments for author File: Comments.pdf

Few minor corrections are needed - I inserted my comments directly in the PDF file "data-2573887-peer-review-v1_reviewed-14082023.pdf"

Author Response

[ Comment 1] The authors have collected a huge data set of bird observations from different internet-based databases and have modelled the distribution of 968 bird species in India. I have checked the most important steps in preparing the data and performing the model. Here are my comments:

Data set and preparation

The authors have used data from open-access citizen science platforms (eBird and GBIF). Generally citizen science data sets come with a number of downsides. However, the authors have done good job by mentioning the downsides in the text and by taking them into account during the data preparation:

- double count were removed;

- spatially weighted data cleaning was applied to the data set;

- very rare species with few observations and occurring in limited geographical area were also removed.

Modelling method

The authors have used MaxEnt, which in my view is appropriate as this method is widely used for such kinds of modelling taks. The method helps creating a good model in cases were reliable absence points are not available.

Model validation

The authors have applied appropriate validation analysis and have provided the statistics in the supplementary materials.

[ Response 1] The authors are thankful for your acknowledgment of our paper, along with your thorough assessment that included favourable observations and considerate remarks about our research. We highly value your input.

[ Comment 2] Eventhough, the model performance statistics look great I would recommend to include a paragraph where the authors describe the limitations of their modelled maps. Modelling birds’ distribution at continental scale using citizen science data with a “presence-only data approach” and only climatic variables generally results in coarse-scale resolution maps. I think the high AUC values are proving just that. Therefore, I would suggest to the authors to list shortly the occasions in which their maps are very useful and the once which are less suitable (for instance local scale planing – estimation of the potential impact on the bird community by a local infrastructure project).

[ Response 2]  We have corrected the manuscript specifying limitation of dataset. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 3] I added few comments/correction directly in the text (in the PDF-file).

[ Response 3]  We have corrected the manuscript as per the comments. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 4] I am satisfied with the quality of the paper and would like to thank the authors for their effort!

I would suggest to publish the paper after taking my comments into account.

[ Response 4] The authors are thankful for your acknowledgment of our paper, along with your thorough assessment. We highly value your input.

Reviewer 2 Report

I think authors could have included some examples of the models obtained, selectig a couple of species with different datasets, to show in every case the final product to have a visual idea of the performance of MaxEnt.

Also, I would like authors include a pagraph warning the users of this data base produced, in some cases or those cases with possible more uncertainty.

Author Response

[ Comment 1] I think authors could have included some examples of the models obtained, selectig a couple of species with different datasets, to show in every case the final product to have a visual idea of the performance of MaxEnt.

[ Response 1]  We have corrected the manuscript by an example with example map. For more information, please see the Fig. 1 and revised text (highlighted in yellow) in the manuscript.

[ Comment 2]  Also, I would like authors include a pagraph warning the users of this data base produced, in some cases or those cases with possible more uncertainty.

[ Response 2]  We have corrected the manuscript specifying limitation and future direction of dataset. For more information, please see the revised text (highlighted in yellow) in the manuscript.

Reviewer 3 Report

Dear Authors,

     This study on bird distribution in India is interesting, but there is very little information and few comparisons with similar analyses from around the world. Even in India, there are some species distribution models (SDMs) available for some bird species. However, the study does not present the bird distribution maps for further analysis or studies.

Abstract:

   Nowadays, distribution maps are not limited to expert opinions and theoretical models. There are many SDMs available on regional and global scales. This assumption needs to be revised, as it is not entirely accurate, even for Asian and African countries.

Summary:

   There are numerous SDMs in biology, for birds and other species. Have you reviewed these articles? Can you compare your study with them? Even the European Bird Breeding Atlas, 2nd edition, utilizes SDMs. Many national Bird Breeding Atlases also employ SDMs.

Methods:

   Previously, the authors mentioned that most of the maps were based on expert opinions and theoretical models. However, here they use as main reference BirdLife International and Handbook of the Birds of the World 2020. Could this method be a limitation for your models? It might exclude new locations for species that are expanding.

   Some of the observations in GBIF had photos. Did you verify the species ID using these photos? These photos can help exclude misidentified species.

   Have you checked the resolution of eBird data? Sometimes the coordinates are not entirely precise.

   Your variables include data from other sources besides WorldClim data. You should mention these additional sources as well.

   In Supplementary Table S1, habitat information is provided. Was this variable used for SDMs? There are no references for this in the Methods section.

   There are no distribution maps presented for this analysis in the main text or supplementary materials.

Line 70: What is the meaning of this number: 2,84,37,430?

Line 89: What does this number: 19,12,725, represent?

Best regards,

Author Response

[ Comment 1] This study on bird distribution in India is interesting, but there is very little information and few comparisons with similar analyses from around the world. Even in India, there are some species distribution models (SDMs) available for some bird species. However, the study does not present the bird distribution maps for further analysis or studies.

[ Response 1]  We have corrected the manuscript with SDM studies in India for different taxa groups. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 2]  Abstract:

   Nowadays, distribution maps are not limited to expert opinions and theoretical models. There are many SDMs available on regional and global scales. This assumption needs to be revised, as it is not entirely accurate, even for Asian and African countries.

[ Response 2]  We have revised the manuscript based on your observation. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 3]  Summary:

   There are numerous SDMs in biology, for birds and other species. Have you reviewed these articles? Can you compare your study with them? Even the European Bird Breeding Atlas, 2nd edition, utilizes SDMs. Many national Bird Breeding Atlases also employ SDMs.

[ Response 3]  We have corrected the manuscript with SDM studies in India for different taxa groups. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 4]  Methods:

   Previously, the authors mentioned that most of the maps were based on expert opinions and theoretical models. However, here they use as main reference BirdLife International and Handbook of the Birds of the World 2020. Could this method be a limitation for your models? It might exclude new locations for species that are expanding.

[ Response 4]  We have revised the manuscript with correct wording. We used Birdlife maps for filtering wrong record and kept genuine vagrant or range expanding species records. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 5]  Some of the observations in GBIF had photos. Did you verify the species ID using these photos? These photos can help exclude misidentified species.

[ Response 5]  We have revised the manuscript. We have only used research grade observations curated by experts on the various citizen science platforms. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 6]  Have you checked the resolution of eBird data? Sometimes the coordinates are not entirely precise.

[ Response 6]  We have revised the manuscript. We have removed, low precision coordinates. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 7]  Your variables include data from other sources besides WorldClim data. You should mention these additional sources as well.

[ Response 7]  We have mentioned the additional variables used in line 120 and 121. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 8]  In Supplementary Table S1, habitat information is provided. Was this variable used for SDMs? There are no references for this in the Methods section.

[ Response 8]  We have not used habitat information provided in Supplementary Table S1, We kept this details for further analysis.

[ Comment 9]  There are no distribution maps presented for this analysis in the main text or supplementary materials.

[ Response 9]  We have corrected the manuscript by an example with example map. For more information, please see the Fig. 1 and revised text (highlighted in yellow) in the manuscript. Rest of the data is available at DOI: 10.5281/zenodo.8221113 once the paper is accepted.

[ Comment 10]  Line 70: What is the meaning of this number: 2,84,37,430?

[ Response 10]  in Line 70 the number 2,84,37,430 indicates exact presence records collected from citizen science platforms. We have revised the manuscript. For more information, please see the revised text (highlighted in yellow) in the manuscript.

[ Comment 11]  Line 89: What does this number: 19,12,725, represent?

[ Response 11]  in Line 89 the number 19,12,725 indicates exact analysis ready presence records after data cleaning procedures. We have revised the manuscript. For more information, please see the revised text (highlighted in yellow) in the manuscript.

Round 2

Reviewer 3 Report

Dear authors,

Thank you for the improved version of the manuscript. It would be helpful to provide a few examples (5-7 species), rather than just one, in your analysis. These examples should showcase some extreme differences when compared to BirdLife's existing maps. This would better illustrate the importance of your work.

Best regards,

Author Response

[ Comment 1] Dear authors,

Thank you for the improved version of the manuscript. It would be helpful to provide a few examples (5-7 species), rather than just one, in your analysis. These examples should showcase some extreme differences when compared to BirdLife's existing maps. This would better illustrate the importance of your work.

Best regards,

[ Response 1] The authors are thankful for your acknowledgment of our paper, along with your thorough assessment that included favourable observations and considerate remarks about our research. We highly value your input.

We have included additional 4 examples to show variation in their ranges considering criteria like range restricted species, wide spread common species and difficult to detect species like owls.