Prediction of Potential Habitat for Korean Endemic Firefly, Luciola unmunsana Doi, 1931 (Coleoptera: Lampyridae), Using Species Distribution Models

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The introduction underlines the need of Luciola unmunsana in South Korea and offers a general background for the ecological value of fireflies. The language is sometimes expressive, though, especially in lines 44, 35, 418—"emotional insects"—which, although culturally relevant, might compromise scientific objectivity. A more objective tone emphasizing ecological and conservation value instead of emotional or cultural qualities would help the introduction. To keep scientific neutrality, the writers should edit lines 44, 35, and 418 to eliminate or reinterpret "emotional insect" and related words.

Although the justification for emphasizing L. unmunsana is well-established, the introduction lacks a critical evaluation of past species distribution modeling (SDM) efforts for fireflies globally. The dearth of studies in South Korea (lines 61, 88, 81) is overemphasized without enough context for the study within the larger international scene. To support their method, the authors should include a quick synopsis of worldwide SDM applications for fireflies or related taxa in lines 61–62 and 88.

Usually, the approaches are rather well described with unambiguous data source and modeling approach explanations. Several important points, though, call for revision or explanation:

39 occurrence points—lines 112–113—are used in the study, a rather small sample for national-scale SDMs. The authors note the small sample size but do not address how this might affect overfitting or model robustness. Lines 112–113 and 424–427 should feature a discussion of the limits of small sample sizes and possible spatial bias in data collecting.

Environmental Variables: We go in great length on the choice and handling of environmental variables. But the use of variables like EVI from 2022 (line 180) models possible habitat based on climate data from 1981–2010, so creating a temporal mismatch. If land cover or vegetation has changed dramatically between these times, this could skew the outcomes. Either by defending the method or, better still, by harmonizing the temporal scope of all variables, the writers must address the temporal inconsistency in lines 125 and 180.

The MaxEnt model was tuned using several feature classes and regularization multipliers (lines 203–207), but the justification for the ultimate parameter choices is not quite evident. To prevent overfitting, the writers should go into more particular on how model selection was done and whether model complexity was assessed (lines 203–207).

For the ensemble model, 1,000 pseudo-absence points (line 226) are generated; the technique for choosing these points is not stated. The writers should specify how pseudo-absences were produced to prevent sampling bias or spatial autocorrelation (line 226).

The results are presented clearly and with suitable use of tables and statistics. Still, several problems need attention:

Although there is no discussion of possible overfitting or the restrictions of depending just on AUC, especially with small sample sizes, the AUC values for both models are reported as "good" or "excellent" (lines 306–307, 315–316). The writers should go over overfitting's risk and give lines 306–307 and 315–316 some thought on reporting extra metrics (such sensitivity, specificity).

The results show as main predictors EVI, hydrological proximity, land cover, and annual precipitation: lines 284–286, 298–300 highlight these aspects. The way response curves (lines 289–294, 367–371) are interpreted, though, is occasionally erratic. The debate, for instance, implies that the reaction to water network analysis rises with distance from water (line 369), so contradicting the theory that proximity to water is advantageous. In lines 289–294 and 367–371 the writers should make clear the directionality and ecological interpretation of these response curves.

The paper claims that field visits verified the presence of L. unmunsana in expected habitats (lines 29–31, 407–410), but it offers no quantitative assessment or methodological detail. Lines 29–31, 407–410 should have the authors state the number of sites visited, the method of confirming presence, and whether absence data were gathered.

Though several points need work, the overall structure of the discussion is rather good:

The authors recognize that MaxEnt could cause overestimation (lines 379–380), but this is not particularly discussed in light of the small sample size and possible spatial bias? To more precisely address these constraints, the conversation should stretch from lines 379–380 and from lines 424–427.

Though it does not address how these predictions should be interpreted or validated before being used for management decisions, the paper asserts that the results provide "basic data for the conservation and usage" of L. unmunsana (lines 35–36, 421–423). Before using these findings to conservation planning, the authors should include a warning note in lines 35–36 and 421–423 regarding the need of more field validation.

Though significant for public involvement, the repeated emphasis on the cultural/emotional value of the species can subtly skew the scientific rigor of the conversation (lines 35, 418). These parts should be changed by the writers to concentrate on ecological and conservation relevance.

Conclusions
The findings and importance of the research are mainly restated in the conclusions. They overstate, though, the dependability and applicability of the models considering their acknowledged constraints. In lines 415–416, 421–423, and 430–431 the writers should temper their assertions by clearly noting the need of more data collecting, higher-resolution variables, and extra field validation.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The topic is imprtant, and there is need for studies on the ecology of endemic species such as Luciola unmunsana, however I strongly advise on applying the following cahnges:

1. I suggest the following new title: Prediction of potential habitat for the Korean endemic firefly, Luciola unmunsana Doi, 1931(Coleoptera: Lampyridae) using species distribution models
2. In the abstract: it starts with: “This study aimed to predict the potential habitats of…”, but it should be structured as follows: brief background information about the species; reasons why such research is needed; aim; very brief description of material (number of occurrences, ext.), methods; results; comment on results.
3. Also in the abstract, edit, the sentence: “Among the input variables, the ecoclimate index built through the Shared Socioeconomic Pathways (SSP) scenario-based detailed climate change data was utilized for climate variables, and non-climate variables were built to reflect the ecological characteristics of Luciola unmunsana, such as topography, land cover, and Enhanced Veg etation Index (EVI)”, the meaning in unclear, probably some part is missing, but also, try to split it in shorter sentences.
4. In the abstract, the phrase “…to be influential in predicting….”, sound confusing too, may be good predictors is better suited here.
5. In Materials and Methods: “Appearance point data”, I suggest changing it to “Occurrence data”, and in the whole text not only here.
6. In Materials and Methods: the thurm “construct”/”constructed” (and its derivatives) is used in many places the text, but in most cases it is not the most appropriate thurm. Please read carefully and substitute it with the appropriate words. Here is an example of unclear expression with repetitions of “construct”: Therefore, a Digital Elevation Model (DEM) with a resolution of 90m×90m was constructed in CGIAR-CSI [34] to construct non-climatic variables affecting the habitat of Luciola unmunsana, followed by slope and shade gradient analysis, and a water network analysis map was constructed using the Environmental Big Data Platform [35] to construct variables on distance to water systems.”.
7. In method section: “The maximum number of background points was set to 10,000, which was typically set to 10,000 if there were more than 10,000 background points [58]”, the sentence is incomplete, and unclear.
8. In method section, about the pseudoabsence points - choosing pseudoabsence points within buffers around occurrences is advisable when working with species expanding their areal of distribution. I strongly recommend implementing this approach in your study. For detailed explanation of the reasons for doing so, please see “Soley‐Guardia, M., Alvarado‐Serrano, D.F. and Anderson, R.P., 2024. Top ten hazards to avoid when modeling species distributions: a didactic guide of assumptions, problems, and recommendations. Ecography, 2024(4), p.e06852.”.
9. Include a table listing all of the environmental variables you use in you analysis, with the respective resolution of the layars you abtained for it, and the respective sources of the data (web site or publication…). Mark in the table how did you use the data for the variable – in building the models or in the analysis after the models were built?
10. In the results section: give the TSS for the MaxEnt model
11. Also in results: “When analyzing the contribution of each variable in the model, the land cover map contributed the most (26%)”, but the map is not variable, correct to land use or land cover variables.
12. “Among the response curves of each variable, the EVI, land cover, water network analysis, and Bio12 were the most important (Table 3)”, please edit, since the response curves cannot be important, while variables can.
13. In results: “In addition, the response curve of the land-cover map…” change “map” to “variables”.
14. Re-phrase the sentence: The importance of each variable in the built model was the highest for EVI (39.3%) followed by water network analysis (25.6%), Bio12 (10.1%), and land cover (8.3% (Table C2).”
15. The thurms “validation” and “verification” are not used correctly, they are not interchangeable, so carefully check trough the text and use them in accordance with their definitions.
16. The meaning of the following is unclear: “The variables with the highest importance in both models were the EVI, hydrographic network analysis, land-cover maps, and Bio12. Therefore, the potential habitat area predicted by the model was analyzed by overlaying the key variables with the predicted potential habitat area using the model with the key variables with the highest importance in both models.”
17. The sections from 3.4.1 to 3.4.3 seam redundant - the analysis by overlap does not give new information, since there are response curves for each of the tree variables used for the overlapping. What new information does this sections give?

In Conclusions: 1) the whole section repats the results and does not summarise them, shorten the section or remove it; There is this part  in the conclusions: “Third, field visits and literature surveys of sites predicted as potential habitats, but not existing sites, such as Geumsan-gun, Chungcheongnam-do, Yeongam-gun, Jeollanam-do, Mudongsan Mountain, Gwangju-si, Gwangju, and Gijang-gun, Busan, confirmed the occurrence of Luciola unmunsana.” But there is no mention of field trips in the methods section, or materials collected by the authors. Describe the field survey in the method section.
Kind regards

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for addressing all the comments 

Author Response

Thank you so much for your meaningful review of our work, your review has improved the quality of our paper.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been significantely, improved. There are only a few sentences which still have confusing structure and need editing:
1)"Among the input variables, the climate variable reflected Luciola unmunsana's ecological characteristics and used the shared sociocultural pathways (SSP) scenario-based ecological climate index, while the non-climate variable used topography, land cover maps, and the Enhanced Vegetation Index (EVI)."
2)  This study is meaningful in establishing the national SDM for the first time for Luciola unmunsana, which is decreasing due to indus trialization and urbanization, and predicting potential habitats by applying various envi ronmental variables reflecting ecological characteristics. 
3)"These studies also considered fireflies as environ mental indicator species and attempted to be used to establish habitat characteristics and conservation areas. Based on this approach, this study analyzed Luciola unmunsana, a Korean species."
Kind regards

Author Response

Thank you very much for your thoughtful comments.

We have revised the three paragraphs you mentioned as follows to convey the content more accurately. Thank you.

 

• Among the input variables, climate variables were based on the Shared Socioeconomic Pathways (SSP) scenario-based ecological climate index, while non-climate variables were based on topography, land cover maps, and the Enhanced Vegetation Index (EVI).

 

• This study is significant in that it is the first to develop a regional SDM for Luciola unmunsana, which is declining due to urbanization. In addition, by applying various environmental variables that reflect ecological characteristics, it contributed to more accurate predictions of the potential habitat of this species.

 

• These studies consider fireflies to be environmental indicator species and actively utilize SDM results in habitat characteristic analysis and conservation area designation. Reflecting this research trend, this study also predicted potential habitats for Luciola unmunsana, a species native to Korea.

Author Response File: Author Response.pdf