We greatly appreciate and thank Francisca Vergara-Pinto for her comments, which have allowed us to clarify several points and further strengthen the introduction, discussion, and conclusions.

Comment 1: To further enhance this work, I suggest some moderate/major changes that could be considered. The suggestions and comments (see PDF) are organised according to the structure of the paper to make the audience more familiar with the topic:

1 Introduction: The authors begin the introduction by stating that “in the context of natural hazards, significant scientific attention on risk perception has focused on earthquakes and tsunami related disasters, while comparatively less attention has been given to volcanology”. The literature offers a different perspective, both internationally and in Chile and Latin America. I recommend reviewing and citing some works that oppose this notion, perhaps noting that “while comparatively minor, volcanology is gaining increased attention.” For example: Iceland (Bird, Gísladóttir and Dominey-Howes 2011; Kokorsh and Gisladottir 2023); Mexico (López-Vázquez 2009); Chile (Vergara-Pinto and Marín 2023; Alegría and Vergara-Pinto 2024); Colombia (Peña et al. 2020). Including this type of background would reinforce the idea of the potential for studying the perception of volcanic risk.

Response 1: Regarding the Introduction, we consider the present paragraph fairly precise (VRP has received less attention in volcanology), although the suggested sentence is also true (that is, gaining attention in volcanology). The proposed references do not oppose this notion (VRP is still receiving less attention in volcanology) but certainly provide evidence of that increasing attention. We added some of those references, including some of them coauthored by Vergara-Pinto.

Comment 2: When referring to levels of risk perception: In the background information, it is clear that VRP is eminently cultural, contextual, and multifactorial. Before this sentence, I suggest pointing out that there are qualitative, quantitative, or mixed ways of approaching VRP and that you consider it optimal to measure “levels” of risk perception (quantifying a factor) and eventually point out difficulties in quantifying certain cultural aspects (for example, quantifying religion or belief variables).

Response 2: With respect to the levels of VRP, we concur with the comment and now explicitly state our preference for a quantitative approach.

Comment 3: When referring to knowledge: Is this factor more related to “Appropiation of scientific information”? I wonder if this factor could distinguish that people's knowledge is the result of a combination of information and experience (other sources, disaster memory, oral tradition, etc.). Or if it only measures access to scientific information.

Response 3: Regarding knowledge, we use the concept as “appropriation of scientific knowledge.” We have added an explanation to clarify this usage and its meaning.

Comment 4: Regarding external trust: Perceiving more risk could sound ambiguous. I suggest clarifying whether you mean that a person may perceive risk more (be more aware of the risks) or perceive that there is more risk (more likelihood of suffering a consequence).

Response 4: With respect to external trust, we interpret the concept of perceiving risk as “being more aware of the risks.” We have included an explanation to clarify this interpretation and its intended meaning.

Comment 5: When referring to the psychometric paradigm and cultural theory of risk: I suggest including examples or results from these applications to help readers understand how this approach aids in interpreting perceptions of risk. For instance, beyond providing statistical data (%), what other insights can psychometric analyses reveal. The criticisms about the cultural theory of risk have been addressed by other authors who have sought to move beyond the general categorization of four archetypes. They suggest that culture is dynamic and, consequently, so are perceptions of risk. Additionally, these works may reference cultural constructs that influence people's understanding of disasters, such as myths, beliefs, or traditions that shape representations of risk (Douglas 1996; Oliver-Smith 1999; García Acosta 2005; Hoffman 2015).

Response 5: Regarding the application of the psychometric paradigm and cultural theory, we added examples similar to those mentioned in the comment and expanded the discussion accordingly, including references such as Siegrist et al. (2005) and Sjöberg (2003).

Comment 6: While there is a substantial body of literature on risk perception and volcanism in the Global North, important references from Chile should also be included. This would demonstrate that your work contributes to an existing line of research on volcanic risk perception in Chile, particularly by introducing new insights from psychometric approaches. Some key references to include: Elderly individuals and place attachment (Díaz, Peña, Valles et al. 2022); Volcanic rurality in transformation (Vergara-Pinto, O’Grady, Fredriksen et al. 2024); Perception and local knowledge of risk at Villarrica volcano (Muena Briones 2023).

Response 6: Additional references based on experiences in Chile have been incorporated, with particular emphasis on those related to the case study.

Comment 7: The section on the study area and historical eruptions offers a detailed description of the volcano and surrounding communities. I suggest concluding by highlighting the effects of population growth, especially regarding land use changes and risks (e.g., Ruiz, Morales, Contreras et al. 2025); and include some insights on volcanic activity after the 2015 event, particularly the yellow alert 2022-2024, and significant changes in the volcano's behaviour since then.  (e.g., Romero et al. 2025)

Response 7: As stated before, some new references have been added, especially some related to the case study.

Comment 8: 2 Methodology: It is important to specify who designed the questionnaire administered in 2016 and 2017. If the authors were responsible for its design, this should be clearly stated. Conversely, if the questionnaire was administered by SERNAGEOMIN staff, evidence of permission to publish the survey data should be provided.

Response 8: Regarding the methodology, we designed the survey when the first author was an M.Sc. candidate, and the survey follows the ethical procedures at the Pontifical Catholic University of Chile. No further permission is required.

Comment 9: When referring to the ‘Occupation’ variable: It is not recommendable to assume that greater access to information equates to greater knowledge. For instance, retired older adults may have accumulated knowledge over their lifetimes, or they may possess local knowledge and disaster memories that aid them in making decisions. This is a point that could be presented as a complementary perspective on the issue of Occupation. Here are some examples: Mt Merapi (Septiana et al. 2019); Nevados de Chillan (Sandoval et al. 2023).

Response 9: Regarding occupation, we appreciate the comment highlighting the role of occupation as a means of acquiring information and increasing knowledge about perceived risks. We agree and acknowledge that retired older adults may possess valuable life experiences that contribute to a better understanding of volcanic processes and their consequences. However, our intention in using occupation as a VRP variable is to emphasize its function as a formal channel for preparedness in the event of a disaster, through access to current scientific and technical information disseminated by companies or institutions. We have incorporated the suggested references and expanded the discussion accordingly.

Comment 10: 3 Results: It would be good for the authors to rethink the structure and order of the Results and Discussion sections, ideally merging them into a single section to avoid repetitions. This restructuring would enhance readers' comprehension and improve the overall coherence of the paper.

Response 10: Results and Discussion. We keep the structure of two separate sections in order to differentiate the discussion from the results. However, we considered the comments and expanded the Discussion.

Comment 11: 4 Discussion: Regarding cultural patterns of VRP levels, the Results section discusses variations on the theme of religion, but here, you could expand on the implications or interpretations. What does this reveal about the population living near the Villarrica volcano? What influence does religion appear to have on the perception of risk? (especially in this area, where people may believe in the Christian God, Chao Ngünechen, or sometimes both).

Response 11: Regarding religion, we understand it in a broad sense, ranging from Christianity and its various denominations to local belief systems, which are sometimes blended through syncretism involving traditional practices. Although no consistent pattern has been identified in predicting risk perception by religion, religious individuals are generally found to be less concerned about risks, which serves as an interpretive basis in this research (Slimak, M. W., & Dietz, T., 2006). We have added explanations about this matter of occupation.

Comment 12: 5 Conclusion: This section summarizes the article, but it could be enhanced with reflections and recommendations on the study topic. I suggest specifying the types of vulnerabilities that may affect specific groups and how tailored risk communication should be in practice. Finally, it is important to reflect on how representative the findings from the 2016-2017 survey are of the current situation at Villarrica volcano. Significant changes - whether positive or negative - in the composition of the population may have occurred, particularly in the aftermath of COVID-19 (e.g., lifestyle migration).

Response 12: The Conclusions section was expanded as suggested. We added comments on the representativeness of the sample and the demographic changes experienced in the area due to the pandemic and migration from major cities such as Santiago of Chile, and Temuco.

Comment 13: The figures and maps are well-done, and the references appear to be accurate. If possible, the questionnaire could be included as supplementary material, or the main themes of the questions could be summarized to provide a clearer overview of what was asked concerning the variables and factors involved.

Response 13: We have added a table listing all survey questions and expanded explanations of the score criteria used. We have also included the raw data and scores as supplementary information.

Comment 14: Overall, the manuscript could be strengthened by addressing more clearly the role of cultural factors in the diverse perceptions held by residents and non-residents of the Villarrica volcano.

Response 14: Again, we greatly appreciate the comments provided by Francisca Vergara-Pinto to support the strengthening of this manuscript.

We greatly thank the comments by Letícia Guimarães, which allow us to clarify some topics and further strengthen the manuscript.

Comment 1: The INTRODUCTION brings a summary of volcanic risk perception (VRP) studies, the concepts of psychometric paradigm and the cultural theory, and the geological setting of the target volcanic system. The text is well written, but the state-of-the-art of VRP studies could be better developed; even considering that VRP has, proportionally, less studies than other geological hazards, there is a set of good published works (especially from the 2000's on) whose findings could be better and briefly presented here (also for supporting the discussion section).

Response 1: We appreciate the comment, although we considered the papers included in a systemic review (Favereau, 2019) and expanded from those is a proper method to get a comprehensive literature review. We kindly request specific examples to those refered by the referee's comment that might appear missed in the introduction. We regardless have included some other papers to keep expanding the literature review and introduction sections.

Comment 2: Also, shouldn’t tephra fall out be included as a potential volcanic hazard for this region (at lines 106-107)? In figure 01, only historical lahar deposits are shown; considering the whole set of volcanic hazards, it would be interesting to include a multiple hazard map.

 Response 2: We highly appreciate the comment. However, we have considered the proximal volcanic hazard of lahars only. This is because lahar flows have been the most impactful volcanic process from Villarrica eruptions. Wind in the area predominantly blows to the east according to atmospheric data. Although we do not neglect the potential occurrence of ashfall impacts, we consider more relevant for the applications of VRP surveys of people highly exposed to locally known and experienced volcanic hazards, such as lahars.

Comment 3: In the METHODOLOGY section the application of the questionnaires and the analyzed social, cultural and economic parameters are well explained, but some additional information on the survey procedure and the statistical approach would greatly improve the manuscript. Considering a greater reliability of the study and its eventual future replicability, I suggest presenting the 28 questions that constitute the survey (not only the "desirable answers" as presented in table 1) and the used score criteria. In section 1.1 (lines 74-76) the authors mentioned that psychometric and cultural theory analysis are usually based on "point-scale systems such as Likert scales or 5-point questions" but it's not clear which is the scoring method applied here (is it a Likert scale?).

Response 3: We have added a table listing all survey questions and provided expanded explanations of the scoring criteria used, which assign a value of 1 when a response aligns with a theoretically optimal VRP, and 0 when it does not.

Comment 4: In section 2.2 (line 172) it is mentioned that "Each survey question is assigned a value of one (1) when the response aligns with theoretically optimal VRP", but how are the answers that diverge from those considered the "desirable answers" in table 1 scored? Presenting the applied questionary and the score method is crucial.

Response 4: Those answers diverging from "desirable" ones are scored as 0. We have provided more explanations on the scoring system.

Comment 5: In figure 3 legend, a better explanation of the "rural sample (ppl)" meaning would be appreciated (what do those numbers mean? the exact number of people interviewed in each area?).

Response 5: We have included in the figure the exact number of people in the different rural areas.

Comment 6: In table 2, the data on “rural” areas refers to all the three municipalities (Pucón, Coñaripe and Villarrica)? Why data from Villarrica region is not presented individualized as the Pucón and Coñaripe data?

Response 6: We agree with the comment and included the sample collected from rural areas, differentiated by municipalities (Pucon, Panguipulli, Villarrica)

Comment 7: In section 2.2 the statistical approach should be better detailed, considering readers and stakeholders that could not be fully familiar with the topic.  After a better explanation of the scoring method in the previous section, a clearer step-by-step method description would be appreciated. Example: (1) the sum of the scores of each response  defines the VRP score of each interviewee: (2) the set of samples (VRP scores of all interviewees) are grouped in clusters considering the k-means analysis; (3) the clusters are defined based on the definition of the best clusters centroids based on standard variation; (4) the standard variation is defined as... (and so on). Here, terminology standardization is important to enhance the understanding of the method and the results obtained.

Response 7: We restructured the Methods section to provide a clearer description of the scoring process (assigning a value of 1 when a response aligns with a theoretically optimal VRP, and 0 when it does not), followed by a detailed step-by-step explanation: (1) calculation of the sum of scores by VRP factors (knowledge information, internal and external trust, and, for residents only, attachment to place) for each interview survey; (2) grouping of the sample sets (resident and non-resident interviewees) into clusters using a k-means analysis to produce a five-tier classification; and (3) ordering of the centroid scores (WCSS) from minimum to maximum, and assigning corresponding labels of Very Low, Low, Moderate, High, and Very High VRP.

Comment 8: For instance:

• k-means clustering usually considers within-cluster sum of squared distances (WCSS), also called variation or standard variation, which is slightly different from the standard deviation. Both terms are used in the manuscript: standard variation (in the methods), and standard deviation (in figures 7 and 8, in section 3.3.2.). If they're not referring to the same thing, and both calculations are applied in the study, I would suggest prioritizing the term "WCSS" in the methods to avoid misunderstandings and make it clearer where each calculation is applied.
• If the "J" term in equation 1 (lines 181 and 183) represents the WCSS, I'd suggest using "WCSS" instead of a new term/definition ("objective function").

Response 8: We agree and highly appreciate this comment. We have included the term within-cluster sum of squared distances (WCSS) throughout the entire manuscript. 

Comment 9: The RESULTS starts with a paragraph whose final sentences (lines 252-254) seem to be outplaced; they would fit better in the introduction, when describing the whole content of the manuscript.

Response 9: We have moved the sentences to the introduction, modifying some grammar for better clarity.

Comment 10: Another necessary textual review is the standardization of terms; for example, in the previous section and in the figures 3 and 6, the authors refer to the localities as Pucón, Coñaripe and Villarrica but in this section (3.2 and figure 5) the authors refer to the localities as Pucón, Panguipulli and Villarica.

Response 10: We agree that the names can be confusing mostly because in the case of Pucon and Villarrica are names of both the town (urban area) and the district/municipality. We have provided more explanation of localities/towns and districts/municipalities in the study area.

Comment 11: Considering that the results are the core of a study, the raw data, the scores for each factor and the final VRP score should be presented (it could be done as supplementary table, considering it is an extensive dataset).

Response 11: We have included the raw data and scores as supplementary information.

Comment 12: From this, some improvements can be made to the graphical presentation of the data, to facilitate visualization and enhance the understanding of eventual patterns and correlations:

• Figures 5 and 6 written legend: please indicate that the results are for residents.

Response 12: We agree and have updated figures 5 and 6.

Comment 13: Figure 6 graphic legend: please clarify the meaning of the sample size symbol - what do those numbers mean (the number of interviewees)?

Response 13: We agree and have updated the legend of Figure 6. The sample size symbol indicates the number of interviewees in the different zones, and we have added the corresponding explanations.

Comment 14: Figures 7 and 8: Y axis titled as "standard deviation" - is this right? Or is it WCSS? If this is the standard deviation, the methods section should include some info regarding this calculation. In any case, neither the WCSS nor the standard deviation provide negative values. Or are these values the clusters’ centroids? In this case, how negative scores could be obtained from the survey? So… what is the meaning of these diagrams? What do the authors want to present? With these diagrams, if they represent standard deviation values, we are simply visualizing the dispersion of scores, so I would suggest a different type of diagram: boxplots could be an interesting and more meaningful way to present the factors' scores. Through boxplots the reader could know the whole range of scores, their dispersion, minimum and maximum scores... and it could include a label with each cluster’s centroids values. With all this information in one diagram, it would be easier to comprehend how each factor is contributing to the final VRP.

Response 14: As stated before, we have included the term within-cluster sum of squared distances (WCSS) throughout the entire manuscript. We again highly appreciate this comment.

Comment 15: Figure 9: this is the high point of the study and the most important information for stakeholders, especially those responsible for risk management. Why did you choose presenting the pie charts as “distribution of cultural groups by VRP levels” and not the opposite way (“distribution of VRP levels by cultural groups”)? Could this change highlight different patterns? The pie charts focused on the VRP levels (the way it is presented) could be hindering a precise characterization of a given social group? For instance: a given group (e.g. 18 to 40 years old residents) have a few people in the high VRP, but them represent the majority of those in this level; the same group (18 to 40 years old residents) have a lot of people in the low VRP (more people than in the high level) but corresponding to the minority in the low VRP. Considering that pie charts display relative proportion of data, the resulting charts would show a bigger area for this group in the High VRP chart and a smaller area in the Low VRP chart, given the idea of more people with higher VRP. But, considering the absolute number of 18 to 40 years old residents, the group has more people with low VRP, so it would also deserve attention. In any case, it is fundamental to present the raw data as supplementary material to support the diagrams and the related discussion.

If the authors agree with the proposed review, the text might also need some revision.

The DISCUSSION section is well written, but in the first part of it the authors are, again, focusing on critics to/problems with the applied approach. To avoid disqualifying the work, such criticisms/problems could be presented only in the introduction, where the pros of the method and the reason why the authors chose to use it should be better detailed, leaving this section exclusively for the discussion of the results and patterns described in the previous item.

Response 15: We highly appreciate this comment regarding this key figure and agree that greater clarity on this subject would significantly enhance the communication of results to stakeholders, particularly in illustrating the “distribution of VRP levels by cultural groups,” as it represents an easily interpretable dataset. Accordingly, we have revised Figure 9 and the corresponding sections describing its results. While the overall result patterns by cultural factors remain consistent, the revisions provide improved clarity throughout the Results and Discussion sections.

Comment 16: At line 375 it is not very clear what the authors mean by “positive behavior”, so this could be better explained.

Response 16: We meant "positive VRP scores" referred to section 3.3.1. We have rephrased the sentence and provided a better explanation.

Comment 17: Providing the score method and the raw data, as requested before, would also be useful in this section, for example, to better understand how negative VRP scores are obtained (as mentioned at line 381).

Response 17: As stated before, we have included the raw data and scores as supplementary information.

Comment 18: The authors discuss the relation between age and education level of non-residents and their VRP score - would their economic characteristics (occupation and economic activity) also influence this result?

Response 18: We expanded the discussion on the influence of occupation on risk perception and incorporated additional literature in the Introduction, including Septiana et al. (2019) and Suwihli et al. (2000).

Comment 19: Lastly, the CONCLUSIONS are well written, concise and straightforward, but may need some revision after the above-mentioned changes.

Response 19: We have updated and rephrased the conclusions as required by the applied changes.

 Comment 20: In summary, the manuscript represents an important contribution to the volcanic risk perception science and to the assessment and management of one of the riskiest volcanic systems in Latin America. A better understanding of exposed population characteristics, their knowledge and risk perception may help delineate educational and communication policies, key parameters for disaster risk reduction. However, some drawbacks need attention to improve the paper and major reviews are required before publication. The main points in this review are: (i) more detailed method explanation; (ii) providing of the applied questionnaires, the score system, and the raw data; and (iii) reviewing of figures, especially those from results section.

Response 20: We highly appreciate the comments and the insightful questions, and the term clarifications. We are certain the changes greatly improve the quality of the manuscript.