Economic Valuation of Geosystem Services in Agricultural Products: A Small-Sample Pilot Study on Rotella Apple and Moscatello Wine

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors, please find my detailed feedback aimed at refining and strengthening your manuscript for its next revision.

 

Title, Abstract, and Keywords

• The title is broad but does not clearly reflect the study’s core methodology (choice experiment with MMNL and LCM) or its pilot nature; it could be more precise (e.g., explicitly noting the economic valuation approach or that it is a small-sample pilot study).
• The abstract lacks detail on the key statistical results (coefficients, odds ratios, or any quantitative indicators). Without numbers, it is hard for readers to gauge the strength of findings.
• The abstract refers to “transparent labelling policies” and “product traceability” but does not clearly connect how the empirical results justify these implications.
• No mention is made of the limitations of the study (e.g., small sample size, pilot nature, respondent knowledge gaps), which is critical to balance the conclusions.
• The keywords are generic; terms like “Mixed Multinomial Logit,” “Latent Class Model,” or “consumer choice experiment” should be included for discoverability.

 

Introduction

• The literature review is imbalanced: while it gives extensive background on geodiversity and geosystem services, there is minimal review of prior economic valuation studies or consumer behavior studies related to abiotic services.
• Several citations (e.g., 3, 19, 20) are repeated or used in multiple contexts without distinguishing their different contributions; this creates redundancy.
• The introduction does not clearly state the research gap. It should clarify whether the novelty is (a) applying geosystem service valuation to consumer products, (b) using discrete choice models, or (c) testing consumer awareness of abiotic services.
• The text repeatedly shifts between geodiversity and soil services without clearly distinguishing them, leading to conceptual ambiguity.
• The research objectives are vague. The paper says “we provide a contribution” but does not explicitly define hypotheses or research questions (e.g., Do consumers value abiotic soil attributes? Is WTP higher for geosystem service-linked labels?).

 

Materials and Methods

• The sampling strategy is weakly justified. 200 participants were selected from a convenience sample (municipal employees, students, etc.) without any discussion of representativeness or potential biases.
• There is no explanation of how attributes and levels were pre-tested to ensure comprehension, despite the later finding that many respondents did not understand them.
• The price levels are presented without justification (e.g., market relevance, consumer familiarity, or testing for range effects).
• The methodology for experimental design generation (400 combinations reduced to four tasks) is underexplained. How were “implausible” or “dominant” combinations systematically eliminated? Was a D-efficient design used?
• The statistical approach is insufficiently justified. While MMNL and LCM are explained theoretically, there is no discussion of model diagnostics, fit statistics (AIC, BIC), or why both models were necessary.
• The Random Utility Theory discussion is textbook-like and lengthy, overshadowing the actual data analysis details. More focus should be on how these theories apply to the study context.
• There is no information about coding of categorical variables, handling of heterogeneity, or interaction terms, which are essential in MMNL and LCM reporting.

 

Results and Discussion

• The tables (4–7) lack clarity. Odds ratios are given, but the text refers to “coefficients” and “increment by one unit” inconsistently. It is unclear if these are log-odds or transformed values.
• The price coefficient being positive and non-significant contradicts economic theory but is not critically examined; the explanation (“to test robustness”) is superficial. Authors should test for model misspecification or scale effects.
• There is no presentation of willingness-to-pay estimates, which is a standard output for choice experiments and would strengthen the applied value of results.
• The no-choice (status quo) option’s high significance is noted but not explored in depth (e.g., could indicate task complexity, attribute misunderstanding, or unrealistic choice sets).
• The LCM results (Tables 6–7) indicate severe heterogeneity and comprehension issues, yet this is treated descriptively rather than analytically. There is no discussion of whether attributes should be simplified or redesigned.
• Discussion repeats introduction material (on labelling, traceability, GS concepts) without fully connecting back to the empirical findings.

 

Conclusion

• The conclusion overstates consumer recognition of geosystem services despite evidence that half of the sample misunderstood attributes. It should be more cautious.
• There is no explicit mention of study limitations, such as the pilot sample size, sample representativeness, or complexity of geosystem concepts for non-experts.
• Policy implications (labelling, education) are mentioned but not directly tied to quantitative results; for instance, what attribute effects suggest consumers might support geological labels?
• There is no roadmap for future research beyond generic education initiatives. More specific directions (e.g., refining attribute framing, testing larger/more diverse samples, incorporating WTP estimates) are needed.
• The study does not discuss external validity. Can these findings be generalized beyond Rotella apple and Moscatello wine?

Comments on the Quality of English Language

The manuscript is understandable but has awkward phrasing, redundancy, and lacks flow in key sections (Introduction, Results, and Discussion).

Author Response

Comment 1: The title is broad but does not clearly reflect the study’s core methodology (choice experiment with MMNL and LCM) or its pilot nature; it could be more precise (e.g., explicitly noting the economic valuation approach or that it is a small-sample pilot study).

Response 1: The title has been modified accordingly.

Comment 2: The abstract lacks detail on the key statistical results (coefficients, odds ratios, or any quantitative indicators). Without numbers, it is hard for readers to gauge the strength of findings. The abstract refers to “transparent labelling policies” and “product traceability” but does not clearly connect how   results justify these implications. No mention is made of the limitations of the study (e.g., small sample size, pilot nature, respondent knowledge gaps), which is critical to balance the conclusions.

Response 2: The abstract has been thoroughly revised following the suggestion of the reviewer to better reflect the scope and findings of the study. The limitations of the study has been added.

Comment 3: The keywords are generic; terms like “Mixed Multinomial Logit,” “Latent Class Model,” or “consumer choice experiment” should be included for discoverability.

Response 3: The keywords have been updated, including those suggested by the reviewer.

INTRODUCTION CHAPTER: The introduction has been extensively revised and simplified. Below are the specific responses to the comments raised by Reviewer 1 

Comment 4: The literature review is imbalanced: while it gives extensive background on geodiversity and geosystem services, there is minimal review of prior economic valuation studies or consumer behavior studies related to abiotic services.

Response 4: The literature in the introduction has been balanced by adding some references on existing consumer studies related to biotic attributes. To our knowledge, previous studies related to abiotic services on the consumer preferences are virtually absent.

Comment 5: Several citations (e.g., 3, 19, 20) are repeated or used in multiple contexts without distinguishing their different contributions; this creates redundancy.

Response 5: Introduction has been extensively revised to address more clearly the topic and the related literature. Nevertheless, some of the repeated citations seem to be necessary.

Comment 6: The introduction does not clearly state the research gap. Its hould clarify whether the novelty is (a) applying geosystemservice valuation to consumer products, (b) using discrete choice models, or (c) testing consumer awareness of abiotic services.

Response 6: Research novelty is envisaged in point (a) and (c). This part of the introduction has been revised accordingly

Comment 7: The text repeatedly shifts between geodiversity and soil services without clearly distinguishing them, leading to conceptual ambiguity.

Response 7: The text referring to soil services as related to geodiversity has been simplified. The rest of introduction clearly explains how geosystem services (including soil services) are related to geodiversity in the same way as ecosystem services are related to biodiversity (as defined in the “CICES”).

Comment 8: The research objectives are vague. The paper says “we provide a contribution” but does not explicitly define hypotheses or research questions (e.g., Do consumers value abiotic soil attributes? Is WTP higher for geosystem service-linked labels?).

Response 8: The research objectives have been revised accordingly

MATERIAL AND METHODS CHAPTER: The chapter has been extensively revised, reorganized, and simplified in accordance with the suggestions provided by Reviewer 1. The text has been divided into two sub-chapter (“2.1 Choice experiment setup” and “2.2 Data analysis”) Below are the specific responses to the reviewer’s comments.

Comment 9: The sampling strategy is weakly justified. 200 participants were selected from a convenience sample (municipal employees, students, etc.) without any discussion of representativeness or potential biases.

Response 9: The sampling strategy is better explained in 2.1

Comment 10: The sampling strategy is weakly justified. 200 participants were selected from a convenience sample (municipal employees, students, etc.) without any discussion of representativeness or potential biases.

Response 10: Explanations has been added

Comment 11: The price levels are presented without justification (e.g., market relevance, consumer familiarity, or testing for range effects.

Response 11: Justification and comments on price leve have been introduced in the text both in section 2.1 and 3.1 “ Multinomial Logit Model”

Comment 12: The methodology for experimental design generation (400combinations reduced to four tasks) is under explained. How were “implausible” or “dominant” combinations systematically eliminated? Was a D-efficient design used?

Response 12: Methodology has been better described in chapter 2.1. Implausible combinations have been selected and removed by filtering all possible profiles using Microsoft Excel. Explanations added to the text in footnotes of chapter 2.1

Comment 13: The statistical approach is insufficiently justified. While MMNLand LCM are explained theoretically, there is no discussion of model diagnostics, fit statistics (AIC, BIC), or why both models were necessary

Response 13: We have added an explanation to clarify why both models are necessary, both in the chapter 2.2 and, in greater detail, in the Results and Discussion chapter

Comment 14: The Random Utility Theory discussion is textbook-like and lengthy, overshadowing the actual data analysis details. More focus should be on how these theories apply to the study context.

Response 14: We have have eliminated the redundant explanations

Comment 15: There is no information about coding of categorical variables, handling of heterogeneity, or interaction terms, which are essential in MMNL and LCM reporting.

Response 15: Information regarding the coding of categorical variables has been added

RESULT AND DISCUSSION CHAPTER

Comment 16: The tables (4–7) lack clarity. Odds ratios are given, but the text refers to “coefficients” and “increment by one unit” inconsistently. It is unclear if these are log-odds or transformed values.

Response 16: This point has been clarified in the text for all tables

Comment 17: The price coefficient being positive and non-significant contradicts economic theory but is not critically examined; the explanation (“to test robustness”) is superficial. Authors should test for model misspecification or scale effects.

Response 17: The positive and statistically non-significant coefficient of price has been addressed in the revised manuscript, with reference to the results presented in Table 4

Comment 18: There is no presentation of willingness-to-pay estimates, which is a standard output for choice experiments and would strengthen the applied value of results

Response 18: We chose not to estimate willingness-to-pay (WTP) measures, as the price attribute was not statistically significant in any of the models. In some cases odds ratios were close to 1 and standard errors relatively high, indicating that price had virtually no effect on respondent choices. Given this, we believed that deriving WTP would be methodologically unsound and would not add meaningful value to the analysis. We have instead focused on the direction and significance of non-monetary attributes, which consistently influenced utility and provide more reliable insights for this exploratory pilot study. Implications, shortcomings and addresses for future researches have been reported in the conclusion chapter.

Comment 19: The no-choice (status quo) option’s high significance is noted but not explored in depth (e.g., could indicate task complexity, attribute misunderstanding, or unrealistic choice sets).

Response 19: Further information and explanation have been given in chapter 3.1

Comment 20: The LCM results (Tables 6–7) indicate severe heterogeneity and comprehension issues, yet this is treated descriptively rather than analytically. There is no discussion of whether attributes should be simplified or redesigned.

Response 20: Discussion about this point has been added in chapter 3.2

Comment 21: Discussion repeats introduction material (on labelling, traceability, GS concepts) without fully connecting back to the empirical findings

Response 21: This point has been revised accordingly

CONCLUSION CHAPTER: The conclusions have been extensively revised. Below are the specific responses to the comments raised by Reviewer 1 

Comment 22: The conclusion overstates consumer recognition of geosystem services despite evidence that half of the sample misunderstood attributes. It should be more cautious.

Response 22: This point have been revised accordingly

Comment 23: There is no explicit mention of study limitations, such as the pilot sample size, sample representativeness, or complexity of geosystem concepts for non-experts.

Response 23: Discussion of study limitations has been added to the conclusions

Comment 24: Policy implications (labelling, education) are mentioned but not directly tied to quantitative results; for instance, what attribute effects suggest consumers might support geological labels?

Response 24: This point has been revised accordingly

Comment 25: There is no roadmap for future research beyond generic education initiatives. More specific directions (e.g., refining attribute framing, testing larger/more diverse samples, incorporating WTP estimates) are needed

Response 25: Discussion of roadmap for future research has been added

Comment 26: The study does not discuss external validity. Can these findings be generalized beyond Rotella apple and Moscatello wine

Response 26: The findings of this study cannot be generalized beyond the studied example. Nevertheless, take with the caution the result show significant attention to soil characteristic and this aspect is certainly worth to be further investigated. These aspects has been added in the conclusion chapter.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript examines the emerging field of geosystem services by applying an experimental choice framework to two Italian niche products—Rotella apple and Moscatello wine—thereby exploring how consumers value soil‐related attributes embedded in agricultural goods. Its ambition to extend ecosystem service valuation into the abiotic domain represents a novel contribution to both ecosystem economics and agrifood traceability debates, as most prior studies have focused on biotic components and conventional geographical labels. The authors’ decision to pilot a Mixed Multinomial Logit Model alongside a Latent Class Model demonstrates methodological rigor and an awareness of consumer heterogeneity, rendering their findings on the salience of “use of soil” and “soil control” attributes particularly compelling.

Scientifically, the study is sound in its theoretical grounding in Lancaster’s attribute theory and Random Utility Theory, and the estimations are transparently reported with odds ratios and significance levels. However, the non‐significant and unexpectedly positive price coefficients point to potential issues in experimental design or attribute framing that merit deeper discussion. The reliance on a convenience pilot sample—municipal employees, academics, volunteers and the self‐employed—limits generalizability and may introduce biases, especially given the high “no‐buy” coefficients indicating attribute misinterpretation by nearly half of respondents. Future iterations should consider stratified sampling and enhanced pretests to ensure attribute comprehension.

The manuscript is generally well organized, flowing logically from introduction to conclusion, but stylistic improvements would enhance readability. Several passages are burdened by lengthy sentences and occasional grammatical slips (for example, agreement errors and typos in both Italian and English terms), which cumulatively distract from the central argument. The authors are encouraged to streamline complex sentences, standardize terminology (e.g., consistently using either “geosystem services” or “GS”), and perform a thorough copyedit.

With respect to methodology, while the elimination of implausible attribute combinations is appropriate, the process by which 400 candidate profiles were reduced to four choice sets requires clearer justification to assure orthogonality and avoid dominance. In results, the interpretation of mixed and latent class outputs is insightful yet would benefit from reporting willingness‐to‐pay estimates to translate odds ratios into monetary values, thereby strengthening policy relevance.

Overall, the paper holds significant promise for advancing the valuation of soil‐based ecosystem services in food products and for informing more transparent labelling policies. Nevertheless, the issues around price effects, sample representativeness, attribute comprehension and linguistic precision must be addressed before publication. I therefore recommend major revisions to refine the experimental design description, elaborate on limitations, correct stylistic and grammatical issues, and enrich the discussion with willingness‐to‐pay analysis and practical policy implications.

Specific sentences requiring additional clarification or information:

1. “The link between agricultural products and GS is a relatively new subject and warrants further investigation.”bPlease specify which aspects (e.g., consumer behaviour, supply‐chain metrics) remain under‐studied and how this study addresses those gaps.
2. “From soil comes more than 99% of human food.” Kindly indicate the data source and clarify whether this refers to caloric intake, mass, or another metric.
3. “Using Stata software, 400 combinations of attributes’ levels for both products were randomly selected considering … the elimination of implausible and dominant alternatives.” Describe the algorithm or criteria employed to ensure statistical efficiency and attribute balance in the final choice sets.
4. “It was assumed that the consumers’ choice was made in a short‐time, so the alternative C (no‐buy option) implies that the utility … is equal to 0.” Define “short‐time” in experimental terms and explain the rationale for normalizing the no‐buy utility to zero.
5. “Our findings suggest that consumers intuitively recognize the contribution of geosystem services to these products.” Please quantify this intuition by providing willingness‐to‐pay estimates or utility differentials to substantiate the claim.

Author Response

The manuscript has been extensively revised throughout, following the suggestions provided by Reviewer 2, in conjunction with the comments made by Reviewer 1. In particular:

• A discussion on potential issues related to the experimental design and the framing of attributes has been added to the manuscript.
• The limitations of the pilot sample and the preliminary nature of the experiment have been explicitly addressed in the discussion, along with suggestions for future improvements.
• The text has been revised for conciseness and stylistic clarity to enhance overall readability.
• The procedures used to extract the choice sets have been clarified and their rationale has been better justified in the revised text.
• All concerns raised by the reviewer have been carefully considered, particularly those regarding the description of the experimental design, the discussion of limitations, the willingness-to-pay analysis, and the policy implications.

Below are the responses to the specific comments provided by Reviewer 2

Comment 1: The link between agricultural products and GS is a relatively new subject and warrants further investigation. Please specify which aspects (e.g., consumer behaviour, supply‐ chain metrics) remain under‐studied and how this study addresses those gaps

Response 1: The aspect the remained understudied is consumer behaviour and the abilty to recognise GS as a value. This aspect has been better specified in the introduction.

Comment 2: From soil comes more than 99% of human food.” Kindly indicate the data source and clarify whether this refers to caloric intake, mass, or another metric.

Response 2: The metric is calories of human food and data source is cited in the text (Pimentel, 2006; ref: [26]). Data source and clarifications has been added to the text.

Comment 3: Using Stata software, 400 combinations of attributes’ levels for both products were randomly selected considering … the elimination of implausible and dominant alternatives.” Describe the algorithm or criteria employed to ensure statistical efficiency and attribute balance in the final choice sets.

Response 3: We employ a full factorial algorithm for the 400 combinations. The elimination of the implausible and dominant alternatives has been done filtering the choiche sets in Microsoft Excel without using optimization algorithm because of the limited and biased nature of our sample. This explanation has been described in the text.

Comment 4: It was assumed that the consumers’ choice was made in a short‐time, so the alternative C (no‐buy option) implies that the utility … is equal to 0.” Define “short‐time” in experimental terms and explain the rationale for normalizing the no‐buy utility to zero.

Response 4: The term short-time is used here to indicate that, when respondents are presented with questions that are difficult to interpret, the most likely outcome is the selection of the no-buy option. This behaviour may reflect either: (i) a short-term non-purchase decision, where individuals might consider buying the product in the future but not among the profiles currently presented; or (ii) a persistent non-purchase stance, where respondents have no intention to buy either now or in the future. Importantly, this does not imply that the associated utility is zero; rather, it suggests that the choice context is cognitively demanding due to the difficulty in interpreting the attributes or the question itself. A clarification of this interpretation has been included in the text.

Comment 5: Our findings suggest that consumers intuitively recognize the contribution of geosystem services to these products.” Please quantify this intuition by providing willingness‐to‐pay estimates or utility differentials to substantiate the claim.

Response 5: The lack of the robust and significant price effect prevented the estimation of willingness-to-pay estimates. Implications, shortcomings and addresses for future researches have been reported in the conclusion chapter.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The revisions made are satisfactory.