Modeling Ecosystem Services and Externalities for Ecosystem Accounting: The Case of Santa Lucia Sub-Basin in Uruguay

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper takes the Santa Lucía sub-basin in Uruguay as the research object, integrates the SWAT model with the SEEA-Ecosystem Accounting framework to model and analyze ecosystem services and agricultural externalities, and explores their correlation. It is of great significance for promoting watershed ecological management and policy formulation in Latin America. The research design is scientific and the application of methods is innovative. However, the paper has deficiencies in data description, model calibration, table presentation, and result analysis, which need to be modified in a targeted manner to improve the rigor and readability of the research.

1. The description of key data sources is incomplete. Table 1 only lists data types and partial sources, but fails to specify the temporal coverage and spatial resolution of crop yield data. It is necessary to supplement the complete attribute information of the data to ensure the traceability of the research.
2. The model calibration and validation process is not rigorous enough. For agricultural productivity and water quality variables, only the "soft calibration" method is adopted due to limited data, but the specific operation steps and rationality verification basis are not explained.
3. The spatial analysis of results is insufficient. Although the study mentions identifying pollution hotspots, it does not conduct quantitative analysis on the spatial aggregation characteristics of nutrient loadings; it only presents the spatial distribution map of ecosystem services without explaining the driving factors of spatial differences. It is necessary to add quantitative spatial analysis methods and clarify the driving mechanism of spatial differences.
4. The discussion on externalities is not in-depth enough. The study distinguishes between non-point sources and point sources of nutrient loadings, but does not analyze the temporal variation characteristics of externalities such as whether there are differences in nutrient loss under different seasons or climate conditions; it does not compare with the externality management policies of similar watersheds in Latin America, lacking policy implication analysis based on regional characteristics. It is necessary to supplement the temporal variation analysis of externalities and cross-regional policy comparison to enhance the practical value of the research.
5. The analysis of research limitations is not comprehensive enough. The paper only mentions the limitations of data quality, point source coverage, and the timeliness of land use maps, while ignoring the limitations of the model itself and indicator selection. For example, the applicability of the SWAT model in small-scale watersheds in Latin America; indicators related to externalities such as soil erosion and biodiversity are not considered.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study addresses the important and timely issue of integrating ecosystem service and externality modeling into ecosystem accounting, a key challenge for Latin American countries, where agriculture is both the foundation of economic development and a significant source of environmental pressure. By examining the Santa Lucía River sub-basin in Uruguay–an important water supply for much of the country's population–the authors address pressing issues of soil and water degradation, as well as the need for improved assessment tools to support environmentally conscious policies. The use of SWAT and the attempt to integrate ecosystem services and agro-ecological externalities within ecosystem asset accounting highlight the practical value of the work and its potential for informing scientifically informed management decisions. However, despite its high relevance, the article requires further development.

 

Comments on the Introduction:

- The paper lacks a detailed overview of the current state of research on the integration of ecosystem services and agro-ecological externalities in the Latin American context. Key research gaps, unresolved issues, and contradictions in existing approaches to modeling externalities in regional agriculture are not identified.

- The current state of application of models like SWAT for the integrated assessment of ecosystem services should be clarified. It would be helpful to clarify the characteristics, limitations, and methodological shortcomings of existing research approaches, data types, spatial resolutions, and the challenges in accurately modeling nutrient flows.

- The statement that SEEA-EA "does not explain temporal variations" is incorrect. SEEA-EA is not intended for causal analysis, but can reflect changes in state and services over time; this statement requires softening.

- The statement that SEEA-EA "does not take into account externalities" is also redundant. Some externalities are reflected as degradation or pressure on ecosystems. - The connection between the described environmental issues and the chosen methodology (SWAT modeling) is not demonstrated. Authors are asked to briefly explain why this tool is appropriate for the task.

- The idea of ​​the importance of water supply for 60% of the population is repeated several times. This repetition should be removed.

- Some sentences are excessively long and overly complex, which impairs readability. Some formulations are generalized and lack quantitative data, although they would be appropriate to justify the significance of the problem.

- Some statements are not accompanied by references, for example, regarding the insufficient effectiveness of measures in the basin.

Comments on the Materials and Methods:

2.1

- This section is excessively descriptive and does not include key hydrological, climatic, and soil characteristics necessary for replicating SWAT modeling.

- It is recommended to describe the main processes and environmental issues in the basin that justify the use of SWAT.

2.2

- The model description is limited to general information and lacks critical details: SWAT version (SWAT2012, SWAT+, modified version); hydrological scheme (HRU method, thresholds); input data (DEM, soils, climate, land use), sources, resolution, and processing.

- It is unclear which ecosystem services are assessed and how SWAT calculates them.

- There is no information on the types of externalities integrated into the analysis.

- Including the number of publications on SWAT and a list of organizations using the model lacks methodological value and clutters the text.

- It is not specified which parameters and model blocks were activated (crop growth, nutrients, sedimentation). There is no information about the time period of the input data, climate stations, or gap processing. The structure of sub-basins and HRUs, their number, and the division criteria are not described.

2.2.1

- Reproducibility deficiencies (critical) - SWAT/SWAT+ and SWATPlusR model versions are not specified, making full replication of the calculations impossible.

- Unclear: it is stated that sensitivity is analyzed only for runoff, but then crop yield and water quality are also calibrated. It is not described whether sensitivity was also analyzed for these parameters.

- It is not explained which dataset is used for calibration at the six stations: runoff or also water quality.

- It is not described why a monthly calibration step was chosen with a daily modeling step; justification is required.

- PSO is mentioned for uncertainty analysis, but how exactly it is applied, the population size, the number of iterations, and the optimization metric are not explained.

- "Soft calibration" is described superficially: acceptance criteria, tolerance ranges, and comparison method are missing. - The use of regional data and national averages for yield results in significant spatial inconsistencies with the accuracy of SWAT and should be thoroughly justified.

- "Biomass appears reasonable" is an unacceptable statement without quantitative comparisons.

- Water quality monitoring stations are not specified.

- Data frequency, processing methods, and the number of available observations are not described.

- How the high variability of nutrient data at a 2-month interval was addressed is not explained.

2.4

- No quantitative justification or calculation example is provided.

- Reproducibility issues (critical) - it is not specified how exactly the values ​​of the variables used as indicators are extracted from SWAT.

- Methodologies for converting SWAT output data into indicators of services and externalities are missing.

- Units of measurement for some indicators are not described, and calculation algorithms are missing (e.g., "water purification," "water availability").

- Threshold values ​​or interpretation criteria are not defined.

- "Water availability" is not disclosed: it is unknown whether this refers to soil moisture, plant available water, water balance, or something else.

- "Nutrients availability" can have different meanings: amount in the soil, in available form, mineral fraction, etc. The authors are asked to clarify which indicator is used.

- How externalities between ecosystems are assessed and how they are integrated into the final analysis is not described. • It is not explained how differences between land uses are taken into account in the model inputs.

2.4.1, 2.4.2, and 2.4.3

- There is confusion between services, conditions, and production indicators. Some indicators relate to agronomic productivity but are presented as ecosystem services.

- The equivalence between actual evapotranspiration (AET) and water availability is incorrect. AET measures water loss, not availability. Scientifically, water availability is linked to soil water content, plant available water, and soil water deficit. AET is incorrectly interpreted as reflecting the amount of water "used for growth"; AET includes evaporation from the soil surface, which has no productive value. There is no description of the AET integration period–season? Year? Growing season? It is not explained how AET expressed as mm/ha is interpreted per unit area (mm/ha is an incorrect unit; perhaps mm or m³/ha?).

- Substitution of concepts: uptake of N and P is an indicator of consumption, not availability. In ecosystem studies, availability is defined as the soil content or the available fraction (mineral N, Olsen P). There is no information on how SWAT calculates uptake: by biomass? by default? were the soil parameters calibrated?

- In SWAT, the process is modeled in a simplified manner and is highly dependent on calibration. This is not reflected. It is not specified whether the proportion of symbiotic fixation versus non-symbiotic fixation was taken into account. It is not explained how the soybean and other legume growing zones are identified in the model (by HRU? by land use?).

- The harvest index is missing: is a default value assumed? Does it vary across crops and pastures? There is no explanation of how the "catchment-level" values ​​were obtained or how scaling is performed. No distinction is made between potential and actual biomass; it is not specified how the issue of overestimation under moisture limitations is addressed.

- The sections do not allow for reproducing the calculations.

- There is no distinction between modeled values ​​and actual observations.

2.5

- This section does not describe the specific procedure for incorporating hydrological services into ecosystem accounts. It only describes the application of SEEA-EA without specifying the methods for transforming model outputs into account elements.

- There are no details on which indicators are distributed among SEEA accounts (extent, condition, supply) and how they are distributed. It is not specified how the indicators are aggregated, normalized, or classified.

- SEEA-EA requires flow reconciliation, unit checks, and temporal comparability–none of these are disclosed.

- Inaccuracies in the description of the extent account and condition account–it is not explained how the boundaries of ecosystem types are determined and on what basis the cartographic data are used. The time base of the extent account is not specified (single year? multi-year dynamics?). The condition account requires a set of biophysical indicators and reference conditions, but the method for selecting them is not described.

- It is unclear how modeled biophysical flows (e.g., water, biomass) are compared with SEEA reporting units. There are no methods to prevent double counting (a critical requirement of SEEA-EA). It is not stated whether allocations were made between ecosystems in mixed HRUs.

- This section does not allow for the replication of the account creation procedure–this is a serious methodological omission for the "Materials and Methods" section.

Comments on the Results:

3.1

- Weaknesses in the presentation of the hydrological calibration: the authors claim that NSE and KGE "exceeded thresholds," but do not provide the actual numerical values ​​in the text. Without the numerical values, it is impossible to assess the quality of the calibration.

- It is not stated whether the criteria of seasonality, flow peaks, and low waters–important indicators for SWAT–were used.

- The statement "the values ​​were similar" is not scientifically acceptable without numerical differences.

 

3.2.1

- Yield and biomass indicators are presented as ranges, but mean values ​​are not provided, coefficients of variation are missing, and spatial distributions are not numerically represented.

- It is not shown how SWAT calculates actual water use by the population–the model generally does not simulate direct water abstraction. 3.2.2 Water Purification Indicators

- Percentage reductions in loads (11% N, 3% P) are inconsistent with the scale of retention–the basis for comparison is unclear.

- A 46 µg/L reduction in P concentration is mentioned, but baseline concentrations are not specified, seasonality is not presented, and the effect at different river sections is not shown.

- The section is overly descriptive, lacking analytical comparisons or a discussion of the causes of observed values ​​or anomalies.

- Model limitations are not discussed, although the services depend on many uncertain SWAT parameters.

- The assessment of externalities is not disclosed, despite being stated in the title.

3.2.3

- It is not described how SWAT calculated nitrogen and phosphorus losses (surface runoff, drainage, erosional loss, dissolved vs. solid form).

- It is not discussed why the lower part of the basin produces the highest emissions.

- There is no connection to actual pollution sources (farms, fertilizers, soils, etc.).

Section 3.3 Hydrological Services and Externalities in an Ecosystem Account

Errors and logical inconsistencies in the condition account:

- In Table 5, water quality indicators are linked only to the "streams" category, not to spatial network nodes, and averaging "over the entire area" is incorrect, since the areas of land classes are not related to the length and discharge of streams. The average values ​​are not representative: rivers have a linear structure and cannot be averaged over land use areas. The authors themselves state that the averages "have no inherent meaning," which is a sign of the incorrectness of the chosen methodology.

- Comparison of concentrations in Table 5 with drinking water standards is incorrect unless it is stated that the data represent water at the watershed.

Comments on the Discussion:

4.1

- The authors claim that SWAT "effectively models hydrology," but do not provide model quality assessments (NSE, R², PBIAS). Justification is required.

- There is no uncertainty analysis of SWAT, other than one superficial phrase. There is no analysis of key sources of error (weather data, soil parameters, calibration quality).

- The assertion that reducing fertilizer application will reduce N and P removal is too general. There is no discussion of the role of soil processes, manure storage, seasonality of fertilizer application, buffer zones, and erosion. - The emphasis is on "hot spots", but it is not indicated how they were defined and with what level of statistical reliability.

4.2

- This section reiterates the theoretical tenets of SEEA without discussing specific study results.

- There are no examples of which specific indicators proved useful or problematic when integrated into SEEA.

- It is not explained how SWAT results are spatially aggregated into SEEA "ecosystem units."

- The limitations of the method are not addressed–the linear structure of rivers versus the areal logic of SEEA, the gap between the modeled processes and the indicators required by SEEA.

- There is no mention of input data quality issues–SEEA requires data comparability over time, but the authors do not discuss: data heterogeneity, differences in temporal resolution, or limited water quality observations.

4.3

- The statement of the study's relevance to LAC regions is descriptive and does not rely on quantitative or conceptual comparisons. There is no analysis of differences in climate regimes, hydrological conditions, institutional environments, or data quality and availability in neighboring countries. This lack of specificity makes the claims unconvincing.

- The authors suggest that the combination of SWAT and SEEA "can be adapted," but do not discuss the real obstacles: the need for detailed soil maps, the difficulty of calibrating SWAT in mountain and tropical basins, and the lack of water quality time series.

- There are no examples of regions or countries where the method is applicable–the reference to LAC is generalized and unsupported by examples, despite the region being extremely heterogeneous. Applicability to, for example, the Andes, Amazon, or Pampas is not specified, despite these being fundamentally different systems.

4.4

- The authors list the limitations but do not assess their impact on the results. There is no sensitivity analysis, parameter uncertainties, or model error ranges.

- The data issues are poorly systematized–the problems are described superficially. It is not specified which water quality indicators are the most unreliable, how much fixed equations distort the results, how often measurements were taken, or what the criteria for data exclusion are.

- The issue of outdated land use is incompletely addressed–the authors acknowledge using a 2018 map, but do not discuss how critical land use change after 2018 is for N and P, which land use classes are most dynamic, or possible ways to update the data. It is not stated why more current remote sensing data were not used.

- No discussion of the limitations of SWAT itself–typical limitations of SWAT are not mentioned–insufficient modeling of P-dynamics, poor representation of erosion in pasture zones, and modeling challenges under high precipitation and data scarcity conditions. The absence of these aspects makes this section incomplete.

Comments on the Conclusion:

- The conclusion does not summarize which specific scientific objectives were achieved and how successfully. There is no connection between the objectives set in the introduction and the results of the study.

- Statements about "feasibility" or "powerful tool" are made without specifying the limitations, assumptions, and robustness of the results. The reliability of the externalities and services estimates is not discussed. There is no discussion of uncertainties and limitations–the conclusion ignores the limitations detailed in the previous section. This creates inconsistency and overstates confidence in the conclusions.

-The phrase about the policy applicability of the approach is too general. There is no clear explanation of what types of decisions the model can improve and the limits to which it can be applied.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Accept in present form

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have submitted a substantially revised version of the article, which, to the extent possible, incorporates all the important comments and recommendations made during the previous review. These comments have been taken into account and reflected in the text, significantly improving the quality of the article. Given the work done, I believe this version of the article meets the journal's requirements and recommend it for publication.

Comments for author File: Comments.docx