Exploring the Behavior of the High-Andean Wetlands in the Semi-Arid Zone of Chile: The Influence of Precipitation and Temperature Variability on Vegetation Cover and Water Quality

Comments 1: Line 180 -The link supplied it does not work. Please provide the available and testable link.

Response 1: We regret the situation. The link was reviewed and confirmed to be correct; however, an error was identified in the formatting (Template) during the editing process. This issue has been addressed in the revised version of the manuscript.

https://developers.google.com/earth-engine/datasets/catalog/landsat

Comments 2: Line 192 – There is the same problem, the link supplied is not reachable.

Response 2: We regret the situation. The link is correct; however, the web page is temporarily out of service.

Comments 3 In the methods you mention you download Landsat 5 collection 2 Please specify if you have used Surface Reflectance, Top of Atmosphere or Raw Images. And please provide the same information concerning Landsat 8 OLI/TIRS. If you have chosen Raw Data please describe which operations of pre-processing you have applied; in line 227 please specify the algorithms used atmospheric correction.

Response 3:  Thank you for this comments. For both collection (Lansat 5 and Lansat 8) the surface reflectances were used. This information has been incorporated into the manuscript (in the section 2.5 Time series analysis, second paragraph).

Comments 4. From line 238 to 250 -Even though you mention the work of Gavilan et al 2019, the process of harmonization Landsat satellites TM and OLI in Landsat 7 ETM have to be describe in details.

Response 4. Thank you for pointing out this important aspect. The description of the harmonization procedure has been complemented in the manuscript (section 2.5 Time series analysis, fourth paragraph). The paragraph highlighted in black is the information that has been added

“To analyze the entire time series and because the image set with which the indices were calculated is from two Landsat satellites (TM and OLI), a harmonization procedure was applied to the images [60,61,62]. Due the operational periods of Landsat 5 and Landsat 8 do not overlap, for harmonization procedure the data from the images from both satellites were transformed to Landsat 7 (ETM). For each pair of images, 217 points were selected (those outside the confidence interval were removed). Using the selected sampling points, two scatter plots were created for each index (TM vs. ETM and OLI vs. ETM). The corresponding fitting equations were obtained using the least squares linear regression method. The harmonization equations are as follows:”

Comments 4. The correlation of water variables with NDVI and NDMI indices determined with the Pearson coefficient is significant only in the case of Fe while with the other components it is either weakly significant or even non-significant. Therefore the conclusions reported from line 443 to 448 should be revisited.

We sincerely appreciate the reviewer’s time and effort in reviewing our manuscript and providing valuable feedback. After carefully reviewing the proposed analysis, it can be stated that, according to the literature (Townend, 2002), the parameters (Cl and Na for NDVI and CE, Cl, Mg, Na, Fe for NDMI) exhibit a statistically significant correlation, similar to Fe, as the p-value is less than 0.05 (5%). The difference is that Fe shows a positive correlation, whereas the other parameters show a negative correlation (i.e., higher values of the indices (NDVI/NDMI) are associated with lower concentrations of the parameter in the water). Furthermore, we acknowledge that there are various defining the strength of correlation. To strengthen the manuscript, the classification of the strength of correlation has been included in the methodology section (2.7 Least-squares regression analysis).

Townend, J. Practical Statistics for Environmental and Biological Scientists; Wiley: Chichester, UK, 2002; ISBN 978-0-471-49665-6.

Comments 5 The article needs to be revised because it addresses some passages too hastily, some of the conclusions need to be modified.

We appreciate the reviewer’s comment regarding to improve the manuscript. In response, we have carefully revised results and conclusion to address this concern. Statistical analyzes on the influence of environmental and terrain factors (wetlands area, slope, altitude, temperature) were incorporated, Analysis of the time series of each wetlands the geological analysis of the area was expanded to identify factors related to the existence of groundwater. We believe these changes enhance the robustness of our manuscript. These analyses have been incorporated into the following section of the manuscript: 3.1. Analysis of temporal behavior of indices with respect to wetland chain precipitation and temperature (Paragraph 6 and 7); Section 3.2: Spatial Analysis of Vegetation Cover by Wetland; 3.2.1. Influence of Terrain Characteristics, 3.3.Linear relationships between vegetation cover and selected water quality parameters and 4.0 Conclusions. The wetlands indices time series were included in Supplementary Materials.

5. Additional clarifications

1.       Due to the modifications made to the manuscript, the following title is proposed: “Exploring the behavior of high-Andean wetlands of the semi-arid zone of Chile: Influence of precipitation and temperature variability on vegetation cover and water quality.”

2.       The information incorporated into the revised version of the manuscript is highlighted in yellow in the text.

Comments 1: This article is a solid traditional time-series analysis of remotely-sensed data with the addition of some hydrochemistry data. It does not make the most of the chemistry data however, and it is limited and inconsistent with the analyses. The authors take time and effort to separate and describe the wetlands as individual pieces then do not perform all analyses on each of them.

The use of chemistry data is limited and there are few conclusions drawn from them. Different chemical signatures might be indicative of particular flow pathways, e.g., surficial, shallow sub-surface, or slower water movement through particular rock types, or perhaps flowed through particular individual wetlands or soil systems. Do the authors have anything to add with the hydrochemistry?

Response 1: We really appreciate the reviewer’s comment. A more complete version of the water quality parameters is available; however, in the initial version, these were not included in the Linear relationships analysis, due to the reduced amount of data available, which decreased to less than 30 data points per parameter after applying the nearest-date filter (15 days) to the satellite images. This situation made it difficult to analyze the correlation with the indices (NDVI/NDMI). However, based on your comment and to better characterize water quality, we have now included a characterization of the temporal behavior of the some water quality parameters (box plot). This analysis generally shows low concentrations of the constituents. (This was incorporated in Section 3.3. Linear relationships between vegetation cover and selected water quality parameters). Moreover, the time series of the other parameters was included in Supplementary Materials.

Comments 2: I found the whole system versus individual wetland is inconsistent. There is some temporal analysis of the wetland system as a whole, but none for the individual wetlands. It would be interesting to find out if the higher elevation wetlands had different chemical signatures, or NDVI/NDMI regressions, than the lower elevation ones. Were there differences based on the slope of the wetland? On the area of each wetland? Was the time delay between climatic and remotely-sensed indices variable between wetlands?

Response 2: Thank you very much for your insightful comment, which undoubtedly strengthens our analysis. Unfortunately, hydrochemical data for individual wetlands is not available. However, based on your suggestion, we conducted NDVI/NDMI regressions for each wetland and included an analysis considering variables such as slope, area, and altitude of each wetland. Additionally, a temporal analysis and correlation with average temperature were carried out. These analyses have been incorporated into the following section of the manuscript: 3.1. Analysis of temporal behavior of indices with respect to wetland chain precipitation and temperature (Paragraph 6 and 7); Section 3.2: Spatial Analysis of Vegetation Cover by Wetland; 3.2.1. Influence of Terrain Characteristics. The wetlands indices time series were included in Supplementary Materials.

5. Additional clarifications

1. Due to the modifications made to the manuscript, the following title is proposed: Exploring the behavior of high-Andean wetlands of the semi-arid zone of Chile: Influence of precipitation and temperature variability on vegetation cover and water quality.

2. The information incorporated into the revised version of the manuscript is highlighted in yellow in the text.

Comments 1: The manuscript mentions the use of Landsat 5 and 8 imagery but does not provide a clear rationale for selecting these datasets, such as their temporal coverage, spatial resolution, and suitability for the study area's wetland characteristics, and the reasons for missing data in 2003, 2012, and 2013. The authors should elaborate on the criteria for dataset selection, the specific preprocessing workflow, and the computational methodology for vegetation indices. This would increase the reproducibility of the analysis

Response 1:

We appreciate the reviewer’s comment regarding to enhance the reproducibility of our study. In response, Landsat 5 and Landsat 8 were selected because their spatial (30 m) and temporal (16-day revisit time) resolutions are well-suited to the study area. Additionally, the operational periods of both satellites coincided with the availability of precipitation and water quality data. For 2003, only one image was available (December 29), which was associated with the summer season of 2004 (Austral Summer, December–March). A similar situation occurred in 2012 and 2013 due to the transition between satellites.

Moreover, the methodology section of the manuscript (2.5 Time series analysis) has been updated to include the information described above, with further details on the harmonization process and a more comprehensive workflow (Figure 2) to improve clarity and reproducibility.

Comments 2: The study highlights a negative correlation between precipitation and vegetation indices (NDVI, NDMI); however, it lacks an in-depth discussion of potential confounding variables such as temperature, groundwater availability, or soil moisture. These factors may interact with precipitation and significantly influence vegetation trends. A multivariate analysis that incorporates these variables would provide a more comprehensive understanding of the vegetation-climate relationship. The authors should explicitly discuss how other environmental factors could modulate the observed patterns.

Response 2: Thank you for highlighting this important aspect. Regarding temperature, we included an analysis (along with a graph in Figure 3) on the long-term temporal variability of the annual mean temperature and its relationship with the indices (NDVI/NDMI) (It has been included in the section 3.1. Analysis of temporal behavior of indices with respect to wetland chain precipitation and temperature). Additionally, a correlation analysis was conducted between the indices and the mean temperature, considering annual averages with one- and two-year lags, as well as summer averages without lag. The results were statistically non-significant (Table 2).

As for groundwater availability and soil moisture, given that the study area is remote and difficult to access, there is no historical data available for these factors. Consequently, based on the existing information, a multivariate analysis was not feasible.

From the geological map analysis, we identified low-permeability rocks (mainly plutonic and metamorphic) in the study area, which suggests limited water infiltration. However, the presence of alluvial fans supports water storage and infiltration, which, in turn, favors the formation and maintenance of wetlands (Harvey, 2018). Additionally, the Las Gualtatas Fault was identified within the study area, potentially acting as a geological barrier or facilitating water infiltration. The geological map and the identification of alluvial fans across the wetland chain have been included in the supplementary files (It has been included in the 3.2.1. Influence of Terrain Characteristics on the wetlands)

Comments 3: The manuscript observes that vegetation cover dynamics vary across wetlands, potentially due to differences in altitude, terrain slope, and additional water inputs. However, the analysis of these variations remains superficial. For example, the reasons some wetlands experienced less vegetation change during the 'megadrought' period are not fully explained. Incorporating a quantitative analysis of these factors, including local hydrology and geological features, would strengthen the discussion. Field data or historical records could be integrated to validate these findings.

Response 3: We appreciate the reviewer’s suggestion to enhance the analysis results. We acknowledge that the scarcity of field data represents a limitation of our study. To address this, we have proposed the implementation of a field monitoring program to the local community as part of a second project phase. Nonetheless, based on the available geological characteristics of the study area and individual wetlands (e.g., area, slope, and altitude), we have refined and expanded the analyses to provide more detailed insights (It has been included in the 3.2.1. Influence of Terrain Characteristics on the wetlands). Additionally, further data were incorporated into the analysis of the megadrought's influence on vegetation cover behavior (3.1. Analysis of temporal behavior of indices with respect to wetland chain precipitation and temperature; second paragraph).

Comments 4: In addition to precipitation, the manuscript briefly mentions correlations between water quality parameters (e.g., EC, Cl, Mg, Na, Fe) and vegetation indices but does not explore their specific impacts on vegetation health. This aspect is critical, particularly in semi-arid wetlands where salinity and mineral content can significantly influence vegetation dynamics. Expanding the discussion to detail how these parameters affect vegetation, supported by relevant studies, would add depth to the analysis. Additionally, the manuscript title should reflect this broader focus on water quality.

Response 4 Thank you very much for your comment, which undoubtedly strengthens our manuscript. Unlike the wetlands of the Chilean Altiplano, which exhibit characteristics of salt flats with high concentrations of water quality parameters, the concentrations in the Claro River are considerably lower and do not reach levels toxic to plants. This is particularly true for sodium (Na+), which, while beneficial to plants at low concentrations, is a well-documented toxic element at high concentrations—for example, Na+ levels of 1.5–2.9 g/L, which are far above the concentrations observed in the study area.

To address and strengthen this point, a boxplot was incorporated into the manuscript to illustrate the concentrations of water quality parameters related to correlation analysis. Additionally, we included in the manuscript a clarification on this situation (Section 3.3. Linear relationships between vegetation cover and selected water quality parameters). (Time series of water quality parameters were included in supplementary files).

Comments 5: While the manuscript provides valuable insights into precipitation impacts on vegetation, it does not clearly outline future research directions or study limitations. A discussion of how the findings inform broader climate change research and potential future work would enhance the manuscript's contribution.

Response 5.

Thank you for pointing out this important aspect. The following has been incorporated into the Conclusion section:

“This exploratory analysis of precipitation impacts on vegetation using remote sensing has proven particularly valuable for characterizing the spatiotemporal dynamics of the wetland chain, a unique and remote mountainous ecosystem with limited accessibility. It has also facilitated the identification of certain environmental, terrain, and water quality factors influencing the wetlands. However, the study faced limitations related to the spatial availability of field data, such as detailed information for individual wetlands, water quality characterization specific to each wetland, and the hydrological characterization of both surface and subsurface contributing flows. Additionally, the need for increased monitoring frequency of chemical parameters and flow measurements has been identified.

Regarding future work, the agricultural community managing the Nature Sanctuary has been approached with a proposal to conduct hydrological and hydrochemical analyses based on field monitoring programs. These efforts aim to enhance the spatiotemporal understanding of the wetland chain and address the identified data gaps”.

Additionally, the relationship between vegetation cover behavior and climate change, linked to the megadrought phenomenon, was strengthened in the results and conclusions.

Comments 6: The manuscript focuses primarily on temporal interactions between precipitation, water quality parameters, and vegetation cover but neglects spatial variability. For instance, the spatial distribution of precipitation and water quality data and their influence on vegetation patterns over time are not adequately addressed. Including a spatial analysis would improve the understanding of these dynamics.

Response 6. Thank you very much for your feedback, which will undoubtedly enhance the understanding of these dynamics. However, given that the study area is mountainous, remote, and difficult to access, water quality data is spatially limited. There is only one monitoring station located downstream of the wetland chain, which was used to characterize the temporal dynamics through the presented method of correlating indices and water quality. Nevertheless, analyses of environmental factors and terrain characteristics (e.g., temperature, area, altitude, and slope of the wetlands) have been incorporated into this revised version, as detailed in the Results section (3.2.1. Influence of Terrain Characteristics on the wetlands).

Comments 7:  Figures such as Figure 1 and Figure 6 require improvement for better readability. Enhancing resolution, labeling, and ensuring clarity in visual representations would significantly improve comprehension.

Response 7: Both figure were improvement. For further detail, a figure illustrating the alluvial fans identified throughout the entire wetland chain has been included as supplementary file.

Comments 8. Formatting errors include the missing space before '[39]' in Line 92 and inconsistent formatting of 'km²' in Lines 92 and 108. These should be corrected for consistency.

Response 8: Thank you, both error have been corrected in the manuscript

5. Additional clarifications

1.       Due to the modifications made to the manuscript, the following title is proposed: Exploring the behavior of high-Andean wetlands of the semi-arid zone of Chile: Influence of precipitation and temperature variability on vegetation cover and water quality.

2.       The information incorporated into the revised version of the manuscript is highlighted in yellow in the text.