Groundwater Pollution Source Identification Based on a Coupled PCA–PMF–Mantel Framework: A Case Study of the Qujiang River Basin

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents an integrated framework combining PCA, PMF, and the Mantel test for groundwater pollution source identification. The topic is relevant to the journal Water, as it addresses groundwater quality, hydrogeochemical processes, and source apportionment under complex hydrogeological settings. However, while the study is generally well-structured and supported by substantial data, there are several issues that require attention before publication.

Major Concerns

1. Introduction

- As the authors claim the transferability of their approach, general overview of groundwater pollution in the world needs to be mentioned before shedding light on the study area.

- Lines 46-48 need proper references.

- Lines 51-54: Classical diagrams are well established and can still be useful. The novelty of the study can be highlighted by only showing the effectiveness of the multivariate statistical methods. I suggest removing this sentence.

- I suggest adding more recent references on using multivariate statistical methods from different regions to highlight its generalizability. Example but not limited to:

[https://doi.org/10.1007/s11356-025-36175-z, https://doi.org/10.1016/j.heliyon.2022.e11308]

- The integration of PCA, PMF, and Mantel test is useful but not entirely novel, as these methods are well-established in hydrogeochemical studies. The authors should better clarify the unique contribution of their work compared to existing PCA–PMF studies.

2. Methodology

- Quality assurance/quality control (QA/QC) measures for chemical analyses should be described (e.g., ionic balance error, duplicate samples, blanks). This is critical for data reliability.

-The limitations of using a single-season dataset should be emphasized earlier in the methodology/results.

- More detail is needed on PMF settings (e.g., how uncertainty values were calculated, error fractions, number of runs, handling of below-detection-limit values).

- For the Mantel test, specify the exact environmental/land-use variables considered and the rationale for their selection.

3. Results and Interpretation

- As the correlation not always implies causations, proper discussion for the results of correlation analysis is required and needs to be supported by references

- Lines 235: what type of anthropogenic activity?

- Line 286: The explained variance is low. What can be the reason?

- Cl is depicted in PC1 and not mentioned in the discussion, meaning that not only dissolution is the main factor of pollution

- Mn²⁺ reflects both natural red-bed enrichment and industrial effluents. This statement needs to be supported.

- Proper discussion of the results of PCA is required and needs to be supported by references.

- More explanation is required on how species distributions in PMF factors were used to assign pollution sources.

- Explain the discrepancies between the results of PMF and PCA.

- The discussion should provide stronger connections to regional hydrogeology and human activity, rather than reiterating statistical outputs.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript (ID: water-3858762) addresses groundwater pollution source identification in the Qujiang River Basin using a coupled multivariate statistical framework.

Comments:

1. The keywords listed (PCA–PMF–Mantel; source apportionment; groundwater; Qujiang River Basin) currently repeat several terms that are already in the title. To increase discoverability, it is advisable to avoid duplicating title words in the keyword list.
2. The authors should explain what is new in this manuscript and that there is no overlapping text or data with their prior publication (Li et al., 2025) on “Hydrochemical Characteristics and Dominant Controlling Factors of the Qujiang River” and another (Liu et al., 2023) on surface-water/groundwater interaction in the Qujiang Basin
3. Although good charge balance is implied by samples plotting near the 1:1 line in the ion-balance diagram (Figure 7b), a numerical charge-balance evaluation for each sample is not provided, and no QA/QC subsection describes it alongside methods and software details.
4. The study area is referred to as the “Qujiang River Basin” initially and sometimes shortened to “Qujiang Basin” in a few places. This is fine after the first mention, but ensure it’s used consistently to avoid any doubt (both refer to the same area). For instance, in the conclusions, it’s called “Qujiang Basin” without “River,” which is acceptable but should be consistent.
5. There is a labeling error in the table numbering. The manuscript contains two consecutive tables labeled “Table 1.” The second of these is actually the statistical summary of hydrochemical indices (appearing after the analytical methods table) , and it should be labeled “Table 2” instead of a second “Table 1.”
6. Formatting issues were noted in how some data are presented, but not the values themselves. For instance, in Table 2, the coefficient of variation for some parameters is listed with a % sign but given as a decimal (e.g., “CV(%) 0.059” for downstream Cl in the parsed text, which presumably means 5.9%). In the final formatting, this should be corrected to either show 5.9% or remove the “%” if 0.059 is the fraction.
7. There are too many significant figures in many places in the manuscript. Some mean values are given to three decimals in the tables (e.g., Na⁺ mean 9.196 mg/L). It should round to 2–3 significant figures.
8. The study measured a solid core suite of analytes: major cations (Ca²⁺, Mg²⁺, Na⁺, K⁺), major anions (HCO₃⁻, SO₄²⁻, Cl⁻), nutrients (NO₃⁻, NO₂⁻, NH₄⁺), indicators (F⁻, As), trace metals (Mn, Ba, Sr), and aggregate COD. This covers natural weathering signatures and some anthropogenic impacts. However, the results and discussion for many of these analytes, which are typically important for groundwater pollution diagnosis, are missing. In addition, other heavy metals like Pb, Cd, and Cr were not listed; if industrial pollution was significant, those might be relevant.
9. The manuscript does reference several other studies throughout. Still, it currently does not provide a direct comparison of this study’s quantitative findings with those of different regions in a consolidated form (e.g., a comparative table). The authors could compile a new comparison table.
10. References 17 (Li, Y. 2022 thesis) and 18 (Yuan, Y. 2022 thesis) might be of limited interest to an international readership. If those particular data are not pivotal, the authors should remove these or replace them with published sources.
11. I recommend the authors consider citing the following recent studies: https://doi.org/10.3389/fenvs.2022.1107465. This reference presents a modern chemometric approach to pollution source apportionment, integrating multivariate statistics with risk assessment, including PCA/PMF. https://doi.org/10.1016/j.marpolbul.2024.116277. This article presents a recent example of a comprehensive groundwater quality assessment in a river basin, utilizing advanced statistical indices and risk analysis.

Overall, this manuscript offers a valuable contribution by demonstrating an integrated approach for groundwater pollution source identification. However, a revision is needed to contextualize the results further and refine the presentation.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

1. The location of the hydrogeological profile needs to be marked in Figure 1.
2. Quality control of water quality data needs to be supplemented.
3. In Table 1, all parameters have been rounded to three decimal places. Please modify according to the detection limit.
4. Figure 4 suggests supplementing the spatial distribution of TDS.
5. Why isn't correlation analysis separated by upstream, midstream, and downstream?
6. Is there no upstream water sample data in Figure 7d?
7. Is the high rotated factor loadings of each PC so low? There should be an error in PCA.
8. Change the title of the conclusion section to 'Conclusion and Suggestions' for better clarity

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

\no comments to the authors

Reviewer 2 Report

Comments and Suggestions for Authors

The author has responded to each of my questions and made all necessary revisions. After reviewing the author's revised manuscript, I have no further questions. Therefore, I recommend that the manuscript be accepted for publication. In my view, the article suits the main aims of the journal very well and will find a broad and interested readership.

Reviewer 3 Report

Comments and Suggestions for Authors

Accept in present form