Evaluating the Evolution of Soil Erosion under Catchment Farmland Abandonment Using Lakeshore Sediment

Round 1

Reviewer 1 Report (Previous Reviewer 1)

The authors revised the manuscript correctly and completely.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report (Previous Reviewer 2)

Dear editor,

The manuscript "Evaluating the evolution of soil erosion under catchment farmland abandonment using lakeshore sediment" submitted in Sustainability Journal. It's an outstanding research with the collaboration of Chinese researchers about soil erosion. The topic is so nice, also I reviewed this article for the third times. The author has included comprehensive corrections and the plagiarism (attached) shows that there is no particular problem in the article. I suggest that the article will be great help to the scientific community in relation to the topics of soil erosion and has the ability to be published in the journal.

Regards

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report (Previous Reviewer 3)

The authors have done a great job improving the quality of the manuscript, but some issues remain and need to be resolved before this work can be accepted. What I concerned most are the limitations in result interpretations due to concerns associated with particle size effect and tracer conservativeness in the depositional environment. The authors may add a new subsection in the discussion to indicate and discuss them. Besides, efforts are needed to improve and expand the current discussion. My specific comments are listed as follows.

·       Line 21: ‘abandoned land’ not ‘AL.’ If an acronym is used in the abstract, it should be spelled out (defined) in the abstract.

·       Lines 44-50: The authors are strongly suggested to present a better introduction for the focus, development, and application of the sediment source fingerprinting approach here. It would be very informative for readers that are not familiar with this approach.

·       Line 165: The authors used fractionation (<0.075 mm) to address the particle size effect on fingerprinting results. This approach is straightforward but has limitations. That is, the use of a single particle size range means that the corresponding source apportionment results would only relate to the fraction being used. For the case of this work, the use of a <0.075 mm fraction means that the fingerprinting results would not say much about the sources of lakeshore sediments with a particle size greater than 0.075 mm. That needs to be indicated in the discussion as Figure 5 showed that a high fraction of sand (>0.063 mm) was observed for the target sample (i.e., lake sediment core, 46.6%). This is a limitation in result interpretations, which merits further discussion.

·       Lines 356-367: The authors need to clarify how they get the results presented in Figure 6. Are these apportionment results from an analysis on a mixture of all sectioned samples of the collected lake core? Or are they just means of the values presented in Figure 7?

·       Lines 363-365: The use of elemental concentrations as tracers to assess the sources of deposited sediments could be problematic. It is likely that tracer signatures of deeper sections of the lake core have changed in the depositional environment due various stressors (e.g., anoxic environments) and no longer reflect the tracer properties of the original sources. Additionally, sediments at deeper sections of the lake core were deposited long time ago (e.g., nearly a century). Their source samples (i.e., catchment soils) may have different tracer signatures then than what is observed now due to natural and anthropogenic changes. This is another limitation in result interpretations, and the authors need to indicate and discuss that in their work.

·       Lines 482-483: As mentioned above, I am not sure how reliable the apportionment results would be for 1930s-1950s as tracer signatures may not be conservative after sediments were deposited for nearly a century.

·       Line 490: Inputs of anthropogenic sources was important but not dominant. According to Figure 7, sediments from fluvial erosion (CB) dominated all the time (i.e., >50%).

·       Line 492: “...had great or little influence...”? What do authors want to address here?

·       Lines 496-497: In most cases fluvial erosion (CB) would be accelerated by a steeper slop gradient (line 503) instead of a smaller one. A better justification is needed here.

·       Lines 504-506: What additional information does Figure 12 add to the existing results and how this would help people understand the sediment dynamics in the studied watershed? The authors are supposed to offer more discussions and thoughts instead of simply presenting trends and numbers.

·       Lines 506-509: This sentence is not clear and needs to be rephrased.

·       Lines 523: Please move the results of GOF and uncertainty analysis to the result section and provide more discussion here on what they indicate for the accuracy and precision of the fingerprinting results and whether they are good or not with some comparisons with other studies. If there is room for improvement, what can be done?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report (Previous Reviewer 3)

The authors have done a good job improving the quality of the manuscript, and I see the paper in good shape now. I recommend the acceptance of this paper in present form for Sustainability.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The authors investigated “The evolution of soil erosion under farmland abandonment in southwestern China: Evidence from spatiotemporal source apportionment of lakeshore sediment in a small agricultural catchment.

The Title is exciting but the Abstract is very complicated. I cannot understand why the authors cannot easily explain the work in the Abstract. Please correct.

Some previous studies also explained the role of abandonment of lands on environmental pollution. Please cite this work:

https://doi.org/10.1016/j.envpol.2020.115883

Please check the paper for English.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear editor,

The manuscript "The evolution of soil erosion under farmland abandonment in southwestern China: Evidence from spatiotemporal source apportionment of lakeshore sediment in a small agricultural catchment" submitted in Sustainability Journal. It's a nice and terrific research with the collaboration of Chinese researchers about soil erosion. The topic is so nice, but there are some problems in this MS that I can't accept. Finally, I present some suggestions for improving the quality of this MS as following:

1.      The title of the article is too long, it is recommended to make it shorter to attract more readers.

2.      In the abstract, please speak briefly about the methodology and the quantitative results. It’s very effective in increasing the readers and citations of the article. It can also help you to increase readers and citations in near future.

3.      The relationship between sustainability and soil erosion is not observed in the introduction. Please pay more attention to this subject.

4.      In study area, it is recommended to mention the spatial information related to the distribution of rainfall, geology, as well as land use.

5.      In figure 4, how do you explain the sudden changes between 1980 and 2000?

6.      The results and discussion are well prepared and deserve congratulations.

7.      In the conclusion, it is recommended to provide operational solutions for local communities in relation to sustainability and soil erosion.

8.      I checked plagiarism detection of this research and the similarity is 33% and there are some concerns; please check the attached file. Be sure to correct the highlighted items in the attached file.

9.      Please be sure that all the references cited in the manuscript are also included in the reference list and vice versa with matching spellings and dates.

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

In this study Wang and others investigated a plateau lake at Southwestern China, calculated lake sediment accumulation rates and assessed sediment sources of lakeshore sediments. This work fits into the remit of the targeted journal, i.e. Sustainability, and is important as quantitative information of sediment sources to rivers and lakes is important for identifying conservation and restoration actions. However, while this study could potentially be a nice contribution, there are some issues in methodology that need to be addressed before this manuscript can be accepted. I am especially concerned about the reliability of the sediment source fingerprint approach used in this work as most conclusions rely on robust sediment source apportionment results. I will list the issues in each procedural step that need to be addressed below.

1. Catchment source sampling

The authors took a sampling campaign for upland sources (lines 137-148) but did not collect any sample for channel sediment sources such as those through fluvial erosion (e.g., streambank) and from channel/near-channel sediment storage. This is not common practice for catchment source sampling design, especially for studies attempting to discern sediment sources according to source types (e.g., land uses). Sampling of representative samples for all potential sources is a prerequisite for the successful application of the fingerprinting technique, and the authors need to clarify why channel sediment sources are excluded (or add more data). Otherwise, relevant interpretations on sediment contributions from different land use types are not convincing.

2. Target sample sampling

It appears that the authors sampled suspended sediments as target samples (lines 242 and 255), but there is no relevant description in the methods.

3. Tracer selection

The authors conducted a range test (lines 236-248) and a PCA pre-screening (lines 249-255) to determine the “optimum composite fingerprints” for measured geochemical tracers. However, I do not believe it is a statistically robust way (please see Palazon and Navas, 2017, https://doi.org/10.1016/j.jenvman.2016.07.058). The authors are suggested to conduct the most commonly used two-step process of Kruskal-Wallis H-test and discriminant function analysis for tracer selection.

4. Source apportionment modeling

No effort was taken to validate model outputs, which is not acceptable for ensuring the reliability and accuracy of apportionment results. Most fingerprinting studies take a Monte Carlo uncertainty test and a model goodness-of-fit as part of a fingerprinting routine, and the authors are suggested to do so. The use of virtual sample mixtures is often considered the most robust way to validate models (e.g., Batista et al., 2022, https://doi.org/10.1007/s11368-022-03157-4), and the authors are strongly recommended to apply this technique in their work.

8. Specific temporal souring analysis for the sediment core

The temporal analysis presented at lines 269-280 does not make much sense to me. First of all, the authors attempted to use a binary model to identify sediments sources for a lake core using two distinct source classification systems (i.e., by geology and land uses) at the same time. However, these two classification systems are not mutually exclusive. For instance, upland sediments from areas with different land uses could be soils characterized by the same rock type (Figure 1). It is scientifically unsound to simply consider sediment from areas with different land uses as “anthropogenic sources” and from different rock types as “natural sources.” Secondly, lake sediment cores are not good target samples to investigate temporal changes in sediment source contributions from different land uses at a decade-to-century scale with the use of geochemical tracers. In the studied case, it is very unlikely that sediments from the bottom of the core can keep the tracer signal of their original sources after a century. Besides, there may also be significant changes in land uses as well as corresponding tracer signals of upland soils at a decade-to-century scale, and the authors did not present any data to clarify that. There is a lack of consideration of tracer conservativeness for this analysis, which makes the discussion at section 4.2 not convincible. If the authors want to discern sediment source contributions from anthropogenic versus natural sources using lake sediment cores, radiometric fingerprinting may be a better choice (e.g., Belmont et al., 2011, https://dx.doi.org/10.1021/es2019109). Lastly, as mentioned above, the authors selected tracers for this analysis simply based on comparing results of a range test and PCA pre-screening, which does not seem to be a statistically robust approach to find a group of tracers with the best discriminating power.

In summary, the authors are suggested to improve most procedural steps of the fingerprinting approach used in this study or demonstrate the reliability of their current approach. I would also recommend authors to review recent advances in sediment source fingerprinting techniques and include that in the introduction. In addition, the manuscript could use more polish with the writing before re-submission as several grammatical errors exist in the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear editor,

The manuscript "Evaluating the evolution of soil erosion under catchment farmland abandonment using lakeshore sediment" submitted in Sustainability Journal as a revision. It's an outstanding research with the collaboration of Chinese researchers about soil erosion. I receive a good revision from authors, and it can be published in Sustainability. But I’m a little concern about plagiarism, because the similarity in the previous version was 33%. Please send me a clean version for checking it.

Best Regards

Reviewer 3 Report

I welcome new data and believe that sediment source fingerprinting can be a reliable approach to assist conservation and restoration efforts. However, there are significant flaws in the procedural steps of the fingerprinting technique used in this work, as I mentioned in the first review. The authors chose not to make any improvement in the revised manuscript, but their arguments are not convincing. I will list them in detail as follows:

1. Sediment source classification

Responses from the authors showed that they did not collect any representative sediment samples for channel sources, which just greatly undermines the reliability of corresponding fingerprinting results. According to section 2.2, all sediment samples were “surface soil samples” (line 201), which cannot be used as surrogates for channel-source sediment samples even though a few of them are close to the river channel. Numerous previous fingerprinting investigations conducted in UK, Australia and US have indicated that sediments from channel sources, such as removal of sediment through processes of fluvial erosion, are dominating for transported sediment within riverine systems (>50%). If the study was conducted without representative channel sources and proper justifications, sediments from channel sources would just be falsely identified as coming from other upland sources. This is a significant flaw in study design, which cannot be resolved by any analysis conducted later.

2. Target samples

I thought the authors collected suspended samples as target samples, but the response from them indicated that they did not. Instead, they just used sediment cores as target samples for both contemporary and historical sediment sourcing information. Unfortunately, sediment cores are not good target samples for contemporary sediment sourcing investigations due to strong concerns associated with tracer conservativeness in long-term depositional environments. The particle sorting effect is another concern, as transported sediments in the riverine systems are often fine sediment, which may not be the case for lake sediment deposits. Figure 5 clearly shows that the particle size distributions between catchment soils and lake cores are very different, which further convinces me that fingerprinting results from lake cores are trivial. The authors reported that most catchment soils have a particle size range of <63 µm. If so, a more robust way to do fingerprinting investigations in their case is to collect suspended sediment samples as target samples and sieve both source and target samples to <63 µm fraction before any other analysis.

3. “Anthropogenic versus natural sources”

The authors’ response on this issue makes me feel that the authors lack a fundamental understanding for the use of sediment source fingerprinting approach. What authors referred as “anthropogenic versus natural sources” is the type of fingerprinting studies discerning sediment contributions from different land uses. However, this type of study is often conducted to discriminate upland sources (or human-activity-derived sources such as croplands, pasture lands) versus channel sources (natural sources such as fluvial erosion). The other type often seeks to determine the relative importance of different geographic locations as characterized by different rock types. These two broad categories have different underlying principles, but the authors just simply mixed them and claimed that to be an approach for “anthropogenic versus natural sources”. If the authors want to defense this approach, they have to add a theoretical context section to demonstrate the validity of their approach.

Reference: Walling 2013, https://doi.org/10.1007/s11368-013-0767-2

4. Other issues

I suggested authors to conduct a KW-DFA test for tracer selection as it is widely considered that KW-DFA is the tracer selection approach with the best discriminating power. In contrast, the PCA pre-screening is often consider less reliable and mostly used as a preliminary assessment for tracer selection. The authors did not reply to my suggestion in their response but addressed something else that is irrelevant. It is true that every method may have its own merits, but if the authors believe that PCA pre-screening is more robust, they need to either approve that or show that this approach has been wildly tested in other cases.

In addition, the author did not conduct anything to validate their model apportionment results. I suggested the virtual sample mixture test to them as it is considered the most robust model validating approach for fingerprinting studies, but the response from the authors show that they did not even take time investigating what it is. The author can also choose the widely used Monte Carlo uncertainty routine, which was conducted in 95% of fingerprinting publications during 2018-2019 (Collins et al. 2020, https://doi.org/10.1007/s11368-020-02755-4). There are many other approaches, and every method could have its own merits, as the author said in the response. However, it is not acceptable to do nothing for model output validation.

In summary, I would recommend authors to dive deeper for a better understanding of fundamentals of the fingerprinting approach and catch up with the latest advances in this technique (e.g., Collins et al. 2020, https://doi.org/10.1007/s11368-020-02755-4 and Evrard et al., 2022, https://link.springer.com/article/10.1007/s11368-022-03203-1). Without these issues being addressed, results and conclusions from this work may just be trivial.