The Impact of Salinization and Wind Erosion on the Texture of Surface Soils: An Investigation of Paired Samples from Soils with and without Salt Crust
Round 1
Reviewer 1 Report
Overview
This paper looks at the potential salt crust mediation of wind erosion soil deflation of fines. They examined this using a space-for-time approach pairing crusted and uncrusted locations at sites presumably on the same soil that have presumably ungone wind erosion to cause hypothesized texture change. The paper is generally well written and straight forward, but there are a couple points of concern for me. The biggest is demonstrating the soundness of the space-for-time assumptions. Soils, even in low-relief alluvial systems can have high spatial heterogeneity, even in short distances (see early works of Burrough). It just make me nervous that in selecting paired sites, the authors didn’t bother to auger a little deeper to make sure the sites being paired looked similar with regards to texture in the subsoil. Without some amount of evidence to better justify the pairing, it’s hard for me to recommend this for publication. I think the authors could still use their current design using a pooled t-test, which would not offer the same statistical power, but as a pedologist who has described thousand of soil profiles, I need a bit more evidence that the assumptions are valid here.
Do these sites flood? If so, the authors would need to demonstrate that water sedimentation processes isn’t also differentially impacting the crusted versus uncrusting paired sub-sites. The same question could potentially arise if these sites were indurated with water for large portions of the year due to other factors (e.g. overuse of flood irrigation).
It would have been interesting to see a bit more effort be put into understanding the drivers of crust strength. Would be an easy multiple regression analysis to understand how varying textures, land uses, and salt contents affect crust strength. Could just do the regression on the crusted sites… The analyses and interpretation could use a bit more thoughtful elaboration in general.
I also wish the authors thought more about the broader implications of the results. It feels like a loose-loose situation described without any context. Salinization is not a positive phenomenon, but it is keeping the soil from blowing away in this case. So, what should be potentially done? If we correct the salinity issue, which seems to have been framed as an issue largely driven by poor irrigation practices, then we end up with dust emissions? Is this just a hopeless duel pronged problem? Have there been attempts to restore these areas for both salinity and dust emissions? What could that look like?
Specific Comments
L93-94. You’ve already established that the Tarim is a dust source in recent text, don’t need to repeat here.
L100. ‘Irrational’ is a bit subjective, maybe change to: ‘Recently, irrigation practices have caused…’
L104-106. Do you mean km2 for the units?
L111-119. Did you all do anything else to make sure there wasn’t inherent differences in underlying soils? Would be nice to have had auger descriptions that ensured the soils didn’t have contrasting underlying textures and salt contents more related to the original depositional parent material that may be driving differences between pairs. Just having quick hand textures of the sub-soils would be enough to provide much more confidence in the space-for-time design assumptions underlying this study design.
Sections 3.1 and 3.2. The power of paired t-tests lies in the ability to focus on the differences between all the pairs as opposed to the pooled means. It’s fine to include some of the pooled statistics, but you should be reporting the distribution of the differences between the pairs as that is the more relevant measure both for the statistic itself and for interpretation. Your statistical descriptions would be more suitable for a pooled t-test design, which may have worked here and shown differences, but is also less powerful.
Table 2. There are several instances of crusted silt loams being paired with a sand this seems like a pretty extreme change for texture. I know that texture coarsening is an established process in wind erosion, but the pairing design here is a space for time design that assumes you are sampling from the ‘same soil’ that has just undergone different erosional winnowing due to modern development of salt crusts as I understand the premise of the study. This goes back to my concerns about the pair selection as it just seems like a really large change in texture since silt loams can have an absolute maximum of 50% sand, but usually have much less. It would just be much easier to believe this if I knew the rest of the soil profile looked similar between the paired samples. If you were to auger down at these paired silt loam – sand sites, and they were both silt loams and sand all the way down to 2m, then your space-for-time assumption would not hold much weight…
L212 typo in describing the D95
Author Response
This paper looks at the potential salt crust mediation of wind erosion soil deflation of fines. They examined this using a space-for-time approach pairing crusted and uncrusted locations at sites presumably on the same soil that have presumably ungone wind erosion to cause hypothesized texture change. The paper is generally well written and straight forward, but there are a couple points of concern for me. The biggest is demonstrating the soundness of the space-for-time assumptions. Soils, even in low-relief alluvial systems can have high spatial heterogeneity, even in short distances (see early works of Burrough). It just make me nervous that in selecting paired sites, the authors didn’t bother to auger a little deeper to make sure the sites being paired looked similar with regards to texture in the subsoil. Without some amount of evidence to better justify the pairing, it’s hard for me to recommend this for publication. I think the authors could still use their current design using a pooled t-test, which would not offer the same statistical power, but as a pedologist who has described thousand of soil profiles, I need a bit more evidence that the assumptions are valid here.
Answer: I understand your concerns, I missed some important information in last version. The sites were chosen based on a criteria : ( ⅳ) the soil texture (5-20 cm) underlying soil sample was same between paired samples in same sites. We assume that the particle size distribution of top soil only was influenced by wind erosion. In fact, the more salt crust sites can be found due to serious soil salinization along the Tarim river Basin, but only 23 sites meet this criteria. I added this information in Material and Method, Line: 113-114. A pooled t-test was used to investigate the statistical difference, line : 146, the test results was similar with paired T- test, only very small difference for P value.
Do these sites flood? If so, the authors would need to demonstrate that water sedimentation processes isn’t also differentially impacting the crusted versus uncrusting paired sub-sites. The same question could potentially arise if these sites were indurated with water for large portions of the year due to other factors (e.g. overuse of flood irrigation).,
Answer: There were flood occurred in history, but a new embankment project reduced the natural river flooding of the Tarim River, and since the beginning of the 1990s, desiccation occurred on the lower reach of the Tarim River because of the high intensity of water resource utilization in the upper reach of the river basin. The lower reaches of nearly 363 km had been dry for 30 years. Thus we think a minor influence by flood.
It would have been interesting to see a bit more effort be put into understanding the drivers of crust strength. Would be an easy multiple regression analysis to understand how varying textures, land uses, and salt contents affect crust strength. Could just do the regression on the crusted sites… The analyses and interpretation could use a bit more thoughtful elaboration in general.
Answer: thanks for your suggestion, we do a multiple regression analysis, a good relationship (R2=0.83) was found tween the clay, silt, pH, thickness of crust and strength (line :238-242), which provide more information to explained the crust strength, line: 330-340.
I also wish the authors thought more about the broader implications of the results. It feels like a loose-loose situation described without any context. Salinization is not a positive phenomenon, but it is keeping the soil from blowing away in this case. So, what should be potentially done? If we correct the salinity issue, which seems to have been framed as an issue largely driven by poor irrigation practices, then we end up with dust emissions? Is this just a hopeless duel pronged problem? Have there been attempts to restore these areas for both salinity and dust emissions? What could that look like?
Answer: thanks for your suggestion, it was very important for this manuscript. I understand what you mean, I miss some important implication. The both soil erosion and soil salinization were serious damage to human environment, it is an undesirable that to increase area of soil salinization to reduce the soil wind erosion. The soil erosion (soil become coarse) recovery was extremely difficult, because that the improvement of soil property was a long process. But the amelioration of saline soils was faster and easier by soil flushing. The leaching of saline soils was simple and easy to use for local farmer, which leads to an increase in the area of cultivated land in Tarim river. However, the wind erosion risks will be increased in soil without salt crust, because a reduction of dust emission by salt crust has been confirmed in our research. It looks like the soil degradation could be improved by saline soil amelioration, but it may introduce the new soil degradation risks (soil erosion).
Therefore, our research results imply that wind erosion risks increased should be considered if the salinity soil was ameliorated and transformed to cropland. We also suggested that some conservation agriculture practice (eg. No-tillage, manure application, residue cover and addition of clay) must be considered to reduce the soil loss by wind erosion Otherwise, the original threat of soil salinization may be transformed to soil erosion, and may be lead to a further degradation of lands, which could provide reference to decision maker.
These information was added in “discussion”, line : 347-376.
Specific comments
L93-94. You’ve already established that the Tarim is a dust source in recent text, don’t need to repeat here.
Answer: revised, I deleted this sentence. Line :93.
L100. ‘Irrational’ is a bit subjective, maybe change to: ‘Recently, irrigation practices have caused…
Answer: revised. Line: 99.
L104-106. Do you mean km2 for the units?
Answer: it’s hm2, I checked it.
L111-119. Did you all do anything else to make sure there wasn’t inherent differences in underlying soils? Would be nice to have had auger descriptions that ensured the soils didn’t have contrasting underlying textures and salt contents more related to the original depositional parent material that may be driving differences between pairs. Just having quick hand textures of the sub-soils would be enough to provide much more confidence in the space-for-time design assumptions underlying this study design.
Answer: the soil texture (5-20 cm) underlying soil sample was same between paired samples in same sites, I have added these information, line : 113. ….
Sections 3.1 and 3.2. The power of paired t-tests lies in the ability to focus on the differences between all the pairs as opposed to the pooled means. It’s fine to include some of the pooled statistics, but you should be reporting the distribution of the differences between the pairs as that is the more relevant measure both for the statistic itself and for interpretation. Your statistical descriptions would be more suitable for a pooled t-test design, which may have worked here and shown differences, but is also less powerful.
Answer: a pooled t-test was used to investigate the statistical difference, line: 146.
Table 2. There are several instances of crusted silt loams being paired with a sand this seems like a pretty extreme change for texture. I know that texture coarsening is an established process in wind erosion, but the pairing design here is a space for time design that assumes you are sampling from the ‘same soil’ that has just undergone different erosional winnowing due to modern development of salt crusts as I understand the premise of the study. This goes back to my concerns about the pair selection as it just seems like a really large change in texture since silt loams can have an absolute maximum of 50% sand, but usually have much less. It would just be much easier to believe this if I knew the rest of the soil profile looked similar between the paired samples. If you were to auger down at these paired silt loam – sand sites, and they were both silt loams and sand all the way down to 2m, then your space-for-time assumption would not hold much weight…
Answer: I understand your concern, it was very important to explain sampling from the ‘same soil’. we have not taken soil sample down to 2m, but we take soil sample based on a criterial: the soil texture (5-20 cm) underlying soil sample was same between paired samples in same sites. It was why only 23 sites was chosen for a 1321 km of river. Line :113.
L212 typo in describing the D95
Answer: revised, line 221.
Reviewer 2 Report
General
If the paper is being submitted to the “Soil erosion control and Land Degradation Neutrality” special issue, then the text needs to make explicit reference to LDN and how this paper contributes to that debate. The special issue is not intended for consideration of soil erosion only.
Up-to-date recent references are needed – many relate to the 1990s and most others are dated to the years before 2016.
Edit text carefully for word spacing, italics, and consistent abbreviations of journal titles; check previous papers in LAND to format text correctly.
How were paired sample sites defined (i.e., how/why were they chosen)? Field sampling rationale needs to be explained.
Please describe the occurrence of salt crusts – were they predominantly situated in minor depressions in the landscape (relative to their adjacent soils without crusts)? How extensive were the crusted areas?
Specific comments
The approximate weight of individual soil samples collected in the field should be noted.
Lines 13-15 – does the second sentence (‘Such results have been reported by many studies’) refer to wind erosion affecting soil surface textural properties, or to wind erosion effects on textural properties being reduced by the presence of salt crusts? Please clarify.
Lines 86 and132-135 refer to chemical analyses which were not reported in a table – the description under Results 3.1 is insufficient. Given that later (line 264) you mention that ‘puffy’ halite crusts were not present, the chemical composition of the salt crusts reported in this study needs to be reported and included in the Discussion.
Lines 135-140 are misplaced. These sentences should appear under 2.2 Soil sampling (i.e., field procedures).
Line 183, Table 1 – column 7 is presumably ‘median’ (not ‘medium’)
Line 196, Table 2 – re-name title of columns 1 and 4 from ‘No.’ to ‘Sample site no.’
Line 208, Fig. 3 – y axis is labelled ‘volume’ – volume of what? Please indicate clearly what is being referred to here.
Line 229, Fig. 5 – would quadratic functions be a better fit than a straight-line correlation?
Lines 281-282 – need to provide potential explanations for formation of clay aggregates. To what extent is flocculation involved? Is organic matter important? If clays aggregate into larger particles, then why is the silt component more important in wind erosion than clays (aggregated into silt-sized particles)? Why are silt contents different between paired samples at a particular site but not between all samples at the regional level? Some more discussion around these kinds of results would be of interest to readers.
Line 266 – ‘…the salt crust showed greater strength with larger aggregates’. Please indicate approximate size/s of these aggregates.
Author Response
If the paper is being submitted to the “Soil erosion control and Land Degradation Neutrality” special issue, then the text needs to make explicit reference to LDN and how this paper contributes to that debate. The special issue is not intended for consideration of soil erosion only.
Answer: I have provide new information to make explicit reference to LDN in “discussion” ,line 347-376. Our research wanted to express that “wind erosion risks increased should be considered if the salinity soil was ameliorated and transformed to cropland. Because the leaching of saline soils was simple and easy to use for local farmer, which was lead to the cultivated land area rose in Tarim river. We suggested that some conservation agriculture practice (eg. No-tillage, manure application, residue cover and addition of clay) must be considered to reduce the soil loss by wind erosion, if the soil was remove the salt or salt crust by soil leaching. Otherwise, the original threat of soil salinization may be transformed to soil erosion, and may be lead to a further degradation of lands. Thus our research results provide present insights to relevant to the implementation of LDN”.
Up-to-date recent references are needed – many relate to the 1990s and most others are dated to the years before 2016.
Answer: I have updated new references in recent years, only five references (before 2000) was retained, which was necessary to support this research.
Edit text carefully for word spacing, italics, and consistent abbreviations of journal titles; check previous papers in LAND to format text correctly.
Answer: I have check carefully through manuscript (include reference list), to meet “LAND” formation.
How were paired sample sites defined (i.e., how/why were they chosen)? Field sampling rationale needs to be explained.
Answer: I have added new information in 113-114
Please describe the occurrence of salt crusts – were they predominantly situated in minor depressions in the landscape (relative to their adjacent soils without crusts)? How extensive were the crusted areas?
Answer: I have added new information, line “113-119”.
Specific comments
The approximate weight of individual soil samples collected in the field should be noted.
Answer: I have added new information in “”, Approximately 2 kg of soil sample was taken at each site. line 119.
Lines 13-15 – does the second sentence (‘Such results have been reported by many studies’) refer to wind erosion affecting soil surface textural properties, or to wind erosion effects on textural properties being reduced by the presence of salt crusts? Please clarify.
Answer: revised, should be “wind erosion affecting soil surface textural properties,” line 15.
Lines 86 and132-135 refer to chemical analyses which were not reported in a table – the description under Results 3.1 is insufficient. Given that later (line 264) you mention that ‘puffy’ halite crusts were not present, the chemical composition of the salt crusts reported in this study needs to be reported and included in the Discussion.
Answer: I have added new figs (2C, 2D and 2E) to exhibited the chemical composition of the salt crusts soil and non-crust soil. Some new content was added in “discussion”, to explicit why the salt crust showed great strength in Tarim river. Line : 281-287.
Lines 135-140 are misplaced. These sentences should appear under 2.2 Soil sampling (i.e., field procedures).
Answer: revised, line : 127-131.
Line 183, Table 1 – column 7 is presumably ‘median’ (not ‘medium’)
Answer: revised, line: 205.
Line 196, Table 2 – re-name title of columns 1 and 4 from ‘No.’ to ‘Sample site no.’
Answer: revised, line 207.
Line 208, Fig. 3 – y axis is labelled ‘volume’ – volume of what? Please indicate clearly what is being referred to here.
Answer: it’s should be “volume of particle”, revised, similar in fig 4.
Line 229, Fig. 5 – would quadratic functions be a better fit than a straight-line correlation?
Answer: we fit the curve used both of linear and quadratic, a better fit for quadratic, line : 233,236.
Lines 281-282 – need to provide potential explanations for formation of clay aggregates. To what extent is flocculation involved? Is organic matter important? If clays aggregate into larger particles, then why is the silt component more important in wind erosion than clays (aggregated into silt-sized particles)? Why are silt contents different between paired samples at a particular site but not between all samples at the regional level? Some more discussion around these kinds of results would be of interest to readers.
Answer: silt contents different between all samples at the regional level, it was attributed to high spatial heterogeneity, a larger difference for silt content along the Tarim river has been reported by previous studies, line :264-265. As a whole, there was two probably reason lead to a difference for silt between paired soil, the first, the soil particles were preferentially depleted by enhanced wind erosion, but clay was confirmed prefer to aggregated into larger sizes. The second, a high silt content was found in our soil samples, which was provide potential erodible particle. A very low SOM (the data was added in Table 1) was found across all soil samples. thus we did not consider a contribution to soil property from SOM. line : 301-313.
Line 266 – ‘…the salt crust showed greater strength with larger aggregates’. Please indicate approximate size/s of these aggregates.
Answer: this information was added, Line : 281-282.
Reviewer 3 Report
1. No research hypothesis is presented in the introduction;
2. The quality of Figure 1 should be better;
3. In the Table1 and Figure 3 significant differences between the treatments are not marked (asterisk or different letters);
4. In the Figures 3 and 4 it is not explained what the vertical bars indicate;
5. In the Figure 5 probability level is not presented;
6. I recommend not to use old literature sources;
7. I advise splitting the discussion into smaller chapters;
8. I recommend moving Figure 6 to chapter "Results".
Author Response
No research hypothesis is presented in the introduction;
Answer: A specific hypothesis is tested: compared to soils without a salt crust, salt crust soils exhibited a finer particle size due to the anti-wind erosion nature of salt crust soils. Line: 83-85.
The quality of Figure 1 should be better;
Answer: a new figure was provided, it looks better, line 120.
In the Table1 and Figure 3 significant differences between the treatments are not marked (asterisk or different letters);
Answer: the significant differences was marked in asterisk in Table 1, the significant differences was texted for fiugre 3. Line 214.
In the Figures 3 and 4 it is not explained what the vertical bars indicate;
Answer: the vertical bars indicated volume of particle (%) in fig 3, and strength and thickness in fig 4 (new fig number is fig 5)
In the Figure 5 probability level is not presented;
Answer: The probability level was added in text, line 234, 235,238.
I recommend not to use old literature sources;
Answer: I have updated new references in recent years, only five references (before 2000) was retained, which was necessary to support this research.
I advise splitting the discussion into smaller chapters; 4.3 soil salinization and soil erosion (Line 347).
Answer: I have added subtitle in “discussion”, 4.1 soil texure (Line 253), 4.2 salt crsut (Line 314) and 4.3 soil salinization and soil erosion (Line 347).
I recommend moving Figure 6 to chapter "Results".
Answer: I have move the figure 6 to chapter "Results", fig 4, line 228.
Round 2
Reviewer 1 Report
Overview
The authors have largely addressed my major concern about the pairing design. Although the 5-20 cm texture is the bare minimum to have confidence in the space for time approach here, I’ll go with it.
I am a little disappointed with the switch to a pooled approach as the paired test is the best statistical approach (since you’ve established your pairing assumptions better). However, the differences in the results are clear enough that the pooled test still gets the job done.
Unfortunately, the new text all needs considerable English editing to be suitable for publication – more comments below.
Specific Comments
L20. Did you switch to a pooled t-test? Please update and make sure this is consistent throughout paper.
L27-28 This sentence doesn’t make sense. Maybe something like: ‘Variation in salt crust strength can influence dust emissions and needs to be considered in future management.’
L84-85 maybe: ‘… due to the erosion-resistant nature..’
L116 something is missing in this sentence.
Section 3.3 needs to be cleaned up. I appreciate that the authors tried out a MLR model, but this needs a mention in the stats section of the methods so that it doesn’t come out of the blue here. The equations are difficult to read. I think if you removed the units from the equations it could clean it up a bit. You could also use an asterisk scheme to indicate which explanatory variables were significant at different alpha values to help interpret the model a bit more.
L265-266. This interpretation is not completely correct as you have both pooled and paired comparisons that indicate shifts in texture between crusted/non-crusted – so not just in paired samples for your results.
L274 This is wordy and the grammar/language is off. You can say this simply as ‘…were protected by salt crusts.’
L282 delete ‘, but’.
L283-284 aggregates
L285 sentence needs to be edited.
L288-289 grammar edits needed.
L 306 A clay amendment…
L301-315 Needs English editing for grammar, tense, subject-verb agreement, and sentence structure.
L321 and entire paper: Please update units to actual superscripts and subscripts for final version.
L339 -340. The statement about the decrease in Rsq needs more context. Can’t expect the reader to remember original result. Maybe say the Rsq decreases by X.XX.
L350-375 needs editing. The content is good but gets lost in the similar English issues noted for the previous sections of new text. This needs to be fixed before publication.
L379. Need to tone down the language. This is not a causal framework so subbing ‘suggesting’ for ‘indicating’ would be an appropriate inferential statement.
L380. Replace ‘anti-wind erosion’ with a more appropriate description: e.g. the protective nature of… In fact, you don’t even need the last part of this sentence – just end it at the end of line 379.
L389-393. Again edits for grammar are needed in the new text.
Author Response
Response for comments
Comments and Suggestions for Authors
Overview
The authors have largely addressed my major concern about the pairing design. Although the 5-20 cm texture is the bare minimum to have confidence in the space for time approach here, I’ll go with it.
I am a little disappointed with the switch to a pooled approach as the paired test is the best statistical approach (since you’ve established your pairing assumptions better). However, the differences in the results are clear enough that the pooled test still gets the job done.
Unfortunately, the new text all needs considerable English editing to be suitable for publication – more comments below.
Answer: the new text (mark in red) has been edited by MDPI's Author Services.
Specific Comments
L20. Did you switch to a pooled t-test? Please update and make sure this is consistent throughout paper.
Answer: We switch to a pooled t-test, and revised a statement throughout manuscript. Line : 166.
L27-28 This sentence doesn’t make sense. Maybe something like: ‘Variation in salt crust strength can influence dust emissions and needs to be considered in future management.’
Answer: Thanks for your suggestion, revised, Line:27-28.
L84-85 maybe: ‘… due to the erosion-resistant nature..’
Answer: revised, Line: 85-86.
L116 something is missing in this sentence.
Answer: we added new information, line:117.
Section 3.3 needs to be cleaned up. I appreciate that the authors tried out a MLR model, but this needs a mention in the stats section of the methods so that it doesn’t come out of the blue here. The equations are difficult to read. I think if you removed the units from the equations it could clean it up a bit. You could also use an asterisk scheme to indicate which explanatory variables were significant at different alpha values to help interpret the model a bit more.
Answer: We added the information in method, Line :152. We remove the units from the equations, and use an asterisk scheme to indicate significant. Line :244.
L265-266. This interpretation is not completely correct as you have both pooled and paired comparisons that indicate shifts in texture between crusted/non-crusted – so not just in paired samples for your results.
Answer: We deleted this sentence.
L274 This is wordy and the grammar/language is off. You can say this simply as ‘…were protected by salt crusts.’
Answer: revised, Line 275.
L282 delete ‘, but’.
Answer: revised, Line:
L283-284 aggregates
Answer: revised, Line: 281.
L285 sentence needs to be edited.
Answer: All new text has been edited.
L288-289 grammar edits needed.
Answer: All new text has been edited.
L 306 A clay amendment…
Answer: revised, Line: 304.
L301-315 Needs English editing for grammar, tense, subject-verb agreement, and sentence structure.
Answer: all of the new text has been edited by MDPI's Author Services.
L321 and entire paper: Please update units to actual superscripts and subscripts for final version.
Answer: I checked the units through manuscript, line 239, 319.322.
L339 -340. The statement about the decrease in Rsq needs more context. Can’t expect the reader to remember original result. Maybe say the Rsq decreases by X.XX.
Answer: this sentence has been edited MDPI's Author Services.
L350-375 needs editing. The content is good but gets lost in the similar English issues noted for the previous sections of new text. This needs to be fixed before publication.
Answer: all of the new text has been edited by MDPI's Author Services.
L379. Need to tone down the language. This is not a causal framework so subbing ‘suggesting’ for ‘indicating’ would be an appropriate inferential statement.
Answer: revised, line 377.
L380. Replace ‘anti-wind erosion’ with a more appropriate description: e.g. the protective nature of… In fact, you don’t even need the last part of this sentence – just end it at the end of line 379.
Answer: revised, we deleted the sentence, line 377.
L389-393. Again edits for grammar are needed in the new text
Answer: all of the new text has been edited by MDPI's Author Services.
Reviewer 3 Report
No comments
Author Response
Thanks for your suggestion.
