Application of Multivariate Statistical Methods for Determining Geochemical Trends of Elements on the Territory of Slovenia

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript „Application of Multivariate Statistical Methods for Determining Geochemical Trends of Elements on the Territory of Slovenia“represents a valuable contribution to the field of soil geochemistry in Slovenia. It is in the scope of the journal Minerals. The manuscript extensively explores the diverse geochemical anomalies within Slovenian soil, offering a regional perspective while focusing on establishing fundamental baseline values and trends critical for future comparative studies.

While the manuscript provides a robust overview of various geochemical associations across different geographical units and areas within Slovenia, it would benefit from further emphasis on the comparison between distinct preparation and analytical techniques employed. Expanding on this aspect in the conclusion could enhance the manuscript's comprehensiveness, offering deeper insights into the implications of different methods on the obtained analytical outcomes.

Here there are some specific comments:

Line 44: What does it mean "environmental" waters? You should delete "environmental“.

Lines 159-163 – Rewrite the sentence. It is not clear.

Line 164- change „compose“ to „ represent“

Line 170 -change „are develped the mostly“ to „predominate“

Line 155-171 - You mix FAO classifications of soil with other classification. If you empasise eg. terra rossa soils, you should explain that this is not a term used in FAO classification. According to the FAO system (FAO, 1974) terra rossa is recognised as Luvisols (Chromic Luvisols), Phaeozems (Haplic Phaeozems or Luvic Phaeozems) and Cambisols. The same for rendzina and some other terms. You should stick to one classification.

Line 244: delete „>“

Line 256: delete „by“

Line 278-279: What do you mean by „enrichment“? Those samples are the same. You mean the contamination of sample by milling in 1993? It is not clear.

Line 555: delete „near mostly“     

Table 1:

 – where are data on QA for the analysis from 1993 for comparison?

-columns DL UL are to close – some nubmers are to close to each other

Figure 2 and  Fig. 3

– colours in legends are not the same as on the map

Comments on the Quality of English Language

The manuscript is written in an easily understandable language, making complex geochemical concepts accessible to a wide audience. It exhibits minor errors that do not detract from the overall clarity and coherence of the content.

Author Response

Reviewer 1

Line 44: What does it mean "environmental" waters? You should delete "environmental“.

Thank you! The remark is accepted.

Lines 159-163 – Rewrite the sentence. It is not clear.

Thank you! The remark is accepted.

Line 164- change „compose“ to „ represent“

Thank you! The remark is accepted.

Line 170 -change „are develped the mostly“ to „predominate“

Thank you! The remark is accepted.

Line 155-171 - You mix FAO classifications of soil with other classification. If you empasise eg. terra rossa soils, you should explain that this is not a term used in FAO classification. According to the FAO system (FAO, 1974) terra rossa is recognised as Luvisols (Chromic Luvisols), Phaeozems (Haplic Phaeozems or Luvic Phaeozems) and Cambisols. The same for rendzina and some other terms. You should stick to one classification.

Thank you very much for the remark. You are right, in the manuscript we have used the FAO classification and classification from Pedogenic map of Slovenia. The Slovenian pedologist system of classification terra rosa classified as Cromic cambisol and not Chromic Luvisol. It is basically taken from the local systematisation of soil type Ćirić-Škorić into FAO classification. The paragraph was changed accordingly.

Line 244: delete „>“

Thank you! The remark is accepted.

Line 256: delete „by“

Thank you! The remark is accepted.

Line 278-279: What do you mean by „enrichment “? Those samples are the same. You mean the contamination of sample by milling in 1993? It is not clear.

Thank you! The remark is accepted.

Line 555: delete „near mostly “

Thank you! The remark is accepted.

Table 1:

– where are data on QA for the analysis from 1993 for comparison?

-columns DL UL are to close – some nubmers are to close to each other

Thank you very much for the valuable comment. The first samples have been sampled during the 1993-94 sampling campaign and analysed after MA and not AQ. Due to this fact, the chemical analyses obtained from 1993-94 are not actually discussed in the manuscript.

Figure 2 and Fig. 3 – colours in legends are not the same as on the map

Thank you! The remark is accepted

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The article by Šajn et al. deals with the multivariate statistical analysis of elements obtained with two digestion methods. The study area is Slovenia. The topic is very interesting and this paper falls within the scope of Minerals.

Although the authors are recognized geochemists in the region and wider, the manuscript in its present form has many weaknesses. The most important fact is that the paper is not well prepared and does not carry a message, i.e. reach a wider readership. It looks more like a presentation of local results.

 

The statistical methods should be addressed a little more:

1) The statistical analyzes are an important part of this paper and the authors should clearly explain how they transform the data, because they combine both digestions in one analysis (e.g. cluster and factor) and that may be flawed. I strongly suggest them to separate the two digestion types and then perform the multivariate analysis separately. Of course, these two groups of samples should be transformed independently. This is mainly because it is difficult to correlate elements with different degrees of extractability. This can also affect the recognition of the stoichiometry of minerals as the ratio of elements may be biased.

2) Please use statistical significance when considering the correlation of variables.

3) The factor analysis needs more details on the methods (e.g., details such as linearity of variables checked by scatter plots, the Kaiser-Meyer-Olkin sample adequacy, the Bartlett sphericity, and the varimax rotation).

4) The authors indicated that they transformed the data prior to statistical analysis using the Box-Cox transformation, a well-known transformation in geochemical studies. However, the authors again use a standardization procedure that is not necessary because the influence of variable units of measurement is already removed by the Box-Cox transformation.

5) Please clearly indicate which type of clustering analysis you used and which type of linkage.

 

I suggest that the authors write more about why this investigation is important in the introduction. It seems that little is written about important issues, such as detecting pollution in areas of high geologic anomalies, what the geochemical background is, etc. I also suggest that the authors do not refer so often to other studies and previous results, thus diminishing the value of their own study.

The study area should be divided into sections (according the natural units, drainage areas, and geological units), otherwise it is difficult to follow the text. Also, the section on study area is too long and should be more compact. I highly recommend this to the authors – it will be more appealing.

The Results and Discussion section should be restructured. It makes sense to mention the concentrations of the two types of digestion, then compare their ratios and then perform a multivariate analysis. Otherwise, it is difficult to follow the text. Please more in detail discuss element associations obtained.

 

Specific comments:

Line 14: Please add which acids were used – it is important to have consistent information, similar to line 15.

Line 16: Please remove “advanced” and also add which one?

Line 33: soil’s

Lines 47-50: Please reformulate – the sentence has no real meaning

Line 65: Please remove “advanced”

Line 112: What does it mean rich soil (organic rich)? Please describe.

Line 197: “Modified aqua regia”

Line 224: Yes, this is probably due to losses of Ge (and probably Ta) during the aqua regia digestion procedure (see Biver and Filella 2019; https://link.springer.com/article/10.1007/s00706-017-2131-5 and Balaram and Subramanyam 2022; https://doi.org/10.1016/j.sampre.2022.100010)  

Line 224: Too high

Line 323: Which type of transformation?

Lines 362-364: These elements were grouped into in-dependent groups using cluster and factor analyses, represented mainly by Factor 2 (eutric cambisol on Paleogene and Cretaceous flysch) – this is not visible from the factor analysis.

Line 366: Which elements?

Line 377: The analysis of Hf using aqua regia makes no sense. Mann et al 2014 (https://doi.org/10.1144/geochem2014-27) showed poor extractability of Zr (i.e. the same situation is for Hf).

Lines 397-412: The main reason is the poor extractability of these elements by means of aqua regia digestion. These terrigenous elements (see also Mann et al 2014) are tightly bound to silicates – mostly Paleozoic shales along the Sava River and elsewhere.

Lines 419: concentration – check

Line 424: Please specify Oligocene tuffs (intermediate and mafic component) – they are the main cause of such association on the east.

Lines 433-446: On the contrary - the clastic rocks in Slovenia contain the highest values of these elements, which can also be found elsewhere. Therefore, it is not appropriate to interpret the aqua regia results of such poorly soluble elements.

Lines 490-498: Some explanation?

Line 499-500: How is possible to explain these differences within the same factor?

 

Figure:

Figure 3: The colors on legend do not match those of the figure

 

 

 

 

Author Response

Reviews 2

The statistical methods should be addressed a little more:

1) The statistical analyzes are an important part of this paper and the authors should clearly explain how they transform the data, because they combine both digestions in one analysis (e.g. cluster and factor) and that may be flawed. I strongly suggest them to separate the two digestion types and then perform the multivariate analysis separately. Of course, these two groups of samples should be transformed independently. This is mainly because it is difficult to correlate elements with different degrees of extractability. This can also affect the recognition of the stoichiometry of minerals as the ratio of elements may be biased.

Thank you very much for the valuable comment, this part has been revised accordingly. In fact, we initially performed the multivariate analyses separately for each digestion method. It turned out that most of the analysed chemical elements correlate well, the coefficient of determination (D =r2*100 [%]) is above 50% depending on the used MA/AR digestion method. The differences in distributions of concentration ratio (CR) according to the digestion method are very different for some elements, which were then included in a unified study of the main geochemical trends.

Indeed, the comparison of MA/AQ concentration ratios was performed as well. The results of their spatial distribution are very interesting but too extensive to be included in this manuscript. Thus, we believe it should be a part of separate manuscript that will be in the future. The main subject of this manuscript are most important geochemical trends (associations) which are very complex and need to be explained in detail.

Here are the main conclusions from the future manuscript (included in the manuscript):

The following concentrations ratios (CR): Rb, Ga, Cr, Al, V, Nb, Cs, Ti, Sn, K (MA/AQ) are highest in the clastites regardless of geologic age. However, there are no significant differences according to the geographical units or catchment areas. The CR of Zn, Cu, Ni, Co, Li, Fe (MA/AQ) are highest in carbonate areas, same as for the first group, there are no significant differences according to the geographical units or catchment areas. CR of Ca, Sr, Ba, Hf, Zr (MA/AQ) are highest in areas with igneous and metamorphic rocks, independent of geologic age. As expected, significant differences are particularly noticeable in the Eastern Alps or the Pannonian Plain or in the Drava catchment area. Ce, Th, La, Y, Sc, Mn. There are not clearly differences according to lithological and geographical units and catchment areas.

The above results also served as a guideline for the inclusion of elements analysed after AQ digestion method (41 MA, 9 AQ). In fact, 9 elements Ag, Hf, Hg, Nb, S, Sr, Ti, Tl and Zr were included in the multivariate analysis. Hg, S and Tl are only analysed using the AQ digestion method. Based on our experience, results obtained for Ag are even better after the AR is performed. The remaining elements Hf, Nb, Sr, Ti, Zr were included after the concentration ratio distribution study. They are actually independent variables.

2) Please use statistical significance when considering the correlation of variables.

Thank you very much for the valuable comment, this part has been revised accordingly. With n=816, all the correlation coefficients mentioned are statistically significant. From my 30 years of experience, I know that real significant correlations coefficient are actually greater than 0.7 (D>50%).

Al, Ba, Be, Cs, Ga, Hf, K, Na, Nb, Rb, Sn, Sr and Ti show a much lower correlation, so it was concluded that they tend to represent " conditionally independent" variables. The selected one are later included in the multivariate statistics.

3) The factor analysis needs more details on the methods (e.g., details such as linearity of variables checked by scatter plots, the Kaiser-Meyer-Olkin sample adequacy, the Bartlett sphericity, and the varimax rotation).

Thank you very much for the valuable comment, this part has been revised accordingly. The distributions of all analysed elements were carefully examined according to both digestion methods, followed by the application of multivariate analyses in order to asure the better results. The authors are aware that at least the minimum criterion (n>=7k) must be met for the application of multivariate statistical analysis. In our case it was n=816/k=50 (factor analyses), which means k>16n.

4) The authors indicated that they transformed the data prior to statistical analysis using the Box-Cox transformation, a well-known transformation in geochemical studies. However, the authors again use a standardization procedure that is not necessary because the influence of variable units of measurement is already removed by the Box-Cox transformation.

Thank you very much for the remark, this part has been revised accordingly. You are completely right; it was used only Box-Cox transformation and all other unnecessary procedures have been removed.

5) Please clearly indicate which type of clustering analysis you used and which type of linkage.

Thank you! The remark is accepted.

I suggest that the authors write more about why this investigation is important in the introduction. It seems that little is written about important issues, such as detecting pollution in areas of high geologic anomalies, what the geochemical background is, etc. I also suggest that the authors do not refer so often to other studies and previous results, thus diminishing the value of their own study.

Thank you very much for the valuable comment, this part has been revised accordingly. Important geochemical trends were identified according to two methods of digestion, the multi-acid method (MA) as total digestion and the partial method (sequential), which does not dissolve silicates (AR). It was found that for most analysed elements there is no difference in the spatial distribution depending on the applied digestion method (MA or AR).

However, some elements are showing quite different distributions regarding the applied digestion method (MA/AR). The best examples are distribution of Al, Cs, Ga, and Rb after MA they are showing distribution like F1 but after AR like F3. Similarity is with Sn that distribute like MA/AR - F1/F4, Be F1/F6 and Na F7/F8.

The study area should be divided into sections (according to the natural units, drainage areas, and geological units), otherwise it is difficult to follow the text. Also, the section on study area is too long and should be more compact. I highly recommend this to the authors – it will be more appealing.

Thank you very much for your suggestion but we believe that this is difficult to implement and too fragmentary as we are dealing with 8 main geochemical trends in 11 geological units, 8 natural units, 4 drainage areas and geological units. The essence of this article is to study the main geochemical trends (associations) across the country, so we will keep the current trends.

The Results and Discussion section should be restructured. It makes sense to mention the concentrations of the two types of digestion, then compare their ratios and then perform a multivariate analysis. Otherwise, it is difficult to follow the text. Please more in detail discuss element associations obtained.

Thank you very much for the valuable comment. As it was explained before, indeed, the comparison of MA/AQ concentration ratios was performed as well. The results of their spatial distribution are very interesting but too extensive to be included in this manuscript. Thus, we believe it should be a part of separate manuscript that will be in the future. The main subject of this manuscript are most important geochemical trends (associations) which are very complex and need to be explained in detail.

Specific comments:

Line 14: Please add which acids were used – it is important to have consistent information, similar to line 15.

Thank you! The remark is accepted.

Line 16: Please remove “advanced” and also add which one?

Thank you! The remark is accepted.

Line 33: soil’s

Thank you! The remark is accepted.

Lines 47-50: Please reformulate – the sentence has no real meaning

Thank you! The remark is accepted.

Line 65: Please remove “advanced”

Thank you! The remark is accepted.

Line 112: What does it mean rich soil (organic rich)? Please describe.

Thank you! The remark is accepted.

Line 197: “Modified aqua regia”

Thank you! The remark is accepted.

Line 224: Yes, this is probably due to losses of Ge (and probably Ta) during the aqua regia digestion procedure (see Biver and Filella 2019; https://link.springer.com/article/10.1007/s00706-017-2131-5 and Balaram and Subramanyam 2022; #https://doi.org/10.1016/j.sampre.2022.100010)  

Thank you! The remark is accepted, we incorporated the proposed papers into manuscript.

Line 224: Too high

Thank you! The remark is accepted.

Line 323: Which type of transformation?

Thank you! The remark is accepted.

Lines 362-364: These elements were grouped into in-dependent groups using cluster and factor analyses, represented mainly by Factor 2 (eutric cambisol on Paleogene and Cretaceous flysch) – this is not visible from the factor analysis.

Thank you! The remark is accepted.

Line 366: Which elements?

Thank you! The remark is accepted.

Line 377: The analysis of Hf using aqua regia makes no sense. Mann et al 2014 (https://doi.org/10.1144/geochem2014-27) showed poor extractability of Zr (i.e. the same situation is for Hf).

Thank you very much for the valuable comment. Hf, and Zr showed very clear distributions but due to the nature of digestion itself, the differences between the defined areas are more emphasized after AR (figure below).

Lines 397-412: The main reason is the poor extractability of these elements by means of aqua regia digestion. These terrigenous elements (see also Mann et al 2014) are tightly bound to silicates – mostly Paleozoic shales along the Sava River and elsewhere.

 Thank you very much for the valuable comment. You are completely right. It was a main purpose to show in the manuscript. This means that the spatial distributions of some elements can be significantly affected by digestion method. Very good examples are distribution of Al, Cs, Ga, and Rb after MA they are showing distribution like F1 but after AR like F3. Similarity is with Sn that distribute like MA/AR - F1/F4, Be F1/F6 and Na F7/F8.

Lines 419: concentration – check

Thank you very much for the valuable comment. The number was referred to average enrichment.

Line 424: Please specify Oligocene tuffs (intermediate and mafic component) – they are the main cause of such association on the east.

Thank you very much for the valuable comment, this part has been revised accordingly. Indeed, it was examined in the database but the elevated levels of already mentioned elements are not found in igneous rocks but only in samples collected near outcrops of metamorphic rocks.

Lines 433-446: On the contrary - the clastic rocks in Slovenia contain the highest values of these elements, which can also be found elsewhere. Therefore, it is not appropriate to interpret the aqua regia results of such poorly soluble elements.

Thank you very much for the valuable comment, this part has been revised accordingly. Indeed, you are right but here we are presenting the main geochemical trends. Aforementioned chemicals are associated into Factor 3 and represent an independent geochemical association.

Lines 490-498: Some explanation?

Thank you very much for the valuable comment, this part has been revised accordingly. The elevated concentrations were found in the vicinity of ironworks (Jesenice, Ravne, Štore), but also in the vicinity of thermoelectric plants and cement plants in Posavje (Trbovlje, Zagorje, Hrastnik).

Line 499-500: How is possible to explain these differences within the same factor?

Thank you very much for the valuable comment. Concentration in calcaric regosols, and chromic cambisols (terra rossa) that form on the above-mentioned rocks, especially in the karst areas of southern Slovenia.

Figure:

Figure 3: The colors on legend do not match those of the figure

The remark is accepted.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

While I appreciate the authors' (technical) efforts to improve the paper, the message for a broader readership was not conveyed and the main objection was not addressed. All that was left was the presentation of the results without an in-depth discussion of these important findings and element groupings – geochemical patterns.

I hope that this will be part of future work, as they promise.

Author Response

Cover letter

Please find below the answers for Reviewer 2, Academic Editor, and Editorial office

Reviewer 2

The authors would like to thank reviewer 2 for his valuable comments. The authors believe that they answer properly to the main objectives of the manuscript.

 

Academic Editor

 

(VI) Special Comments from the Academic Editor for your consideration: "The section 2 was not modified considering the reviewer 2 comments "Also, the section on study area is too long and should be more compact. I highly recommend this to the authors – it will be more appealing". In my opinion I sugest removing all the descriptions in lines 167-207. The authors only remove some figures."

 

Thank you very much for your remark. The authors indeed agree with your suggestion. The description in lines 167-207 is completely removed from the manuscript since this description is not so relevant for the data interpretation.

Lines 229-230: The following sentence “Slovenia is characterized by carbonate rocks and the corresponding karst surface (70% of territory).” Was removed from the manuscript since the authors would like to avoid repetition.

 

Editorial office

(VII) The editorial office also re-checked the similarity of your revised manuscript and suggested further revising section 2, to reduce the similarity rate. Please find the similarity checking report via the below link:

https://we.tl/t-86C9riEPmu.

 

Regarding the Section 2, the authors revised this chapter based on Reviewers 1 and 2 and also the Academic Editor suggestions. The description between lines 167-207 is completely removed from the manuscript.

Similarity checking report marks are in many parts of the manuscript inappropriate according to our opinion. In many cases, there are marked a part of sentences or several common words, chemical symbols, equations, SI units, names of standard chemical procedures, etc. that cannot be considered as similarities at all. Some absurd similarities probably marked with software need to be critically revised by the Academic Editor. The authors believe that is not something that should be considered as similarities.

However, the authors reviewed all marked similarities throughout each chapter in the manuscript. Please, find our answers below:

1. Introduction

Lines 36-41: It is taken from the referred literature, the sources of soil contamination with trace elements cannot be changed. This is a fact!

Lines 42-98 This is a fact! The section is cited with appropriate references.

2. Study area

Line 100 – 110 the geographical description can’t be changed. The geography of Slovenia is as it is. The summarised version is taken from the referred literature.

Line 151 Half of the sentence is marked, it is a fact. Appropriate reference is mentioned.

Lines 155-162 phrase “natural units” and the name of units are national classifications that can't be changed.

Lines 167-207 were removed, as described before.

Lines 229-230: The sentence was removed.

Lines 230-248 The standard geographical terms or geographical units of Slovenia or standard determination of soil profiles (such as A-C, A-R) can’t be renamed.

3. Materials and methods

Lines 251 – 260 the standard sampling procedure or sample preparation, number of samples can’t be changed.

Lines 304 – 332 chemical symbols, name of laboratory, ISO standard, standard analytical treatments, instrument for analyses, name of acids, and our routine to send the samples for analyses in random order can’t be changed.

Table 1 – how are the authors supposed to write that they calculated accuracy and precision?

Standard SI units or chemical symbols, and numbers can’t be changed.

Lines 346- 383 the standard steps before the application of statistical treatment, calculation of some parameters, equations, conclusions, etc. can’t be rephrased or changed in the manuscript. Appropriate citations are already inserted in the manuscript.

Lines 385 – 404 The name of the software or name of parameters, equations and their explanations can be changed in the manuscript.

Line 426 the expression for concentration is marked twice, this can’t be changed.

Table 2 – Chemical symbols (Ta, Tb, Te) can’t be changed.

Lines 444 – 506 appropriate citations referred to the manuscript. Part of sentences, chemical symbols can’t be changed.

4. Results and discussions

Lines 561 – 665 mainly, some particular expressions or part of sentences, chemical elements are isolated by software. The authors believe that is not something that should be considered as similarities.

Lines 715 – 723 This conclusion is made based on the result, some similar facts are already published previously, which might be a reason for finding the similarities.

Lines 744 – 782 Based on our opinion the marked phrases or symbols for chemical elements shouldn’t be a matter of marked similarities.

5. Conclusion

Lines 808 – 829 Similar to the previous comment, the marked similarities shouldn’t be considered as similarities because part of the sentences, chemical symbols, main conclusions, city names, etc. can’t be changed in the manuscript.

Author contributions and Fundings - The name of the Slovenian research organisation and research programs can’t be changed.

Appendix A – the shortcuts for descriptive statistical parameters are standard shortcuts.

Author Response File: Author Response.pdf