Novel Approach to Freshwater Diatom Profiling and Identification Using Raman Spectroscopy and Chemometric Analysis

Round 1

Reviewer 1 Report

The present manuscript entitled "Novel approach to freshwater diatom profiling and identification using Raman spectroscopy and chemometric analysis" by Pinto et al. deals with the use of Raman spectroscopy for species identification of diatoms. The authors compare 2 different evaluation methods (chemometric analyses), a partial least squares regression discriminant analysis (PLS-DA) and an artificial neural network (ANN) approach. The simple and rapid determination of microorganisms for aquatic monitoring using spectroscopic data (FTIR and Raman) has now been successfully demonstrated several times. Detailed work on diatoms, such as the work presented here, complements these efforts very usefully. Thus, I consider the results shown here to be very valuable and would appreciate their publication.

For the most part, the text is written in an easy-to-understand and fluid manner. A few passages are difficult to follow for readers without prior knowledge of the statistical methods used. However, this does not diminish the significance of the results. The introduction provides a good overview of the topic. The literature used, especially the review articles, are partly outdated and should be updated with more recent literature. However, references to previous work on the identification of species of microalgae and microorganisms in general are missing here, as this is not a new idea. Material and methods section is very concise and contains at first glance all relevant information. However, detailed characteristics of the 3 sampled waters with regard to nutrient status, light ect would be desirable, partly also necessary for the evaluation (see below).

For the Results and Discussion section I have some concerns. The present results show that individual species can be determined with the available data. However, the accuracy of the species determination is rather low, so that the method can be regarded at most as an addition to the microscopic determinations. This should also be formulated in this clarity. In addition, it is not clear whether the identification of each species is due to its physiological state. Individual species are known to respond very differently to abiotic (and biotic) conditions, resulting in differences in biochemical cell composition and cell physiology. This leads to the following questions: Are the species compositions different in the 3 water bodies? Are these differences due to growth conditions (abiotic such as light, nutrients and so on or biotic)? How do Raman data of a single species differ in the different waters? These questions should be discussed in detail and possibly accompanied by supporting information. With reference to these notes, it should be made clear that the present models are limited exclusively to the ecosystems used here and also to the conditions prevailing at the time of sampling. A general statement as to whether the method also works under different growth conditions has yet to be evaluated.

Furthermore, all statements made regarding "physiological status" should be revised, as this is not the focus of the manuscript, nor have they been studied in detail. This applies to the abstract as well as to the conclusion. The statements are currently misleading.

Some detailed remarks:

Figure 2: The difference in baseline drift shows up very clearly. Looking at e.g. the integrals, the question arises how these were calculated. Were integrals to the ground line or to virtual baselines used here?

Figure 2: The legend says that 800 to 1660 cm-1 is approximately 860 to 1680. Why is the entire spectrum not displayed according to M&M (800 to 1700)?

All figures: The abbreviations may be easier to recognize if written as normal species names, such as CSTE C. ste. . There should be enough space in the figures. Also the wave numbers should be written instead of a, b etc. (Figure 2).

179 please check numbering of support information (here its referred to S2 but there is no S1)

Line 220 ff: What do these numbers stand for? The nomenclature (width (W), area (A) and frequency (F) - should be explained beforehand.

250 ff: Are the biofilms sampled under the same light conditions on the same day? If not, how can a light-induced change in pigmentation be excluded? The distribution of the species within each water body should be shown in tabular form (unfortunately I do not have access to Table S1)

253 Not all carotenoids are used for defence mechanisms (fucoxanthin is used for light harvesting!). If you refer to ß-carotene this paragraph should be revised and the special role of this molecule should be explained

299 “…since the same number of spectra were obtained per species…” This info should go into M&M

375 “…can be used to better understand the physiological status of the different…” please delete (see above)

Author Response

We would like to thank the careful review of the manuscript and the suggestions made, which allowed us to improve its quality.

Comment 1: The literature used, especially the review articles, are partly outdated and should be updated with more recent literature. However, references to previous work on the identification of species of microalgae and microorganisms in general are missing here, as this is not a new idea.

Response: The literature was updated as indicated, to address the concerns raised.

Comment 2: Material and methods section is very concise and contains at first glance all relevant information. However, detailed characteristics of the 3 sampled waters with regard to nutrient status, light etc would be desirable, partly also necessary for the evaluation (see below).

Response: The lakes are similar in in hydrogeomorphic characteristics and environmental conditions; they also have an interlinked water flow. This information was added to the text. A table summarising physico-chemical parameters measured in water samples was also added to the Supplementary material.

Comment 3: For the Results and Discussion section I have some concerns. The present results show that individual species can be determined with the available data. However, the accuracy of the species determination is rather low, so that the method can be regarded at most as an addition to the microscopic determinations. This should also be formulated in this clarity.

Response: We acknowledge and appreciate your observation, this information was added in the text for clarity.

Comment 4: In addition, it is not clear whether the identification of each species is due to its physiological state. Individual species are known to respond very differently to abiotic (and biotic) conditions, resulting in differences in biochemical cell composition and cell physiology. This leads to the following questions: Are the species compositions different in the 3 water bodies? Are these differences due to growth conditions (abiotic such as light, nutrients and so on or biotic)? How do Raman data of a single species differ in the different waters? These questions should be discussed in detail and possibly accompanied by supporting information. With reference to these notes, it should be made clear that the present models are limited exclusively to the ecosystems used here and also to the conditions prevailing at the time of sampling. A general statement as to whether the method also works under different growth conditions has yet to be evaluated.

Response: There were some differences among the lakes in species composition as indicated in Table S2 of supplementary information. However, these three lakes were located within the same park, in close proximity. Furthermore, they exhibited similar hydrogeomorphic characteristics and an interconnected water circulation system. Lakes 1 and 3 were also very similar in physico-chemical parameters. Though the idea is very interesting, differences of single species across systems were not investigated in this work as they were out of the scope of the study and are more suitable for a bigger dataset encompassing more sampling sites. The fact that taxa identification was still possible over local variation in conditions is also a suggestion of the promising potential of the approach. Raman spectroscopy is very sensitive and able to detect structural molecules useful to distinguish among taxa. This study was a first investigation of the usefulness of this approach in a small area. The next step will be to enlarge the number of sites and ecosystems to refine its use under different environmental and growth conditions and select the most useful Raman spectra to generalise the application. General statements were revised and this discussion was added to the manuscript as per reviewer indication.

Comment 5: Furthermore, all statements made regarding "physiological status" should be revised, as this is not the focus of the manuscript, nor have they been studied in detail. This applies to the abstract as well as to the conclusion. The statements are currently misleading.

Response: All the statements regarding “physiological status” were revised as indicated here and in comments above for clarity.

Comment 6: Figure 2: The difference in baseline drift shows up very clearly. Looking at e.g. the integrals, the question arises how these were calculated. Were integrals to the ground line or to virtual baselines used here?

Response: All the integrals were calculated according to the ground line of each band. The figure just shows the raw spectra without any adjustments, for illustrative purposes.

Comment 7: Figure 2: The legend says that 800 to 1660 cm-1 is approximately 860 to 1680. Why is the entire spectrum not displayed according to M&M (800 to 1700)?

Response: The first band recorded was approximately at the position 867 cm^-1 and the last band recorded was approximately at the position 1656 cm^-1. Before and after this interval, no bands appeared on the spectra. This was corrected in the Material and methods section.

Comment 8: The abbreviations may be easier to recognize if written as normal species names, such as CSTE C. ste. . There should be enough space in the figures. Also the wave numbers should be written instead of a, b etc. (Figure 2).

Response: Figure 2 was changed according to the suggestions. For the remaining figures, we tried to change the acronyms to the normal species names, however, the figures became too large and difficult to read.

Comment 9: 179 please check numbering of support information (here its referred to S2 but there is no S1)

Response: The numbering has been corrected; thank you detecting this.

Comment 10: Line 220 ff: What do these numbers stand for? The nomenclature (width (W), area (A) and frequency (F) - should be explained beforehand.

Response: The explanation was added to the Materials and Methods section for clarity.

Comment 11: 250 ff: Are the biofilms sampled under the same light conditions on the same day? If not, how can a light-induced change in pigmentation be excluded? The distribution of the species within each water body should be shown in tabular form (unfortunately I do not have access to Table S1)

Response: All the biofilms were sampled in the same day under similar conditions; this was clarified in the methods section.

Comment 12: 253 Not all carotenoids are used for defence mechanisms (fucoxanthin is used for light harvesting!). If you refer to ß-carotene this paragraph should be revised and the special role of this molecule should be explained

Response: The paragraph was rewritten according to the suggestions.

Comment 13: 299 “…since the same number of spectra were obtained per species…” This info should go into M&M

Response: This information was included in the Materials and Methods section. The statement was rewritten for clarity.

Comment 14: 375 “…can be used to better understand the physiological status of the different…” please delete (see above)

Response: All the statements regarding physical status were revised and eliminated according to the previous suggestions.

Reviewer 2 Report

Review of paper „Novel approach to freshwater diatom profiling and identification using Raman spectroscopy and chemometric analysis”

Authors are presenting interesting aspect and possibilities of Raman spectroscopy in diatom identification. Paper is interesting but same important data should be included:

-          For application point of view authors should add to results and discuss information about the acquisition and processing time of RS of one slide – how much time takes obtain taxa list for one slide.

-          What with possibilities of storage of “slides” for future comparison, both for light microscopy and Raman.   

-          Authors should include same results about “dead” specimens (empty frustules/vales). They were present in sample or not? If the will be present RS will be not identify diatom. This is very interesting aspect from a quantitative point of view, especially when compere LM and RS.

-          In my opinion for now presented method of diatom identification have no practical application but can be interesting for e.g. cell condition identification and should be studies base on laboratory cultures.

-          I found same mistakes with diatom names. Unfortunately I was unable to check Table S1 – I was not included to ms. file.

-          Few additional comment are included in ms. file.

Comments for author File: Comments.pdf

Author Response

We would like to thank the careful review of the manuscript and the suggestions made, which allowed us to improve its quality.

Comment 1: For application point of view authors should add to results and discuss information about the acquisition and processing time of RS of one slide – how much time takes obtain taxa list for one slide.

Response: The acquisition of each spectrum takes thirty seconds, i.e.(10s each acquisition with three accumulation in total (see the methods section line, subsection Raman spectroscopy). The spectra processing time is about 2 mins.

Comment 2: What with possibilities of storage of “slides” for future comparison, both for light microscopy and Raman.  

Response: Here Raman spectroscopy was applied to fresh diatoms, after an identification of each oxidised sample was done. However, Raman spectroscopy can also be applied to mounted slides and the oxidised slides used with the light microscopy can also be easily kept.

Comment 3: Authors should include same results about “dead” specimens (empty frustules/vales). They were present in sample or not? If the will be present RS will be not identify diatom. This is very interesting aspect from a quantitative point of view, especially when compere LM and RS.

Response: Empty frustules and valves were present in the collected samples, however, they were not used for acquisition of Raman spectra. We clarified this in the Material and Methods section.

Comment 4: In my opinion for now presented method of diatom identification have no practical application but can be interesting for e.g. cell condition identification and should be studies base on laboratory cultures.

Response: We plan to study the practicability of this method to other scenarios, including laboratory cultures. We thank your interesting suggestions.

Comment 5: I found same mistakes with diatom names. Unfortunately I was unable to check Table S1 – I was not included to ms. file.

Response: The mistakes with diatom names were corrected. Thank you for pointing them out.

Comment 6: Few additional comment are included in ms. file.

Response: We acknowledge the careful review of our manuscript and we answered all the comments as suggested.

Round 2

Reviewer 1 Report

Concerns the 2 round in the review process of the manuscript "Novel approach to freshwater diatom profiling and identification using Raman spectroscopy and chemometric analysis".

The authors have tried to implement the comments to the best of their knowledge. Nevertheless, I have an important remark concerning a passage in the text which should definitely be revised:
L260 ff: The new section concerning carotenoids is not quite correct in content. Since the work deals exclusively with diatoms, this should also be noted here. Diatoms do not normally have a xanthophyll cycle that converts violaxanthin into zeaxanthin. Instead, diadinoxanthin is converted into diatoxanthin. These pigments are also not a defense mechanism against ROS (like the Mehler reaction) but prevent the formation of ROS. This is very well documented in the literature, references #51 and #52 should be adjusted accordingly. In addition, I am of the opinion that the pigments mentioned play only a minor role. The resulting conclusions are very vague and not experimentally validated. The main pigment fucoxanthin is not mentioned. The low carotenoid absorption could, after all, also be caused by low fucoxanthin concentrations. This would lead to a completely different conclusion, because then e.g. possible light stress would have to be discussed.

And a small formal hint: The species names should also be written in italics in the additional information Table S2.

After revision I consider the manuscript ready for publication

Author Response

We would like to thank the reviewer for the careful review of the manuscript and the suggestions made, which allowed us to improve its quality.

Comment 1: L260 ff: The new section concerning carotenoids is not quite correct in content. Since the work deals exclusively with diatoms, this should also be noted here. Diatoms do not normally have a xanthophyll cycle that converts violaxanthin into zeaxanthin. Instead, diadinoxanthin is converted into diatoxanthin. These pigments are also not a defense mechanism against ROS (like the Mehler reaction) but prevent the formation of ROS. This is very well documented in the literature, references #51 and #52 should be adjusted accordingly. In addition, I am of the opinion that the pigments mentioned play only a minor role. The resulting conclusions are very vague and not experimentally validated. The main pigment fucoxanthin is not mentioned. The low carotenoid absorption could, after all, also be caused by low fucoxanthin concentrations. This would lead to a completely different conclusion, because then e.g. possible light stress would have to be discussed.

Response: The discussion was corrected accordingly and the references were rectified.

Comment 2: And a small formal hint: The species names should also be written in italics in the additional information Table S2.

Response: Changes were made accordingly.

Reviewer 2 Report

Dear Authors,

Dear Editorial Office,

Thank you for taking into account my comments. For the second version I still have same remarks:

-       L158 – s.l. – should no italics; Czarnecki. 1994 – should be: Czarnecki, 1994

-       Figure 1 – I’m not sure proper identification of N. gregaria -  the valve shape is not typical fore species and is extremely small. Similar for Ctenophora pulchella, Navicula notha  - pictures are very low quality – authors should verify all species just again !

-       From whale text should be deleted or added spaces, for example L74, 77, 78.

-       What was total time of spectrum scanning for one slide ? In other words wat was the total time of diatom identification using RS?

Best regards

Author Response

We would like to thank the reviewer for the careful review of the manuscript and the suggestions made, which allowed us to improve its quality.

Comment 1: L158 – s.l. – should no italics; Czarnecki. 1994 – should be: Czarnecki, 1994

Response: The sentence was corrected. Thank you for the remark.

Comment 2:  Figure 1 – I’m not sure proper identification of N. gregaria -  the valve shape is not typical fore species and is extremely small. Similar for Ctenophora pulchella, Navicula notha  - pictures are very low quality – authors should verify all species just again !

Response: The species were verified and rectified in all the samples. Some photographs in figure 1 were changed for clarity.

Comment 3: From whole text should be deleted or added spaces, for example L74, 77, 78.

Response: The whole text was verified and corrected.

Comment 4: What was total time of spectrum scanning for one slide? In other words wat was the total time of diatom identification using RS?

Response: It depends on the number of species per sampling site and how many individuals are assessed per species. The acquisition of each spectrum takes a total of thirty seconds (ten seconds and three accumulations, see Materials and Methods). As an example, for a sample containing 20 different species and spectra acquisition for 15 individuals per species, the procedure will take about two and a half hours.