Dynamics of Hydrology and Anaerobic Hydrocarbon Degrader Communities in A Tar-Oil Contaminated Aquifer

Round  1

Reviewer 1 Report

The paper presents interesting multi-annual  research. It is written in comprehensible and correct language.

However, there are some inaccuracies related to the NGS sequencing methodology. The authors confuse the sequencing based on 16S rRNA amplicons with transcriptomics. It cannot be said that transcriptomics or transcriptomes have been used, while the method is based on DNA isolation, DNA amplification and DNA sequencing. However, I believe that the error occurred at the stage of describing the sequence when uploading to publicly available databases and does not affect the results obtained, but it does mislead the reader and should be corrected.

It would be more convenient to read the paper if the methodology described more extensively in this paper, instead of referring readers to their own previous papers (high level of self-citation). I understand that such a procedure makes the work seem concise, more straightforward and avoids duplication of the same text, but it loses its reading value.

The following remarks do not impair the positive value of the manuscript:

Section: 2.2 Molecular analyses

ln 106 - 127: 

- Please describe more precisely the DNA extraction method.

- How many subsamples were taken

- Why qPCR was performed only on DNA material, not the mRNA (it could better reflect if chosen genes were actively involved in toluene degradation)

- the methodology for the sequencing and bioinformatics analysis is given as an external source. Please provide at least the name and sequences of the primers used. It is very uncomfortable to read the manuscript, in which the authors often refer to other/previous papers.

- which version of mothur was used

- why 97% similarity cutoff was chosen, while the current standard is at least 99% . Using 97% some different species will be undistinguishable. Previously, 97% were used because the computation limits, which now is not an issue.

- primer sequences were removed from reads?

- there are some inconsistencies: sequencing was performed on the 16S rRNA amplicon library, but the description on the GEO NCBI database says that  the library source = transcriptomic and the description says “small rRNA were amplified from total nucleic acids extracted from environmental samples using specific PCR primers”. I assume that you have isolated total DNA (not RNA) and amplify DNA fragments (not RNA, again) of 16S rRNA gene region. So there is no transcription since you isolate DNA, amplify DNA and sequence DNA.

Otherwise you should isolate total RNA, perform reverse transcription (RNA -> cDNA) and then you can say about “transcriptiomics”.

- again, looking at SRA NCBI portal (e.g. your SRX031255 submission), there is a mistake (same as above). You didn’t performed transcriptomics.

- Greengenes database is outdated and no longer recommended.

Section: 3.2

Fig.4 – It would be more readable if the results were presented in separate charts. Separate presentation of genera would simplify the comparison. It would be also interesting to present the total number of identified genera.

Author Response

REV1

The paper presents interesting multi-annual research. It is written in comprehensible and correct language.

Thank you for your comment and advices to help us further improve the manuscript. Please see below and in the main text the reply to your comments. We hope you will find them adequate to address all of your concerns.

However, there are some inaccuracies related to the NGS sequencing methodology. The authors confuse the sequencing based on 16S rRNA amplicons with transcriptomics. It cannot be said that transcriptomics or transcriptomes have been used, while the method is based on DNA isolation,

DNA amplification and DNA sequencing.

We are well aware of the difference between transcriptomics and amplicon sequencing and, in fact, are absolutely confident that the term “transciptomics” was never used in the original manuscript. We understand that the reviewer’s confusion probably originates from our information deposited in the GEO archive, which were then somehow automatically transferred to SRA as “transcriptomics”. To alleviate this problems and to avoid future misunderstandings, the following actions were taken:

– The entries in GEO were edited to better reflect that DNA and not RNA was sequenced, an update to the sample analysis procedure was posted, as well as updates to the fields “Library strategy”, “Library source” and “Library selection” (now updated as “other”).

– The SRA entries were edited to reflect that the sequences are not from “transcriptomics” and the fields “Strategy” were updated as “genomic”. We hope that these changes are satisfactory to the reviewer.

However, I believe that the error occurred at the stage of describing the sequence when uploading to publicly available databases and does not affect the results obtained, but it does mislead the reader and should be corrected.

That is correct, we hope the actions described above help to mitigate the problem.

It would be more convenient to read the paper if the methodology described more extensively in this paper, instead of referring readers to their own previous papers (high level of self-citation). I understand that such a procedure makes the work seem concise, more straightforward and avoids duplication of the same text, but it loses its reading value.

We agree and a substantial amount of information has been added to the material and methods section, especially to “2.2. Molecular analyses”. Please see the revised manuscript for approval.

The following remarks do not impair the positive value of the manuscript:

Section: 2.2 Molecular analyses

ln 106 - 127:

- Please describe more precisely the DNA extraction method.

Done

- How many subsamples were taken

Triplicate sediment subsamples from each depth of the intact sediment liners were extracted for DNA extraction. Clarity is now added to the text

- Why qPCR was performed only on DNA material, not the mRNA (it could better reflect if chosen genes were actively involved in toluene degradation)

Our lab has a strong routine in working with environmental extracts of both DNA and RNA from soil and subsurface sediments. For the Flingern sediments investigated here, however, total RNA yields and quality have never been satisfactory to allow for robust RT-qPCR work or transcriptome sequencing. So for the work presented here, RT-qPCR was not an option, unfortunately. We hope that the reviewer can accept this.

- the methodology for the sequencing and bioinformatics analysis is given as an external source. Please provide at least the name and sequences of the primers used. It is very uncomfortable to read the manuscript, in which the authors often refer to other/previous papers.

Done

- which version of mothur was used

Version 1.34.3, info has been added to the text

- why 97% similarity cutoff was chosen, while the current standard is at least 99% . Using 97% some different species will be undistinguishable. Previously, 97% were used because the computation limits, which now is not an issue.

We still trust that a 97% sequences similarity cutoff is the current standard for OTU classification. For an authoritative and recent opinion, we suggest to read Knight et al. “Best practice for analyzing microbiomes”, Nature Reviews Microbiology, 16: 410–422 (2018). We are also aware that the higher 99% cutoff suggested by the reviewer is more and more frequently used in recent oligotyping studies, but we do not think that this would have been very useful in the present study.

In fact, we differentiate and interpret mainly at the family level, and only touch on the genus level for a few selected populations. Finally, minor differences in ecological interpretation are usually reported when comparing 97 and 99% cutoffs, apart from increased abundance of singletons (e.g., Poretsky et al., PLoS One. 2014; 9(4): e93827). Therefore, we hope that these arguments can convince the reviewer of the merits of using a 97% cutoff in our present study.

- primer sequences were removed from reads?

Yes, barcode primers were removed, as this is the best practice of the mothur pipeline we used. An additional clarification has been added to the text. The original 16S PCR primers, however, were not removed from the contigs, as this would complicate T-RF prediction. We are aware that the removal of amplification primers is important when depositing full-length type-strain quality 16S rRNA sequences for taxonomic purposes. However, since this was not our aim in the present work, we did not find it crucial to remove 16S primer sequences.

- there are some inconsistencies: sequencing was performed on the 16S rRNA amplicon library, but the description on the GEO NCBI database says that the library source = transcriptomic and the description says “small rRNA were amplified from total nucleic acids extracted from environmental sample susing specific PCR primers”. I assume that you have isolated total DNA (not RNA) and amplify DNA fragments (not RNA, again) of 16S rRNA gene region. So there is no transcription since you isolate DNA, amplify DNA and sequence DNA.

As already explained above, all inconsistencies have been fixed in both GEO and SRA repositories. To avoid further confusion, the accession number of the SRA record is now given in the revised manuscript, instead of the original GEO access.

Otherwise you should isolate total RNA, perform reverse transcription (RNA -> cDNA) and then you can say about “transcriptiomics”.

We agree but again, the original manuscript never claimed that transcriptomics was done. This must have been a misunderstanding.

- again, looking at SRA NCBI portal (e.g. your SRX031255 submission), there is a mistake (same as above). You didn’t performed transcriptomics.

all entries have been corrected in GEO and SRA. Thanks for spotting this!

- Greengenes database is outdated and no longer recommended.

The Greengenes database was not used for OTU classification, this was done with the much more updated version of SILVAngs. This is explained in the manuscript. Even in our previous analyses of these data (e.g., Pilloni et al. PLOS ONE, 2012), the parts of the Greengenes pipe were only used for quality/length trimming of the sequences. In the present manuscript, Mothur was used for these operations, while the updated taxonomic framework of SILVA was used for classification. The text should be clearer now, thanks for catching that.

Section: 3.2

Fig.4 – It would be more readable if the results were presented in separate charts. Separate presentation of genera would simplify the comparison. It would be also interesting to present the total number of identified genera.

Thank you for this comment, however, we would like to argue to not dissect this complex composite figure. The reason is that we feel that Fig. 4 is a very powerful chart since, while showing the general composition and variability of the total microbial community, it also highlights the changes in those specific genus- or family-level populations that are dynamic and important for the geochemistry in situ. Please keep in mind that our study is not focused on NGS alone and remains multidisciplinary, combining chemistry, hydrology, isotope analysis and microbial community characterization. Therefore, we consider the analysis of the microbial community as one of several tools used to describe the complex ecology of the system. The current figure 4, in our opinion provides just the right amount of detail needed to describe the most relevant microbiome dynamics, while many other secondary or low-abundance taxa are not resolved. We hope the reviewer can agree to these arguments.

Reviewer 2 Report

This is an interesting study where the authors investigated the effects of hydraulic dynamics on anaerobic degrader communities in an oil-contaminated aquifer. The manuscript is well written and was easy to follow. My major concern is the way how the impacts of hydraulic dynamics on microbial communities were analyzed and discussed. I did not get a complete sense of the effects of hydraulic variations on the degrader community; the results rather indicate more of temporal variations within a fixed plume boundary. The hypothesis needs to be specific with more context on functional redundancy in a dynamic hydraulic environment (see my specific comment below). Supplementary figures were cited in several places, which may distract readers, and some of these can be included in the main text. I recommend changing the focus of this study more towards the temporal dynamics (inter-annual variations) in plume biogeochemistry rather than directly relating this to hydraulic dynamics unless more experimental evidence are added.

Introduction

L56: Elaborate more on the ‘reactive plume-fringe’ concept;

L68: cite the previously characterized study

L68-69: Be more specific about what specific attributes of the degrader communities are expected to change and why. Also, the idea of ‘functional redundancy’ and its relationship with hydraulic dynamics are not introduced.

L71-72: Move to the conclusions

Materials and methods

Add more details on sampling design such as number of replicates, distance between the sampling points, and abiotic environment of the plumes.

L92-93: Is different sampling time a concern for detecting the localized centric zones of plumes? How did other groundwater recharge phenomena affect this?

L104: Briefly describe how these were done.

Results

May be add a section to characterize the plume boundaries (upper, lower, and deeper) and nature of contaminations.

Fig. 2 and 3: Did the plume boundaries (upper and lower ends) not change in different years? These figures only show temporal variation in BTEX and microbiome data.

L170: Rephrase

L257-263: Can different zones (upper, lower, and deeper) be indicated in this graph?

L273-274: Why then PCA was not used for the pooled ‘All depths’ graph?

Discussion

L302-305: This is not clear from Fig. S1. The 2009 figure shows spatial variability in BTEX concentrations. The other two figures (2006 and 2007) just show contour lines. Also, what is the experimental evidence that the variability in spatial distribution of plumes does not impact the vertical biogeochemistry; specially when this was not something measured in other sampling locations!

L331: Is ‘prior to the plume dynamics’ indicating the sampling done before the fluctuating hydraulic phenomena?

L360-363: Major limitation of the study! The idea of ‘plume shift’ is introduced here but all the figures show a fixed plume boundary!

Author Response

REV2

This is an interesting study where the authors investigated the effects of hydraulic dynamics on anaerobic degrader communities in an oil-contaminated aquifer. The manuscript is well written and was easy to follow. My major concern is the way how the impacts of hydraulic dynamics on microbial communities were analyzed and discussed. I did not get a complete sense of the effects of hydraulic variations on the degrader community; the results rather indicate more of temporal variations within a fixed plume boundary.

The hypothesis needs to be specific with more context on functional redundancy in a dynamic hydraulic environment (see my specific comment below). Supplementary figures were cited in several places, which may distract readers, and some of these can be included in the main text. I recommend changing the focus of this study more towards the temporal dynamics (inter-annual variations) in plume biogeochemistry rather than directly relating this to hydraulic dynamics unless more experimental evidence are added.

We appreciate the reviewer’s constructive criticism of our work and will do our best to alleviate the concerns. First, it is undisputed that temporal variations in plume microbiota were observed. While this alone is interesting and has rarely been reported, we are also absolutely positive that also hydrological and biogeochemical dynamics of the plume system were observed and are adequately documented in the results. In contrast to the reviewer’s interpretation, Figs. 1 and 2 unambiguously document that (1) a fluctuation of the groundwater table of +/- 50 cm occurred over the 5 years of sampling; and that (2) also the plume core and sulfidic lower plume fringe shifted up and down by several dm over the same time. The discussion then carefully puts both observations into context and tries to argue, if and how both observations could be connected. We cautiously conclude that a connection is not an unlikely scenario (L348), while at the same time carefully explaining the contribution of other controlling factors possibly causing the observed dynamics (L349-365).

Although we strongly oppose the reviewer’s perception of a static plume boundary, which was clearly not the case over the years, the reviewer is fully correct that a final mechanistic conclusion is not possible without further direct experimental evidence. However, such evidence could only be provided via a controlled experimental setting, and not from field monitoring where a fully reliable control is essentially not possible. As also this was already explained in the original manuscript (L405, L419), we have now hope the reviewer can reconsider our very cautious lines of argumentation, and can accept that we prefer to not rewrite the manuscript with a focus on biogeochemical dynamics alone.

Introduction

L56: Elaborate more on the ‘reactive plume-fringe’ concept;

Further explanatory detail has been added.

L68: cite the previously characterized study

References have been placed

L68-69: Be more specific about what specific attributes of the degrader communities are expected to change and why. Also, the idea of ‘functional redundancy’ and its relationship with hydraulic dynamics are not introduced.

We have added expected community rearrangements to the hypothesis. However, we would like to leave the functional redundancy context to the discussion.

L71-72: Move to the conclusions

Done

Materials and methods

Add more details on sampling design such as number of replicates, distance between the sampling points, and abiotic environment of the plumes.

More details on the number of replicate sediment subsamples and DNA extractions has now been added. However, we would like to stress that only one high-resolution depth resolved montotring well was installed, so there is no lateral distance between sampling points. More detail on the depth resolved sampling was added.  

L92-93: Is different sampling time a concern for detecting the localized centric zones of plumes? How did other groundwater recharge phenomena affect this?

Monitoring by the site owner was not strictly synchronized with our own sampling campaigns, but always occurred within 1-2 months in advance. Fig. S1 shows that the plume system was laterally highly static, so we have no concerns in adequately detecting the centric zones.

L104: Briefly describe how these were done.

Done

Results

May be add a section to characterize the plume boundaries (upper, lower, and deeper) and nature of contaminations.

The Flingern plume and it’s compartments have been intensively described previously [Refs 12-15]. We prefer not to add too much repetitive detail here. We believe the reader can easily grasp the existence of an upper and a lower plume fringe, as well as a plume core. The nature of the contamination is described in the introduction and in the M&M.

Fig. 2 and 3: Did the plume boundaries (upper and lower ends) not change in different years? These figures only show temporal variation in BTEX and microbiome data.

We regret that the reviewer has obtained an impression of a fixed plume boundary. However, in contrast to this impression, we are confident that sufficient data is shown to prove that not just the GWT, but also the plume fringes were non-static. This is especially apparent in Fig. 2a and 2 c, where both toluene and sulfide peaks are moving up and down by at least 20 – 30 cm over the years.

We can only speculate that the reviewer has come to this impression by the fact that the shaded area of the lower plume fringe is averaged over the years in Figs. 2 and 3. We see no other way to more optimally present the comprehensive data in figures summarizing the data over the years.

However, we would like to stress that the legend of the figures clearly explains that “the extent of the fluctuations … are indicated (vertical arrow at GWT)” and that “The zones of the upper and lower plume fringes are averaged over the entire study period…”. We hope that the reviewer can accept this line of argumentation and presentation. We have now added further detail to the figure legends to better resolve these issues.

L170: Rephrase

Sentence has been modified

L257-263: Can different zones (upper, lower, and deeper) be indicated in this graph?

Depth and toluene are indeed included as vectors in Fig. 5a, which allows to delineate certain associations between sampling depth, plume compartments and plume microbiota. However, the plume compartments themselves cannot be indicated as zones in the graph, as they themselves were not variables in the statistical data matrix. As the different plume zones are then resolved in the following graphs (b, c, d), we argue that the inclusion of another CCA or PCA graph comprising all plume zones would not add further valuable information. No further change conducted.

L273-274: Why then PCA was not used for the pooled ‘All depths’ graph?

CCA was first used to confirm significant associations of taxa abundances with depth and other quantitative variables between plume zones. Overall, the differences between plume zones were much stronger than the differences between time points within plume zones. Hence, if samples were to be analyzed across the different depths in a single PCA, the differences between time points within plume zones would be overshadowed by the strong differences between the plume zones. Therefore we decided to carry out individual PCAs for each plume zone to better illustrate the shifts in community composition over time and the taxa that mainly contributed to these shifts. We hope this better explains our choice of statistics.

Discussion

L302-305: This is not clear from Fig. S1. The 2009 figure shows spatial variability in BTEX concentrations. The other two figures (2006 and 2007) just show contour lines. Also, what is the experimental evidence that the variability in spatial distribution of plumes does not impact the vertical biogeochemistry; especially when this was not something measured in other sampling locations!

The graphs of Fig. S1 have been handed over to us by the site owner. It is clearly stated that the data was not generated by us, but by different monitoring offices, which is also why they offer a distinct level of detail. We would like to bring to the reviewer’s attention that the 2006 and 2007 figures also contain BTEX concentrations, next to the contour lines inferred via Kriging. This information has now been better highlighted in the figure legend.

Clearly, there was variation in BTEX concentrations over the years, and we never claimed that these were not. The point we would like to make here is simple: The HR-MLW was always centrally placed within the plume! We believe this is sufficiently documented by this supplementary figure.

To the second aspect, we are fully aware that we have no experimental evidence. This could only be provided via a laboratory microcosm experiment, which would clearly be beyond the scope of the present manuscript. We feel that we adequately address this limitation and inspiration for future work in the discussion (L405, L419).

L331: Is ‘prior to the plume dynamics’ indicating the sampling done before the fluctuating hydraulic phenomena?

Yes. More detail has been added to avoid confusion.

L360-363: Major limitation of the study! The idea of ‘plume shift’ is introduced here but all the figures show a fixed plume boundary!

We do not agree, please see our response to the major comments to the results above.