An 18S V4 rRNA metabarcoding dataset of protist diversity in the Atlantic inflow to the Arctic Ocean, through the year and down to 1000&thinsp;m depth

Egge, Elianne; Elferink, Stephanie; Vaulot, Daniel; John, Uwe; Bratbak, Gunnar; Larsen, Aud; Edvardsen, Bente

doi:https://doi.org/10.5194/essd-13-4913-2021

Articles | Volume 13, issue 10

https://doi.org/10.5194/essd-13-4913-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

Extreme environment datasets for the three poles

https://doi.org/10.5194/essd-13-4913-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 10

Data description paper

|

26 Oct 2021

Data description paper |

| 26 Oct 2021

An 18S V4 rRNA metabarcoding dataset of protist diversity in the Atlantic inflow to the Arctic Ocean, through the year and down to 1000 m depth

Elianne Egge, Stephanie Elferink, Daniel Vaulot, Uwe John, Gunnar Bratbak, Aud Larsen, and Bente Edvardsen

Download

Final revised paper (published on 26 Oct 2021)
Preprint (discussion started on 10 May 2021)

Interactive discussion

Status: closed

RC1:
'Comment on essd-2021-133', Anonymous Referee #1, 09 Jun 2021

General comments

In this paper, the authors constructed an matabarcoding dataset of marine protists communities in the Northern Svalbard region of the Arctic Ocean. These data comprise samples collected at some stations from the surface to the 1000 m depth every two to three months. The total number of amplicon sequence libraries is huge and the data would be enough to address a variety of ecological questions. Environmental metadata was also prepared for each of the sampling event. I agree that the dataset can be valuable as the study area is key to understand the connectivity between Arctic and Atlantic Oceans. However, I have some serious concerns regarding the sampling and sequencing strategies used in this study.

Major issues

In my opinion, sampling strategy is inappropriate to evaluate seasonal change or size-dependence of the microbial assemblages. For example, as the sampling locations are distributed among season, it would be difficult to decipher if the observed variation of microbe is owing to the season or just to the location. Similarly, size fractionation was not consistent across seasons. Samples were taken from 3-180 µm fraction in Jan to Mar, while 3-10, 10-50, and 50-200 µm fractions were applied for May to Nov. Sequencing platform was also inconsistent across samples (ie., MiSeq and HiSeq) and the different criteria were used for the downstream sequencing processing. Thus, the users will have difficulty in interpreting their results as they have to consider the possible effects of the different location, size fraction, and sequence platform. These methodological discontinuities would collectively diminish the overall quality of the dataset, and consequently the strength of the conclusion of the analysis.

The accuracy of the eukaryotic community profiling strongly depends on the choice of primer pair. I checked the ability of the primer set used in this study by in silico PCR analysis using a primer test tool (Silva TestPrime: https://www.arb-silva.de/search/testprime/), and found that the primer may amplify only 2% of known haptophyte sequences and 0.5% of Rhizaria sequences in the Silva SSU database. As these lineages are important components of marine microbial eukaryotes, the sequencing libraries were potentially biased due to the mismatches of some specific species. Although it is impossible to amplify all the rRNA genes in the environments, some primers have shown to be highly universal for both eukaryotes and prokaryotes (Parada et al., 2015, 10.1111/1462-2920.13023; McNichol et al., 2021, 10.1128/mSystems.00565-21). These primer may be a standard method for monitoring overall communities of prokaryotes and eukaryotes across space and time. Although there is no golden standard in a choice of the primer pair so far, it's worth pointing out.

Specific comments

L8-9: Please specify size range of each of picoplankton, nanoplankton, etc.

L59-65 and the related metadata: The sampling conditions (location, depth and filter size) were bit complicated and sometimes inconsistent. For example, readers cannot recognize which size fractions and depths were applied for each season and site from this information (metadata is not suitable to see this kind of information). I recommend the authors to make a table summarizing the sampling site, season, depth and size fraction.

L88-89: A bit difficult to find out what the authors mean (i.e., "opposite patterns" of what?)

L129: The authors should justify the use of this primer set. As described earlier, this primer may not be a universal for some of the eukaryotic lineages, such as Haptophyta and Rhizaria.

L146-147: Please specify sequence length (e.g., 150 bp, PE) of MiSeq and HiSeq sequencing.

L203: It is difficult to know if there is a seasonality of eukaryotic communities from the Figure 4. I would recommend the authors to add an ordination plot such as NMDS to overview the variation of the community composition across season, size, and depths.

Environmental data file: Some parameters lack the unit (e.g., counting data of bacteria and virus).

Citation: https://doi.org/10.5194/essd-2021-133-RC1
- AC1: 'Reply on RC1', Elianne Egge, 03 Jul 2021
  
  Dear reviewer. Thanks for taking the time to thoroughly review our paper. We would here like to give an initial response to your major concerns, and get back to you with our final response once we get the comments from the other reviewer and the editor.
  In Arctic waters, processes connected with the ice edge such as melting and stratification, are crucial for the primary production and microbial life. Therefore, we aimed to get as close as possible to the ice edge during each cruise, rather than sampling from a fixed location. The temporal change in taxonomic composition in our metabarcoding data reflects what is previously known from morphological studies on the succession of the major groups of microplankton in the Arctic – few autotrophs in winter, a spring bloom dominated by diatoms, Phaeocystis and Micromonas, followed by a high proportion of dinoflagellates in summer. At all sampling locations there is polar night, i.e. the sun is below the horizon from autumn until spring. This pattern could thus not have been produced only by variation in location.
  We agree that the inconsistent size-fractionation is unfortunate. This was due to the different space available to us on the vessels on these cruises. However, we consider it unlikely that the taxonomic composition in the 3-180 micron fraction would be qualitatively different from the composition in 3-200 fraction (consisting of 3-10, 10-50 and 50-200).
  Regarding different sequencing platforms and downstream processing – in the cases where the same sample was sequenced with both HiSeq and MiSeq, the taxonomic composition was similar between the two platforms. This can be seen in the interactive version of Figure 4 found in the Shiny app, where the taxonomic compositions produced by MiSeq and HiSeq are shown separately for the samples in question. We are therefore confident that variation in sequencing method does not introduce a bias. The Shiny app is now permanently available at: https://micropolar-protists.metapr2.org/ .
  With regards to primer choice, it is always a question whether to use published primers that are already tested, or to try to improve on existing primers, but risking that the changes make them less efficient. The primers from Piredda et al. 2017 have been carefully tested. Allowing for 1 mismatch, they amplify 94% of Haptophyta, and 58% of Rhizaria. However, we agree that possible primer bias should be kept in mind when interpreting the data.
  Finally, please keep in mind that the purpose of this paper is purely to present data. We are not concluding anything regarding the seasonal and depth variation in taxonomic composition. Those type of analyses will be presented in follow-up papers.
  
  Citation: https://doi.org/10.5194/essd-2021-133-AC1
RC2: 'Comment on essd-2021-133', Anonymous Referee #2, 14 Jul 2021

The Arctic is undergoing major changes and data collection there is always challenging. Thus, Arctic datasets are of great value and interest. The manuscript of E. Egge and coauthors includes, as the title states, an 18S V4 rDNA metabarcoding dataset of protist diversity in the Atlantic inflow to the Arctic Ocean, through the year and down to 1000 m depth. This dataset is important and relevant. Yet, this dataset has major problems: there is inconsistency in the sampling points and the filtration procedures, as those change every month. There are also some other minor concerns in this version of the manuscript. Please see comments below.

-Concerns about the project: Samples were taken in 2014. Why is it important to publish the data, now in 2021, and in this journal? Is the dataset complete, or do you have more data from more recent samplings? May this dataset overlap with other datasets retrieved from other polar expeditions in this area?

-Concerns about the sampling stations: The distribution of the sampled stations seems chaotic . Why is not more consistent? Please explain the criteria for selecting the stations. Please provide information on the ice cover of each month (e.g. ice cover maps). What was the distance between the station and the ice cover? Please include this information in the dataset.

-Concerns about the size fractions: Why does the size of the fractions vary among months? This must be justified. Maybe the authors aimed to increase the resolution of the plankton fractions in May, August and November. The only consistent and comparable fraction is the 0.45-3 µm fraction. This is a strong point of the dataset and should be highlighted in the article. A proposal to increase the robustness of the results would be to join the 3-10 µm, 10-50 µm and 50-200 µm samples into one (3-200 µm). In total, there would be only two size fractions (0.45-3 µm and 3-180/200 µm), but will be possible to compare the larger size fractions.

-Concerns about the sequencing: There is little consistency in the sequencing process. Why were the samples sent to two different sequencing centers? Why two different sequencing machines (HiSeq vs MiSeq)? Why were some samples replicated? Why was the number of PCR cycles changed in some samples? Etc.

-Concerns about the dataset: When the project is searched at the ENA browser (https://www.ebi.ac.uk/ena/browser/view/PRJEB40133) the result is: “No records were found for PRJEB40133.” This means the project does not exist, and thus the dataset is not available. Dataset is 155 samples (140+15), and 199 sequencing events. At the ASV document, there are only 198 columns, so one sequencing event is missing. The environmental table includes many variables, but accessory information is lacking on what each variable means, and its units. The latitude and longitude is wrong at the first rows.

Specific comments:

-How to refer to samples: In this manuscript it happens that individual DNA samples (where each sample corresponds to a particular station, depth and fraction) are named as "sample_sizefract". This nomenclature is misleading because it appears to refer to a size fraction in general (and thus encompasses multiple samples). I propose to refer to all DNA samples simply as "sample" or as "DNA sample". Explain this in the text if necessary. To avoid further confusions, I recommend name any other type of sample (replica, niskin bottle, chlorophyll ...) with its particular specification.

-L.40: Please explain better about the challenges. Include if necessary, arguments about adverse weather.

-L.71-77: The information in this paragraph is confusing. If light is important, the dataset should include: sampling time, daylight hours, and number of hours of sunshine during the sampling day.

-Section 2.2.3: Please say that the results of Inorganic nutrients and Chlorophyll a are in Figure 2. Where is the methodology on chlorophyll measurements, cell counts, nutrients and other environmental variables?

-Section 3.1.1: Did you use a rosette? How many bottles per rosette? What time were the casts released?

-L.105-108: There is repeated information in this section.

-L.106, 107, 111: Why is the buffer temperature important?

-L.109: was the nylon mesh previously sterilized?

-L.124: “Subsequently 4 μL RNase was added”. Why?

-L. 129-130: Please briefly comment on the advantage of this primer over others that are also commonly used.

-L.143-151: This section needs many clarifications, e.g:

-L.143-145: Why were the samples sent to two different sequencing centers? Why two different sequencing machines (HiSeq vs MiSeq)?

-L.146-147: What kind of issues did you have with Illumina Miseq in 2015, and why are they relevant? Why was it only done with two runs? Include "center" next to “GATC”, since as it is it leads to confusion.

-L.148: What were these “few samples”? and why were replicas made?

-L.149-150: And why increasing the cycles was a solution? Why not pooling samples?

-L.151: Table 1 should include: type of platform used (Hiseq / MiSeq), site (Oslo / Germany), PCR cycles (25-30)…

-L.161-162: Are there versions of the PR2 databases? which one have you used?

-L.168: better explain this “merged” and why.

-L.169: Why doesn't the 3-180 µm fraction have 40,000 reads, as the others? Why 3-180 µm fraction in Figure 3 has variable number of reads?

-L.170: please indicate if this subsampling was made with a specific function in R

-L.176-178: For consistency, previous sections should explain where the sequencing data is (ENA, link…).

-L.181: I guess "44 samples from niskin bottles" is more accurate (instead of "44 Niskin samples")

-L.182: at this point the reader cannot understand what this code means: "May_P4_net_10_50 failed"

-L.182: “These samples are in the following referred to as ’sample_sizefract”.Where?

-L.181, 185, 187 and others: change "sample_sizefract" to "sample" or "DNA sample". See general comments.

-L.185: please explain what is ENA.

-L.187: why they were merged? Please explain this.

-L.191: the removal of singletons was not mentioned in section 4.2

-L.192 and 193: I imagine that when you say “size fraction” you mean “sample”.

-L.194: I recommend to separate the numbers with ";" instead of "," .

-L.195-196: Please clarify this point.

-L.200: how was the slope calculated?

-L.202: why is this correlation important?

-L.204-205: both figures, Figure 4 and Figure A1, show: metabarcoding reads, ASV, division and class levels. Please rephrase.

-L.206-207: Is the name of the fractions (pico-, nano- and micro-) important? Then explain them at the beginning of the manuscript, and include them in datasets and graphs.

-L.209: Please explain why in Table A2 some groups do not have any >0 values.

-L.210: Please explain which groups are heterotrophic or parasitic.

-L.211: At this point data is "relative abundances", not "read abundances".

-L.210-255: please explain where the values of “richness” comes from.

-L.219: please explain which groups are phototrophs

-L.227: please explain which Division / class corresponds to Diatoms

-L.257: This dataset is descriptive and does not include patterns or dynamics.

-L.258: This dataset is not about food webs.

-L.256-260: An interesting argument to add is that this dataset is a baseline for future studies aiming to determine temporal changes.

-L.262: https://www.ebi.ac.uk/ena/browser/view/PRJEB40133 says: No records were found for PRJEB40133.

-L.263: I recommend to remove this part "corresponding to the size-fractionated plankton samples"

-All figures and tables: captions need improvement. Should provide more information and guidance. Here some comments:

-Figure 2: some profiles are missing, e.g.: there are at least 6 stations in March, and at least 4 go deep. Something is wrong with the y-axis: Is it logarithmic?

-Figure 3: change “sample_sizefract” by “sample” or “DNA sample”. Figure caption needs improvement. The reader here is lost. Avoid using references like "asvtab3_merged_subsamp_readnum.txt". Better if you make other types of references, for example: "see details at Table X ..."

-Figure 4: remove codes and put an understandable legend (e.g. "1m" instead of 0001). If in the bars, the order of protist groups is from left to right, the same order should be in the color code of the legend, but it is the other way around. The two fractions 3-180 µm and 3-10 µm are very different, and should appear separately (different columns).

-Figure A1: this figure needs to be linked with the main text.

-Table A1: include the N (number of samples) included in each size fraction.

-Table A2: Please explain why in Table A2 some groups do not have any >0 values.

Citation: https://doi.org/10.5194/essd-2021-133-RC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Elianne Egge on behalf of the Authors (30 Aug 2021) Author's response Author's tracked changes Manuscript

ED: Publish as is (02 Sep 2021) by David Carlson

AR by Elianne Egge on behalf of the Authors (20 Sep 2021)

Short summary

Here we present a dataset of DNA sequences obtained from size-fractionated seawater samples from the Arctic Ocean that are used to identify taxonomic groups of unicellular plankton. This dataset can be used to investigate the diversity and distribution of plankton groups both by season and by depth and thus increase our understanding of the factors influencing the dynamics of this important part of the Arctic marine ecosystem.