Potential Role of Gastrointestinal Microbiota in Growth Regulation of Yellowtail Kingfish Seriola lalandi in Different Stocking Densities

Round 1

Reviewer 1 Report

Thank you for submitting your manuscript for consideration in Fishes. The premise of the study is interesting and serves as a basis for additional research which could be useful for yellowtail kingfish farming. That being said, there are several areas where the manuscript needs to be revised. In particular, the microbiota data analysis and results needs work. There are too many issues with terminology to mention all. I would suggest re-reading a few of the manuscripts already referenced (Walburn et al. 2019 and Horlick et al. 2020) and/or having whoever analyzed the data perform a more thorough review of how all microbiota analysis/data are presented. However here are a few examples:

“…the principles of dominance, commonality and difference.” – these principles are mentioned more than once in the paper but they are not explained. The specific criteria used to determine “core” microbiota should be mentioned. “Dominant KEGG pathways” are also described and presented in a figure with no description as to how these were determined to be “dominant”.

“the change rule of core…” – I am not understanding what “change rule” means here

PCR methods are not well described. I suspect they did their V3-4 PCR, followed by clean-up, and finally indexing PCR, but this is not clear.   

I am not aware of a Illumina HiSeq PE250 system. Perhaps the authors meant to mention a specific HiSeq model (2500? 3000? 4000?), however, to my knowledge, paired end reads of 250 bp each are not available on any HiSeq. That would be a relevant read length for a MiSeq. Did the authors actually mean a MiSeq or 2x125bp on a HiSeq?

“Raw data obtained from high throughput sequencing were processed by splitting, splicing, filtering, and extracting, then…” – the specific tools used for each step need to be mentioned and generally the entire methods section needs to be revised to be more clear. I suspect by splitting the authors mean splitting by barcoding (demultiplexing), by splicing perhaps trimming (trimming off primers? Specific length?), at some point there should be merging of forward and reverse reads, etc.

“Data for all samples were homogenized based on…” – I suspect this is talking about data normalization in the form of rarefaction? If that is the case, please mention the specific # sequences that samples were rarefied too.

The data resulting from microbiota studies are typically not normally distributed and generally reject many of the assumptions that would allow for parametric tests to be utilized. The statistical analyses likely need to be modified. Additionally, it’s unclear if paired tests were perform however if you are evaluating changes over different timepoints, the statistical test used needs to consider that the samples are not independent.

Beta diversity analysis (incorrectly listed as “correlation analysis”) should include some kind of statistical analysis. Additionally, the PCA plot is difficult to interpret. Colors and/or shape should be used to differentiate between different variables of interest.  In addition to this, generally in the figures there needs to be more clear and consistent labelling – instead of “A”, “B”, “C” (or “A2”, “B2”, etc.) these should be labelled as the specific treatments e.g. “high”, “medium”, “low”.

“microbiota genes” is used in reference to the pathway prediction that was performed. This is not a measure of microbiota genes, this is only predictions of pathways based on the 16S sequencing data.

Figure references become incorrected after 4.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The content of the submitted manuscript is scientifically sound and gives interesting information on the potential gastrointestinal microbiota role for rearing yellowtail kingfish Seriola lalandi. However there are some points to be revised:

Results

Lines 177: Table 1: third column final body weight

Line 179: different density groups

Line 185: Alpha diversity: The authors mention about the trend of Chao 1 index. Why did you not consider the commonly used Shannon index?

Line 214: A1, B1 and C1 represented three different treatments at the beginning of this trial, while A2, B2 and C2 represented those three treatments at… You should repeat this caption in all the figures.

Line 225: Epsilonbacteraeota was significantly different in stomach among three density groups (in Fig. 3a only between the low density group and the others) be more precise.

Line 240: Fig 3a is cut. It lacks the gut part in the histogram.

Line 246: dominant phyla

Line 293-299: The principal component analysis is shown in Fig. 6. The  figure 6 should be enlarged and described better; it is not clear and  does not explain what you describe in the text. “structures of microbiota ............. more similar at the end of this trial, while those in pyloric caecum and gut between low- and medium-density group? (also C1P and C2P are very similar) were all more similar, respectively. Describe better Fig. 6 in the text. With the growth of fishes, structures of microbiota in stomach of fishes farmed in low- and high-density groups changed more obviously? (why more obviously…. The change was more evident?) than those in pyloric caecum and gut in three groups. PCA does not explain all the variability but only a part of it 34.5% PC1 and 17.8% PC2. You should talk about this in the discussion.

Line 305: C2P, A2G, B2G and C2G represented the stomach, pyloric caecum and gut samples at the end of…….

Line 310 and 314: Fig.7a and Fig.7b respectively.

Line 334: Better to include Fig. 8a in the text.

Line 340 and 344: Fig. 8a and 8b respectively instead of Fig.9a and 9b.

Discussion

Line 381: “the ratio of Firmicutes to Bacteroidetes in stomach and pyloric caecum of fishes farmed in the medium-density group were higher” with respect to the other groups.

Line 431: “……yoghurt to reduce aflatoxin poisoning and to produce bacteriocins….”

Line 435-436. “The relative abundance of Lactobacillus, was relative stable in all groups. There were  no significant differences among three density groups at the beginning of this trial. (data not shown).

Line 437-443: “ Influenced by density, relative abundances of Bifidobacterium and Dongia in stomach, Bifidobacterium, Sphingomonas, Klebsiella and genus represented by MND1 in gut significantly increased in the medium density group, and the relative abundance of Weissella in pyloric caecum and  Ruminiclostridium in gut significantly decreased in the medium density group.

After a minor revision the paper can be accepted.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Thank you for making modifications to your manuscript. The methods regarding next-generation sequencing are improved and should be more helpful to readers.

However there still needs to be further work done with how microbiota data are analyzed and presented. In addition, this manuscript still needs to go through English language editing. It’s possible that many of my issues with the way data are presented and parts where I am not understanding what is being written could be resolved with modified wording.  

In my previous comments, I suggested that there were many issues with how the microbiota data were presented and provided only a few examples to be revised. These examples were addressed, however these were only examples and I would suggest the authors take time to read through other manuscripts that contain microbiota data and modify their own manuscript to better present these data.

I do have a few responses to the modifications/author responses from this first round of review:

Point 1: “…the principles of dominance, commonality and difference.” – these principles are mentioned more than once in the paper but they are not explained. The specific criteria used to determine “core” microbiota should be mentioned. “Dominant KEGG pathways” are also described and presented in a figure with no description as to how these were determined to be “dominant”.

Response 1: In the section of “Results”, inorder to explain the “dominant”, we changed “The top-ten phyla” to “The domiant (top-ten according to the relative abundance) phyla” in the first sentence of the first paragraph in “3.3”, changed “All dominant genera” to “All dominant (top-ten according to the relative abundance) genera” in the first sentence of the third paragraph in “2.3”, and changed “… were the main pathways in all samples (Fig. S5)” to “… were the dominant (top-fifteen according to the relative abundance) pathways in all samples (Fig. S5)” in the second sentence of the first paragraph in “3.5”. We selected the dominant, shared and differential microbiota as “core microbiota”, so, we mentioned that “…the principles of dominance, commonality and difference.” in the manuscript. If you do not agree with this modification and explanation, you are welcome to communicate with us.   

“Top ten according to the relative abundance” meaning according to the average relative abundance across all samples? If yes, should be revised to “Top ten based on average realtive abundance”. Although the “core microbiota” in the context of bacterial communities can be determined based on a number of different criteria, it broadly means members that are shared between some kind of groups (either different environments or in the case of this publication, the different sample groups) indicating they are permanent members of that community, likely with some kind of important role. While there is flexibility with how these core microbiota are defined, the “principles of dominance, commonality and difference” are not only unclear but do not seem to make sense for defining a core microbiota. In fact, it is worded in a way that is inherently contradictory due to the “commonality” and “difference” aspects of that definition, which implies that your core microbiota is defined as including both bacteria that are shared but also not shared between sample groups. Perhaps if there was further explanation I would feel the criteria used to identify these core genera are sufficient, but as it stands it seems as though your core microbiota include any bacteria that is either in:

a. your “top ten” (top ten in quotation makes because the list of the top 10 bacteria based on relative abundance actually includes more than 10 genera) bacteria with the highest relative abundance (highest average relative abundance

b. top ten genera with significant difference in relative abundance (which appear to not just be the top ten most abundant bacteria subjected to differential abundance testing but maybe all genera subjected to this testing, and then the top ten in terms of relative abundance chosen?)

and then also included Lysobacter and Ruminiclostridium with no previous mention of these bacteria, except for them being present in a single plot where they appear to have some change in relative abundance in specific sample types/groups over time point. In addition it does not appear that any bacteria fitting in either of these criteria allows them to be considered part of the “core microbiota” since there are some taxa (e.g. Sutterella and Blautia) which appear to fit these criteria based on the text and figure 5, but or not included in the list of core microbiota on line 519-523.

Please clearly define a set of criteria for which you are defining the core microbiota and ensure that it is conducive to the broad definition of core microbiota. Additionally if my interpretation of the existing criteria and list of core microbiota are correct, I would highly suggest you modify this. Perhaps choose all bacteria that are present in all samples (across time points, stocking densities, sections of GI tract) above a certain relative abundance.

If the authors disagree and still feel that their criteria is logical, please provide an argument for such, and adjust your list of core microbiota to include any bacteria fitting these criteria (e.g. incorporating Sutterella, Blautia, and any others not already included) or provide explanation as to why this list is curated in this way.

Point 2: “the change rule of core…” – I am not understanding what “change rule” means here

Response 2: We think that the “change rule” can be understood as “ the trend of change”.

I would not understand “change rule” to mean “the trend of change” nor do I feel many other readers would. If “change rule of core genera…” is meant to mean “Differences in the relative abundance core genera between sample groups…” then it should be written as such.

Point 8: Beta diversity analysis (incorrectly listed as “correlation analysis”) should include some kind of statistical analysis. Additionally, the PCA plot is difficult to interpret. Colors and/or shape should be used to differentiate between different variables of interest.  In addition to this, generally in the figures there needs to be more clear and consistent labelling – instead of “A”, “B”, “C” (or “A2”, “B2”, etc.) these should be labelled as the specific treatments e.g. “high”, “medium”, “low”.

Response 8: We analyzed the differences of all samples in PC1 and PC2 as the statistical analysisi for Beta diversity analysis. And, we changed “3.4 The correlation analysis” to “3.4 Beta diversity analysis”. The plot in PCA (Figure 6a) represented each sample and was labeled with relevant letters and number. And, these labells were marked in figure 6a. In order to make the “Notes” more clear and considering the consistent labelling, we changed “Notes: …” to “Notes: A1S, B1S and C1S represented the stomach samples in low-, medium- and high-density group at the beginning of this trial, while A1P, B1P and C1P represented the pyloric caecum samples in low-, medium- and high-density group at the beginning of this trial, and A1G, B1G and C1G represented the gut samples in low-, medium- and high-density group at the beginning of this trial, respectively. Likewise, A2S, B2S, C2S, A2P, B2P, C2P, A2G, B2G and C2G represented the stomach, pyloric caecum and gut samples in low-, medium- and high-density group at the end of this trial, respectively.” in Figure 6a.

I am not familiar with statistical analysis for beta diversity being applied on principal components themselves. I have only seen researchers conduct statistical tests on the beta diversity data itself (dissimilarity or distance matrix), rather than the PCA values which are just used to plot. Some other statistical analysis (maybe either PERMANOVA or ANOSIM) should be performed.

Looking at this plot again, I see now that there is only 1 sample per sample type, collection time, and group. From the methods, I would think that you’d have 3 for each of these, e.g. 3 samples obtained from the stomach of fish in the high density group at day 0. Why are these data not shown on the plot?

Although I do see that you describe the labelling system in the legend, I do feel it would be better to use colors and/or shapes to differentiate between the variable of interests. For example, stomach samples from day 0 could be shaded squares, stomach samples from day 90 could be open squares, samples from the high density group could be blue, from low density group could be red, etc. Additionally, throughout the manuscript I do still feel as though samples in figures should be labelled to be more informative, where appropriate. E.g. instead of “A”, “low density” or “low”, instead of “1”, “Day 0”. These in part personal preference, however I feel this would greatly improve the figures. I will leave this up to the editor to decide whether this is something that should/should not be adjusted.

 

Previously I said that since some statistical analyses are conducted over time using the same experimental units, these tests need to incorporate the fact that data points are not independent. This has not been incorporated in the manuscript. Either the authors should explain why this is not necessary or the statistics should be performed using a test that considers these data are not independent. The non-parametric Freidman would potentially be a good choice for this.

Author Response

Please see the attachment.

Author Response File: Author Response.doc