Environmentally-Driven Variation in the Physiology of a New Caledonian Reef Coral

Round 1

Reviewer 1 Report

The authors used coral samples from poorly studied areas in New Caledonia and analyzed the correlation between environmental factors and biological (genetic and physiological) features, and discussed the importance of the results as samples taken before the reefs were severely damaged by annual severe bleaching events. They describes the re-analysis of the original data published elsewhere using different methods. In my opinion, although the data from poorly studied coral reefs are ecologically and biologically important, there is room for improving the way of presenting the methods and results, as well as their interpretation of the results.

 

My major concern is the design of the statistical analyses. It seems to me that some of the response variables are highly correlated with each other, e.g. time and light, which may affect the results of regression analyses. I also wonder whether ANOVA based analyses were suitable for this study. The author just mentioned “data were non-normally distributed” at line 236 and I could not judge from the manuscript how they tested and examined the methods they used were appropriate. I would suggest the authors consider alternative way, e.g. using only selected explanatory variables, for which variance inflation factors are calculated to minimize the effect of multicollinearity, and generalized linear mixed models or similar instead of variance analyses.

 

 

Line 28, “the sampled corals were already presenting signs of climate change- induced stress before onset of ASB”

In my opinion, this type of logic is scientifically inappropriate. This would accommodate ANY changes in this time period as signs of climate change: Without control data, e.g. corals with no eventual damage by annual bleaching for comparison, this conclusion should be avoided.

 

Line 116, “There was an interest in determining the EP that were most influential in…”

I think that this paragraph is better to be moved to the Discussion section.

 

Line 128, “Indeed, 128 this temporal variation has generally invalidated the use of housekeeping genes in real-time PCR-based target mRNA analyses of corals [41] and was an impetus for this research project.”

I could not find how the authors measured, calculated and conducted standardization of mRNA expression levels in the Materials and Methods section and appendix. Especially standardization/normalization is critical when symbiont density within a host is variable among samples. Please describe relevant information without need to bother the readers for looking into previous publications.

 

Line 173 “Environmental parameters have been ranked in accordance with their effect on coral physiology, from most to least significant (see appendix for explanation of the ranking process.)”

In my view this is misleading and the ranking should be reconsidered, since in principle p-value does not represent the size of the effect. Table 2 should be modified for the same reason.

 

Line 345, “were the molecular data not normalized to accommodate differing host/endosymbiont ratios over time (see protocol details in Mayfield et al. [17] & references therein.), the gene expression data would have been inherently biased”

As mentioned earlier, please describe the methods of normalization in the Methods section.

 

Line 418, “SRA found it to be the most 418 influential gene in building a predictive model for light vs. dark behavior”

Please show where the data are presented (with Figure/Table numbers).

Author Response

Reviewer #1

 Summary: The authors used coral samples from poorly studied areas in New Caledonia and analyzed the correlation between environmental factors and biological (genetic and physiological) features, and discussed the importance of the results as samples taken before the reefs were severely damaged by annual severe bleaching events. They describes the re-analysis of the original data published elsewhere using different methods. In my opinion, although the data from poorly studied coral reefs are ecologically and biologically important, there is room for improving the way of presenting the methods and results, as well as their interpretation of the results.

Authors’ response to reviewer #1’s summary: Thank you for dedicating your time to reading our manuscript and making key comments needed to improve its content and clarity. We have done our best to address all issues and concerns you have raised; please see our detailed responses below.

Major comments

 Reviewer#1 major comment #1: My major concern is the design of the statistical analyses. It seems to me that some of the response variables are highly correlated with each other, e.g. time and light, which may affect the results of regression analyses. I also wonder whether ANOVA based analyses were suitable for this study. The author just mentioned “data were non-normally distributed” at line 236 and I could not judge from the manuscript how they tested and examined the methods they used were appropriate. I would suggest the authors consider alternative way, e.g. using only selected explanatory variables, for which variance inflation factors are calculated to minimize the effect of multicollinearity, and generalized linear mixed models or similar instead of variance analyses.

Authors’ response to reviewer #1’s major comment #1: This is an excellent point and suggests that we should have better explained the rationale behind our analysis. In fact, the high degree of collinearity amongst response variables (which is evident from the PCA) was, in addition to the lack of normality, the reason why we used similarity-based approaches, namely PERMANOVA, instead of traditional ANOVA. We did, however, include MANOVA in the main text as a comparison, despite this approach not being well suited to this dataset. Because others may have the same concern as you as to why ANOVA was used with non-normally distributed data, we have removed the MANOVA from the analysis, focusing only on the more robust PERMANOVA, in which neither non-normally distributed data nor multicollinearity are a concern. Partial least squares (PLS) was also undertaken since it was, in fact, developed for the analysis of highly collinear datasets. However, we had failed to state this in the methodology and have now done so.

You are also correct in noting that we failed to state how normality was assessed. We have now done so (Shapiro-Wilk tests), while simultaneously describing the rationale for the analytical approach as follows: “It is important to note that, because the data were not normally distributed (Shapiro-Wilk p<0.01) and the RV tended to correlate with one another, the more common MANOVA was not undertaken. In contrast, both PERMANOVA and PLS are robust against non-normally distributed datasets featuring high degrees of multicollinearity.” In fact, the only parametric statistics found in the article are the MANOVAs calculated as part of the discriminant analysis, and we only interpreted these differences when they were validated by PERMANOVA.

Reviewer#1 minor comment #1: Line 28, “the sampled corals were already presenting signs of climate change- induced stress before onset of ASB” In my opinion, this type of logic is scientifically inappropriate. This would accommodate ANY changes in this time period as signs of climate change: Without control data, e.g. corals with no eventual damage by annual bleaching for comparison, this conclusion should be avoided.

Authors’ response to reviewer #1’s minor comment #1:  This is a good point; better not to end the abstract on pure speculation. Instead, we have reworded this sentence to instead be an avenue for future research: “Whether this degree of sub-cellular stress reflects cumulative climate change-driven impacts or, instead, a stress-hardened phenotype, will be unveiled through assessing the fates of these corals in wake of increasingly frequent marine heatwaves.”

Reviewer #1 minor comment #2: Line 116, “There was an interest in determining the EP that were most influential in…”I think that this paragraph is better to be moved to the Discussion section.

Authors’ response to reviewer #1’s minor comment #2: this is a good point. Actually, since we never discuss the pCO2 effects (we had originally planned to factor them in.), we have simply removed the sentence in question.

 Reviewer #1’s minor comment #3: Line 128, “Indeed, 128 this temporal variation has generally invalidated the use of housekeeping genes in real-time PCR-based target mRNA analyses of corals [41] and was an impetus for this research project.” I could not find how the authors measured, calculated and conducted standardization of mRNA expression levels in the Materials and Methods section and appendix. Especially standardization/normalization is critical when symbiont density within a host is variable among samples. Please describe relevant information without need to bother the readers for looking into previous publications.

Author’s response to reviewer #1’s minor comment #3: Yes, we belabor this point (which most coral biologists actually DISAGREE with!) in both the methods AND the discussion without actually saying how it was done (short of citing our past articles). I have now added a sentence to describe the method in the Materials and Methods: “It is worth noting here, though, that endosymbiosis-tailored protocols developed in our prior works [41-42] were adopted to ensure that spatio-temporally variable host/endosymbiont nucleic acid ratios did not bias gene expression analyses. After first normalizing gene expression (inverse-log raw threshold cycle [Ct] values) to recovery of an exogenous RNA spike (Solaris®, Thermo-Fisher Scientific), we then normalized the dinoflagellate spike-normalized gene expression values to the aforementioned Symbiodiniaceae GCP to control for the fact that samples with high endosymbiont densities would inherently demonstrate higher dinoflagellate gene expression levels.”

 Reviewer #1 major comment #2: Line 173 “Environmental parameters have been ranked in accordance with their effect on coral physiology, from most to least significant (see appendix for explanation of the ranking process.)” In my view this is misleading and the ranking should be reconsidered, since in principle p-value does not represent the size of the effect. Table 2 should be modified for the same reason.

Authors’ response to reviewer #1’s major comment #2: You are correct in noting that p-value alone is not a good indicator of effect size, which is why had paired it with the partial least squares (PLS) misclassification rate. To strengthen this, however, we have now taken a new approach. We conducted multi-dimensional scaling of the 10 response variables and exported the first four coordinates (stress=0.09) to undertake distance-based linear modeling via PLS to uncover the percent variation in coral physiology explained by each environmental parameter. These values are now presented in Table 2 in addition to the PERMANOVA p values and PLS misclassification rates. We no longer attempt to rank the environmental parameters; we simply show the data.

Reviewer #1 minor comment #4: Line 345, “were the molecular data not normalized to accommodate differing host/endosymbiont ratios over time (see protocol details in Mayfield et al. [17] & references therein.), the gene expression data would have been inherently biased” As mentioned earlier, please describe the methods of normalization in the Methods section.

Author’s response to reviewer #1’s minor comment #4. Please see our response above. We have now summarized the normalization strategy in the main text now.

 Reviewer #1 minor comment #5: Line 418, “SRA found it to be the most 418 influential gene in building a predictive model for light vs. dark behavior” Please show where the data are presented (with Figure/Table numbers).

Authors’ response to reviewer #1’s minor comment #5: These data are found in Table 3 (as footnotes, since this approach was essentially used to validate the others.

Reviewer 2 Report

The manuscript entitled 'Environmentally driven variation in the physiology of a New Caledonian reef coral' provides a thorough investigation into the influence of 13 environmental parameters on 10 response variables of the globally ubiquitous coral Pocillopora damincornis. The authors provide a concise introductory to the project, followed by a detailed and comprehensive methodology (linked to previous work to avoid repetition) and a well presented results section, where the data is laid out logically. The paper is well written and hence the authors logic is easy to follow, although I encourage the authors to do a thorough check as there were a few minor grammatical errors remaining (e.g. line 327 remove the additional however in the text). The discussion was also concise, focusing on the key messages with paragraphs 2 and 3 (in the discussion), highlighting important consideration points in which the data was interpreted. Lastly, the data were thorough and appropriately analysed, which gives further confidence to the studies findings. Hence I recommend the paper for publication.

Author Response

Authors’ response to reviewer #2’s summary: Thank you for critically reviewing our article. We are pleased that you see its merits, despite certain grammatical errors and typos. We have now corrected the one you pointed out, as well as several others. Are you to read this article again, you will find we have taken reviewer #1’s suggestion to NOT constantly send the poor reader to the appendix and the prior literature every other sentence, but actually reiterate some things in the main text to help the flow. We have also taken his/her suggestion to remove many of the less appropriate statistical analyses (effectively put in there are a demo in the last version). Upon these revisions, we now hope this article will be suitable for publication in Oceans. Thank you again for taking the time to read this work.  

Reviewer 3 Report

The manuscript by Mayfield and Dempsey deals with spatial and temporal variation in physiological response variables in Pocillopora. The work comprises a partial re-assessment of earlier analyzed (and published) samples supplemented with new samples from the same expedition. The rationale for the current study is different from the previously published report, so the current work has merits for coral science. However, the manuscript is confusing in its aims and is not presented in the most logical way. For example, part the rationale for the current study is presented in the methods section, e.g., lines 116-131 and lines 178-183.

I recommend to simplify the setup of this paper. It is an analysis on 89 samples from Pocillopora damicornis, examining how 10 molecular RV’s respond to 13 EV, with the main aim to set a pre-bleaching baseline for future reference when corals in these areas are studied in association with bleaching, a baseline that takes into account intraspecific variability. Specifically, the analysis focuses on EV that are associated to photosynthesis, as these are presumed the most influential on the RVs chosen. So, I suggest to introduce and present the work in this way, rather than making too much reference to the earlier analysis on these samples. I suggest to skip Table 1.

 

Also the methods section needs revision:

Methods lines 185-191 seem not in agreement with how results are presented. The methods sections suggests that five multivariate methods have been used: PCA, DCA featuring MANOVA, PERMANOVA, SRA and PLS. However, in Table 2, MANOVA is presented as a stand-alone result (not in association with DCA) and SRA is missing in the results (only one mention in the discussion).

Building upon this, I doubt whether all analyses presented are necessary to support the message you want to conceive. I suggest only to include the analyses that are relevant to your discussion (mostly the PERMANOVA and the univariate analysis, I think, maybe with PCA as an in between step). If more is needed, I suggest that it is better described which analysis is used to answer what question. Following up on that, maybe not all result tables and figures are needed.

It will help to add a table with brief descriptions of the biological role of the ten RVs that were used.

Somewhat confusing is the information on the number of samples (biopsies) taken and analyzed. Lines 131-135 suggest that less than 36 night dive samples have been taken (4-9 replicates per location, 4 locations). However, in line 143, the number of 36 is mentioned. I also do not understand the difference between 120 samples in line 147 and 140 samples in line 143, and I do not understand the 52 out of 120 in line 147. Please, stick to the sample numbers that have been analyzed for this paper.

 

The discussion is well written, but limited. A lot of work has been done on HSP in corals by other groups, which is totally ignored. The discussion could also benefit from comparison with diurnal patterns in gene expression in other phototrophic organisms. A lot of work has been done on photosynthesis-related physiological measurements on corals during a diurnal cycle (i.e. PAM data), which could be aligned to your gene expression patterns.

The number of self-citations must be reduced considerably. I counted 47 out of 76 references in which one of the authors is involved, that is outrageous and akin to citation-pushing. An example: where you cite reference 24-28, only numbers 24 and 26 will do the job.

 

A few small comments:

 

Line 189: “undercover” should be “uncover”

Line 222: delete “First”

Line 233 ...analyze in different ways

Line 327: there is one “however” too much

Lines 315-319: does this text relate to molecular responses to stress or to gene expression in general? In the former case, it should be mentioned, in the latter case, I do not understand the addition between brackets in lines 318-319.

Line 365: how can something be “generally high in 100% of samples”?

Lines 365-367: “mean real-time PCR 365 threshold cycle (Ct) values of 22±1.6, 24±1.6, and 23±1.7, respectively (see normalized, 366 transformed, &/or standardized values in Figures 2-4 & the OSDF.)” Why are these data mentioned here? If this is an important finding for your narrative, these data should be shown in the results section and put into context, so that we can also see that these Ct values represent a high expression.

Line 390: ... shifts in gene expression in symbionts are indicative...

Fig 4: either use 50% or 1.5 fold, now it is mixed. This info is missing for panel e.

Author Response

Reviewer #3

 

Summary: The manuscript by Mayfield and Dempsey deals with spatial and temporal variation in physiological response variables in Pocillopora. The work comprises a partial re-assessment of earlier analyzed (and published) samples supplemented with new samples from the same expedition. The rationale for the current study is different from the previously published report, so the current work has merits for coral science. However, the manuscript is confusing in its aims and is not presented in the most logical way. For example, part the rationale for the current study is presented in the methods section, e.g., lines 116-131 and lines 178-183. I recommend to simplify the setup of this paper. It is an analysis on 89 samples from Pocillopora damicornis, examining how 10 molecular RV’s respond to 13 EV, with the main aim to set a pre-bleaching baseline for future reference when corals in these areas are studied in association with bleaching, a baseline that takes into account intraspecific variability. Specifically, the analysis focuses on EV that are associated to photosynthesis, as these are presumed the most influential on the RVs chosen. So, I suggest to introduce and present the work in this way, rather than making too much reference to the earlier analysis on these samples. I suggest to skip Table 1.

 Authors’ response to reviewer #3’s summary: Thank you for taking the time to critically review our manuscript and make suggestions to improve its readability. I am glad you were able to sift through all of the background info to correctly deduce what we set out to achieve. We had thought that a lengthy treatise on our prior analysis to frame the story, as well as because the editor actually asked for us to justify how this work differed from our prior New Caledonia publication, but readers could simply read that prior work (i.e., no need for so much repetition and focus on the prior work). In fact, I really like your summary sentence and have reworded it to feature at the end of our Introduction. I also noticed that, as you pointed out, the best justification for the approach was actually in the methods, and that effectively repeated what was stated in the Introduction. You will now see that the justification for the approach that was previously in the Methods is now at the end of the Introduction (which is now much shorter thanks to your suggestion).

Regarding Table 1, we have left it in, but we have removed the final column that mentioned prior results since most of those were never discussed in this manuscript. Namely, we just want to show the reader how we “binned” or categorized the various environmental parameters. However, this could easily be moved to the appendix if you and the other reviewers deem it necessary. After all, it will be clear from the online supplemental data file how we categorized the various response variables.  

Also the methods section needs revision:

Reviewer #3 major comment #1: Methods lines 185-191 seem not in agreement with how results are presented. The methods sections suggests that five multivariate methods have been used: PCA, DCA featuring MANOVA, PERMANOVA, SRA and PLS. However, in Table 2, MANOVA is presented as a stand-alone result (not in association with DCA) and SRA is missing in the results (only one mention in the discussion).

Authors’ response to reviewer #3’s major comment #1: This is a good observation and one that reviewer #1 pointed out, albeit it in a different light. For one, we have generally removed MANOVA since it is NOT the best approach for this work. Secondly, SRA was only used to effectively corroborate results (it is a footnote in Table 3). As you pointed out, PERMANOVA was essentially our #1 approach since it is the most robust. However, reviewer #1 wanted us to emphasize partial least squares (PLS) MORE since it is the best method (besides PERMANOVA) for addressing inter-response variable multi-collinearity. You will now see in the paragraph in question that we have (hopefully) better outlined the strategy. It wasn’t simply: “Let’s throw every approach in the book at it and see what happens,” but more meant for self-validation. I have now re-written this paragraph to more carefully explain why we used the approaches we did: “PCA was also used to depict relationships among RV, while DCA was used to quantify the differences evident from PCA in two ways: calculation of 1) Wilks’ lambda (i.e., multivariate ANOVA [MANOVA]) and 2) a partial least squares (PLS) model misclassification rate (% of samples misclassified by the associated predictive model). The former was presented in the figures but not interpreted since, unlike PLS, MANOVA is poorly suited for highly collinear datasets featuring non-normally-distributed data (Shapiro-Wilk p<0.01) such as the one produced herein. In addition to PLS, PERMANOVA is also robust against non-normally distributed datasets featuring high degrees of multicollinearity since the similarities (i.e., distances) among samples, rather than absolute differences in levels of individual RVs, are instead tested in response to the EP(s) of interest. In a separate PLS in which the first four coordinates from a multi-dimensional scaling (MDS) analysis of standardized data from the 10 coral RV were the model Y terms, each EP was tested as a predictor, and the percent variation in the multivariate trait space was calculated (NIPALS fit with kfold validation of 7). SRA was only used to assess multivariate effects for light and time (see appendix.) and was undertaken mainly to corroborate findings from the non-parametric univariate analyses of the 10 RV over time and light levels (discussed in the following paragraph).”

Reviewer #3 major comment #2: Building upon this, I doubt whether all analyses presented are necessary to support the message you want to conceive. I suggest only to include the analyses that are relevant to your discussion (mostly the PERMANOVA and the univariate analysis, I think, maybe with PCA as an in between step). If more is needed, I suggest that it is better described which analysis is used to answer what question. Following up on that, maybe not all result tables and figures are needed.

Authors’ response to reviewer #3’s major comment #2: Since Reviewer #1 actually was unconvinced of the statistical power of our analysis, we have left most analyses in. however, please see our response to your previous, similar comment; since we DID deliberately include some redundant analyses (to basically double-check things), we now describe this better. Otherwise, the reader will wonder why we essentially did two very similar analyses that gave us the same answer. We had wanted to design a ranking system to determine which environmental parameter is MOST influential on coral biology, but reviewer #1 did not like this idea, so we have rewritten large sections of the methods to accommodate both his/her comments, as well as your own.

Reviewer #3 major comment #3: It will help to add a table with brief descriptions of the biological role of the ten RVs that were used.

Authors’ response to reviewer #3’s major comment #3: Although we had something similar in our prior work with these samples, I do think it would be good if readers weren’t constantly having to go back and forth among various publications. As of right now, the response variables are simply explained in text form in the appendix, so now I’ve modified Table 1 to where it includes all environmental parameters AND response variables (with a brief description of their full name and function). Hopefully, this will now make the manuscript easier to read.

Reviewer #3 minor comment #1: Somewhat confusing is the information on the number of samples (biopsies) taken and analyzed. Lines 131-135 suggest that less than 36 night dive samples have been taken (4-9 replicates per location, 4 locations). However, in line 143, the number of 36 is mentioned. I also do not understand the difference between 120 samples in line 147 and 140 samples in line 143, and I do not understand the 52 out of 120 in line 147. Please, stick to the sample numbers that have been analyzed for this paper.

Authors’ response to reviewer #3’s minor comment #1: this was originally done to better distinguish this work from our prior work (requested by the editor), but we see where it could be confusing. Also, there were 26 colonies sampled in both day and night for which data were produced (52 samples); there were actually additional night dive colonies whose day time samples were poorly preserved. This is why there were 59 day and 30 night samples. I have tried to do a better job explaining this in the revised methods as follows: “Each biopsy was frozen at -150°C in a liquid nitrogen vapor shipper (MVE Chart) upon return to the small craft vessel, and RNAs, DNAs, and proteins were extracted from these coral biopsies as described previously [17] after export under a CITES permit to A.B.M. to Taiwan. Specifically, 30 colonies sampled during both day and night (n=60 biopsies), as well 33 additional biopsies sampled only during the day underwent extractions (n=93). Of these, four of the daytime biopsies yielded degraded RNA and were not analyzed, resulting in a final sample size of 30 night and 59 daytime samples (of which 52 represented matched-pair sampling of the same 26 colonies).

Reviewer #3 major comment #4: The discussion is well written, but limited. A lot of work has been done on HSP in corals by other groups, which is totally ignored. The discussion could also benefit from comparison with diurnal patterns in gene expression in other phototrophic organisms. A lot of work has been done on photosynthesis-related physiological measurements on corals during a diurnal cycle (i.e. PAM data), which could be aligned to your gene expression patterns.

Author’s response to reviewer #3’s major comment #4: There is indeed much work done on coral and endosymbiont HSP levels, and we deliberately chose to avoid discussing it because we honestly do not trust the data integrity (see second paragraph of the Discussion.). However, it could be fruitful to look at PAM or other photosynthesis data, since there is a rich literature on that. Although PAM data will not tell you about gene expression, reactive oxygen species levels will, and it is possible that diel fluctuations in ROS levels could be driving some of the temporal variation in stress marker levels in particular. I have now written a new section in the Discussion as follows: “It is also possible that diel fluctuations in reactive oxygen species (ROS) levels resulting from even normal levels of photosynthesis-driven oxygen evolution from the endosymbionts to the hosts [74] could have driven not only changes in expression of photosynthesis-targeted genes but also the aforementioned stress markers (in both compartments); indeed, oxidative stress is a key regulator of both gene expression and protein levels in reef corals [75]. Although cu-zn-sod levels and ubiq-lig levels were higher at night, it is worth mentioning that corals were sampled only an hour after sunset; corals were no longer photosynthesizing yet this nevertheless could reflect the peak period of ROS production given the inherent lag time between oxygen generation and macromolecular damage. Future work should attempt to sample later in the evening to determine whether stress marker levels eventually reach the lower daytime levels well before the morning.”

Reviewer #3 major comment #5: The number of self-citations must be reduced considerably. I counted 47 out of 76 references in which one of the authors is involved, that is outrageous and akin to citation-pushing. An example: where you cite reference 24-28, only numbers 24 and 26 will do the job.

Authors’ response to reviewer #3 major comment #5: This is a good suggestion and, honestly, most referenced statements are common knowledge and thus require no reference at all. I have now removed a good number of the references, as well as added some new ones to support new arguments, to attempt to give the article a more balanced read. The total number of references is now only 49. We did have to self-cite at certain point since this is essentially the use of an SOP that we are testing around the world to increase comparability across coral studies.

A few small comments:

Reviewer #3 minor comment #2: Line 189: “undercover” should be “uncover”

Author response to reviewer #3’s minor comment #2: Good catch! The suggested change has been made.

Reviewer #3 minor comment #3: Line 222: delete “First”

Author response to reviewer #3’s minor comment #3: I think the sentence in question must no longer exist.

Reviewer #3 minor comment #4: Line 233 ...analyze in different ways

Author response to reviewer #3’s minor comment #4: I do not think this sentence exists in the manuscript anymore.

Reviewer #3 minor comment #5: Line 327: there is one “however” too much

Author response to reviewer #3’s minor comment #5: Wow, two reviewers caught this error! The suggested change has been made.

Reviewer #3 minor comment #6: Lines 315-319: does this text relate to molecular responses to stress or to gene expression in general? In the former case, it should be mentioned, in the latter case, I do not understand the addition between brackets in lines 318-319.

Author response to reviewer #3’s minor comment #6: To my knowledge, no molecular work has been done on a New Caledonian coral period. We also used “vague” wording like “molecular physiology” because these sentences are referring to gene expression AND the other molecular response variables (RNA/DNA ratio and Symbiodiniaceae GCP), in other words, not ONLY gene expression. We just focused more on gene expression in the remainder of the Discussion since we thought it would be of more interest to others.

Reviewer #3 minor comment #7: Line 365: how can something be “generally high in 100% of samples”?

Author response to reviewer #3’s minor comment #7. Concentrations were high in all samples. We have revised this statement to reflect this better wording.

Reviewer #3 minor comment #8: onceLines 365-367: “mean real-time PCR 365 threshold cycle (Ct) values of 22±1.6, 24±1.6, and 23±1.7, respectively (see normalized, 366 transformed, &/or standardized values in Figures 2-4 & the OSDF.)” Why are these data mentioned here? If this is an important finding for your narrative, these data should be shown in the results section and put into context, so that we can also see that these Ct values represent a high expression.

Author response to reviewer #3’s comment #8: We had included these in the event that someone familiar with qPCR would read the article, but you are correct that it could be seen as strange to present seemingly “new” data in the Discussion. Therefore, we have deleted the absolute values.

Reviewer #3 minor comment #9: Line 390: ... shifts in gene expression in symbionts are indicative...

Author response to reviewer #3’s minor comment #9: The suggested change has been made.

Reviewer #3 minor comment #10: Fig 4: either use 50% or 1.5 fold, now it is mixed. This info is missing for panel e.

Author response to reviewer#3’s comment #10: The information is missing for panel e because there was no effect in this test. The second point is a good one, though, and we have made the suggested change.

 

Round 2

Reviewer 1 Report

The authors have made a considerable effort to respond to my comments in the previous reviewing process. I have minor comments on the data presentation.

I still have confusion about the interpretation of Table 2.

I suggest that actual p-values (raw values) for all the test results be described instead of thresholds (i.e. P<0.01, N.S.) for readers to interpret the results more evenly. Raw p-values will help to avoid the confusions, e.g. I could not judge if reef exposure “p<0.05” was higher than 0.045 and classified as N.S. In addition, according to the legend, “p=0.05” for salinity should be classified as “N.S”.

It seems to me that the “Conclusion” column is misleading as neither in the legend nor main text sufficient information is provided, and the subjective conclusions do not fit to the table of results, e.g. distinctions among “Weak effect”, “No significant effect”, “No effect” are unclear. I suggest that the column of “Conclusion” is not necessary and be removed.

Similarly, Table 3 can be improved by showing raw p-values for all cells and removing the “Conclusion” column. It can also be potentially helpful for readers to raise a question about the needs of adjustment for multiple comparison, considering the type of tests done using the similar datasets.

Author Response

Please note that I included this same text in the cover letter to the editor (in case you'd prefer to download a document and read them).

Summary: The authors have made a considerable effort to respond to my comments in the previous reviewing process. I have minor comments on the data presentation.

Author response to reviewer #1’s summary: Thank you for taking a second look at our article, and, in fact, we had already begun to change Table 2 in accordance with your suggestions prior to even receiving these comments. Please read on for details, and thanks again for your efforts in improving this article.

Reviewer #1’s minor comment #1: I still have confusion about the interpretation of Table 2. I suggest that actual p-values (raw values) for all the test results be described instead of thresholds (i.e. P<0.01, N.S.) for readers to interpret the results more evenly. Raw p-values will help to avoid the confusions, e.g. I could not judge if reef exposure “p<0.05” was higher than 0.045 and classified as N.S. In addition, according to the legend, “p=0.05” for salinity should be classified as “N.S”.

Author response to reviewer #1’s minor comment #1: This is a good suggestion and will help with the ranking we had originally aimed to present. First, you are correct; technically, by our a priori definition, p=0.05 is not statistically significant, so this has been revised accordingly. Unfortunately, PRIMER does not provide exact p-values below 0.01. If the value is below 0.01, it simply states "0.01" unless it is closer to 0.001, in which case it states "0.001." In other words, you effectively get a range (e.g., 0.001-0.01) since it is based on randomly permutating the data. However, I did think of another, potentially useful way to essentially “break ties” for two environmental parameters with the same PERMANOVA p-value range by creating a parameter, the “predictor score,” that factors in the two partial least squares-based metrics: % variation explained (higher=better) and % misclassified (lower=better). Basically, I added the % variation explained to the model accuracy (1-% misclassified) to give a sense of how influential each environmental parameter was in serving as a predictor of coral condition. As you will see, this resulted in a few differences in the relative ordering. I also got rid of the “Conclusions” column since it was poorly defined (as you pointed out astutely) and, more importantly, not actually very important for our story.

Despite these efforts, I have still taken your initial suggestion and have tried to steer clear of using the term “rank;” despite this essentially having been done in the background. The environmental parameters are ordered first with respect to the PERMANOVA p-value range and secondly with respect to the aforementioned predictor score but I only read into this for the non-statistically significant findings. For instance, I am not confident that temperature is more important than colony color; they have the same PERMANOVA p-values, and while temperature has a higher predictor score, it also explained a higher percentage of the variation. In other words, depending on the model benchmark used, the relative ordering might change. The important thing to note for this article is that light/time were near the top of the list.

Reviewer #1 minor comment #2: It seems to me that the “Conclusion” column is misleading as neither in the legend nor main text sufficient information is provided, and the subjective conclusions do not fit to the table of results, e.g. distinctions among “Weak effect”, “No significant effect”, “No effect” are unclear. I suggest that the column of “Conclusion” is not necessary and be removed.

Author response to reviewer #1 minor comment #2: See our response above. Briefly, the suggested change has been made, with a new metric added instead of the “conclusion,” which was poorly defined (and, in some cases, not actually supported robustly by the data).

Reviewer #1 minor comments #3-4: Similarly, Table 3 can be improved by showing raw p-values for all cells and removing the “Conclusion” column. It can also be potentially helpful for readers to raise a question about the needs of adjustment for multiple comparison, considering the type of tests done using the similar datasets.

 Author response to reviewer #1’s minor comments #3-4: Because these two comments are actually linked, please allow me to address them in tandem. The reason we did not originally present actual raw p-values or even use annotation (e.g., *p<0.05, **p<0.01, etc.) is because false discovery rates were used given the large number of comparisons: 10 response variables vs. 13 environmental parameters (130 tests). However, perhaps such a stringent approach is not needed for Table 3 since in this instance we are only looking at one environmental parameter: sampling time. Therefore, I have gone back and included the actual non-FDR-adjusted p-values in the table. Note that for these univariate tests, a lower alpha was set (0.02 vs. 0.045 for the multivariate analyses). Also note that, for p-values below 0.0001, the statistical software (JMP) does not provide an exact value, so, in these instances, “p<0.0001” has been used instead of the exact value.

Reviewer 3 Report

I appreciate your honest and cheerful reply to the review. The manuscript has improved considerably.

There are a few minor points in the revised text that I would like to see modified before final acceptation.

Introduction, lines 60-63, I suggest to change to:

In an attempt to develop a more rigorous understanding of environmental drivers of variability [cf. 13-15] in the molecular physiology of this model coral species [16], we performed an additional analysis on the previously published data [12]. Hereby, we focused not only on spatial heterogeneity (among locations), but also on interspecific (inter-genotype) variation within locations and on temporal variation within genotypes (intra-genotype).

Line 63: “modeling” is not an appropriate term here. You can change to: “relating”

Line 64: “three EP” should be “13 EP”

Table 1: the last note ( c. Lower number of statistical bins than when both host species were analyzed simultaneously (Tables A1-2).) is not relevant anymore and should be omitted.

In the final edit, Table 2 should appear in the results section

Section 2.3 should be rewritten as follows:

Ten RV were analyzed in each of 89 coral biopsies (n=60 light & 29 dark samples; see Supplemental materials for details). These RV included the RNA/DNA ratio (a proxy for total gene expression), the Symbiodiniaceae genome copy proportion (GCP; a molecular proxy for endosymbiont density [22]), and expression of four host coral and four Symbiodiniaceae mRNAs. Based on prior transcriptomic analyses [23-24], these mRNAs were hypothesized to be environmentally sensitive and useful in physiology-based resilience model building. The 8 mRNAs featured genes encoding proteins involved in metabolism, photosynthesis, the stress response, and other key cellular processes.

Author Response

Reviewer #3 summary: I appreciate your honest and cheerful reply to the review. The manuscript has improved considerably. There are a few minor points in the revised text that I would like to see modified before final acceptation.

Authors’ response to reviewer #3’s summary: Thank you for taking another look at our manuscript and for providing comments and suggestions to improve its flow. We have done our best to address your concerns. Please read on for details.

Reviewer #3 minor comment #1: Introduction, lines 60-63, I suggest to change to:

In an attempt to develop a more rigorous understanding of environmental drivers of variability [cf. 13-15] in the molecular physiology of this model coral species [16], we performed an additional analysis on the previously published data [12]. Hereby, we focused not only on spatial heterogeneity (among locations), but also on interspecific (inter-genotype) variation within locations and on temporal variation within genotypes (intra-genotype). 

Authors’ response to reviewer #3’s minor comment #1: The suggested change has been made with a small correction; technically, in this analysis we looked ONLY at one species, so we focused on intraspecific (inter-genotype) AND intra-individual (i.e., intra-colony) variation over space and time. We do plan to add in other pocilloporid species in the future (i.e. inter-specific comparisons), though.

Reviewer #3 minor comment #2: Line 63: “modeling” is not an appropriate term here. You can change to: “relating”

Authors’ response to reviewer #3’s minor comment #2: The suggested change has been made.

Reviewer #3 minor comment #3: Line 64: “three EP” should be “13 EP”

Authors’ response to reviewer #3’s minor comment #3: Good catch! The suggested change has been made.

Reviewer #3 minor comment #4: Table 1: the last note ( c. Lower number of statistical bins than when both host species were analyzed simultaneously (Tables A1-2).) is not relevant anymore and should be omitted.

Authors’ response to reviewer #3’s minor comment #4: The suggested change has been made.

Reviewer #3’s minor comment #5: In the final edit, Table 2 should appear in the results section

Authors’ response to reviewer #3’s minor comment #5: That is a good point. I dragged it to the beginning of the Results and will ensure that it falls within the Results section when it goes to proof.

 Reviewer #3’s minor comment #6: Section 2.3 should be rewritten as follows:

Ten RV were analyzed in each of 89 coral biopsies (n=60 light & 29 dark samples; see Supplemental materials for details). These RV included the RNA/DNA ratio (a proxy for total gene expression), the Symbiodiniaceae genome copy proportion (GCP; a molecular proxy for endosymbiont density [22]), and expression of four host coral and four Symbiodiniaceae mRNAs. Based on prior transcriptomic analyses [23-24], these mRNAs were hypothesized to be environmentally sensitive and useful in physiology-based resilience model building. The 8 mRNAs featured genes encoding proteins involved in metabolism, photosynthesis, the stress response, and other key cellular processes.

Authors’ response to reviewer #3’s minor comment #6: This is a good suggestion; rather than constantly sending the reader to the appendix, put the important bit front and center. The suggestion change has been made.