Coupling of palaeontological and neontological reef coral data improves forecasts of biodiversity responses under global climatic change

Reef corals are currently undergoing climatically driven poleward range expansions, with some evidence for equatorial range retractions. Predicting their response to future climate scenarios is critical to their conservation, but ecological models are based only on short-term observations. The fossil record provides the only empirical evidence for the long-term response of organisms under perturbed climate states. The palaeontological record from the Last Interglacial (LIG; 125 000 years ago), a time of global warming, suggests that reef corals experienced poleward range shifts and an equatorial decline relative to their modern distribution. However, this record is spatio-temporally biased, and existing methods cannot account for data absence. Here, we use ecological niche modelling to estimate reef corals' realized niche and LIG distribution, based on modern and fossil occurrences. We then make inferences about modelled habitability under two future climate change scenarios (RCP4.5 and RCP8.5). Reef coral ranges during the LIG were comparable to the present, with no prominent equatorial decrease in habitability. Reef corals are likely to experience poleward range expansion and large equatorial declines under RCP4.5 and RCP8.5. However, this range expansion is probably optimistic in the face of anthropogenic climate change. Incorporation of fossil data in niche models improves forecasts of biodiversity responses under global climatic change.

1. "Modern" baselines. What do you define as "modern"? Is it present data (i.e., 2018/2019), data from the preindustrial era, or some other time? This might seem like splitting hairs, but the climate (and thus environment) is changing so fast that coral reef distribution may not be representative of current settings, but rather the averages from 20 years ago, or 100 years ago. It should be explicitly stated in the methods what you mean by "modern" and better yet, would be to see if your ENM data change significantly (or are at least different) if you use 2018 data vs. 2000 data vs. pre-industrial data. These results would be interesting in their own rights and would be a good test for a) the sensitivity of the ENM to input data and b) the shifting baselines that reefs are currently subjected to. If this has already been tested sufficiently by other groups, it would also be ok to just cite those works.
2. Saturation State. Other reviewers have already pointed this issue out, but we still do not feel it has been properly addressed. The authors themselves state (correctly) that SST is the primary control on Z-coral reefs, but that light availability, saturation state (Ω), salinity and nutrient levels are also important (lines 144-147). Yes, they are "secondary factors" and thus not AS important as SST, but as the ENM shows, several of these factors are still important environmental variables (lines 138-140). While it is understood that saturation state is not recorded directly in the fossil record, most of the variables are generated by a climate model anyway. At the very least it would be good to incorporate modern saturation states to see if they influence the ENM. Feely et al., 2009 (https://doi.org/10.5670/oceanog.2009.95) have done a nice job of reconstructing modern, pre-industrial, and future saturation states that could be used (and there are probably better models that have come out in the last decade). If this test shows that Ω values do not change the ENM, then they can ignore it for the past… but if they change, then the authors should at least highlight that this is an unknown parameter that should be included in future models, especially since many GCMs are not including carbonate chemistry.
3. Ecological variables. Many of the ecological variables discussed are quite variable within a 1.25° x 1.25° grid cell (e.g., depth or salinity can change dramatically on a shelf edge reef vs. lagoonal reef). These factors can have significant effects on a reef ecosystem. At the very least, the authors should note that this local variability will not be captured by their ENM and would be significant. Better yet would be to do the ENM at a higher resolution for some significant regions (e.g., the coral triangle), but the model or computing power may be limited. Also, what about seasonality or temperature extremes? Some papers have shown that temperature extremes might be linked with adaptability (e.g., Palumbi et al., 2014; DOI: 10.1126/science.1251336).
• Line 15-17: The last interglacial is not the fossil record. We suggest rephrasing to: "The paleontological record from the Last Interglacial (LIG; 125,000 years ago), a time of global warming, suggests that reef corals experienced poleward range shifts and an equatorial decline relative to their modern distribution." • Line 70: missing punctuation in the brackets • Line 84: Please define what is meant by "ecotype-level". It is good to be explicit in the methods section.
• Lines 95-97: Suggest you include an explicit statement saying that the zooxanthellate characteristics are assumed to be the same as the modern for the fossil coral taxa (obvious yes, but an assumption nonetheless). For example: Kiessling  • Lines 315-316: It would also be good to mention that stressed coral reef ecosystems are a lot less likely to survive other stressors, such as pollution, overfishing etc.
• Line 338: Please define/explain Variable Contribution analysis (or provide a citation) • Lines 338-341: Bathymetric variability should also be discussed here, either in terms of the lack of ENM resolution and the tectonic changes that have occurred; i.e., LIG bathymetry is generally similar to modern bathymetry, but many places (e.g., Jamaca or Vanuatu) have experienced nontrivial uplift due to tectonics.
• Line 389: replace "(i.e. absence)" with "(i.e. absence of reported fossil collections/occurrences)" • Line 406: Suggest adding "anthropogenic" before "environmental impacts" demonstrate the value of utilising fossil reef coral data to improve biodiversity risk assessment through consideration of the realised niche under varied climate states." All of these goals address important questions. As far as I know, goal 2 and 3 are novel and this paper provides important results. Goal 1 was previously explored by Kiessling et al (2012) [which I was a coauthor].
All in all, I think that, although this paper tackles an important issue, it does not meet its potential because it is confusing. The focus on occurrences and ENM methods and not the questions posed is a big part of this confusion for me. Take for example, the LIG and modern diversity gradients. Although these are major parts of the the paper (goals 1 and 2), no where is diversity shown, only numbers of occurrences and numbers of habitable cells.
Major comments: I understand the motive for studying the relationships between coral distributions and climate. I don't think a convincing case for using niche models is made in the introduction. No doubt there are sampling issues with the past that cannot be fully ameliorated by sub-sampling. But framing the current paper as a methodological correction to Kiessling et al limits the current paper's appeal and impact. ENM are needed for a much more important reason than correcting Kiessling-a niche model is necessary for the forward prediction into future climate scenarios. I suggest adding in a paragraph into the introduction, at about line 73, to point out that there is no other way to make a prediction into the future than with a well trained model. Adding in fossil occurrences from the last interglacial, when the earth was warmer than today, aids in the modeling effort because modern occurrences alone sample a small range of possible climatic variation. More sampled variation in temperatures and occurrences will lead to a higher fidelity model.
Geographic ranges surely are species specific, therefore I can't see how lumping occurrences of all coral species together can say anything about the range shifts from LIG to the modern. However, those ENMs are the only option for predicting the coral occurrences in the future. Would species specific ENMs be too noisy? I find understanding Figure 2 to be confusing. There are many combinations of training dataset and modeled pattern that are hard to tease apart. For one, I'm surprised that the modern-trained modern habitability is so different from the observed modern occurrences. The peak habitability in the southern hemisphere is 5 or more degrees off of the peak occurrences in the raw data. This mismatch makes me wonder if the number of habitable cells is informative for diversity. What is the pattern of latitudinal diversity in this dataset anyway?

07-Mar-2019
Dear Mr Jones On behalf of the Editors, I am pleased to inform you that your Manuscript RSOS-182111 entitled "Coupling of palaeontological and neontological reef coral data improves forecasts of biodiversity responses under global climatic change" has been accepted for publication in Royal Society Open 6 Science subject to minor revision in accordance with the referee suggestions. Please find the referees' comments at the end of this email.
The reviewers and handling editors have recommended publication, but also suggest some minor revisions to your manuscript. Therefore, I invite you to respond to the comments and revise your manuscript.
• Ethics statement If your study uses humans or animals please include details of the ethical approval received, including the name of the committee that granted approval. For human studies please also detail whether informed consent was obtained. For field studies on animals please include details of all permissions, licences and/or approvals granted to carry out the fieldwork.
• Data accessibility It is a condition of publication that all supporting data are made available either as supplementary information or preferably in a suitable permanent repository. The data accessibility section should state where the article's supporting data can be accessed. This section should also include details, where possible of where to access other relevant research materials such as statistical tools, protocols, software etc can be accessed. If the data has been deposited in an external repository this section should list the database, accession number and link to the DOI for all data from the article that has been made publicly available. Data sets that have been deposited in an external repository and have a DOI should also be appropriately cited in the manuscript and included in the reference list.
If you wish to submit your supporting data or code to Dryad (http://datadryad.org/), or modify your current submission to dryad, please use the following link: http://datadryad.org/submit?journalID=RSOS&manu=RSOS-182111 • Competing interests Please declare any financial or non-financial competing interests, or state that you have no competing interests.
• Authors' contributions All submissions, other than those with a single author, must include an Authors' Contributions section which individually lists the specific contribution of each author. The list of Authors should meet all of the following criteria; 1) substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) drafting the article or revising it critically for important intellectual content; and 3) final approval of the version to be published.
All contributors who do not meet all of these criteria should be included in the acknowledgements.
We suggest the following format: AB carried out the molecular lab work, participated in data analysis, carried out sequence alignments, participated in the design of the study and drafted the manuscript; CD carried out the statistical analyses; EF collected field data; GH conceived of the study, designed the study, coordinated the study and helped draft the manuscript. All authors gave final approval for publication.
• Acknowledgements Please acknowledge anyone who contributed to the study but did not meet the authorship criteria.
• Funding statement Please list the source of funding for each author.
Please ensure you have prepared your revision in accordance with the guidance at https://royalsociety.org/journals/authors/author-guidelines/ --please note that we cannot publish your manuscript without the end statements. We have included a screenshot example of the end statements for reference. If you feel that a given heading is not relevant to your paper, please nevertheless include the heading and explicitly state that it is not relevant to your work.
Because the schedule for publication is very tight, it is a condition of publication that you submit the revised version of your manuscript before 16-Mar-2019. Please note that the revision deadline will expire at 00.00am on this date. If you do not think you will be able to meet this date please let me know immediately.
To revise your manuscript, log into https://mc.manuscriptcentral.com/rsos and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with Decisions". Under "Actions," click on "Create a Revision." You will be unable to make your revisions on the originally submitted version of the manuscript. Instead, revise your manuscript and upload a new version through your Author Centre.
When submitting your revised manuscript, you will be able to respond to the comments made by the referees and upload a file "Response to Referees" in "Section 6 -File Upload". You can use this to document any changes you make to the original manuscript. In order to expedite the processing of the revised manuscript, please be as specific as possible in your response to the referees. We strongly recommend uploading two versions of your revised manuscript: 1) Identifying all the changes that have been made (for instance, in coloured highlight, in bold text, or tracked changes); 2) A 'clean' version of the new manuscript that incorporates the changes made, but does not highlight them.
When uploading your revised files please make sure that you have: 1) A text file of the manuscript (tex, txt, rtf, docx or doc), references, tables (including captions) and figure captions. Do not upload a PDF as your "Main Document"; 2) A separate electronic file of each figure (EPS or print-quality PDF preferred (either format should be produced directly from original creation package), or original software format); 3) Included a 100 word media summary of your paper when requested at submission. Please ensure you have entered correct contact details (email, institution and telephone) in your user account; 4) Included the raw data to support the claims made in your paper. You can either include your data as electronic supplementary material or upload to a repository and include the relevant doi within your manuscript. Make sure it is clear in your data accessibility statement how the data can be accessed; 5) All supplementary materials accompanying an accepted article will be treated as in their final form. Note that the Royal Society will neither edit nor typeset supplementary material and it will be hosted as provided. Please ensure that the supplementary material includes the paper details where possible (authors, article title, journal name).
Supplementary files will be published alongside the paper on the journal website and posted on the online figshare repository (https://rs.figshare.com/). The heading and legend provided for each supplementary file during the submission process will be used to create the figshare page, 8 so please ensure these are accurate and informative so that your files can be found in searches. Files on figshare will be made available approximately one week before the accompanying article so that the supplementary material can be attributed a unique DOI.
Please note that Royal Society Open Science charge article processing charges for all new submissions that are accepted for publication. Charges will also apply to papers transferred to Royal Society Open Science from other Royal Society Publishing journals, as well as papers submitted as part of our collaboration with the Royal Society of Chemistry (http://rsos.royalsocietypublishing.org/chemistry).
If your manuscript is newly submitted and subsequently accepted for publication, you will be asked to pay the article processing charge, unless you request a waiver and this is approved by Royal Society Publishing. You can find out more about the charges at http://rsos.royalsocietypublishing.org/page/charges. Should you have any queries, please contact openscience@royalsociety.org.
Once again, thank you for submitting your manuscript to Royal Society Open Science and I look forward to receiving your revision. If you have any questions at all, please do not hesitate to get in touch. The manuscript has now been thoroughly reviewed and the reviewers have made a number of important points that, if addressed, will significantly improve the impact of the eventual published version. I agree with all the comments and suggestions made by the reviewers and encourage the authors to consider them carefully.
Reviewer comments to Author: Reviewer: 1 Comments to the Author(s) This paper is a nice analysis of zooxanthellate coral (Z-coral) biogeography using Ecological Niche Modelling (ENM). We like the author's use of the ENM to assess the completeness of the fossil record as well as their future predictions. It is clear that this manuscript has already been through several rounds of revision and so should be accepted once a few additional things are addressed.
Moderate/Major Issues 1. "Modern" baselines. What do you define as "modern"? Is it present data (i.e., 2018/2019), data from the preindustrial era, or some other time? This might seem like splitting hairs, but the climate (and thus environment) is changing so fast that coral reef distribution may not be representative of current settings, but rather the averages from 20 years ago, or 100 years ago. It should be explicitly stated in the methods what you mean by "modern" and better yet, would be to see if your ENM data change significantly (or are at least different) if you use 2018 data vs. 2000 data vs. pre-industrial data. These results would be interesting in their own rights and would be a good test for a) the sensitivity of the ENM to input data and b) the shifting baselines that reefs are currently subjected to. If this has already been tested sufficiently by other groups, it would also be ok to just cite those works.
2. Saturation State. Other reviewers have already pointed this issue out, but we still do not feel it has been properly addressed. The authors themselves state (correctly) that SST is the primary control on Z-coral reefs, but that light availability, saturation state (Ω), salinity and nutrient levels are also important (lines 144-147). Yes, they are "secondary factors" and thus not AS important as SST, but as the ENM shows, several of these factors are still important environmental variables (lines 138-140). While it is understood that saturation state is not recorded directly in the fossil record, most of the variables are generated by a climate model anyway. At the very least it would be good to incorporate modern saturation states to see if they influence the ENM. Feely et al., 2009 (https://doi.org/10.5670/oceanog.2009.95) have done a nice job of reconstructing modern, pre-industrial, and future saturation states that could be used (and there are probably better models that have come out in the last decade). If this test shows that Ω values do not change the ENM, then they can ignore it for the past… but if they change, then the authors should at least highlight that this is an unknown parameter that should be included in future models, especially since many GCMs are not including carbonate chemistry.
3. Ecological variables. Many of the ecological variables discussed are quite variable within a 1.25° x 1.25° grid cell (e.g., depth or salinity can change dramatically on a shelf edge reef vs. lagoonal reef). These factors can have significant effects on a reef ecosystem. At the very least, the authors should note that this local variability will not be captured by their ENM and would be significant. Better yet would be to do the ENM at a higher resolution for some significant regions (e.g., the coral triangle), but the model or computing power may be limited. Also, what about seasonality or temperature extremes? Some papers have shown that temperature extremes might be linked with adaptability (e.g., Palumbi et al., 2014; DOI: 10.1126/science.1251336).
Minor issues: • There were too many acronyms (especially similar acronyms) and this often made the text difficult to understand. Please remove unnecessary acronyms (e.g., LBG, LHG), leaving only the common, or highly used ones (e.g., SST, DEM, LIG). A good general rule is to limit acronyms to ~3 -5 in a paper/grant). • Throughout the text there are places where comparative adjectives are used (e.g., "larger" in line 245, "more favorable" in line 261, etc.) but there is only one thing in the sentence. Please go through the text and make sure that if you are using a comparative adjective to compare 2 things, it is clear what both of those things are (i.e., larger than what?).
• Line 15-17: The last interglacial is not the fossil record. We suggest rephrasing to: "The paleontological record from the Last Interglacial (LIG; 125,000 years ago), a time of global warming, suggests that reef corals experienced poleward range shifts and an equatorial decline relative to their modern distribution." • Line 70: missing punctuation in the brackets • Line 84: Please define what is meant by "ecotype-level". It is good to be explicit in the methods section.
• Lines 95-97: Suggest you include an explicit statement saying that the zooxanthellate characteristics are assumed to be the same as the modern for the fossil coral taxa (obvious yes, but an assumption nonetheless). For example: Kiessling  • Lines 315-316: It would also be good to mention that stressed coral reef ecosystems are a lot less likely to survive other stressors, such as pollution, overfishing etc.
• Line 338: Please define/explain Variable Contribution analysis (or provide a citation) • Lines 338-341: Bathymetric variability should also be discussed here, either in terms of the lack of ENM resolution and the tectonic changes that have occurred; i.e., LIG bathymetry is generally similar to modern bathymetry, but many places (e.g., Jamaca or Vanuatu) have experienced nontrivial uplift due to tectonics.
• Line 389: replace "(i.e. absence)" with "(i.e. absence of reported fossil collections/occurrences)" • Line 406: Suggest adding "anthropogenic" before "environmental impacts" • Fig. 3: These plots are definitely difficult to read, the blue should be lightened… it is almost impossible to distinguish blue cells from black cells unless you're in perfect lighting conditions. It's great that you have checked for colorblindness, but it just looks like some yellow and grey dots in black squares to someone of normal vision. Since this is the second time a reviewer has highlighted issue, you should change the colors.

Reviewer: 2
Comments to the Author(s) In this paper, the authors use modern and paleontological (Last interglacial [LIG]) occurrence data to niche model the distributions of reef corals. Using these niche models, they then predict the response to two future climate scenarios. The goal of this paper is to predict how corals will respond to future climate change. A model of coral distributions is needed in order to make this prediction. And the authors use fossil and modern distributions to train their model. The paper has three stated goals: "Here, we use ENM to: (1) evaluate the extent to which the LIG equatorial decline of reef corals is the result of fossil bias or a genuine loss in habitability; (2) quantify the variation in geographic range of reef corals during the LIG and under future climate scenarios, with implications for understanding the evolution of the reef coral latitudinal biodiversity gradient; and (3) demonstrate the value of utilising fossil reef coral data to improve biodiversity risk assessment through consideration of the realised niche under varied climate states." All of these goals address important questions. As far as I know, goal 2 and 3 are novel and this paper provides important results. Goal 1 was previously explored by Kiessling et al (2012) [which I was a coauthor].
All in all, I think that, although this paper tackles an important issue, it does not meet its potential because it is confusing. The focus on occurrences and ENM methods and not the questions posed is a big part of this confusion for me. Take for example, the LIG and modern diversity gradients. Although these are major parts of the the paper (goals 1 and 2), no where is diversity shown, only numbers of occurrences and numbers of habitable cells.
Major comments: I understand the motive for studying the relationships between coral distributions and climate. I don't think a convincing case for using niche models is made in the introduction. No doubt there are sampling issues with the past that cannot be fully ameliorated by sub-sampling. But framing the current paper as a methodological correction to Kiessling et al limits the current paper's appeal and impact. ENM are needed for a much more important reason than correcting Kiessling-a niche model is necessary for the forward prediction into future climate scenarios.
I suggest adding in a paragraph into the introduction, at about line 73, to point out that there is no other way to make a prediction into the future than with a well trained model. Adding in fossil occurrences from the last interglacial, when the earth was warmer than today, aids in the modeling effort because modern occurrences alone sample a small range of possible climatic variation. More sampled variation in temperatures and occurrences will lead to a higher fidelity model.
Geographic ranges surely are species specific, therefore I can't see how lumping occurrences of all coral species together can say anything about the range shifts from LIG to the modern. However, those ENMs are the only option for predicting the coral occurrences in the future. Would species specific ENMs be too noisy? I find understanding Figure 2 to be confusing. There are many combinations of training dataset and modeled pattern that are hard to tease apart. For one, I'm surprised that the modern-trained modern habitability is so different from the observed modern occurrences. The peak habitability in the southern hemisphere is 5 or more degrees off of the peak occurrences in the raw data. This mismatch makes me wonder if the number of habitable cells is informative for diversity. What is the pattern of latitudinal diversity in this dataset anyway? You can expect to receive a proof of your article in the near future. Please contact the editorial office (openscience_proofs@royalsociety.org and openscience@royalsociety.org) to let us know if you are likely to be away from e-mail contact. Due to rapid publication and an extremely tight schedule, if comments are not received, your paper may experience a delay in publication.
Royal Society Open Science operates under a continuous publication model (http://bit.ly/cpFAQ). Your article will be published straight into the next open issue and this will be the final version of the paper. As such, it can be cited immediately by other researchers. As the issue version of your paper will be the only version to be published I would advise you to check your proofs thoroughly as changes cannot be made once the paper is published. Below, we respond to each individual point raised by the associate editor and reviewers. Our modifications to the original manuscript are attached in a tracked format, and a "clean" version is also submitted. Comments from the referees herein are in italics, while our responses are in bold.
The main results and conclusions from our original submission are unchanged following this revision. We hope that these changes and responses will satisfy the associate editor and reviewers, and that our MS can now be accepted for publication. First of all, we would like to thank the associate editor for their involvement with this manuscript. We hope that our responses below satisfy their concerns, as well as those of the reviewers.
Reviewer comments to Author: Reviewer: 1 To begin with, we would like to thank reviewer 1 for their time, constructive criticism and clear efforts to improve this manuscript. We have taken their comments wholeheartedly on board and have made several amendments to our work based on their insight. Please see details below. These results would be interesting in their own rights and would be a good test for a) the sensitivity of the ENM to input data and b) the shifting baselines that reefs are currently subjected to. If this has already been tested sufficiently by other groups, it would also be ok to just cite those works.
We have adjusted our text accordingly to specify that the modern data is of pre-industrial climate conditions. Whilst we agree with the reviewers that it would be interesting to essentially use a timecalibrated ENM, this would be a study or perhaps even two in its own right. It would require filtering all occurrences to specific collection times and relating it to climatological means for that date. This would be a substantial undertaking and is not within the scope of this particular study. What we have done here is use pre-industrial climate conditions as it represents the long-term climate which resulted into the biogeographic pattern we observe today. To address this issue we have now generated a modern-trained ENM with aragonite saturation state (Ω) included. As a sensitivity test we generated difference maps of the geographic projections for the modern from ENMs with and without Ω included. We also ran variable contribution tests to see the importance of Ω in our ENMs. We found that at a global scale, predictions are essentially the same, however, there are some differences at a few isolated coastlines. In terms of variable contribution, our analyses confirms that Ω is indeed of secondary significance, with similar contributions to the ENM as irradiance variables. As the reviewer notes and our results confirm, although Ω may not be as important as bathymetry and sea surface temperature, it is not insignificant. Throughout the MS we have now tailored the text so it is in line with these results.

3.
Ecological We raise this issue in Supplementary Material 2 (first paragraph), particularly in regards to salinity. However, we have added an additional couple of lines expanding on this. Whilst we agree with the reviewer that it may have significant impacts at localised scales, the spatial scale needed to capture such impacts would be on the scale of metres, opposed to kilometres. Unfortunately, this level of resolution for environmental data at a global scale is simply not available at this time. There are several works using remotely operated vehicles to capture data at this resolution (particularly on cold-water corals), but as far as we are aware, these usually focus on localised transects. We hope to see higher-resolution data from general circulation models in the future which can accurately account for climate at higher resolution. However, it should also be considered that there may be a higher level of uncertainty associated with future or past climate predictions at such resolutions, and available computing power is restricted. In regards to the last point concerning acclimatization, it is correct that the historical climate experience of some reef coral populations makes them more adaptable to temperature extremes (even within the same species). However, our models treat reef corals as a single entity (ecotype), and whilst we could introduce seasonal variables into our models, it might have significant impacts on any model projections to other time-slices due to overfitting and multicollinearity. A mechanistic ENM approach would be much more beneficial for the type of analyses reviewer 1 is suggesting, rather than a correlative one.

Minor issues:
• There were too many acronyms (especially similar acronyms) and this often made the text difficult to understand. Please remove unnecessary acronyms (e.g., LBG, LHG), leaving only the common, or highly used ones (e.g., SST, DEM, LIG). A good general rule is to limit acronyms to ~3 -5 in a paper/grant).
We have taken the reviewers concern on board and removed both the acronyms LBG and LHG. Amended, thank you.

•
Line 15-17: The last interglacial is not the fossil record. We suggest rephrasing to: "The paleontological record from the Last Interglacial (LIG; 125,000 years ago), a time of global warming, suggests that reef corals experienced poleward range shifts and an equatorial decline relative to their modern distribution." Amended, thank you for the suggestion.

• Line 70: missing punctuation in the brackets
Amended, thank you.

• Line 84: Please define what is meant by "ecotype-level". It is good to be explicit in the methods section.
Added.
• Lines 95-97: Suggest you include an explicit statement saying that the zooxanthellate characteristics are assumed to be the same as the modern for the fossil coral taxa (obvious yes, but an assumption nonetheless). For example: Kiessling  Whilst we agree, we are specifically referring to those variables that can be determined for the fossil record from currently available GCM outputs (second half of the sentence).

• Line 186: semicolon not needed
Removed, thank you.

• Lines 315-316: It would also be good to mention that stressed coral reef ecosystems are a lot less likely to survive other stressors, such as pollution, overfishing etc.
Thank you for raising this important point. We have added: "Nevertheless, it should also be borne in mind that reef corals, and coral reef ecosystems, are far less likely to survive other stressors, such as overfishing and pollution, when residing in marginal habitats. "

• Line 338: Please define/explain Variable Contribution analysis (or provide a citation)
Added computed within biomod2 and citation for reference.
• Lines 338-341: Bathymetric variability should also be discussed here, either in terms of the lack of ENM resolution and the tectonic changes that have occurred; i.e., LIG bathymetry is generally similar to modern bathymetry, but many places (e.g., Jamaca or Vanuatu) have experienced non-trivial uplift due to tectonics.

Added two sentences within the paragraph on this matter: "Whilst LIG bathymetry is generally similar to that of the modern at a global scale, several areas have experienced localised tectonic uplift (e.g. Huon Peninsula, Barbados, Sumba and Vanuatu (McCulloch and Esat, 2000)). However, this is not likely to impact our results, based on our sea-level sensitivity test (see supplementary material 2)."
• Line 389: replace "(i.e. absence)" with "(i.e. absence of reported fossil collections/occurrences)"

Figure Edits
*Please make the below corrections in the supplemental material as well.

• Figs. 1 & 2: Define the dashed lines and greyed area in the plots (presumably it is the tropics/equator but the reader should not need to guess).
Implemented.

• Fig. 2: Please define what the binary thresholds in the caption mean as well as the different scenarios (i.e., LIG, RCP4.5). Figures should be able to stand alone.
Implemented.

Reviewer: 2
Comments to the Author(s) In this paper, the authors use modern and paleontological (Last interglacial [LIG]) occurrence data to niche model the distributions of reef corals. Using these niche models, they then predict the response to two future climate scenarios. The goal of this paper is to predict how corals will respond to future climate change. A model of coral distributions is needed in order to make this prediction. And the authors use fossil and modern distributions to train their model.

To begin with, we would like to thank reviewer 2 for their time, constructive criticism and clear efforts to improve this manuscript. We have taken their comments wholeheartedly on board and have made several amendments to our work based on their insight. Please see details below.
The paper has three stated goals: "Here, we use ENM to: (1) evaluate the extent to which the LIG equatorial decline of reef corals is the result of fossil bias or a genuine loss in habitability; (2) quantify the variation in geographic range of reef corals during the LIG and under future climate scenarios, with implications for understanding the evolution of the reef coral latitudinal biodiversity gradient; and (3) demonstrate the value of utilising fossil reef coral data to improve biodiversity risk assessment through consideration of the realised niche under varied climate states."

All of these goals address important questions. As far as I know, goal 2 and 3 are novel and this paper provides important results. Goal 1 was previously explored by Kiessling et al (2012) [which I was a coauthor].
All in all, I think that, although this paper tackles an important issue, it does not meet its potential because it is confusing. The focus on occurrences and ENM methods and not the questions posed is a big part of this confusion for me. Take for example, the LIG and modern diversity gradients. Although these are major parts of the the paper (goals 1 and 2), no where is diversity shown, only numbers of occurrences and numbers of habitable cells.
We actually show sample in bin diversity within Figure 1. The first aim refers to using ecological niche modelling to predict where we should or should not have fossil occurrences based on predicted habitability (not so much diversity). In regards to the second aim, we are simply using ecological niche modelling to show similarities between modern and LIG latitudinal habitability, and what this could suggest for the latitudinal biodiversity gradient during the LIG, and under future climatic conditions. Ecological niche modelling is not a tool for predicting biodiversity, but the distribution of habitable localities. Rationally, we can make some inferences on the impact of change in habitability for biodiversity, but not quantify how many of species x or y we should have.

Major comments: I understand the motive for studying the relationships between coral distributions and climate. I don't think a convincing case for using niche models is made in the introduction. No doubt there are sampling issues with the past that cannot be fully ameliorated by sub-sampling. But framing the current paper as a methodological correction to Kiessling et al limits the current paper's appeal and impact. ENM are needed for a much more important reason than correcting Kiessling-a niche model is necessary for the forward prediction into future climate scenarios.
We don't feel this is a fair comment. This paper does not set out to correct Kiessling et al. (2012), but compliment it. For example, we concurthat there was a higher diversity at higher latitudes during the LIG. However, the combination of predicted habitability at equatorial latitudes during the LIG and the known sampling bias within tropical latitudes, suggests the equatorial decline might be artificial. We have added a couple of sentences to the introduction (relating to the next point) to bolster the need for niche models as highlighted in this comment.
I suggest adding in a paragraph into the introduction, at about line 73, to point out that there is no other way to make a prediction into the future than with a well trained model. Adding in fossil occurrences from the last interglacial, when the earth was warmer than today, aids in the modeling effort because modern occurrences alone sample a small range of possible climatic variation. More sampled variation in temperatures and occurrences will lead to a higher fidelity model.

We have added "Future projections trained solely on modern-occurrences may underestimate habitability due a restricted range in possible climatic variation. Through incorporation of the fossil record this variation is increased, and might improve future projections by reducing the amount of novel climate conditions faced from a modern-only trained model."
Geographic ranges surely are species specific, therefore I can't see how lumping occurrences of all coral species together can say anything about the range shifts from LIG to the modern. However, those ENMs are the only option for predicting the coral occurrences in the future. Would species specific ENMs be too noisy?
Our study is not species specific, but at ecotype-level as we specify in the methods. Running this type of global analysis at species-level would require higher resolution environmental data and fairer sampling between different species to achieve meaningful predictions based on a correlative approach. It is also worth mentioning that within different populations of the same species you may have different environmental tolerances due to historical experience (e.g. thermal regimes; Howells et al. (2013)), which ought to be considered if focusing efforts at species-level, at which point mechanistic models may prove more fruitful. In this paper we are not looking at specific geographic ranges sizes of species but the range size of suitable habitats. Figure 2 to be confusing. There are many combinations of training dataset and modeled pattern that are hard to tease apart. For one, I'm surprised that the modern-trained modern habitability is so different from the observed modern occurrences. The peak habitability in the southern hemisphere is 5 or more degrees off of the peak occurrences in the raw data. This mismatch makes me wonder if the number of habitable cells is informative for diversity. What is the pattern of latitudinal diversity in this dataset anyway?

Within this figure there are only two training datasets (a) modern-trained and (b) LIG-trained. The separate curves on each plot show different threshold selections. We have amended figure captions in the hope to make this clearer. These figures do not represent diversity in any way, and we do not refer to them in such a manner. We are simply looking at how latitudinal habitability changes under different climate scenarios, and what implications this may have for diversity. For example, if habitability is greater during the LIG at equatorial latitudes than today, it doesn't make sense that we would have an equatorial decline, at least due to abiotic reasons. In regards to the mismatch between habitability and occurrences, one should not expect a perfect match. To provide a few examples of why:
-You may find more occurrences in site A opposed to site B due to sampling -Site A might have a higher level of connectivity than site B and receive more coral larvae -Site B might be more environmentally suitable but is impacted by biotic and anthropogenic interactions -Distribution is impacted by historical reasons such as limits to dispersal We thank the reviewers for the kind and constructive comments they have provided to strengthen this manuscript. We feel that it has improved the quality of the manuscript considerably and look forward to any further comments. estimates ranging from no significant difference, to as much as 2°C warmer than today [26-28]. 67 Kiessling et al. [13] proposed that this temperature rise was the driver of the dramatic range shifts 68 observed in LIG reef corals. Consequently, these authors suggested that anthropogenic global 69 warming today means that an equatorial decline of reef corals is likely to follow their current 70 poleward expansion, which has important implications for conservation, such as which geographic 71 regions should be prioritised. However, the fossil record is inherently biased both spatially and 72 temporally, impacted by incomplete sampling, variable fossil preservation, erosion, and burial. 73 Although existing subsampling methods, such as those applied by Kiessling et al. [13], can 74 ameliorate problems associated with uneven 'raw' occurrence data (e.g. ref. [29]), they cannot 75 account for the absence of data. ENM offers the prospect of mitigating for biases caused by the 76 uneven distribution of known occurrences (i.e. raw data) by measuring habitat availability, 77 providing insights into the potential geographic distribution of organisms through consideration of 78 their observed realised niche. Pleistocene Scleractinia. These datasets were then filtered to exclude data not identified to species 99 level, as well as further quality control checks (e.g. synonymisations). We based our analyses on 100 zooxanthellate corals, herein referred to as reef corals, which host unicellular organisms known as 101 zooxanthellae in their tissue, forming a photosymbiotic relationship [35]. Assignment to 102 zooxanthellate class focused on the principle of uniformitarianism for fossil coral taxa. 103 Azooxanthellate (those without a photosymbiotic relationship) and apozooxanthellate (those 104 which may or may not sustain a photosymbiotic relationship) corals were excluded from both 105 databases. Further quality control was performed on the OBIS dataset, whereby occurrences found 106 to occur on land or at depths greater than 100 m were removed under the assumption that there 107 were errors in their geographic coordinates, species identification, or are in fact fossil occurrences. 108 In addition, we omitted data collected from rapid ecological assessments to improve the credibility 109 of species level identification, as well as those from coral transplant studies. The PBDB data were 110 filtered to exclude occurrences which cannot be reliably assigned to the LIG. Duplicate occurrence 111 records of the same species from exactly the same coordinates were also removed from both 112 datasets.  34,35], and that can be viably 142 determined for the fossil record from GCM outputs, were chosen to reflect abiotic niche 143 characteristics (SM2 : Table SM2). Initially, these variables were considered at several temporal 144 levels: annual, seasonal, and monthly. However, to reduce co-linearity between variables and 145 prevent over-fitting, we retained the combinations of climate variables with a Pearson's pairwise 146 correlation coefficient of less than 0.7 (SM2: Figure SM2 As previously noted, short-term observations of distribution dynamics can be inherently biased by 162 a taxon not occupying its full geographic or environmental space. However, an additional 163 fundamental question is whether environmental niches are conserved over evolutionary time scales 164 [45]. To evaluate this, we utilised the R-package ECOSPAT [46] to quantify measures of niche 165 margin dynamics (unfilling, expansion, and stability). We also performed niche overlap analyses, 166 testing for niche equivalency and similarity of the reef coral niches occupied across two temporal 167 windows (LIG and modern). The ECOSPAT niche overlap analysis applies kernel smoothers to 168 the densities of species occurrences in gridded environmental space, using a randomisation test 169 framework to compare observed equivalency and similarity to that anticipated under a null 170 hypothesis [46]. Both niche equivalency and similarity are quantified in terms of Schoener's D 171 [47], with values ranging from 0 (least equivalent or similar) to 1 (most equivalent or similar). We 172 compared the subsampled occupied niches of LIG and modern reef corals using a principal 173 component analysis (PCA-env), testing the null hypothesis by randomly sampling 5000 replicates 174 of both LIG and modern niches. The PCA-env was calibrated with all five environmental variables 175 implemented in the construction of our ENMs, and allowed us to assess differences in the occupied 176 niche of reef corals from the LIG and modern (see ref. [48]  In order to measure distribution dynamics, we first generated 24 binary maps from the ensemble 196 models using two different training datasets (LIG reef corals and modern reef corals) and three 197 different binary thresholds for the four climate scenarios (LIG, modern, RCP4.5 and RCP8.5). We 198 selected three thresholds for binary conversion: the 5 th -, 10 th -, and 20 th -percentiles of the lowest 199 tail of habitat suitability values (5LPT, 10LPT, 20LPT) associated with the occurrence data for 200 each model to provide insight into habitability dynamics (see SM2 for further details). As short-201 term observations are intrinsically biased and perhaps incomplete in both environmental and 202 geographic space, we combined binary predictions generated from both LIG and modern training 203 datasets. This was done for each time-slice/climate scenario to establish a multi-temporal 204 approach, estimating the full geographic localities reef corals may occupy. We then assessed 205 whether this approach improved the models' capability to predict the occurrence of both LIG and 206 modern localities, which would have implications for forecasting under future climate scenarios. 207 The latitudinal range and habitability of reef corals for past, present, and future climatic scenarios 208 was computed as a function of the sum of suitable cells, within 5° latitudinal bins, for each model 209 class' projections (LIG, modern, and combined). 0.9), good (0.7-0.8), fair (0.6-0.7), and poor (0.5-0.6), whereas a score of < 0.5 is worse than one 216 would expect from a random model [58]. Evaluation scores for TSS range from -1 to 1, where a 217 8 score of 1 demonstrates perfect model performance, whilst < 0 is considered worse than a random 218 model [57]. As we used two ENM constructs (LIG-and modern-trained), and generated respective 219 forecasting and hindcasting (backtesting) projections, we calculated the AUC scores using LIG 220 occurrences and modern occurrences as an evaluation dataset for model projections to other 221 climate scenarios (i.e. modern-trained and LIG projection; LIG-trained and modern projection). 222

Geographic Predictions 254
The results from binary habitability maps (Figure 2; raw, binary, and clamping mask maps 255 available in SM2: Figure SM15 SM3). More specifically, results from the combined projections indicate that LIG equatorial 283 regions (-5 to 5°) were more habitable than during the modern, with 34 (5LPT), 19 (10LPT), and 284 40 (20LPT) additional habitable cells predicted. In addition, habitability is generally greater at 285 higher latitudes in both hemispheres for the LIG than in the modern, for all binary thresholds, 286 suggesting that reef coral range size was likely larger during this time. To provide example, for 287 5LPT, only 3 cells were predicted for latitudinal bin 40-45° in the modern northern hemisphere, 288 yet 14 cells were predicted for the LIG (Figure 2; SM3). 289 290 Under RCP4.5 and RCP8.5, the models predict a considerable decrease in habitability (for all 291 binary thresholds) within the tropics with a 5% (5LPT), 20% (10LPT), and 59% (20LPT) loss in 292 habitable cells for RCP 8.5 relative to the modern (SM3), notably in the centre of the hyper-diverse 293 Coral Triangle (Indo-Pacific Ocean) (Figure 3). In contrast, habitability increases significantly at 294 higher latitudes (25-45°) in the northern hemisphere (Figure 3), with a 36% (5LPT), 40% (10LPT), 295 and 43% (20LPT) increase in habitable cells from the modern (SM3). South of the equator, there 296 is a similar gain at higher latitudes (25° to 45°), with a 30% (5LPT), 40% (10LPT), and 46% 297 (20LPT) increase in habitable cells (SM3). However, the clamping masks, particularly for RCP8.5, 298 highlight some uncertainty for equatorial tropical latitudes due to one environmental variable being 299 outside the range used for model calibration (SM2: Figure SM21-22). This finding highlights the 300 need to incorporate fossil data from various climatic regimes to reduce uncertainty when projecting 301 to future climate scenarios, which represent climate conditions outside modern experience [59]. 302

303
The LIG has occasionally been considered as an analogue for the impacts of global warming on 304 species' distribution [13, 60,61]. At face value, the LIG fossil record of reef corals depicts a global 305 biogeographic pattern that is considerably dissimilar to that observed in the present day (Figure 1). 306 Whereas today, both species abundance and diversity are highest within the tropics, with an 307 equatorial dip, the LIG record shows a poleward shift in occurrence records and species diversity, 308 with both increasing at higher latitudes in the northern hemisphere, and a prominent equatorial 309 decline [24,25]. Support for this observed pattern is also provided when subsampling methods are 310 applied to ameliorate biases pertaining to uneven latitudinal sampling [13]. However, using ENM, 311 our findings provide little support for an equatorial decline in habitability for reef corals during 312