Dear Dr. Yates,
You have provided a lengthy and convincing response to the reviewers‘ criticism. I am therefore inclined to accept your paper for publication in Biogeosciences.
However, some arguments you raise in your response letter do not show clearly in the revised manuscript and the text is still not to the point and somewhat disorganized. I must therefore insist on another revision emphasizing the points listed below. Importantly, stay closer to the data at hand and be less speculative.
To my understanding you have observed high a high-diversity and largely healthy coral assemblage at Hurricane Hole and measured some physico-chemical parameters and bleaching/non-bleaching in two coral species. Data and analyses are missing that would prove that your specific habitat may be classified as a refuge.
The arguments in your paper are interesting but not iron-clad. You can only provide evidence that shading by mangroves helps the corals, but the effects of saturations state and temperature variability is quite speculative and should be condensed.
Please consider carefully the following points:
1. Change of title. I agree with you that a question would be a bad title, but the reviewers concerns suffice to raise serious concerns about generalizing the role of mangroves as refuges. So a change of your title is necessary. Either you change the title to a more neutral term (e.g., Diverse coral communities in mangrove habitats suggest a novel refuge from global warming) or you are more specific (e.g., Mangrove habitats as coral refuges in St. John, U.S. Virgin Islands).
2. The buffering effect of mangroves from ocean acidification is still not convincing. In your rebuttal letter you agree that the shading effect may be prevalent but you keep discussing at length the acidification issue, for which the evidence is still meager. Overall you seem to be arguing that the stress in mangrove habitats (temperature variability, storms, runoff etc.) makes corals more adaptive to even more stress. This should be perhaps be laid out more clearly and be supported by suitable references.
3. You report many numbers and tests in the main text that could be summarized in tables. At the same time some of your tests are not to the point. Importantly, you report at length (l. 402-422) the different bleaching sensitivity of the two coral species, when you should rather provide a rigorous test on the effect of shading by mangroves. So either perform standard medical tests such as log-odds ratios (not possible individually for C. natans because there are zeros in one category) or a binomial test on the probability that the observed differences are due to chance. With your numbers in table 8 I get:
Unshaded shaded
bleached 121 16
unbleached 69 73

The odds ratio is thus 8.5 meaning unshaded corals are eight times more prone to bleaching than shaded corals. This is the sort of information that may help the reader to follow your argument that mangroves are refugia. If you have appropriate data, you could also compare bleaching between mangrove systems and open reefs (the latter could be from the literature).
4. Please omit all but one statement of “more research is needed”
Specific comments: Sections 3.2 and 3.3 are especially hard to read as they report much data in a somewhat disorganized fashion. Please summarize important data in a table or refer to figures and condense the text by at least 30%.
Section 4 also requires reorganization. Please start with a succinct summary of previously identified refuges and then put your findings in this context. Avoid repetitions.
Finally, I agree with referee 1 that the absence of species-abundance data is regrettable and your argument that the morphology hinders quantitative assessments is weak. There are methods (e.g. line-transect method) to deal with this. Please try to compare some of the semi-quantitative data from nearby reefs with those in your mangrove system. This is important to judge which corals are best suited to survive in the stressed mangrove habitats. Importantly absent from your mangrove systems are the two Acropora species which are (or were) important Caribbean reefs builders. You may want to hint to this observation more explicitly, because it means that some corals may survive in mangroves but not the ones that may save the reefs.
Minor issues: No numbering of references please. Update references to tables and figures. These are often wrong.
Consider omitting Fig. 5. Very hard to read and only visualizing Table 7.

Wolfgang Kiessling (editor)

Dear Dr. Yates,
Thank you for your in-depth response to my previous comments. You have addressed all my comments in a satisfactory way and I am pleased to accept your paper for publication in Biogeosciences.

I have nevertheless indicated technical corrections because
1. The heterogeneity figures prominently in the results (3 sections) but there is no mention of it in the introduction or methods. You may want to introduce the concept before reporting the results.
2. You may want to thank the reviewers for the insightful reviews
No further review is required and I trust that you accommodate these very minor changes.

Thank you for submitting your work to Biogeosciences.

Best regards

Wolfgang Kiessling