The authors attempt in this manuscript an approach to the assembly processes of isolated communities, emphasizing the influence of their phylogenetic history, which they hypothesize has a primary role along with species functional traits.
Using a multivariate hierarchical generalized linear mixed model (implemented following the HMSC framework) they analyzed the abundances of 59 reef fish species from four oceanic islands during a 7 year long time series.
The various results of the application of this model are interpreted to suggest that community assembly of reef fishes in the South-western Atlantic islands has been strongly determined by phylogeographical processes over the environmental filtering.
The discussion draws together these various results and argues for the prevalence of the species capacity to reach and establish in isolated sites over environmental stochasticity. The different strains of the analysis results are hypothesized as expressions of the process of assemblage of four isolated communities, and they interpret the environmental covariates influence on species abundances, which also provides some insight on the importance of species niche on the community assemblage in these islands.
I think this manuscript contains a very interesting research which addresses a relevant topic. I have indeed enjoyed reading the manuscript and learning about the many complexities of fitting a hierarchical, multivariate, GLMM, with Bayesian inference to communities’ abundance data. I think the manuscript is a good fit for this journal and only have some minor comments that I would recommend were checked before its publication.

Line 7: I would rephrase the sentence: “This poses a challenge because community assembly is embedded in a spatio-temporal context …” What poses a challenge? Maybe something like: “Getting a proper insight of this is challenging…” or “Understanding its exact role poses a challenge…”
Line 18: “A proxy” TO “primary productivity”. Also, along the manuscript, the authors referred to Chlorophyll A sometimes as “chlorophyll-a” and sometimes as “chlorophyll a”. Consider revising.
Line 51: I might be mistaken but I would say this citation comes from MacNeil and Connolly, 2015
Line 51: “Annual changes in abundance are, on average, more variable in more abundant species, following Taylor’s power law” Not sure how this sentence fits this paragraph
Line 63: Insert “previous” to refer to this study.
Line 71: I find this to be a weird phrasing. I would correct “with some being efficient rafters” to “some of them being efficient rafters”
Line 71: I’d rephrase this to “and the accumulation of species with low dispersal ability end up contributing to speciation” instead of “with”
Line 95: Focuses.
Line 99: I suggest replacing “where diversification processes were” by “are characterized by diversification processes that were”
Line 129: Are all these criteria met? Or only one of them? I find the "or" confusing.
Line 132: “By doing so”. I would specify what you mean by “so”. “By filtering the species pool” or something like that.
Line 161: Is this the first time you use MHW for marine heat waves? If so, expand the acronym.
Line 187: “available at”
Line 268: I understand you have done four models for each year to account for delayed and inmediate effects of covariates, but I don’t really understand the sentence starting in line 269, do you mean covariates DO coincide with the sampling periods and islands, except for AR’s, for which you used data of previous year’s spring?
Line 272-274: “This is because as we considered surveys starting in 2013, the species identified on AR for this year were probably still responding to the spring season of the previous year”: Consider reviewing this sentence. “Since surveys started in 2013, we considered the species identified on AR this year were probably still responding to the previous year’s spring season” ??
Line 277: “actually the squared Pearson correlation coefficient between fish abundance and model prediction” I’m sorry I don’t get this sentence, I feel the “actually” is what’s confusing.
Line 469: OF ”these oceanic islands”
Line 499: “We therefore hypothesize that the response of reef fish communities to the marine environment in the southwestern Atlantic islands is mainly the result of basin-scale vicariant (Tethys vicariance and closure of the Panama Isthmus) and dispersal filters that shaped species distribution” I do not quite get how you conclude this from the two previous sentences in the discussion, I copy them: “Species adaptation to the production of a large number of highly dispersive larvae could also enhance within-site species coexistence, as suggested by the observed low species association (see Fig. 5). Eventually, high colonization efficiency and chance extinction in a benign environment would play an important role on the short-term variation in reef fish abundances”. Might the first sentence “Species adaptation to the production…” be better phrased as “Species production of a large number…”?
Line 519: I find this sentence in which the authors referred to the alignment with a previous study a bit poorly built.
Line 556: “resulting from” A “regime shift…”
Lines 618-639: I appreciate this particular paragraph in which authors have commented on some ins and outs of the HMSC model, I find it helps understand the limits of the application they have made of this statistical framework.
Minor inquiry on the methods used in this research: I hope I didn’t miss it if this was already explained in the text, but I missed some information on a possible model selection. When designing the model specified in the Material and Methods section, did you compare different model designs? Or is the model you present the only version you have explored? In other applications of such models I have seen it is a complex step, but at some point might be very informative to explore different model definitions, such as with different selections of traits or environmental covariates or with/without the phylogeny dimension. Since you have been able to reach satisfactory results with apparent robustness, I understand your model’s design was probably best.

I appreciate the author's effort to adapt their manuscrip to the revisions proposed. The corrections have mostly addressed my comments and I think the manuscript has been improved overall. I only have a minor comment for the authors to consider before the manuscript is published.
I feel the first paragraph of the discussion should be more explicetly focused on your findings and their relevance, maybe it is possible to merge your two first paragraphs into a shorter summary of the singularity of your work's main results.
Line 657 in the final document: ‘abundance’ is misspelled