The aesthetic value of reef fishes is globally mismatched to their conservation priorities

Reef fishes are closely connected to many human populations, yet their contributions to society are mostly considered through their economic and ecological values. Cultural and intrinsic values of reef fishes to the public can be critical drivers of conservation investment and success, but remain challenging to quantify. Aesthetic value represents one of the most immediate and direct means by which human societies engage with biodiversity, and can be evaluated from species to ecosystems. Here, we provide the aesthetic value of 2,417 ray-finned reef fish species by combining intensive evaluation of photographs of fishes by humans with predicted values from machine learning. We identified important biases in species’ aesthetic value relating to evolutionary history, ecological traits, and International Union for Conservation of Nature (IUCN) threat status. The most beautiful fishes are tightly packed into small parts of both the phylogenetic tree and the ecological trait space. In contrast, the less attractive fishes are the most ecologically and evolutionary distinct species and those recognized as threatened. Our study highlights likely important mismatches between potential public support for conservation and the species most in need of this support. It also provides a pathway for scaling-up our understanding of what are both an important nonmaterial facet of biodiversity and a key component of nature’s contribution to people, which could help better anticipate consequences of species loss and assist in developing appropriate communication strategies.


COMMENT ON THE TITLE
The title advertises an evaluation of the "global mismatches" between aesthetics and conservation threat. In the valuation literature, the term mismatch is widely used, in particular to describe the mismatches between the supply and demand for an object of valuation. In this manuscript, the term is not used in the text, neither in the introduction nor in the discussion, and having it in the title may be misleading for the readership. Besides, you also use the term "match" to describe a pair of images being compared to one another, and the term "mismatch" may also be confusing. I would suggest to slightly change the title to avoid this potential double confusion. In addition, the mismatch between the aesthetic value and the conservation status is postulated to be a concern, but this may not be the case (see my comments on the discussion).

COMMENTS ON THE INTRODUCTION
L68: what do you mean by "means of NCP" ? consider rephrasing? L95-96: can you be a bit more specific here and explain what you mean by "visual feature"? L.101: the sentence "to accurately identify patterns and predict values on images" is confusing. What "value" is being predicted and to what is this value assigned? As this sentence is key to describing the advantages of CNN, I would recommend to be crystal clear on what CNN does that other approaches don't, and how it operates in more concrete terms.
L.108: "public members" feels odd (but note that I'm not a native speaker). I would suggest "13,000 respondents from the broad public". L.112-119: In this paragraph, you justify the use of coral reef fishes as a study object by demonstrating their importance first for material NCP (economic activity and industry), regulating NCP (functioning of coral reef), and then only for non-material NCP (aesthetic values). I believe it would make more sense to reverse the argument, as you nicely explain in the paragraphs above that assessments of aesthetic values and, more broadly, nonmaterial NCP are lagging behind material and regulating ones. I would therefore recommend to start with their importance in non-material terms with greater emphasis (i.e. what makes them the perfect object of study to investigate non-material NCP and aesthetic value in particular; diversity of morphologies, aesthetic values, opportunities for recreation, and probably more?), then the regulating (their contribution to the functioning of ecosystem) and then their contribution to material NCP. Besides, there is a slight contradiction between the introduction and the interpretation of the results, because reef fish species are presented as key components for economic activity (L.113-115) to justify using them as a study object, when "only" a fifth of them according to Fig.5b are used for the fishing industry. Rearranging the paragraph would restore the balance in the argumentation a bit.

COMMENTS ON THE RESULTS
Fi.1a panel (ii): the screen capture of the box capture the red underlining of Word/Powerpoint, which should be removed.
L.131-142 & L167-169: these sentences rather belong to material and methods, but I let the authors and/or the editor appreciate whether they are essential to the presentation of the results.
L.138: the term "judge" is rather unusual. I would recommend using "respondent" and stick to it throughout the text and in the supplement L.193: specific "predicted aesthetic values"; the values have not been described so far in the text, and I believe this is an important aspect to clarify. The valuation literature is full of examples of different valuation methods, and giving a few words on the Elo method would be critical, in my humble opinion.

COMMENTS ON THE DISCUSSION
The paper highlights that the aesthetic value of fish is inversely proportional to their threat status and to their ecological originality/distinctiveness/usefulness (e.g.L.293). Playing the advocates' devil, isn't that a good thing that the aesthetic, intangibles values associated non-material NCP are complementary to the material and regulating NCP provided by those fish? This complementarity is an exciting finding in my opinion, as a broad spectrum of fish species are valued for complementary reasons, which makes the case for paying attention to multiple aspects to preserve them: ecological, aesthetic, and economic. To frame my question differently: what would happen if the fish species that were appreciated for their aesthetics were also highly valued by the industry, and were the most original ones and essential to the functioning of ecosystems? Worse, if they were also threatened, there would be a greater risk of an anthropogenic allee effect, whereby rare + beautiful species would suffer from an even greater demand from the broad public.
L.443-446: how is this statement backed-up by the results, and what do you mean by "less well-represented in the assessments"? I would recommend greater clarity in this statement, as it is not straightforward that data deficient species are less attractive than species present in the IUCN assessments. Besides, isn't this contradictory to the earlier statements in this paragraph about charismatic and aesthetically attractive mammal species receiving greater conservation attention (L.441-442)? Here, you demonstrate that the most aesthetically pleasing species are not particularly threatened, and will therefore not necessarily need greater conservation attention.

COMMENTS ON THE METHODS
L. 566-567: given the fact that the evaluation of the 9 features in an image is based on saturation and lightness (L.665), how did you make sure that your photoshop correction did not bias the results in any way? How many pictures did you correct that way? I do acknowledge that some corrections may have been necessary, but given the later evaluation of the image characteristics, why not discarding the images that were deemed in need of a photoshop correction?
I could not find the information on the number of "matches" undertaken by each species in the dataset. I don't doubt that with over 13,000 respondents evaluating 30 matches, the distribution of matches over all species is hypothetically even, but I was curious to know if this was the case.
It is a bit surprising that no ethical statement is provided concerning the use of a social survey. Usually, an informed consent is provided by the respondent, acknowledging that the participant undertakes the survey on a voluntary basis, that anonymity is safeguarded (this is the case here), that the respondent can opt out at no cost, and that the objective of the research has been clearly stated. Was such a consent from the participants clearly formulated?