General comments

I find this research using AGI to evaluate the effect of long-term warming and deoxygenation on contemporary habitat useful. It uses the AGI an index that represents the o2 supply to demand ratio for maintenance activity. It is handy as it requires few data somewhat easily accessible. The authors show how this index can be used to assess species vulnerability to environmental changes using only species-specific biogeographic data of 47 species. An interesting point, is that the authors show that tendencies and mean changes alone (warming, deoxygenation and mean changes in AGI) do not suffice to predict species vulnerability within their present habitat, but rather the quantity of habitat volume close to AGIcrit as show by the CDF of the AGIs. They also show the high inter-species variability in terms habitat preferences and critical thresholds greatly influence the changes in viable habitat. Indeed the mean changes do not reflect species-specific changes in habitat viability. It is also very interesting to present the results by degrees of global warming.

A few improvements could be made to facilitate the reading of the results (see specific comments):

• More systematic presentation of the results
• Some methods of calculation are not given
• A bit more clarity is needed regarding the definition of some terms or choice of wording (e.g. habitat viability, potential habitat, AGI^rel vs. ΔAGI…). I suggest they all be defined in the method section.

Questions regarding the AGI need to be discussed.

Specific comments

In general, a more systematic presentation of results is need to ease the reading and further support the demonstration. In particular, a more systematic presentation of the figure (to facilitate the reading, so the reader doesn’t have to go back and forth in the main text. Also more consistency when choosing the warming level, scenario, etc. when presenting the figure in the main text. If you start presenting results for the levels of warming (Fig. 3 and 4) , please do so for the rest of the manuscript. Even with figures in the supplemental.

Abstract

Line 17-20: not clear, please rephrase.

Methods and data

• The choice of keeping a constant value for j1 and j2 across all species is indeed convenient and confers great value to the AGI and has been somewhat evaluated in the Clarke et al 2021 per comparison to the metabolic index. However, I still believe that sensitivity analysis to j1 and j2 would be useful to demonstrate the added power of the AGI since this index is quite sensitive to parameters j1 and j2. Given the formula and the scale to which it is applied, I imagine AGI^rel at global scale will be only weakly affected by the choice of j1 and j2. But for species species-specific AGI, it is less certain, in particular for species from equatorial and tropical areas. In particular, it will affect AGI_crit and so the changes in volume of viable habitat (AGI>AGI_crit) and possibly the slope of the CDF used to assess species vulnerability to changes in AGI, as you demonstrate that changes in viable habitat are species-dependant. So sensitivity analysis of AGI^rel and various AGI would be useful to demonstrate the degree of independence of the AGI to these parameters and would add great value to the results.
• As the AGI is comparable the metabolic index and given that it has recently been shown that it cannot be applied to certain species such as D. gigas or other species performing vertical migrations (Seibel & Birk, 2022), I wonder if the same limitations may apply to the AGI ? which case this type of species should be excluded from the study. Also it may need to be discussed in the manuscript.

Line 139 : which data ? O2, T, salinity ?

Line 156 : please detail a bit more. You mean global mean SST reached by 2100 ?

Results

Line 174 : « habitat viability » suggests you refer to where AGI>AGIcrit, but you refer to AGI^rel. It can be confusing. AGI^rel would indicate « potential viable habitat » ?

Also, « AGI^rel reduction » is incorrect. AGI is either negative or positive reflecting a decrease or increase in AGI between t0 and t1. Please rephrase.

Line 180-191 : A figure to show this would be better.

Line 201 : «A relative reduction in habitat viability […] we expect a reduction in habitat viability ». Please rephrase. See comment above (line 174) relative to « habitat viability ».

Line 207 : Please detail somewhere how the contribution of po2 and T to AGI and AGI^rel is calculated.

Line 209 – 216 : Not clear what the difference is between the calculation method of the contribution of po2 and T is between line 208-210 and 212-216. For instance, « the AGI^rel due to T is -xx % for the epipelagic» (line 209) and « an average 87 % of AGI^rel is driven by… warming » (line 212). What is the difference be the two ? please detail calculation.

Figure 2 : please provide the same map as Fig. 2 (and C2) for 1.5 °C and 3°C to be consistent with the remainder of the paper.

Fig C3 caption : « AGI^crit as the minimum in-habitat AGI value, the 5th percentile, the 10th percentile, the 15th percentile and the 20 th percentile ». Word missing ?

Line 242-246 : in the text, changes in viable habitat are expressed in terms of habitat loss, but in the referenced figure (Fig. 3) changes in viable habitat are expressed in terms of remaining habitat. Please be consistent.

Line 256: please define “absolute loss”.

Figure 3 : not clear how the different models/scenarios are represented or used for the calculation of changes in viability.

Fig. 5 : for which degree of warming ? Scenario ? Period of AGI ? Please precise in the figure caption. Also C4 is presented for a 3°C global warming. For the purpose of the demonstration, I understand that the chosen level is not determining, but consistency between figs within the same result section would be better to support the demonstration. Also distribution of AGI +/.

Line 316: “The correspondent linear equation taken across all depth realms is 𝑣𝑜𝑙𝑢𝑚𝑒 𝑙𝑜𝑠𝑠 (%) = 7.31 ∗ vulnerability − 0.10.” not useful.

Line 322: ΔAGI is AGI(t1) – AGI(t0) ? please define.

Fig 6 : please provide same for 1.5 and 3 °C. Also, why only SSP5-8.5 ?

Line 336-337 : any hypothesis regarding those two species ?

Technical comments

Line 43 : ref Bopp et al. 2013 is about CMIP5

Line 139 : please specify « all data »

Line 201 : « reduction in AGI^rel », see comment above.

Line 204: « AGI^rel reduction » see comment above.

Line 229 : « decrease in AGI^rel », see comment above. See also line 231, 239

Line 240 : habitat volume is where AGI>AGI^crit ? Please precise.

Line 291-296 : Please precise which period of the AGI is used for the PDF.

Line 302 : Only → only

Line 303 : In → An

Line 306 : indicates → remove s

The manuscript «Impact of deoxygenation and warming on global marine species in the 21^st century» assesses the potential effects of changes in ocean temperature and pO₂ on the viability of current marine habitats. The computation of a metabolic index (the AGI), which depends on temperature, oxygen and few easily-accessible species-specific parameters, allows the authors to explore the species’ (47 selected species) vulnerability to environmental changes during the current century. Environmental data rely on six ESM. The authors first discuss the projected global and regional changes on ocean temperature, pO₂, and AGI by degrees of global warming. Then, they present the loss/gain of viable habitat volume with respect to current habitats as future AGI index falls below a critical value considered to be a threshold for holding aerobic activity. Moreover, authors assess the “vulnerability” of a species by computing the probability density function of AGI computations for each species, and evaluate the volume of viable habitat loss with respect to reductions in AGI. Authors discuss the novelties (including vertical and seasonal variability, species representativeness,..) and limitations (temporal resolution, deoxygenation underestimation, adaptation capability of species,...) of their approach. They finally highlight the key points of the study.

As I understand from the manuscript, three-dimensional monthly data (temperature, oxygen content, and salinity) from six ESM simulations (bias and drift corrected) following historical and two (low and high) emission scenarios were used to determine the habitat viability of 47 different marine species. Species were selected in order to be representative for different climatic zones, sizes and vertical levels. Viability was analyzed by using the AGI index, which depends on temperature and pO₂ along with two species-specific parameters (they can be determined by using the species distribution data), and which gives information on the aerobic state of an ecosystems and, hence, of the species. Global and regional variation of temperature, pO₂ and AGI, as well as the relative contribution of each of AGI drivers (i.e., AGI due to pO₂, and AGI due to temperature) were evaluated. Then, the relative change of AGI with respect to current situation was assessed giving information of the habitability loss/gain for each species. From the probability density distribution of each species’ AGI, a “vulnerability” of each individual species was computed. This exercise highlighted that the vulnerability of a species not only depends on the volume loss (volume below a critical value of AGI) but also on the habitat volume in sub-critical values of AGI. According to the loss in habitable volume and the vulnerability, species were classified in three groups; highly affected, resilient, and vulnerable species. From these results, authors discussed the limitations of the study and highlighted the main messages.

The study is scientifically relevant and worth publishing as it introduces the potential utility of a metabolic state index like AGI to evaluate how species will be affected by changes in their environment. I only have some comments which, I hope, are useful to improve the manuscript.

I have some doubts understanding the vertical distribution of the layers considered. As I understand, horizontal distributional data is extended in the vertical over the depth range of species distribution. However, this is not exactly the case as some species crosses the limits of epipelagic and mesopelagic layers as they are defined in the text. This is especially true for some mesopelagic species that live in the deepest limit of the layer to below, while others are considered in the same layer but lives closer to epipelagic waters. I think a detailed analysis (maybe out of the intention of this study) should include the “real” depth range for each species as informed in their distributional data. Moreover, as has been recently pointed out (Seibel & Birk, 2022), organisms carried out vertical migration have specific metabolic constraints that make them difficult to assess using current metabolic indexes. Do the authors think this might be plausible as well for AGI? How this affects results? I think these issues should be commented in the text.

In my opinion, some lines should be added to the discussion section to proper discuss the implications of the results. What differences were found between scenarios? How this new method can be help in managing fisheries for example in the future? Although I recognize this can be out of the scope of the paper, I think the manuscript would benefit from a short discussion on these issues.

Regarding the conclusions of the paper, I think the authors could make a more narrative presentation of the key messages.

Finally, I found some sections hard to understand, especially, section 3.4. which in fact I consider it is the ‘key’ of the manuscript. I feel that some more information is needed, maybe in a supplementary methodology section, to better explain the computation of vulnerability.

Specific comments about the text and figures:

Line 9: I would change to: “...the observed and projected warming and deoxygenation of the world’s ocean in the 21^st century may strongly affect marine species’ habitats.”.

Line 11: Change “We” to “in a particular location, to assess…”.

Lines 19 – 20: I find this can be reprhased.

Line 21: I think this “is” should be after “epipelagic species”.

Line 43: Bopp et al., 2013 is about CMIP5, not SSP5-8.5 scenario.

Line 46: I think this is not about “impacts” but just trends. In addition, I would split here the paragraph, and start a new one with temperature.

Line 72: “to a species,”.

Line 105: I think considering the nomenclature of mesopelagic and bathypelagic is problematic. As only 200 to 1000 m is considered, I would say mesopelagic layer. Bathypelagic is usually considered for depths ranging 1000 m to below.

Line 132: It is in fact the 200 to 1000 m depth range representative of the species considered? Daily migration can have some effects?

Line 160: In figure 1 it is indicated a 20-year running.

Line 203: “is generally larger in”.

Line 211: I would remove “Globally”, it seems redundant here.

Line 223: “uncertain” refers to multimodel uncertainty?

Line 223: I don’t understand what regions are referred to with “eastern-boundary equatorial upwelling regions”.

Line 229: “to some small parts”.

Line 240: I would substitute or remove “local extinction”. Organisms may also move or adapt to new conditions.

Lines 256 to 264: This is somehow expected; big losses in large-distributed species would account for small relative losses.

Line 286: What “realized loss” is referred to?

Line 304: I think some concepts here should be better explained/introduced, like ΔAGI.

Line 316: This is an important point of the study that can be put upfront in the discussion/conclusions.

Line 350: It might be good to include a discussion on the implications of the results that complement the discussion of the limitations of the work.

Figure 1:

- I think “transparent blue and red” may be changed to “light blue and red”.

- Indicate that AGI is given in percentage in the first time it is called within the caption.

- Though I like the idea of presenting results relative to global warming, I think it is somewhat more difficult to read because of the length of y-axis. I suggest to increase the readability of the figure to change the order to two lines; in the top line, panels a) and b) are displayed, and in a second line below that, panel c) is displayed. Something like (sorry for the bad picture…).

Very minor, but color of Fig. C1 is blue but it is indicated in the figure caption to be dark grey.