Review Boscolo-Galazzo et al., EGUSphere

Dear editor,

I have carefully read the manuscript by Boscolo-Galazzo and co-workers you asked me to review. Although I am very enthusiastic about their approach and think that the manuscript is clearly written, there are two main issues that need to be addressed before this manuscript can be accepted for publication in BG. The second of these major comments, may lead to substantial changes in the manuscript, including the conclusions of this paper. In short, the authors need to substantiate their current claim that phylogeny and calcite chemistry are related by a more rigorous approach. In addition, I also listed some things that are less urgent, but do need attention.

Looking forward to see the authors’ response to my comments,

Lennart de Nooijer

Major comments:

1. The statistical approaches should be explicitly described. This includes the exact way the regression analyses were done and the way the MC simulation was applied. Please provide the necessary details. Line 254: do the ‘calibration coefficients’ refer to the uncertainty in the estimated constants from the regression analysis? It would be interesting to know which of the sources of the uncertainty contribute most to the overall CI.
2. The evolutionary perspective on differences in Mg/Ca-temperature relationships is very interesting (and slightly annoying for myself: I was just working on something similar…). However, the analysis of offsets in Mg/Ca and del18O versus evolutionary kinships is not making things clearer. Figures 4 and 5 actually suggest that offsets occur here and there, with no single ‘branch’ containing all the offset results. It may help to show which lineages fall towards the left-upper side of the average line and which towards the right-lower side (Figure 6). This would mean that three groups can be compared. But with the current figures, it remains very difficult to see if the species with a larger offset from the average (Figure 6) cluster together. I think this needs to be shown more rigorously if the conclusion of this exercise (i.e. that evolution is somehow responsible for the variability in Mg/Ca versus del18O) can hold. Also: the black color of some lines in figure 4 and 5 suggest that they are not offset from the average Mg/Ca-del18O, but with the red question mark beside them, we actually don’t know, right? This gives the false impression that more species than is maybe the case, fall around the average Mg/Ca-del18O relationship. Can figures 4 and 5 not be combined? If done, it will become clear that the spinose/ non-spinose divide doesn’t match the offset/ non-offset divide. This actually makes me most suspicious of the conclusion that there is a link between phylogeny and calcite chemistry (at least on this level).

Minor comments:

Line 30: remove ‘of’

Line 32: The authors argue in the discussion that deviations from the average Mg/Ca-del18O correlation is mainly due to inter-species variability in Mg/Ca (rather than variability in del18O). This may be the case, but interspecies differences in oxygen isotopes may still contribute to the variability shown in figures 6 and 7. Therefore, it may be more accurate here to say ‘systematic offsets in the Mg/Ca-del18O relationship’ or something similar.  

Line 175: I guess you combined the ‘Mg/Ca-temperature calibrations’

Lines 205-208: it may be easier to simply state that you corrected the Mg/Ca_cc for past seawater Mg/Ca and pH (see 2.4.2). I guess it doesn’t matter for this procedure what Mg/Ca-temperature calibration you use (since you use it twice in ‘opposite’ directions).

Line 239: I don’t understand this: it seems that the effect of pH, salinity and temperature were taken into account twice. In lines 218-234 it was also described that the long-term changes in salinity and pH were corrected for. Conversion to temperature was already mentioned in line 238.

Line 240: what exponential coefficient? For the G. ruber Mg/Ca-pH sensitivity?

Line 250: what does this re-fit mean precisely? I understood that the Mg/Ca_cc were already corrected for salinity and pH…

Figure 2: so, since the different corrections are plotted ‘on top’ of each other, the order in which they are stacked matters for the suggestion that only part of the data could be accounted for with these corrections. The order could also be such that the upper left part of the data cloud falls within the calibration lines. It is not important for the paper, but perhaps the authors could try to change this? The applied corrections are actually vectors pushing the original calibration in different directions…

Figure 2: I don’t understand the Mg/Ca_sw =2.5. That is (almost) half that of today, but in the methods there lacks a justification for this. Instead, the methods describe how the reconstructed change in Mg/Ca is used to correct the Mg/Ca_cc. It makes me wonder in figure 2, then what this line actually signifies: the caption suggests that the change in Mg/Ca_sw from 15 Ma till now was used to correct the Mg/Ca of the foraminifera, but this would not (necessarily) result in a constant offset (as the orange line here suggests).

Figure 2B: what does the ’m = -2.08’ mean?

Line 284-285: ‘out of’ should be ‘from’.

Line 311: I tried to look it up, but couldn’t find how many datapoints were generated in total. The 62% sounds a higher percentage than it looks to me, but there may simply be many datapoints obscured by the density of the center of the cloud.

Line 352: should be ‘addition’

Line 434 and on: what do the ratios between parentheses mean?

Line 442: reference to Fig 5 is incorrect here. Also, references to figure 5 in the text of the manuscript appear after references to figure 6.

Figure 4: what are ‘offset’ spinose species? If it refers to species that have a del18O-MgCa relationship that deviates too much from the average relationship (figure 6), then what exactly makes it offset? What is the threshold value for such a distinction and what is this based on?

Line 539: should be ‘as in Figure 6’.

General comments

This paper makes a positive contribution to our understanding of d18O and Mg/Ca as temperature indicators. The authors make use of a suitable dataset to address a number of knowledge gaps from both proxies. However, their final conclusions rely on a number of underlying assumptions some of which are still subject to uncertainties that are not clearly disclosed. My main concern is explained under "specific comments" below but once this is covered or acknowledged in some more detail I recommend that the manuscript be accepted for publication.

Specific Comments

d18O-temperatures: I’m aware that the online tool of Gaskell and Hull (2023) allows for a limited number of d18Osw corrections and that Rohling et al. (2021) (R_21) is a perfectly valid option. However, not only is d18Osw beyond the Pleistocene (or even just the LGM) quite uncertain, but also a recent benthic d18O deconvolution from Rohling et al. (2022) (R_22  https://doi.org/10.1029/2022RG000775) suggests heavier d18Osw during most of the last 15 million years compared to R_21, with the independent model output from De Boer et al. (2010) sitting somewhere in between R_21 and R_22. Meanwhile, clumped isotope derived d18Osw suggests even heavier values for parts of the Miocene (e.g.  https://doi.org/10.1029/2020PA003927).

For these reasons, I wonder if it’d be possible to at least account for some of this uncertainty in your d18O-temperatures (larger vertical error bars?) and whether this might bring some of your “discordant" samples back into the uncertainty envelope of Figure 3. Alternatively, you could replot a few more versions of Figure 3 with different d18Osw corrections as offered in Gaskell and Hull (2023). Either way, I think it would be important to at least acknowledge these uncertainties in d18Osw before giving the reader the impression that a high degree of confidence exists behind the non-thermal corrections mentioned in the text. Moreover, this has implications for e.g. your hypothesis in Section 4.1.

Technical corrections

229: This line suggests a very similar (or equivalent?) Mg/Casw than that one derived in Rosenthal, Bova and Zhou (2022) where they also combined Zhou et al., 2021 with the fluid inclusion data from Brennan et al., 2013. If it is then it might be more clear for the reader to cite the Rosenthal and Bova (2022) record directly (https://www.ncei.noaa.gov/access/paleo-search/study/36413).