Using B isotopes and B/Ca in corals from low saturation springs to constrain calcification mechanisms

Ocean acidification is expected to negatively impact calcifying organisms, yet we lack understanding of their acclimation potential in the natural environment. Here we measured geochemical proxies (δ11B and B/Ca) in Porites astreoides corals that have been growing for their entire life under low aragonite saturation (Ωsw: 0.77–1.85). This allowed us to assess the ability of these corals to manipulate the chemical conditions at the site of calcification (Ωcf), and hence their potential to acclimate to changing Ωsw. We show that lifelong exposure to low Ωsw did not enable the corals to acclimate and reach similar Ωcf as corals grown under ambient conditions. The lower Ωcf at the site of calcification can explain a large proportion of the decreasing P. astreoides calcification rates at low Ωsw. The naturally elevated seawater dissolved inorganic carbon concentration at this study site shed light on how different carbonate chemistry parameters affect calcification conditions in corals.

1) The authors cast their study in large part as a test of the "recently developed bioinorganic model" of McCulloch et al. (2012). However, in my view, this "model" is already outdated, and by placing too much emphasis on testing this model, the authors are underselling their study. The "model" itself is more of a list of assumptions rather than what is typically meant by the word "model". It is simply a way to convert pH cf (derived from d 11 B) to W cf and calcification rate by assuming that (i) DIC cf is 2X DIC sw , (ii) that [Ca 2+ ] cf is constant and within a few % of [Ca 2+ ] sw , and (iii) that bulk coral calcification is a simple function of W cf . In 2012, when the "model" was proposed, (i-iii) were not known and the techniques did not exist to test them, and thus this was a reasonable list of assumptions that, at the time, enabled coral calcification rates to be interpreted in relation to d 11 B. However, new techniques have been developed since then, and our understanding of calcifying fluid chemistry has advanced beyond these early assumptions. B/Ca and d 11 B can now be used in tandem to derive DIC cf and [CO 3 2-] cf . This means that (i) no longer needs to be assumed, which is the most relevant point here because the authors have the B/Ca data to clearly show that assumption (i) is not valid. Additionally, we now know that [Ca 2+ ] cf is not constant, but responds to pH sw in some species Comeau et al. 2018). These points do not take away from the authors' data or their study overall; I am merely saying the authors should go directly to interpreting their d 11 B and B/Ca data together, rather than restricting themselves to testing an outdated model that uses only d 11 B.
2) The authors should go more in-depth in their thinking about DIC cf . Specifically, the authors cast DIC in terms of "DIC cf /DIC sw ", but without justification or a physiological reason why the organism would increase DIC by a specific multiple of the seawater concentration. I understand that this approach has been used before (McCulloch et al. 2012, but DIC sw was relatively constant in those cases, meaning that those previous authors could have presented their work in terms of absolute DIC cf (or DIC cf = DIC sw + X). The present study is fundamentally different in that DIC sw does vary substantially between the control and treatment locations. Therefore, the present study is unique in that interpretations differ dramatically for absolute DIC cf and DIC cf /DIC sw (compare Fig. 2C with Fig. S1). I actually see this as an opportunity for the authors to test some existing ideas about DIC cf and its relation to seawater chemistry, so this is more of a suggestion to incorporate some additional interesting aspects to the manuscript, rather than a criticism. For instance,  found strong relationships between DIC cf and DIC sw in laboratory experiments, which obviously differs from the present findings, but sets up an insightful comparison of laboratory and field results.
3) It should be described clearly that the information provided by d 11 B and B/Ca is [CO 3 2-] cf , not W cf . The two differ depending on [Ca 2+ ] cf , which the authors have no data to constrain. In my view, the text and figure labels should all be "[CO 3 2-] cf ", and not "W cf ". If the authors really wish to use "W cf ", then it needs to be clearly explained in the main text and figure captions that W cf can only be estimated here by making an assumption about [Ca 2+ ] cf . 4) There is a substantial amount of very recent literature that is not cited, but that is highly relevant to this study (DeCarlo et al. 2017, a, Ross et al. 2018bComeau et al. 2018;Cornwall et al. 2018). I recognize that these papers were published in the past year, and that they are all from our group. Thus, I am not trying to push our own interpretations or force the authors to cite our work, but rather I want to make sure the authors are aware of these papers because they represent recent applications of d 11 B and B/Ca in tandem, new techniques to quantify W cf and [Ca 2+ ] cf , insights into controls on calcifying fluid chemistry, other field data to compare with the present study, and codes for computing systematic and non-systematic uncertainties associated with the combined d 11 B and B/Ca proxy approach. It is at the authors' discretion if, and how, to integrate these references into their manuscript. 5) Methods: some clarification of the proxy calculations is needed. Interpretations of the boron-based proxies are sensitive to salinity, temperature, d 11 B sw , and [B] sw (see DeCarlo et al. 2018b). Since some of these parameters vary between ojos and control sites, careful consideration is required. I commend the authors for measuring seawater [B] and d 11 B, as these could depart from typical values within the ojos. However, more information is needed. How many water samples were taken and at how many different times? The seawater [B] is presented as "~430 µM", but how much did it vary? The seawater d 11 B was different between the ojos and control sites, which must influence the pH cf interpretations. How influential is this difference in seawater d 11 B (i.e. how much would results change if d 11 B was assumed constant)? Since the seawater [B] and d 11 B could change over time (and this is presumably not captured by the water sampling), some assessment of their potential influence on the results is needed. Another potential issue is that the authors used a constant pK B (averaged from the two sites) for the pH cf calculations, but site-specific pK B for the B/Ca calculations. I do not understand this approach, and I wonder how much it influences the results. 2) Abstract: The statement "were found growing in natural settings where the aragonite saturation state is very low" should be revised because this study did not find these corals. The corals and the setting were described by some of these authors previously; the present study applies geochemical proxies to the previously-described corals.
3) Abstract: "decades" -is evidence presented here that these corals have been alive for multiple decades? 4) Introduction: I suggest not to cite IPCC for ocean acidification affecting coral calcification. There are many primary articles to support this statement.  Comeau et al. 2018). 11) "The implementation and application of a combined d 11 B-B/Ca, with B/Ca serving as a proxy for CO 3 2concentration and, subsequently, DIC cf is a relatively new approach 29 " I think reference 29 is a typo here because that study did not use B/Ca to determine [CO 3 2-] or DIC.
12) Need to be consistent with "acclimation" and "acclimatization". 13) Methods: "First seawater boron concentration was determined based on the sitespecific pK B assuming boron concentration at the site of calcification equals to seawater boron concentration." I assume the authors mean "First, seawater borate…". -Thomas M. DeCarlo Reviewer #2 (Remarks to the Author): Wall et al., have utilized δ11B and B/Ca measurements in Porites astreoides in order to derive the calcifying pHcf, DICcf and subsequent Ωcf. The idea was to understand the acclimation of corals in natural low Ωsw environment from species collected in the NMP at Puerto Morelos. In order to achieve that, they used 2 close locations with different environmental settings and Ωsw. Corals from both locations exhibit up-regulated pHcf, DICcf with more up-regulation at the low pH centers, however, Ωcf for both locations differed and the low pH locations presented a lower Ωcf compared to control. This difference is consistent with the lower calcification (~37%) observed at the low pH locations but this lower calcification is not fully explained by the decrease in Ωcf (model 44%).
Due to the growing interest of the boron based proxies in corals, this paper adds a stone to the growing evidence of a dynamic calcifying fluid (pHcf, DICcf) in natural environments modulating the calcification response. This paper highlights the need for a better understanding of the symbiotic relationship of corals under different environments.
It terms of analysis, it would be great to have some standards presented in the paper to give more confidence to the data, even if they look consistent.
Minor points: Line 26: It is a bit confusing here. "whether these corals", may be change for "in what extent these corals have acclimated".
Line 28: "increase the pH" may be change for "increase the pH and subsequent Ωcf" since Ωcf is likely the main trigger for calcification. Line 54: May be add "and upregulate their Ωcf".
Line 91: They do not have similar temperature or salinity, 1-2 °C lower and 3psu lower in the "low pH centers" compared to centers. Do you have an idea of the seasonal variability? Can you test whether those differences could explain a part of your variability?
Line 98: Could the acclimation come from this high Ca content as well?
Line 181: This result is central in your paper might be good to explain a bit more at line 439.
Line 197: "thereby modulating Ωcf and calcification rate" Line 210: May be add a line saying that the temperature is not the main driver because people will look for this relationship, and that will reinforce your arguments.
At "low pH sites" the temperature is lower than at control sites, since a decrease in pHcf and DICcf is observed with increasing temperature, temperature in our case can't explain the decrease of pHcf and DICcf.
Line 217: I think this paragraph needs to move or towards the conclusion or in the introduction.
Line 262: Is there any data for your corals?
Line 265: So Ωcf explains 44% of the calcification rate variability, all the arguments here are likely to modulate the Ωcf but since it is only 44%, any ideas what are the independent (from the calcifying fluid) triggers for the 56% left?
Line 284: Can you explain why it is controlled by the adjacent epithelium?
Line 290: May be add a sentence to conclude more smoothly.
Line 320: How did you measure your seawater samples?

Reviewer #1
Many of the comments of reviewer #1 are related hence we grouped these comments and responded to them together. We then respond to additional comments separately.
The manuscript, "B isotopes and B/Ca of Porites astreoides corals as indicators of calcification processes at low aragonite saturation" investigates the calcifying fluid chemistry of corals growing in naturally low-saturation state seawater. The main findings are that calcifying fluid carbonate ion concentration responds to seawater carbonate chemistry, and this explains some of the calcification response under low saturation state. Although the sample sizes are relatively small, the results provide important new insights into the sensitivity of coral calcification to ocean acidification. In particular, these results from a natural environmental gradient differ from recent results in controlled laboratory experiments, which raises interesting questions for the community to address. I think the data are worthy of publication, but I do have some major suggestions for the manuscript.
Major comments: 1) The authors cast their study in large part as a test of the "recently developed bioinorganic model" of McCulloch et al. (2012). However, in my view, this "model" is already outdated, and by placing too much emphasis on testing this model, the authors are underselling their study. The "model" itself is more of a list of assumptions rather than what is typically meant by the word "model". It is simply a way to convert pHcf (derived from δ11B) to Ωcf and calcification rate by assuming that (i) DICcf is 2X DICsw, (ii) that [Ca2+]cf is constant and within a few % of [Ca2+]sw, and (iii) that bulk coral calcification is a simple function of Ωcf. In 2012, when the "model" was proposed, (i-iii) were not known and the techniques did not exist to test them, and thus this was a reasonable list of assumptions that, at the time, enabled coral calcification rates to be interpreted in relation to δ11B. However, new techniques have been developed since then, and our understanding of calcifying fluid chemistry has advanced beyond these early assumptions. B/Ca and δ11B can now be used in tandem to derive DICcf and [CO32-]cf. This means that (i) no longer needs to be assumed, which is the most relevant point here because the authors have the B/Ca data to clearly show that assumption (i) is not valid. Additionally, we now know that [Ca2+]cf is not constant, but responds to pHsw in some species Comeau et al. 2018). These points do not take away from the authors' data or their study overall; I am merely saying the authors should go directly to interpreting their δ11B and B/Ca data together, rather than restricting themselves to testing an outdated model that uses only δ11B.
2) The authors should go more in-depth in their thinking about DICcf. Specifically, the authors cast DIC in terms of "DICcf/DICsw", but without justification or a physiological reason why the organism would increase DIC by a specific multiple of the seawater concentration. I understand that this approach has been used before (McCulloch et al. 2012, but DICsw was relatively constant in those cases, meaning that those previous authors could have presented their work in terms of absolute DICcf (or DICcf = DICsw + X). The present study is fundamentally different in that DICsw does vary substantially between the control and treatment locations. Therefore, the present study is unique in that interpretations differ dramatically for absolute DICcf and DICcf/DICsw (compare Fig. 2C with Fig. S1). I actually see this as an opportunity for the authors to test some existing ideas about DICcf and its relation to seawater chemistry, so this is more of a suggestion to incorporate some additional interesting aspects to the manuscript, rather than a criticism. For instance, Comeau et al. (2018) found strong relationships between DICcf and DICsw in laboratory experiments, which obviously differs from the present findings, but sets up an insightful comparison of laboratory and field results.
3) It should be described clearly that the information provided by δ11B and B/Ca is [CO32-]cf, not Ωcf. The two differ depending on [Ca2+]cf, which the authors have no data to constrain. In my view, the text and figure labels should all be "[CO32-]cf", and not "Ωcf". If the authors really wish to use "Ωcf", then it needs to be clearly explained in the main text and figure captions that Ωcf can only be estimated here by making an assumption about [Ca2+]cf.
4) There is a substantial amount of very recent literature that is not cited, but that is highly relevant to this study (DeCarlo et al. 2017, a, Ross et al. 2018bComeau et al. 2018;Cornwall et al. 2018). I recognize that these papers were published in the past year, and that they are all from our group. Thus, I am not trying to push our own interpretations or force the authors to cite our work, but rather I want to make sure the authors are aware of these papers because they represent recent applications of δ11B and B/Ca in tandem, new techniques to quantify Ωcf and [Ca2+]cf, insights into controls on calcifying fluid chemistry, other field data to compare with the present study, and codes for computing systematic and non-systematic uncertainties associated with the combined δ11B and B/Ca proxy approach. It is at the authors' discretion if, and how, to integrate these references into their manuscript. 9) It would be more appropriate to plot DICcf in the main text. As the authors note, DICcf/DICsw changes mainly due to DICsw, not DICcf. Previous studies have interpreted DICcf/DICsw as indicative of calcification processes because in those studies DICsw was relatively constant. Conversely, in the present study, DICcf/DICsw really just tells us about the environment, not differences in the calcification process between treatment and control corals.
We thank the reviewer for recognizing the importance and relevance of this paper and in particular, for his valuable comments that helped to improve and better structure our manuscript. We also thank for referring us to the recently published very relevant papers in his comments #1-4, 10, 9. These references were not used in the original manuscript primarily because the manuscript has been written before most of these papers were published and has been under review for a while. Yet we certainly agree that this are very relevant aspects and we are including these important concepts, mechanisms and references in the revised version.
First of all, in the revised version we now focused only on the model where both δ 11 B and B/Ca are applied together. In the results section we now refer just to this newer approach (Line 162: "We used the combined pH cf and the B/Ca-derived DIC cf concentration at the site of calcification for both ojo and control corals."). We also adjusted the discussion by only talking about the ability to use our dual-geochemical proxy data to elucidate underlying mechanisms and potential drivers. We updated the introduction and when we talk about the model we used the opening of this paragraph to state (Line 244): "The application of our dual geochemical proxy data to model coral growth (e.g. IpHRAC 21,33 ) allowed us to further pinpoint potential mechanisms of how external seawater conditions affect internal calcifying conditions and ultimately skeletal growth." We used this to directly enter the discussion on DIC sw variations and their role in modulating or not DIC cf (more to this topic see below).
We moved the old model to the supplements (Fig. S1) because it is used to support the point at the beginning of our discussion (Line 190): "Interestingly at this field site ojos with low aragonite saturation state had elevated DIC sw , but this did not result in higher DIC cf concentrations in the calcifying fluids, indicating a decoupling of internal and external DIC concentrations.", and used this opportunity to further emphasize this point. This also is in-line with the comment on DIC cf and we specifically addressed this issue now in more detail in Line 251: "…At our study sites, DIC sw significantly and extensively differs between sites, with higher values at the low Ω sw sites (in average 2790 µmol kg -1 compared to control average DIC sw of 2050 µmol kg -1 ). … If corals modify internal DIC cf by simply upregulating DIC cf from the external concentrations baseline, we would expect higher DIC cf values for the ojo corals where DIC sw is higher. Under such assumption the elevated DIC cf compensates for the slightly lower pH cf effect on Ω cf and calcification rates would essentially be similar between sites (Fig.  S1)." We also adjusted Figure 2 by adding the supplementary figure of DIC sw here, so it already clearly visualizes that DIC cf /DIC sw observed response to seawater Ω sw is mainly due to the substantial and significant change in seawater DIC sw pinpointing towards the above mentioned decoupling of seawater DIC and internal DIC cf . We added this to the discussion (Line 259): "However our data clearly demonstrates that DIC cf is not directly linked to external concentrations and can differ significantly from that of seawater 25,33,52 (reported DIC cf -upregulation values range from 1.6-3.2 33,44 , with the ojos corals at our sites are below the lower end), and this impacts Ω cf (more precisely CO 3 2-) and calcification. A recent study under laboratory conditions with Stylophora pistillata 53 observed that changes in DIC sw concentration modulates internal pH cf regulation, with higher external DIC cf facilitating higher internal pH cf observing a clear correlation between seawater DIC sw /H + sw and pH cf . Since DIC sw at the ojos is significantly higher than at the control sites, one could expect this compensates reduced pH cf up-regulation induced only due to changes in seawater pH sw . Yet we do not see a strong correlation between pH cf and seawater DIC sw /H + sw , suggesting different drivers for pH cf regulation in Porites astreoides compared to those observed in Stylophora pistillata 53 . Nevertheless, the change in pH cf and the limited ability to upregulate DIC cf beyond seawater DIC sw at the ojos corroborates the observed calcification rate decrease at the ojos." Regarding comment #3 we agree that we need to be more precise here and clearly state that our Ω cf is based on the two derived carbonate chemistry parameters from our proxies and not based on the more relevant Ω cf derived from CO 3 2and Ca 2+ cf . Plotting the data against CO 3 2only, would not show the observed relationship since our conclusion is that the COMBINED even though limited change in pH cf and CO 3 2-(expressed as Ω cf ) concentration along the Ω sw gradient contributes to the change in calcification rate. We want to specifically make this point for reasons mentioned by the reviewer but also to point out that we need to be cautious with relationships that may not be significant yet in combination with each other can provide some explanation of/new insights to observed patterns. This relationship pinpoints towards the points raised in the reviewer's comment #4 that Ca 2+ cf may be changed along the Ω sw gradient and likely drives the observed decline in calcification rate. In the revised version we specifically address this issue and we thank the reviewer for encouraging us to take the discussion into this direction: eg. We now state in Line (275): "Recent studies identified internal calcium (Ca 2+ cf ) regulation as an additional player in coral calcification responses and emphasized that regulation of Ca 2+ cf can contribute to a corals' resistance to future ocean changes [42][43][44]54 . In this sense, the good agreement of our model with the observed calcification response potentially indicates that internal average steady-state calcium concentrations (Ca 2+ cf ) are lower at the ojos by some proportion that is related to the pH cf changes, since our model based on pH cf and CO 3 2cf can reproduce the observed calcification decline. This suggests a strong link between Ca 2+ cf and pH cf and supports the idea of a plasma-membrane Ca-ATPase 55,but see 56 responsible for pH cf regulation. However, it is possible that pH cf but also Ca 2+ cf was regulated by additional and/or different ion transport mechanisms (e.g. potentially ion exchangers, Ca 2+ -channels) 56,57 . " and in (Line 288): "The present study also indicates that the acclimation process in different corals encompassed some degree of flexibility in terms of the relative role of pH cf and DIC cf regulation in increasing the Ω cf , with some individuals compensating by adjusting their internal pH cf and others primarily by DIC cf modulation. This may also be true of the role of Ca 2+ cf upregulation. The relative amount, source and transportation pathways of DIC, H + and Ca 2+ to the site of calcification are still not fully understood 58 and transport processes may differ between individual corals." 5) Methods: some clarification of the proxy calculations is needed. Interpretations of the boronbased proxies are sensitive to salinity, temperature, δ11Bsw, and [B]sw (see DeCarlo et al. 2018b). Since some of these parameters vary between ojos and control sites, careful consideration is required. I commend the authors for measuring seawater [B] and δ11B, as these could depart from typical values within the ojos. However, more information is needed. How many water samples were taken and at how many different times? The seawater [B] is presented as "~430 M", but how much did it vary? The seawater δ11B was different between the ojos and control sites, which must influence the pHcf interpretations. How influential is this difference in seawater δ11B (i.e. how much would results change if δ11B was assumed constant)? Since the seawater [B] and δ11B could change over time (and this is presumably not captured by the water sampling), some assessment of their potential influence on the results is needed.
All water samples (in total 8) were first analyzed on ICP-MS Finnigan Element XR for boron concentration measurements following Krupinski et al. (2017) before they were further analyzed on a Neptune multi-collector ICP-MS at National Cheng Kung University, Taiwan for seawater δ 11 B. Within these set of water samples (n=8 for both control and ojo; 3 ojos and 5 control) the variation was rather low (2%) and agree with typical open ocean B concentrations reported ([BT] = 432.9 µmol kg -1 ± 2; (Lee et al., 2010)). Since also no differences could be found between sites it did not require adjustments in terms of boron concentration in our calculations. Our seawater δ 11 B composition, though, indicate minor differences in composition between the different sites investigated in this study. Hence, we used our dataset to illustrate the relationships of changing seawater δ 11 B and pH cf when we left the different pKBs for each site constant. We applied a range of δ 11 B that encompasses the average measured δ 11 B per site but also seawater isotopic composition beyond this level ranging from 38.55 to 39.45 ‰ and recalculated pH cf from the entire dataset (Fig. 1A). This allowed us to decipher the combined role of site specific pK B and seawater δ 11 B for a range of skeletal δ 11 B (Fig. 1B). In general, the δ 11 B derived pH cf decreases with increasing seawater δ 11 B. Changes in seawater δ 11 B in the corals surrounding will either over or underestimate pH cf and calculated changes in pH cf are with 0.019 to 0.023 pH units per 0.3 change in δ 11 B sw (Fig. 1C; the average difference between our sites; or 0.056-0.065 for the entire δ 11 B sw range tested) in the range observed pH cf for the investigated corals 8.88-8.18 ( Fig. 1A; ∆pH cf = 0.8; or sd per individual ranging from: 0.04-0.13). Of course such changes add some uncertainty and potentially also contribute to the fact that our model only explains 44% of the observed variation. We added this into the Method section in Line 428: "We note here, that the local variability in carbonate chemistry at the ojos and hence, associated changes in pK B and seawater δ 11 B can add some uncertainty to the derived pH cf and over-or underestimate its actual value. To test the sensitivity to changes in pK B we used our dataset and recalculated pH cf values. We applied a range of seawater δ 11 B sw that encompasses the average measured δ 11 B sw per site but also seawater isotopic composition beyond this level ranging from 38.55 to 39.45 ‰ and recalculated pH cf (supplementary figure S3A). This allowed us to decipher the combined role of site specific pK B and seawater δ 11 B sw for a range of skeletal δ 11 B (supplementary figure. S3B). In general, the δ 11 B derived pH cf decreases slightly with increasing seawater δ 11 B. Changes in seawater δ 11 B sw in the corals surrounding will either over or underestimate pH cf and calculated changes in pH cf range from 0.019 to 0.023 pH units per 0.3 change in δ 11 B sw (supplementary figure S3C the average difference between our sites; or change from 0.056-0.065 for the entire seawater δ 11 B sw range tested). Compared to the pH cf range (8.2-8.8) derived from individually measured skeletal δ 11 B values such changes are minor ( Fig. 1A; ∆pH cf = 0.8; in contrast to the individual coral's pH cf standard deviation of 0.04-0.13, supplementary table S2)." Another potential issue is that the authors used a constant pKB (averaged from the two sites) for the pHcf calculations, but site-specific pKB for the B/Ca calculations. I do not understand this approach, and I wonder how much it influences the results.
We apologize but this was actually an error in the manuscript. We used different pK B for both pH cf and B/Ca calculations.

Minor comments:
1) I suggest revising the title. At present, it sounds as if the study develops a new technique to study coral calcification based on boron isotopes and B/Ca. But those techniques have already been developed. This study applies those techniques to an interesting natural gradient in seawater chemistry.
We changed the title to: Using B isotopes and B/Ca in corals from low saturation springs to constrain calcification mechanisms 2) Abstract: The statement "were found growing in natural settings where the aragonite saturation state is very low" should be revised because this study did not find these corals. The corals and the setting were described by some of these authors previously; We changed this to (Line 22): "Here we measured geochemical proxies (δ 11 B and B/Ca) of these previously described Porites astreoides corals that have been growing for their entire life (decades) under conditions of low seawater aragonite saturation (Ω sw : 0.77-1.85) 1 ,.." 3) Abstract: "decades" -is evidence presented here that these corals have been alive for multiple decades?
Yes, based on annual bands analyzed by Crook et al 2013 we have some that are at least 30 years old. 4) Introduction: I suggest not to cite IPCC for ocean acidification affecting coral calcification. There are many primary articles to support this statement.
We changed the citation.

5) Second sentence of the introduction: Ref 15 does not show calcification changes, as is
implied. Barkley et al (2015) is the appropriate reference showing calcification changes at this site.
Reference changed 6) Introduction: "Number of tropical corals (40%)"? Is this species or something else? 7 The referred meta-analysis by Chan and Connell used different studies (fulling their criteria that studied different but also the same species) and treated all individually (mixing species, sites, etc). In the revised version we deleted the percentage to not imply that it refers to 40% of all tropical coral species or anything else. We just state that (Line 45): "These efforts have provided strong evidence that the calcification rates of a large number of coral species investigated to date will decline in response to projected pCO 2 17 ." 7) "Most studies suggest that seawater pHsw is the main driver affecting pHcf": References to support this statement?
We adjusted the text here and added references studying the relationship between internal and external seawater pH. 8) I do not think it is appropriate to say, "slightly different salinity". Even 1-3 PSU differences can be meaningful. Just report the actual differences.
Change to (Line 96): ".. and have consistently lower salinity (2-4 psu lower than ambient), …" 11) "The implementation and application of a combined δ11B-B/Ca, with B/Ca serving as a proxy for CO32-concentration and, subsequently, DICcf is a relatively new approach 29" I think reference 29 is a typo here because that study did not use B/Ca to determine [CO32-] or DIC.
Correct the reference and add the new references 12) Need to be consistent with "acclimation" and "acclimatization". Changed 13) Methods: "First seawater boron concentration was determined based on the site specific pKB assuming boron concentration at the site of calcification equals to seawater boron concentration." I assume the authors mean "First, seawater borate…".
Yes that is correct, we now just referred to the publication where the procedure and all the details are provided.

Reviewer #2 (Remarks to the Author):
Wall et al., have utilized δ11B and B/Ca measurements in Porites astreoides in order to derive the calcifying pHcf, DICcf and subsequent Ωcf. The idea was to understand the acclimation of corals in natural low Ωsw environment from species collected in the NMP at Puerto Morelos. In order to achieve that, they used 2 close locations with different environmental settings and Ωsw. Corals from both locations exhibit up-regulated pHcf, DICcf with more up-regulation at the low pH centers, however, Ωcf for both locations differed and the low pH locations presented a lower Ωcf compared to control. This difference is consistent with the lower calcification (~37%) observed at the low pH locations but this lower calcification is not fully explained by the decrease in Ωcf (model 44%). Due to the growing interest of the boron based proxies in corals, this paper adds a stone to the growing evidence of a dynamic calcifying fluid (pHcf, DICcf) in natural environments modulating the calcification response. This paper highlights the need for a better understanding of the symbiotic relationship of corals under different environments.
We thank the reviewer for an accurate summary of our work and major findings and for recognizing the importance and relevance of this paper.
1. It terms of analysis, it would be great to have some standards presented in the paper to give more confidence to the data, even if they look consistent.
A figure showing the analyses of standards is added to the supplemental material (Fig. S2).

Fig. S2: Repeated LA-MC-ICPMS analyses of pellets of coral standard Jcp-1 and boric acid standard NBS951.
Individual runs consisted of 40s background and 60s ablation data collection. All individual run data have been normalized to the mean of all NBS951 (0.0±0.4 ‰; 2se) data to allow for testing the reproducibility of both standards' measurements (Jcp-1: 24.0±0.4 ‰; 2se). Note: no drift correction has been applied. Repeated measurements of glass standard NIST610 vs. NBS951 pellet yielded a mean δ 11 B for NIST610 of 0.3±0.2 ‰ (2se, n=160).
Minor points: 2. Line 26: It is a bit confusing here. "whether these corals", may be change for "in what extent these corals have acclimated". This change has been made in the revised version of the manuscript.
3. Line 28: "increase the pH" may be change for "increase the pH and subsequent Ωcf" since Ωcf is likely the main trigger for calcification.
We changed it to (Line 26): "…addressing their ability to manipulate the carbonate chemistry of the calcifying fluid, including the pH (pH cf ), dissolved inorganic carbon (DIC cf ) concentration, and subsequently carbonate ion concentration (CO 3 2cf )." We used CO 3 2instead of Ω cf considering the first reviewers comment that more precisely with our method we characterized CO 3 2cf and not Ω cf . 4. Line 47: May be add Ries et al., 2009. We added the reference when we talk about (Line 47): "… that certain coral species were able to maintain high calcification rates or even benefit from elevated pCO 2 2,18-20 suggesting … ".
The paragraph focuses on pH cf upregulation capacity of corals and the d 11 B proxy and not the potential consequences such as changes in Ω cf . However, only a few sentence further below (now Line 65) when we start addressing the corals ability to up-regulate DIC cf we discuss the change in Ω cf . We adjusted this sentence to clarify that the upregulation includes both pH and DIC (Line 69): "Together -the potential to upregulate DIC cf and pH cf -allows for higher carbonate ion concentrations at the site of calcification and hence a higher Ω cf that facilitates calcification 32,23 ." 6. Line 91: They do not have similar temperature or salinity, 1-2 °C lower and 3psu lower in the "low pH centers" compared to centers. Do you have an idea of the seasonal variability? Can you test whether those differences could explain a part of your variability? Line 210: May be add a line saying that the temperature is not the main driver because people will look for this relationship, and that will reinforce your arguments.
The temperatures at the ojos are typically within 1 degree of the lagoon background temperatures and are cooler than the ambient water in summer and warmer in winter. This is because the groundwater has more constant temperatures year-round and the pre-discharge mixing with seawater in the subsurface results in these trends. If the temperature offsets are averaged over a year, then the differences are negligible (typically less than a degree cooler). We adjusted the sentence now in Line 93 to: "… from sites that have similar light conditions differ marginally in temperature (less than 1°C lower at the ojos averaged over all seasons with temperatures cooler than ambient in summer and slightly warmer in winter), have consistently lower salinity (2-4 psu lower than ambient), and are considerably different in Ω sw (supplementary table S1) 1,37 ." Hence, temperature is not likely to impact our data since our sampling integrated over a few years of growth. Moreover, as the first reviewer pointed us to recent new publications: At "low pH sites" the temperature is lower than at control sites, since a decrease in pH cf and DIC cf is observed with increasing temperature (Ross, Decarlo, & Mcculloch, 2019), temperature in our case can't explain the decrease of pH cf and DIC cf ." We now also explicitly mention this in the discussion in Line 308: "Environmental factors may also affect pH cf and DIC cf explaining some of the observed differences between the ojo and ambient corals at our study site. Studies have shown that a decrease in pH cf and DIC cf is associated with increasing temperature 44 , yet at our sites the temperatures at the ojos is actually lower, on average, than at control sites." 1 0 The salinity in contrast is indeed consistently lower by 2-4 psu (no seasonal trends). We note that both temperature and salinity were included in calculating the environmental saturation. Clearly since only 41% of the observed lower calcification can be explained by the decrease in Ω cf , it is possible that other factors also contribute, and the lower salinity could be one such factor. Having said that laboratory experiment with the same species at different pH but constant salinity show a decline in calcification (Albright & Langdon, 2011) as we observe, hence the role of salinity cannot be large. However, Ω cf covaries with salinity so it is hard to quantitatively evaluate the salinity effect separately. For an extended model (calcification ~ Ω cf + salinity) the AIC increased for the simplest model (calcification ~Ω cf ) of 1.74 to 1.80 and slightly failed to be significant for the extended (p=0.0695) suggesting Ω cf is the best predictor for the measured calcification response. Regardless we note that previous studies show that corals tended to be more tolerant to salinity than temperature stress (Berkelmans, Jones, & Schaffelke, 2012;Chui & Ang, 2017, and references therein). Although, different corals species may have different tolerance levels to reduced salinity but, in most cases, measured effects are seen only at salinities lower than < 26psu (Berkelmans et al., 2012;Chui & Ang, 2017;Kerswell & Jones, 2003), which we do not encounter in our study site. We discussed the salinity issue now in Line 312: "Salinity might also influence regulation processes, yet the measured average values (32.2 psu) as well as the salinity range measured (26-36 psu) 37 at the springs can be tolerated by corals and the long-term exposure to such conditions may have allowed them to develop mechanisms to better cope and adapt to this variable environment 65,66 ." 7. Line 98: Could the acclimation come from this high Ca content as well?
This is certainly possible, yet we see that our sentence here is a bit misleading. We mentioned that calcium is higher at the seeps when normalized to salinity. In total concentration -which is more relevant for coral calcification -calcium concentration at the ojos is in average similar between control and spring (even at the lower salinity of in average 32.2ppm for all ojo centers as well as the range of average salinity values of 31.1-33.9 ppm). Clearly this similar calcium concentration was not sufficient to mitigate the effect of the lower carbonate ion since the derived internal saturation conditions was still lower at the ojo's. We adjusted this part now in Line 101:" The spring water is characterized by lower pH, higher dissolved inorganic carbon (DIC), higher total alkalinity (TA) but similar calcium concentration compared to the ambient conditions away from the spring influence." We now also included a discussion on calcium regulation following the first reviewers comment that also relates to this questions (see point: 8., 12.).
8. Line 181: This result is central in your paper might be good to explain a bit more at line 439.
We now moved this part more towards the end of the discussion also ending the discussion referring to this result. We now state (Line 288): "The present study also indicates that the acclimation process in different corals encompassed some degree of flexibility in terms of the relative role of pH cf and DIC cf regulation in increasing the Ω cf , with some individuals compensating by adjusting their internal pH cf and others primarily by DIC cf modulation. This may also be true of the role of Ca 2+ cf upregulation. The relative amount, source and transportation pathways of DIC, H + and Ca 2+ to the site of calcification are still not fully understood 58 and transport processes may differ between individual corals." And close the discussion with (Line 318): "…Our geochemical model approach assumes steady-state equilibrium conditions; however, the rates of the various transport processes involved in regulating the chemistry of the calcifying fluid will ultimately dictate the calcification response 67 and these rates may differ 1 1 between individual coral genotypes further contributing to the offsets between the model output and observations." 9. Line 197: "thereby modulating Ωcf and calcification rate" We adjusted the sentence and added now in Line 184: "…thereby modulating CO 3 2cf , Ω cf and calcification rate." Also considering the first reviewers comment.
10. Line 217: I think this paragraph needs to move or towards the conclusion or in the introduction.
We now integrated this paragraph into our conclusion, ending now with (Line 325): "In this study we utilized a dual geochemical proxy approach (δ 11 B & B/Ca) to constrain calcifying fluid carbonate chemistry in P. astreoides corals that spent their entire life (decades) under acidified low Ω sw conditions. We found that at the pH cf for corals at the low Ω sw was slightly lower than at the ambient conditions indicating lack achievement of optimal calcification conditions. We also determined that pH cf and DIC cf are independently regulated corroborating the calcification response in P. astreoides. The study provides new insights into calcification responses of P. astreoides under changing environmental conditions and sheds light on the potential of corals to acclimate 33,47,52,68,69 . Using the geochemical proxies in combination with the bio-inorganic model brought forward by McCulloch and colleagues 33 we could explain 41% of the variability in coral growth rates along a Ω sw gradients. The variability which is not explained indicates that additional physiological and environmental processes contribute to the control of calcification rates in natural environments. This provides promising new avenues towards studying acclimation and adaptation potential of long-lived marine invertebrates such as corals." 11. Line 262: Is there any data for your corals?
Yes -symbiont densities were higher in corals at low saturation, but symbiont types did not differ. We now integrated this finding into our discussion in Line 296: "Corals at the ojos harbour a higher density of symbionts 59 that may potentially account for the higher energy demands for pH cf up-regulation resulting in the relatively small difference in the internal conditions (pH cf , DIC cf ) we see." 12. Line 265: So Ωcf explains 44% of the calcification rate variability, all the arguments here are likely to modulate the Ωcf but since it is only 44%, any ideas what are the independent (from the calcifying fluid) triggers for the 56% left? This is a very good point. We tried to improve our argumentation to make clear (which we did not do in an explicit way) why these parameters can also contribute to the unexplained variation. It is right that they on the one hand will also contribute to the steady-state internal conditions (thus, 41 % explained variability) that we derive from our geochemical data, yet these parameters may also modulate rates of transport processes and thus ultimately the calcification response. We now clearly stated in the revised version that we need to distinguish between these contributions. Another important independent factor that with our approach we could not address is internal calcium regulation. Calcium regulation is certainly another regulating parameter and thus, contribute to the unexplained variation. This argumentation and information is now given in the manuscript. In Line 275 we address the role of calcium regulations: "Recent studies identified internal calcium (Ca 2+ cf ) regulation as an additional player in coral calcification responses and emphasized that regulation of Ca 2+ cf can contribute to a corals' resistance to future ocean changes [42][43][44]54 . In this sense, the good agreement of our model with the observed calcification response potentially indicates that internal average steady-state calcium concentrations (Ca 2+ cf ) are lower at the ojos by some proportion that is related to the pH cf changes, since our model based on pH cf and CO 3 2cf can reproduce the observed calcification decline. This suggests a strong link between Ca 2+ cf and pH cf and supports the idea of a plasma-membrane Ca-ATPase 55,but see 56 responsible for pH cf regulation. However, it is possible that pH cf but also Ca 2+ cf was regulated by additional and/or different ion transport mechanisms (e.g. potentially ion exchangers, Ca 2+ -channels) 56,57 . " Now we moved the other potential contributor to the end of the discussion we close this part (Line 313): "Overall, these environmental and biological parameters may be responsible for the observed internal conditions but likely also affect rates of processes, and thus contribute to the unexplained component in our relation between the calcification and the geochemically derived DIC cf and pH cf ." 13. Line 284: Can you explain why it is controlled by the adjacent epithelium?
During revision of the manuscript, we integrated this part (on DIC concentration and that it lies a certain confined concentration range (Fig. S1) controlled by the adjacent epithelium) into another part (addressing DIC upregulation and the role of the external concentrations levels on the upregulation potential Line 251) changing not only the wording but also deepening the discussion. In general it has been shown that corals aboral epithelia is in tight association with the coral skeleton, and from immunolabelling experiments (Barott, Perez, Linsmayer, & Tresguerres, 2015;Zoccola et al., 2004) that ion-transport associated channels/enzymes/exchangers are located also within this epithelia and allowing the epithelia to regulate transport processes.
14. Line 290: May be add a sentence to conclude more smoothly.
We hope, we did this by restructuring and ending differently in the revised text.
15. Line 320: How did you measure your seawater samples?
Samples were first analyzed on ICP-MS Finnigan Element XR for boron concentration measurements following Krupinski et al. (2017) before they were further analyzed on a Neptune multi-collector inductively coupled mass spectrometer at National Cheng Kung University, Taiwan. The boric acid standard acid standard IAEA-B-1 was used as the reference standard (e.g. 39.77 ‰) to determine the δ 11 B of the samples, reproducibility (± 0.25‰) using a standard-sample-standard bracketing technique following Wang et al. (2010). We added this information now at Line 347: "Water samples were also taken for seawater boron concentrations (measured on a ICP-MS Finnigan Element XR following Krupinski and colleagues 71 ; ~430 ± 8 µM, with no difference between ojos and control) and a boron isotopic composition (δ 11 B sw ) of 39.15 (1sd = 0.12; n = 3) for the control site and 38.85 (1sd = 0.17; n = 5) for the low pH ojos. Boron isotopic samples were analysed on a Neptune multi-collector inductively coupled mass spectrometer at National Cheng Kung University, Taiwan, using the standard-sample-standard bracketing technique 72 . The boric acid standard IAEA-B-1 was used as the reference standard (e.g. 39.77 ‰) to determine the δ 11 B of the samples, reproducibility (± 0.25‰)." 16. Line 334: What are those modifications? 1 3 We modified the text so that it is clear what were the main modification now in Line 367: "Specifically, we used Multiplier and Faraday cups simultaneously to collect data for B 10 and B 11 (both on multiplier) as well as C 12 (Faraday cup). This allows us to derive B/C and δ 11 B from the same skeletal material. Similar to previous work the cones were cleaned on a regular basis (every 2-4 days). …" 17. Line 360: Is it due to the natural variability or to the method? Do you have any standards ran in liquid during your session?
The statement in this line refers to the occurrence of two different skeletal enteties in corals and that they are naturally variable. To account for this, we used this approach to make sure that in all corals we cover a similar proportion of both skeletal enteties. We added additional information to make this more clear (Line 389): "By this approach we expect to cover a representative sample set and minimize the natural variability in skeletal enteties and cover similar proportions in the different corals (assuming that COC to fibre ratio in coral grown under various environmental conditions stays constant)."