We are deeply grateful of this review by Dalton Hardisty and his careful reading. These comments will be very helpful to improve the manuscript.

RC1.1. About the papers published in 2026 (Wang et al. 2026; He et al. 2026).

AC1.1. We agree that these two papers must be cited in the manuscript, and considered for statistical analysis and discussion. We will add the oxycline depth and oxycline oxygen value to the statistical analysis. The two papers will be cited, and the oxycline role will be discussed.

RC1.2. About the circumstances in which the proxy is quantitative, and the different approach of this paper.

AC1.2. We agree that our message of the limitations of the quantitative use of the proxy may not be clear enough, leading to potential confusion. Our statistical analysis of a larger dataset confirmed the major role of dissolved oxygen on foraminiferal I/Ca values. However, it is very difficult to determine precisely when, where, and why I/Ca can or cannot be used quantitatively for [O₂].

The main difference in our approach compared to the previous studies is our use of p25[O₂], which has a more robust statistical definition. As a result, the data points are more evenly spread along the x-axis (Fig. AC1.S2). We examined the influence of the scattered relationship between the proportion of iodate and the concentration of dissolved oxygen (Fig. 3a and 3b), as well as the effect of seawater temperature on iodate incorporation, assuming the relationships obtained for abiotic calcite (referred “modelled” below) are valid for foraminiferal calcite. A comparison of the data and modelled values shows that most of the field data are within the expected range, even though some data are above or below the modelled I/Ca ratio (Fig. AC1.S2b and Fig. AC1.S2d). Data with a positive anomaly are observed in the Atlantic Ocean, but it is unclear whether this observation is real or biased by spatially heterogeneous sampling. Further multi regression analysis did not allow us to identify additional factors affecting foraminiferal I/Ca ratios (Fig. S6), making it very difficult to quantify the limits of using I/Ca as a quantitative proxy.

We will add this precision in the last discussion section and in the conclusion to summarise our position on the proxy: a semi-quantitative proxy to highlight oxygenation change at subsurface.

RC1.3. About the role of mixing on I/Ca in relation with local O₂.

AC1.3. It is true that mixing has a major impact on inorganic iodine speciation in the ocean, partially because of the difference in oxidoreduction timing. This could produce low iodate concentration in highly oxygenated water, which results in low I/Ca. However, the mixing solely cannot explain the observed dispersion. For example, despite of the low foraminiferal I/Ca in the Mediterranean Sea, no evidence of oxygen deficient water has been observed. The lowest reported oxygen value in the Alboran Sea is about 170 µmol/kg (Packard et al., 1988). Considering the challenge of modelling inorganic iodine speciation depending on oxygen, we have used direct measures in the 0–500 m layer. This approach is expected to partly account for the mixing, which could be responsible for an important part of the scattering, together with primary productivity iodate uptake.

We will add explanations about mixing and primary productivity and how it is thought to be partially accounted for in the inorganic iodine speciation model throughout the manuscript.

RC1.4. About Figure S2: Relationship between iodate and oxycline depth.

AC1.4. We agree with this suggestion. To better discuss the influence of oxycline, we will add oxycline depth and oxygen value at oxycline to the tested parameters influencing foraminiferal I/Ca. We will add the information in the discussion.

RC1.5. About Figure S2: Relationship between I/Ca and latitude.

AC1.5. We have also found a pattern between foraminiferal I/Ca and oxygen (Figure S5). However, it is possible that this pattern is related to the spatial distribution of data points. Most OMZs investigated with I/Ca are found at low latitudes, and oxygen depletion would trigger iodate reduction. Similar pattern is also found in p25[O₂] as a function of latitude (Figure S2). Moreover, a temperature effect has been shown in iodine incorporation in abiotic calcite (Zhou et al., 2014). Thus, it is difficult to correlate directly the influence of latitude to I/Ca, as other parameters co-vary with it. We will mention this in the new version of the manuscript in section 2.4.

RC1.6. About the sentence “This mismatch between oxygen and iodate concentration may at least partly explain the apparent presence of iodate in waters here defined as oxygen-depleted”.

AC1.6. We agree that we meant a potential presence of iodide was possible in oxygenated water. This sentence will be corrected.

RC1.7. About Figure 4 and the use of samples younger than 8 ka and not 2 ka.

AC1.7. We agree that foraminiferal I/Ca ratios dated at 8 ka do not reflect the current oxygen concentration. This selection is to examine the spatial coverage of the data at this stage. Figure 4 highlights the new data from the Mediterranean Sea and the Arabian Sea. Some samples were dated around 6 ka BP. We think it is important to show high I/Ca in the Mediterranean Sea.

RC1.8. About Figure 1 and the colour bar.

AC1.8. The colour of the symbols in Figure 1a and 1b correspond to the I/Ca colour scale in the legend. We choose a linear colour scale maxed at 10 µmol/mol because all core-tops I/Ca are found in this range (0–10 µmol/mol). We will add ticks to the colour bar to clearly present a linear scale.

We will also use corrected I/Ca to take into account reductively cleaned samples (+30% as suggested by reviewer#2).

RC1.9. About the iodate concentrations and temperature from each site to obtain the predicted abiotic I/Ca and Figure 6.

AC1.9. Due to scarce inorganic iodine speciation, it is not possible to define a concentration in iodate at foraminiferal I/Ca sites. This is why we choose to represent foraminiferal I/Ca data and a prediction model of abiotic calcite based on inorganic iodine speciation in different oxygen contexts. The model considers different inorganic iodine speciation, distribution, and the equation of I/Ca depending on iodate concentration at given temperature from Zhou et al. (2014). The colour scale of foraminiferal I/Ca data is the mean temperature in the upper 500 m 0.25°x0.25° around sampling site. The objective was to see if a pattern in temperature was visible with more/less iodine incorporated when temperature is low/high. But no pattern is visible.

As you mentioned, it is possible to compare in another way, by modelling expected abiotic I/Ca at given iodate concentration and temperature. A linear interpolation for temperatures between 6, 19 and 33 °C is needed. Extrapolation of equation for temperatures below/above 6/33 °C may be too uncertain. The range 6–33 °C offers the ability to compare a lot of points in various oceanographic contexts.

By considering iodate concentration and temperature, we model the predicted abiotic I/Ca independently of foraminiferal I/Ca dataset. Then, we plot the expected abiotic I/Ca as a function of p25[O₂] (Fig. AC1.S1).

The result is similar to the reconstructed range in Figure 6, with I/Ca values between 0 and 6 µmol/mol.

As the conclusions do not differ and the results are very close to the modelled abiotic I/Ca from Figure 5, we don’t think the addition of this figure is necessary.

RC1.10. About the highly elevated values of I/Ca >6 µmol/mol in aragonitic core-tops.

AC1.10. We agree that aragonite-rich sediment may have a higher I/Ca. In the new data presented in this study, the highest ratio is found in SL95 core (6.86 µmol/mol). SL95 is located in the Gulf of Sirte, area with high aragonite concentration due to the presence of Halimeda (Reitz et De Lange 2006). Remaining aragonite needles after cleaning may result in higher I/Ca. We investigate the influence of the presence of aragonite with binary mixing model between aragonite and foraminifera. We consider an I/Ca of 10 µmol/mol in aragonite, similar as the ratio measured in scleractinian corals (Sun et al. 2023). For foraminiferal I/Ca, we consider an expected value of 4 µmol/mol from the new calibration. Aragonite contribution would need to be around 40% to account for an I/Ca between 6 and 7 µmol/mol. Sr/Ca of the mixture would be 4.8 mmol/mol considering a ratio of 10 mmol/mol in aragonite. Whereas the Sr/Ca from SL95 core-top is 1.345 mmol/mol. The contribution of aragonite alone cannot explain the measured value. We will add this point and will develop argument of possible iodine source in the discussion.

RC1.11. About the figures if I/Ca is considered to be quantitative.

AC1.11. We do not intend to present foraminiferal I/Ca as a quantitative proxy for oxygen as explained above. A new Figure S8 (visible with Fig. AC1.S2) will be added for the visualisation of global trend.

AC1.12. We have also updated the statistical analysis. In the first version, plankton tow samples where integrated in the dataset for statistical analysis. It has been removed. Now, only the corrected I/Ca from core-tops samples is used.

References:

He, Ruliang, Alexandre Pohl, Xingliang Zhang, et al. 2026. « A Reversed Latitudinal Ocean Oxygen Gradient in the Proterozoic Eon ». Nature Geoscience 19 (3): 325‑30. https://doi.org/10.1038/s41561-025-01896-w.

Reitz, Anja, et Gert J. De Lange. 2006. « Abundant Sr-Rich Aragonite in Eastern Mediterranean Sapropel S1: Diagenetic vs. Detrital/Biogenic Origin ». Palaeogeography, Palaeoclimatology, Palaeoecology 235 (1‑3): 135‑48. https://doi.org/10.1016/j.palaeo.2005.10.024.

Sun, Yun-Ju, Laura F. Robinson, Ian J. Parkinson, et al. 2023. « Iodine-to-Calcium Ratios in Deep-Sea Scleractinian and Bamboo Corals ». Frontiers in Marine Science 10 (novembre): 1264380. https://doi.org/10.3389/fmars.2023.1264380.

Wang, Xubin, Rosalind E. M. Rickaby, et Zunli Lu. 2026. « A Deep Dive into the Planktic Foraminiferal I/Ca in Global Core-Tops ». Global and Planetary Change, mars, 105412. https://doi.org/10.1016/j.gloplacha.2026.105412.

At first, we are deeply grateful to the anonymous reviewer for the numerous insightful comments and careful reading. These comments are very helpful in improving the manuscript.

RC2.1. About the correction for reductive cleaning.

AC2.1. We agree that the correction method was not explicitly mentioned when it should have. The new samples from this study followed the short cleaning protocol from Barker et al. (2003), which does not contain a reductive cleaning. This will be clarified in the manuscript.

Correction of + 30% will be applied to samples cleaned with Reductive + Oxidative steps and, a column in the dataset will summarise the cleaning method used for the sample. This update does not change the main message of this work but slightly changes the equations. We will update all the figures, equations, and statistical analysis.

RC2.2. About the factors affecting I/Ca aside from oxygenation.

AC2.2. We agree with the reviewer. Factors other than oxygen do influence inorganic iodine speciation. And, if partially accounted for in the inorganic iodine speciation model, it should be mentioned in the text. Our model may not fully cover the range of possible inorganic iodine speciation in different oceanic contexts. This will be mentioned in greater detail in results, discussion and conclusions sections.

RC2.3. About the proposed p25 calibration.

AC2.3. We have proposed p25 calibration considering the difficulty to define minimum oxygen concentration (Lu et al. 2020).  The definition of minimum oxygen concentration can be influenced by the dataset used. This, in turn, is affected by sampling seasons, potential measurement issues, and database management. During our database creation, we encountered several inaccurate oxygen values. We aim at creating I/Ca dataset with the oxygen as reliable as possible to eliminate outliers. We will add more interpretation around this to improve the clarity of the manuscript.

RC2.4. About Figure 3 and the use of [iodate]/[total inorganic iodine].

AC2.4. Inorganic iodine (iodate + iodide) concentration is not constant in the ocean. So, a higher total inorganic iodine concentration may lead to higher iodate concentration, and so, to a higher I/Ca in oxygenated water. This effect is tested by using ± 1s range for the total inorganic iodine concentration as shown in Figure 3c in the present version of the manuscript.

As the main message of the figure is to present a different concept to remove the influence of total inorganic iodine concentration and not a comparison between min[O₂] and p25[O₂], we think it is not necessary to modify the figure. However, a sentence will be added in section 2.2 to guide the reader and not see Fig. 3 as a comparison of p25 and min[O₂].

RC2.5. About the fit potentially derived by new data from ODP Site 977 and MD99-2341.

AC2.5. To answer this question, we examined the cleaning-corrected I/Ca but, excluding our Mediterranean and Gulf of Cadix data. p25[O₂] still gives the best fit, with data from V14-70 site in Southern Benguela (Lu et al., 2020a) ending up in the same domain as low Mediterranean I/Ca. This result suggests that the aforementioned data is not responsible for the fit.

RC2.6. About the difference of I/Ca with calcification depth.

AC2.6. Depth dwelling of different species is an important assumption in paleo-studies. Figure AC2.S1 shows our examination of the dataset by ocean, presenting Atlantic, Pacific and Indian oceans as a function of both p25 and min[O₂] to evaluate the influence of the definition of dissolved oxygen (p25 [O₂] or min[O₂]). We only present core-top values dated at younger than 2 ka to be coherent with the other figures.

The pattern of foraminiferal I/Ca with calcification depths is visible at moderately oxygenated sites (100 < p25 < 180 µmol/kg) in the Pacific Ocean. With min[O₂], this pattern is not clearly observed at this oxygen range because of the differences with p25[O₂]. The pattern is visible at low oxygen concentrations (0 < min[O₂] < 50 µmol/kg).

We propose to use figure AC2.S1 presenting ocean sub-datasets as a function of p25 instead of the global Figure S4 presented in the first version of the manuscript. We will update the text accordingly.

RC2.7. About the comparison of p25 and min[O₂] with figures 3 and 5.

AC2.7. We agree that the comparison is interesting. While the comparison is important, we think it should not be in the main manuscript as it could lead to the misinterpretation that low I/Ca in high oxygen concentration are isolated, which is not the case. As this is a dataset issue, we propose to add the figure AC2.S2 in the supplementary. We will add the comparison with min[O₂] as Fig. S7 and fix the text accordingly.

RC2.8. About the p25[O₂] calibration and the determination of the existence of an OMZ in the past.

AC2.8. We think the proxy may still be used to determine the presence of an OMZ in the past. However, we argue that low I/Ca may occur in oxygenated place, and the absolute I/Ca value alone cannot be interpreted. The trend of the I/Ca, however, may be used at a site to capture the change in oxygenation dynamic in the past.

RC2.9. About shift from plural to singular lines 15-17 and issue line 120.

AC2.9. Grammar shift from plural to singular lines 15-17 and issue “of the two core-top studied” line 120 will be checked.

RC2.10. About the exclusion of sites from the calibration.

AC2.10. To our understanding, Lu et al. (2020) did not exclude data. In their work, they showed that using oxygen values from a 0.25°x0.25° area around sampling position helped to account for local oxygen variability (instead of using the closest oxygen profile). Hess et al. (2025) explicitly mentioned excluded values. They argued that shallow water sites (<~500 m water depth) near the southern tip of Africa from Lu et al. (2020) are prone to important local oxygen variability. They also excluded Site RR1313 2MC (545 m water depth) for the same reason. We choose not to exclude any samples, as these sites with variability over small spatial scales can help identify specific oxygen or iodine behaviour.

RC2.11. About spatial resolution of the iodine speciation dataset.

AC2.11. Inorganic iodine speciation data are sparse, as shown by Winkelbauer et al., (2023). We are using the same dataset, but with different treatment. We first select data shallower than 500 mbsl. We then only select data with iodide and iodate measures. We select total inorganic iodine in the ±1σ interval to avoid heavily iodine depleted/enriched areas. Lastly, we remove the Bermudas inshore data as they originate from a highly restricted area. We are left with 1692 points and oxygen statistics were recovered for n = 1551 points. The majority of the measures are close to surface as shown by the z-axis of Figure 3a and 3b (mean depth = 113 m, median depth = 75 m). Mean p25[O₂] of the subset is 152 µmol/kg with standard deviation of 95 µmol/kg. Latitudes from the subset span from 70°S to 70°N. These statistics ensure a good coverage of different oxygenation contexts. Figure AC2.S3 presents the map of the subset. Data from the Indian Ocean were requested and will be present in the new version of the manuscript.

RC2.12. About Figure 2 and missing dotted red lines and the percentiles represented.

AC2.12. We will fix the figure and indicate the percentiles represented in the description.

RC2.13. About the importance of calcification depth on I/Ca and the mention of layer of calcification.

AC2.13. We acknowledge that not all species used in this study are mixed-layer calcifiers. Depth-dwelling layers will be specified in the methods section.

RC2.14. About Figure 3 and the comparison between p25[O₂] and min[O₂]

AC2.14. We would like to clarify that the primary objective of Figure 3a and 3b is not to directly compare the statistical performance of p25 and min[O₂], but rather to illustrate two different representation of inorganic iodine behaviour. Figure 3b is the commonly described threshold behaviour of iodate reduction at low oxygen concentrations. Figure 3a is the more continuous relationship based on inorganic iodine speciation.

We agree that these two plots do not allow us to compare whether there is a best fit between p25/min[O₂] and inorganic iodine speciation as the y-axis are different.

The visible threshold in Figure 3b only allows iodate reduction in very low oxygen concentration (around 10-20 µmol/kg) which is highly incompatible with a linear calibration curve.

We will correct the sentences for the comparison between Fig. 3a and 3b in the results section.

RC2.15. About Mediterranean Sea values in Figure 3a and 3b.

AC2.15. We thank the reviewer for this suggestion, the figure will highlight the Mediterranean Sea values.

RC2.16. About min[O₂] at specific sites.

AC2.16. In the Gulf of Cadix, minimum oxygen is 190 µmol/kg, and 168 µmol/kg for Alboran Sea. This will be specified in section 2.3.

RC2.17. About missing reference of Figure S5 lines 344-345.

AC2.17. Yes, Table 2 and Figure S5 are reporting similar results, with Figure S5 being the visual representation of Table 2. We will also mention Figure S5 here.

RC2.18. About the mathematical jump for iodate to abiotic I/Ca.

AC2.18. We will add explanations of the mathematical and geochemical reasoning in the revised manuscript section 3.3.

RC2.19. About Figure 5 and the regression lines.

AC2.19. We agree, the line not related to the x-axis will be removed.

RC2.20. About Figure 6 and the regression lines.

AC2.20. We agree, it will be removed.

RC2.21. About the linear regression being an “appropriate” or “simplest” assumption.

AC2.21. We agree, linear regression is a reasonable choice here, until a justification for another regression type is found. We will fix the sentence.

RC2.22. About vital effect only affecting points outside of the 95% prediction interval (lines 421-423) and the utility of p25[O₂].

AC2.22. In our point of view, the data points at moderate p25 and low I/Ca are not the only one affected by potential vital effects. We were thinking of vital effect as a deviation from the abiotic integration of iodine. This deviation is highlighted in this part of the figure, as the others are within the 95% prediction interval, and we cannot differentiate vital effect (iodine incorporation) from iodate variability (water column linked process).

The problem we encountered with min[O₂] is that it is not representative of what is happening in the area, it may sometimes be adequate or representative of oxygen depletion, but it is also an extreme value. This may be caused by extreme event in a specific area or by an error in the oxygen dataset. We agree that I/Ca is closely related to oxygen depletion through inorganic iodine speciation. But in order to model the I/Ca, we needed a statistically reliable variable to compare the two approaches.

RC2.23. About recommendations for Figure 7.

AC2.23. We agree that a red line will help follow the point. The figure and caption will be fixed.

RC2.24. About raising by 30% the Mediterranean samples, placing them above 0% line in Fig. 7.

AC2.24. Yes, it is true. However, the short cleaning used didn’t integrate reductive cleaning. Another parameter may be responsible for this deviation.

RC2.25. About “from the OMZ” grammar issue line 436.

AC2.25. We will fix the sentence with “from within OMZs”. We were not referring to a specific OMZ.

RC2.26. About the term “provenance in the foraminiferal test”.

AC2.26. We were referring to the possible contrasts of iodine concentration in the calcite lattice. Differences may exist in chambers, and so, on the development stage of the foraminifera. It could also be a reason for the deviation observed. It will be more explicit in the discussion, section 3.5.

RC2.27. About emphasis on the vital effect, not enough on iodine speciation.

AC2.27. We say that remaining dispersion may result from vital effect or inorganic iodine speciation, and more details will be added in the discussion, sections 3.3, 3.5 and in the conclusion.

RC2.28. About the use of the proxy in semi-enclosed basins.

AC2.28. We agree that we did not intend to say it was only applicable in semi-enclosed basins. We meant it was also possible to use the proxy in semi-enclosed basins. We will rephrase the corresponding sentence in the conclusion.

RC2.29. About the dataset and recreation of the figures.

AC2.29. Figure 5a was using minimum_Oxygen column. We agree that the provided table could be more user-friendly, and that a note will help to understand all parameters. We will provide which column is used in which figure together with the description of the column.

AC2.30. We have also updated the statistical analysis. In the first version, plankton tow samples where integrated in the dataset for statistical analysis. It has been removed. Now, only the corrected I/Ca from core-tops samples is used.

References:

Chance, Rosie, Keith Weston, Alex R. Baker, et al. 2010. « Seasonal and Interannual Variation of Dissolved Iodine Speciation at a Coastal Antarctic Site ». Marine Chemistry 118 (3‑4): 171‑81. https://doi.org/10.1016/j.marchem.2009.11.009.

Hess, Anya V., Yair Rosenthal, Xiaoli Zhou, et Kaixuan Bu. 2025. « The I/Ca Paleo-Oxygenation Proxy in Planktonic Foraminifera: A Multispecies Core-Top Calibration ». Geochimica et Cosmochimica Acta, janvier, S0016703725000298. https://doi.org/10.1016/j.gca.2025.01.018.

Lu, Wanyi, Alexander J. Dickson, Ellen Thomas, Rosalind E. M. Rickaby, Piers Chapman, et Zunli Lu. 2020. « Refining the Planktic Foraminiferal I/Ca Proxy: Results from the Southeast Atlantic Ocean ». Geochimica et Cosmochimica Acta 287 (octobre): 318‑27. https://doi.org/10.1016/j.gca.2019.10.025.

Luther, George W. 2023. « Review on the Physical Chemistry of Iodine Transformations in the Oceans ». Frontiers in Marine Science 10 (février): 1085618. https://doi.org/10.3389/fmars.2023.1085618.