the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Feb 2026
Planktonic foraminifera Iodine/Calcium ratio: is it a proxy for dissolved oxygen in the ocean?
Abstract. Direct observations indicate a declining trend in ocean oxygen concentrations, which is not quantitatively captured by models. The complexity of oxygenation variability, linked to both physical and biochemical parameters, must be investigated across different climate contexts. Foraminiferal iodine-to-calcium (I/Ca) ratios have emerged as a proxy for subsurface oxygen concentrations although its capacity for quantitative reconstructions remains to be elucidated. We provide a new database, including the first results from the Mediterranean Sea and new samples in the Arabian Sea together with the parameters of biochemistry (nutrient concentration, pH, chlorophyll, oxygen, net primary productivity), physical and geographical (temperature, salinity, latitude, distance from the coast, water depths) and diagenesis potential (depth in core, sediment age) to better understand the proxy behaviour. Considering the strong spatiotemporal variability in dissolved oxygen in subsurface ocean, we propose to use statistically robust 25th percentile of oxygen concentration (p25 [O2]) instead of minimum concentration in the upper 500 m in the water column. Our results affirm that oxygen concentration is the primary driver of foraminiferal I/Ca and we propose a new calibration equation of foraminiferal I/Ca against p25 [O2]. Based on a new database, we identify a complex relationship between p25 [O2] and iodine speciation, which is one of the main sources of scatter. A comparison with synthetic calcite reveals that planktonic foraminiferal tests can incorporate either more or less iodate at high p25 [O2] than abiotic calcite, probably due to “vital effects”. The Mediterranean Sea samples present a wide range from low to high I/Ca (0.6 to 6.9 µmol/mol) in well-oxygenated water, which cannot be explained solely by authigenic calcite precipitation. Our results highlight the complex behaviour of the I/Ca proxy, while reinforcing its semi-quantitative reconstruction in palaeoceanography.
- Preprint
(2040 KB) - Metadata XML
-
Supplement
(7751 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-622', Dalton Hardisty, 25 Mar 2026
-
RC2: 'Comment on egusphere-2026-622', Anonymous Referee #2, 07 Apr 2026
I agree with Dalton that this review of the factors affecting I/Ca is a needed contribution for the development of this proxy. The factors the authors have chosen are extensive and appropriate. I appreciate the effort the authors have put into compiling a database with core-top I/Ca values and these factors, which will be very helpful moving forward, as well as their statistical approach. Though they are a relatively small part of the manuscript, I particularly appreciate the look at the effects of photosymbionts and authigenic overgrowths.
The following is a list of my concerns and suggestions.
Correction for reductive cleaning: The Barker et al. (2003) protocol (line 125) uses a reductive cleaning step, which Zhou, Hess et al. (2022) showed causes a ~30% lowering of I/Ca. Can the authors confirm whether they included that step? If so, a correction is needed to make measured values comparable to other core-top data used in the analyses, without which the statistical analyses and calibration can’t be accurate.
Factors affecting I/Ca aside from oxygenation: As Dalton noted in his review, it’s important to acknowledge that oxygen is not the only factor determining iodine speciation. While I agree that the spread in iodate concentrations at a given oxygen value likely leads to a lot of the spread in the I/Ca measurements (paragraph beginning line 353), other factors aside from oxygen concentrations of that water parcel also affect iodate concentrations (e.g., productivity, different rates of oxidation vs reduction, mixing). This concept is key to the manuscript and needs to be better explained/supported (see lines 86-87, 147-149, 347-352).
Proposed p25 calibration: The calibration of I/Ca to p25 instead of [O2]min is a major deviation from previous calibrations, and therefore needs to be explained in detail and well supported. The authors compare and contrast the relationship between iodate and these two oxygen parameters (Fig. 3), but the comparisons are not always apples-to-apples and the logical step from the iodate-O2 comparison to I/Ca could use more explanation. More specifically,…
- Conceptually, the use of [iodate]/[total iodine] rather than simply [iodate] needs more explanation/justification. If foraminifera incorporate iodate but not iodide, and iodine=iodate+iodide, how does the total iodine affect I/Ca? Importantly, Figure 2a and 2b use different y-axis parameters, making it impossible to compare the relationship between iodate and the two oxygenation parameters. What would 2a look like with [iodate] on the y-axis or 2b look like with [iodate]/[total iodine]?
- It seems that the reason p25 gives the best fit with I/Ca is driven very strongly by the new data from sites ODP977 and MD99-2341; in Figure S1’s p5 plot (the closest presented to the [O2]min which would be p0), it looks like those are 2 of the 3 sites with data that fall below the main data cloud at ~200 umol/kg O2. This is especially problematic if the 30% correction for reductive cleaning has not yet been applied. Aside from that, is there anything unusual about these sites?
- Hess et al. (2025) showed that I/Ca of different species varies with depth, with different patterns in OMZs and the open ocean: deeper dwellers have lower I/Ca values than shallow dwellers in areas with an OMZ (driven by iodate reduction below the forams in the OMZ), and the opposite in areas without an OMZ (driven by productivity iodate reduction in the surface). In Figure S4 and lines 330-331, we lose that pattern, which makes me think [O2]min makes more conceptual sense than p25.
- Figures 3 and 5 make a nice comparison between p25 and [O2]min. It would strengthen the story substantially to continue that comparison through to Figure 6 by adding a version using [O2]min.
- Importantly, with the proposed p25 calibration, I/Ca cannot be used to determine that an OMZ existed in the past, arguably the main utility of the proxy, because I/Ca values <1 umol/mol can occur at almost any p25 value. This would be an unfortunate development for the proxy, and I find I’m not yet convinced by the arguments for using p25.
Line-specific comments:
- Lines 15-17: Grammar check, it looks like there’s a shift from plural to singular
- Lines 81-83: This sentence should also reference Lu et al. (2020), who first suggested this for Atlantic sites. Have the sites that Lu et al. (2020) and Hess et al. (2025) excluded from their calibration plots also been excluded from the plots herein?
- Line 120: “of the two core-top studied” has a grammar problem. Possibly “core-tops studied”?
- Lines 157-158: Can the authors expand on what that iodine speciation data looks like, i.e., spatial resolution? In my experience, this type of data is typically geographically sparse and also has low vertical resolution, with possibly only a few points in the upper 500m. Can the authors explain whether they encountered these problems and, if so, how they addressed them?
- Figure 2:
- There are no dotted red lines, it seems the O2min line is solid. Is the p25 line actually not shown?
- Please be more specific about which p’s the gray lines represent.
- Lines 125-129: most of these species are mixed-layer calcifiers, but not all. Given the importance of calcification depth on I/Ca (Hess et al., 2025), the authors should include that information
- Figure 3:
- To make the comparisons in lines 244-246 and 248-249, the fit and prediction interval (red line and band in part a) need to also be shown for part b. However, the data in part b show a threshold, as has been widely described in the literature (most recently Wang et al., 2026), so of course they will have a poorer exponential fit. With this in mind, it doesn’t seem that these plots can show whether [O2]min or p25 is a better fit for I/Ca.
- To help us follow the description in lines 251-255, can the Mediterranean Sea values be differentiated somehow?
- Line 264: missing period.
- Line 276: What is the [O2]min at these sites?
- Lines 344-345: It seems like this should reference Figure S5, is that right?
- Lines 358-359: I’m not following the mathematical jump from the iodate plot to the I/Ca numbers presented here. If the authors are saying that the entirety of the I/Ca variability (I think that’s where the 1-5 umol/mol number comes from?) can be attributed to iodate variability, this is a very strong assertion and needs more support.
- Figure 5:
- The x-axes in parts a and b are different, and the data are plotted using those axes, but it is not possible to accurately plot both the red and blue line on both plots. The red line should be removed from part a and the blue line should be removed from part b.
- Figure 6:
- As with Figure 5, the line from Hess et al. (2025) can’t be shown on this figure, as the x-axis is for a different measurement.
- Line 414: “areas”
- Line 416-417: “we consider the linear regression is an appropriate approximation” A slight word of caution that “appropriate” is perhaps too strong here. Linear is certainly the simplest way to characterize this empirical relationship, but as we learn about the proxy, we may find another type of characterization is most appropriate. Of course, I leave this to the author’s discretion.
- Lines 421-423: It seems very unlikely that only the data points at moderate p25 and low I/Ca would be affected by vital effects. For this to be the case, I would expect the data to be from species only sampled in these data points, which I think is not the case? If the species from these samples have values that make sense in other parts of the curve, it seems unlikely it would be vital effects. Since many of the low I/Ca at moderate p25 points have lower [O2]min (as shown by gray lines in Fig. 5a), is another possibility that the p25 parameter is not the best way to understand the oxygenation effect on iodate concentration? To continue the compare-and-contrast that’s been done between [O2]min and p25 thus far in the manuscript, I wonder what Figure 6 would look like with [O2]min and the threshold value from Fig. 3b.
- Figure 7:
- I suggest making the black line red to match the same line in earlier figures.
- Caption describes the blue dashed lines as “grey dotted lines”
- Interestingly, if samples were reductively cleaned and this has not yet corrected, raising the I/Ca values 30% would place most of them above the 0% line.
- Line 436: Something is going on grammatically here. Perhaps “from the OMZ” or “from within OMZs”? I’m not sure which OMZ the authors are referring to in this paragraph and the next.
- Line 506: “at the surface”
- Line 509-510: I’m sorry, I’m not understanding the phrase “is hardly needed to distinguish provenance in the foraminiferal test”
- Line 528-530: Again, too much emphasis on the catchall “vital effects” and not enough information on the other factors that drive iodine speciation.
- Line 536-537: I see that the authors are saying that the Mediterranean data herein shows the proxy can be used in semi-enclosed basins, but the proxy can also be used in the open ocean. If the authors did not mean to imply it can’t perhaps a slight rephrasing would help clarify. Looking at the Mediterranean data in Figure 4, with Sites ODP977 and MD99-2341 having low I/Ca despite high p25, I wonder if we actually need a caveat for enclosed basins, that I/Ca values can be low in these settings despite high oxygen, something not observed for open-ocean sites.
Finally, I was curious to see which points are which in Figure 5a and tried to recreate it from the database, but after quite a while of trying I wasn’t able to. Can the authors point me to the column for [O2]min 0-500m that was used to make this plot? I suggest adding a legend for the columns, with more explanation.
Citation: https://doi.org/10.5194/egusphere-2026-622-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 267 | 102 | 15 | 384 | 32 | 43 | 71 |
- HTML: 267
- PDF: 102
- XML: 15
- Total: 384
- Supplement: 32
- BibTeX: 43
- EndNote: 71
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This is a much-needed examination of the controls on the foraminiferal I/Ca proxy. The new data are of high quality and so is the global I/Ca compilation and associated statistical analyses.
A few important considerations before publication:
1. Since 2026, 2 papers were published (Wang et al., 2026 and He et al., 2026) that are highly relevant to this topic. These papers should be integrated into the discussion and analyses of the paper at hand. In particular, this includes the impacts of oxycline depth and latitude on I/Ca.
2. The authors need to clarify under what circumstances they suggest the proxy is quantitative, if at all. My general reading of the authors text as well as my own impression from this paper and others in the field is that there are large limitations on the degree to which I/Ca can be used predict O2, but I don't think those limitations are clearly outlined. For example, several other papers identified I/Ca ranges linked to O2 ranges, but not explicit O2 values. It should be explained how this approach does or does not differ from these past more conservative approaches and provide a more prescriptive breakdown of when, where, and why I/Ca can or can’t be used quantitatively for O2, if at all. For example, Line 242: “The objective of this representation is not to find the best fit, but rather to determine whether the relationship between iodate proportion and dissolved oxygen content could explain the variation in foraminiferal I/Ca-based oxygen reconstruction.” Here the authors express that the I/Ca vs O2 equations are intended solely to explore potential relationships. However, I feel that this apparent intention is lost in the Discussion since it is not mentioned again although multiple equations are presented. I find this confusing and fear that readers will over extrapolate such equations to define O2 from I/Ca. Regardless of whether I correctly understand the authors intentions, clarifications should be presented appropriately in the discussion as a caution to readers for when and when not to apply I/Ca as quantitative for O2 reconstructions.
3. How would mixing impact the I/Ca proxy relationship with local O2? The IO3 vs O2 relationship has a major mixing component and so I/Ca is expected to as well in addition to vital effects, diagenesis, etc. See Cheng et al., 2024, which demonstrates the challenges of modeling iodate in the ocean with reduction and oxidation dependent on solely O2. See also the Wang et al., 2026 paper referenced above which shows mixing signals and oxycline depth. Mixing is not discussed as an impact on iodate in ODZs. See Moriyasu et al., 2020; Hardisty et al., 2021; Evans et al., 2020, which together make a case for mixing impacts on iodate in the Tropical North Pacific ODZ. This is also summarized in Cheng et al., 2024. This is because iodide oxidizes slowly, so low iodate signals can be mixed and integrated over large areas of ODZs without a clear local O2 dependency.
https://doi.org/10.5194/bg-21-4927-2024
https://doi.org/10.1016/j.epsl.2020.116676
https://doi.org/10.1029/2019GB006302
https://doi.org/10.1002/lno.11412
Detailed comments:
Figure S2: Wang et al., 2026 was published recently. It demonstrates a relationship between iodate and oxycline depth. Is the same relationship observed in the data here? I recognize that the Wang 2026 paper was published after the manuscript in hand was submitted, but that paper is highly relevant to this one, so should be incorporated into the discussion. https://doi.org/10.1016/j.gloplacha.2026.105412
Figure S2: He et al., 2026 also demonstrate a relationship between I/Ca and latitude that mimics the same relationship seen for IO3 and latitude. Does the dataset assembled here also reveal that pattern? https://doi.org/10.1038/s41561-025-01896-w
Line 351: “This mismatch between oxygen and iodate concentration may at least partly explain the apparent presence of iodate in waters here defined as oxygen-depleted.” Is it not the other way around: i.e., that the presence of iodate in low O2 waters explains the mismatch between O2 and IO3?
Figure 4: Why are data from samples younger than 8kyr used here? Elsewhere 2kyr is used. It seems difficult to expect a comparison between I/Ca from samples from last 8kyr and contemporary water column O2.
Figure 1: Do the color of the symbols on the map in part A marking locations correspond to I/Ca legend? Their colors overlap but the values all seem very low and thus inconsistent with the subsequent graphs of I/Ca. I’m wondering if the maximum values on the color scale are too large thus giving the impression of uniformly low I/Ca? If this is the case, adjust the color scale range so the spread in I/Ca is more clearly visualized.
Line 395 and Figure 6: Zhou et al., 2014 also demonstrated a temperature dependency between I/Ca and iodate in medium (see their supplement). It looks like this was utilized but a uniform temp of 19 deg was used, though the color bar shows a large range of temperatures. It would be helpful to use the iodate concentrations and temperature from each site at the depth of p25(O2) to plot the synthetic calcite predicted I/Ca value using the temperature dependent I/Ca equations from Zhou et al., 2014 and plot these against the actual measured I/Ca values. This could be done for each site and demonstrate the relationships, if any. This could aide in interpreting deviations from “ideal”.
Line 408: The highly elevated values of I/Ca >6 µmol/mol are not necessarily specific to foraminiferal calcite. It is noteworthy that similarly elevated values are observed for core-top aragonite (see Hardisty et al., 2017 data from Great Bahamas Bank), which may reflect partition coefficient differences for aragonite or other environmental influences. Also see Lu et al., 2022 for a similar discussion from benthic forams.
https://doi.org/10.1016/j.gca.2022.06.001 and https://doi.org/10.1016/j.epsl.2017.01.032
Figures: If the authors intend to present I/Ca as quantitative for O2 under any circumstance, it would be useful to make another map like Figure 1A showing marine O2 (showing p25O2 in this case) but with the fill of the circles marking sites shaded with the predicted O2 based on the I/Ca measured at this site. This would provide a global visualization not apparent in the other figures for which regions follow the predicted trend line and which do not.