the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2025
Past Ocean surface density from planktonic foraminifera calcite δ18O
Abstract. Density of seawater is a critical property that controls ocean dynamics. Previous works suggest the use of the δ18O calcite of foraminifera as a potential proxy for paleodensity. However, potential quantitative reconstructions were limited to the tropical and subtropical surface ocean and without an explicit estimate of the uncertainty in calibration model parameters. We developed the use of the δ18Oc of planktonic foraminifera as a surface paleodensity proxy for the whole ocean using Bayesian regression models calibrated to annual surface density. Predictive performance of the models improves when we account for inter-species specific differences.
We investigate the additional uncertainties that could be introduced by potential evolution of the δ18Oc-density relationship with time (from the last glacial maximum (LGM) to the preindustrial (PI)) through the combination of past isotope enabled climate model simulations and a foraminiferal growth module. We demonstrate that additional uncertainties are weak globally, except for the Nordic Seas region.
We applied our Bayesian regression model to LGM and Late Holocene (LH) δ18Oc foraminifera databases to reconstruct annual surface density during these periods. We observe stronger LGM density value changes at low latitudes compared to mid latitudes. These results will be used to evaluate numerical climate models in their ability to simulate ocean surface density during the extreme climatic period of the LGM.
The new calibration has great potential to be applied to other past periods and to reconstruct the past temporal evolution of ocean surface density.
- Preprint
(3235 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 13 Sep 2025)
-
RC1: 'Comment on egusphere-2025-2459', Anonymous Referee #1, 11 Jul 2025
reply
Caley et al. calibrate planktic δ18O from core tops to surface density and assess the uncertainty via Bayesian modelling. The authors test the calibration with the results of isotope enabled models and apply the method to foraminiferal δ18O values from the Last Glacial Maximum and the late Holocene. The attempt to translate δ18O directly into density is certainly worthwhile since we have probably an order of magnitude more δ18O data available compared to combined δ18O/temperature reconstructions. However, the paper has methodological and transparency issues that need to be addressed.
1. Representation of mean ocean density. There are surface density changes related to local/regional SST and SSS changes and mean ocean density changes related to ocean volume. Part of the local/regional density changes will be related to mean ocean salinity due to volume changes with sea level. For example, Duplessy et al. (1991) estimated that the smaller LGM ocean volume led to ~1 psu higher salinity (and hence a significantly higher global ocean LGM density). In principle, foraminiferal δ18O contains information on sea level via the ice effect (albeit with a higher slope as usual evaporation -precipitation changes). The δ18O ice effect, however, is removed before the LGM density reconstruction. Can the LGM reconstructions really reflect absolute density or have the authors rather reconstructed the density changes due to local/regional changes in SST and SSS? Perhaps I am missing something here, but in my view the mean ocean density changes corresponding to mean ocean salinity changes due to ice/ocean volume changes have to be added to the LGM values since the used foraminiferal isotope data do not contain this information and the method does not account for it. This point requires further clarification.
2. Uncertainty of density reconstruction. With respect to the previous point, the uncertainty in the density reconstruction due to ocean volume changes should be implemented into the error analysis. With respect to the uncertainty of δ18O seawater reconstructions (line 72-75), I recommend citing the work of Schmidt (1999), because this author provides a reliable error estimate for δ18O sea water reconstructions. Also, I miss an assessment if the total error of the method is small enough to distinguish glacial densities from the modern ones, particularly if the global warming bias in the modern reference data is considered.
3. Ice effect correction. The authors cite an ice effect correction of either 1.0 (line 365) or 1.05 (line 421). Please clarify why different numbers have been used or correct. I also find it appropriate to cite Labeyrie et al. (1987) (see their Fig. 5) in this context, as they for the first time provided robust evidence for an ice effect on the order of 1 o/oo.
4. Transparency. The documentation of the data sources is not sufficient. For the LGM compilation the authors mention „additional data“ which are not specified anywhere in the paper. In the „Code and data availability“ section, it is stated that „The additional LGM and δ18Oc dataset will be available as a supplement“. Unfortunately, the supplement is not available to me. The „Obligations to authors“ states that „ A paper should contain sufficient detail and references to public sources of information to permit the author's peers to replicate the work.“ As a reviewer, I am unable to replicate the work because neither the data/supplement nor all the sources are available to me. Also, „Copernicus Publications requests depositing data that correspond to journal articles in reliable (public) data repositories, assigning digital object identifiers, and properly citing data sets as individual contributions“ (from https://www.climate-of-the-past.net/policies/data_policy.html). I strongly suggest that the authors adhere to this policy.
5. Transport of foraminiferal shells with currents. Currents can transport foraminiferal shells and the isotope signals they carry over relatively large distances. Based on typical current speeds, it can be estimated that planktic foraminifera may be transported several degrees latitude within their lifetime. While one can argue that the effects will be minimal because the ambient water mass is transported with the shells, discrepancies between recorded δ18O and calculated δ18O may occur if foraminifera/water masses are subducted, if water masses are mixed or in the vicinity of fronts, with the filaments from upwelling regions or close to freshwater plumes. I suggest that the authors consider and discuss shell transport/expatriation in addition to seasonality and vertical migration.
6. Abstract. „We developed the use of the δ18Oc of planktonic foraminifera as a surface paleodensity proxy for the whole ocean...“. As the authors show obviously not for the Nordic Seas and hence not for the global ocean surface.
7. Effect of mixing/bioturbation (line 91-95). The paper by Köhler and Mulitza (2024) mainly deals with the detection of the carbon ion effect, not with bioturbation. Bioturbation will have a significant effect on most core tops used in this study. At typical mixed layer depths of 5-10 cm, deglacial/glacial material will be mixed with the Holocene layer below a sedimentation rate threshold of about 2 cm/kyr (see for example Broecker, 1986), mid-Holocene material (including monsoonal related salinity/density changes) at even higher sedimentation rates. For most of the MARGO core tops, there seams to be a weak stratigraphic control.
8. Global warming in modern hydrography. The fact that all core top calibrations are affected by global warming (line 140) is not a good justification for its use. There might be products like the World Ocean Atlas that integrate over longer time periods and therefore contain less global warming signals. This issue should at least be discussed, since global ocean warming approaches the magnitude of the deglacial warming and the bias can be considerable.
9. Stability of the δ18Ow salinity relationship. The authors have tested the stability of the δ18O/salinity relationships with the results of model simulations for the LGM. I find the choice of the time slice not ideal. In the tropics and subtropics (the majority of the ocean area), the strongest precipitation changes (and hence changes in surface δ18O and salinity) occur in the early to mid-Holocene with the strengthening of the Monsoon (see for example Weldeab et al. 2007). This is the time when I would expect changes in δ18O of the freshwater endmember for example due to the amount effect and hence a potential instability of the δ18O/Salinity relation. The authors should have access to isotope-enabled model runs representing the mid-Holocene (e.g., Shi et al. 2023, co-authored by M. Werner).
10. Direct comparison to modelled LGM density. Why has the LGM foraminiferal-based density reconstruction not directly been compared to modelled LGM density? The models are considered good enough to test the stability of the δ18O salinity relation, why are they not good enough to compare with the density reconstruction directly?
Broecker, W. S.: Oxygen Isotope Constraints on Surface Ocean Temperatures, Quat. res., 26, 121–134, https://doi.org/10.1016/0033-5894(86)90087-6, 1986.
Duplessy, J.-C., Labeyrie, L., Anne, Maitre, F., Duprat, J., and Sarnthein, M.: Surface salinity reconstruction of the North Atlantic Ocean during the LGM, Oceanologica Acta, 14, 311–324, 1991.
Labeyrie, L. D., Duplessy, J. C., and Blanc, P. L.: Variations in mode of formation and temperature of oceanic deep waters over the past 125,000 years, Nature, 327, 477–482, https://doi.org/10.1038/327477a0, 1987.
Köhler, P. and Mulitza, S.: No detectable influence of the carbonate ion effect on changes in stable carbon isotope ratios (δ13C) of shallow dwelling planktic foraminifera over the past 160 kyr, Clim. Past, 20, 991–1015, https://doi.org/10.5194/cp-20-991-2024, 2024.
Schmidt, G. A.: Error analysis of paleosalinity calculations, Paleoceanography, 14, 422–429, https://doi.org/10.1029/1999PA900008, 1999.
Shi, X., Cauquoin, A., Lohmann, G., Jonkers, L., Wang, Q., Yang, H., Sun, Y., and Werner, M.: Simulated stable water isotopes during the mid-Holocene and pre-industrial periods using AWI-ESM-2.1-wiso, Geosci. Model Dev., 16, 5153–5178, https://doi.org/10.5194/gmd-16-5153-2023, 2023.
Weldeab, S., Lea, D. W., Schneider, R. R., and Andersen, N.: 155,000 years of West African monsoon and ocean thermal evolution, Science (New York, N.Y.), 316, 1303–1307, https://doi.org/10.1126/science.1140461, 2007.
Citation: https://doi.org/10.5194/egusphere-2025-2459-RC1 -
RC2: 'Comment on egusphere-2025-2459', Anonymous Referee #2, 22 Aug 2025
reply
General comment
Caley et al. investigate the use of planktonic foramifera δ18Oc as a surface paleodensity proxy for the whole ocean. For that, the authors applied three Bayesian regression models on δ18Oc datasets to reconstruct surface paleodensity for the late Holocene (LH) and the Last Glacial Maximum (LGM). Using isotope-enabled models of different complexities, Caley et al. investigated the additional uncertainties that are introduced by the potential evolution of the δ18Oc-density relationship with time (i.e., from LGM to LH). Except for the Nordic Seas, the authors demonstrated that additional uncertainties are weak globally (for a LGM to LH climate change).
The objectives and the method of this study correspond to the scope of CP. The study is easy to follow, and I took pleasure to read it. I have some minor comments related to the datasets, the evaluation, and the comparison with LGM results.
Major comments
- The description of the LGM d18O dataset lacks details, especially, on the additional data from more recent studies (lines 130-131), which are not available with the paper (or I missed them). For the revised paper, I suggest the reviewer to provide the compilation of all the data (δ18O for LH and LGM + ocean datasets) they used for this study, with the appropriate references inside.
- For the evaluation of the residuals under LH climate (Figure 1), why is there a like a threshold in observed data at a value of 28. Is it a problem with the data? I think it should be discussed because it gives the largest density residuals. Moreover, the authors do not discuss the strongest residuals in the Mediterranean Sea, which are probably influenced by bias in net freshwater fluxes and thermoaline circulation. The authors could use this recent study from Ayache et al. (2024, https://doi.org/10.5194/gmd-17-6627-2024).
- Lines 328-329: I suggest to try with the yearly δ18Osw values from ECHAM5/MPI-OM to really know the effect of seasonality on calculated d18Oc. With the current comparison, it be cannot excluded that the differences between the results using ECHAM5/MPI-OM and iLOVECLIM is due to lower resolution of iLOVECLIM or other missing/biased processed in this lower resolution model.
- For the evaluation for LGM results, the residuals in the Nordic Seas are stronger with ECHAM5/MPI-OM than with iLOVECLIM (Figure 5). This point should be discussed more in details by the authors. Moreover, I would like to see some evaluation of the reconstructed density anomalies between LGM and LH (Figure 7a). Are there other reconstructions? Or can the authors compare those results with modeled LGM-LH surface densities?
- I would like the authors to put into context their results regarding other climate periods. The authors state that additional uncertainties due to the evolution of the δ18Oc-density relationship with time are globally weak (lines 45-46). However, this is true, except for the Nordic Seas, for a LGM-to-LH change. It has not been proven for another period, such as the Last Interglacial (110-130 ka). Considering mid-Holocene period (6 ka) raised by the reviewer #1, the changes in δ18O of seawater are rather small (+0.5‰ maximum, only, in the western Pacific Ocean according to Shi et al., 2023 and Cauquoin et al., 2019) compared to the LGM ones.
- Generally, the units are missing in the labels of figures’ axes. Please check all the figures. Also, the panel labels are used on one complete column or row in Figures 1, 3, 5, 6 or are absent for Figure 4. Please add a letter label for each panel in the figures.
Specific comments
- Lines 74-75: to quantify past ocean density and dynamics.
- Lines 148: give the units for δ18Oc (and relative to which standard) and ρ (sigma-theta relative to a density of 1029 kg/m3?).
- Section 2.4.1: specify that iLOVECLIM is an intermediate-complexity model, whereas ECHAM5/MPI-OM is an Earth System Model.
- Figure 1: This is for LH period I suppose?
- Lines 235: explain a bit more that is ELPD.
- Figure 3: give the p-values.
- Figure 4: Only for LH period?
- Row (a) of Figure 5: the legend for the LGM values is not clear.
- Line 421: 1 or 1.05‰?
- Lines 455-456: By applying a Bayesian regression hierarchical model to LGM and LH δ18Oc foraminifera databases, we reconstructed LGM and LH annual surface density and found stronger LGM density…
References
Ayache, M., Dutay, J.-C., Mouchet, A., Tachikawa, K., Risi, C., and Ramstein, G.: Modelling the water isotope distribution in the Mediterranean Sea using a high-resolution oceanic model (NEMO-MED12-watiso v1.0): evaluation of model results against in situ observations, Geosci. Model Dev., 17, 6627–6655, https://doi.org/10.5194/gmd-17-6627-2024, 2024.
Cauquoin, A., Werner, M., and Lohmann, G.: Water isotopes – climate relationships for the mid-Holocene and preindustrial period simulated with an isotope-enabled version of MPI-ESM, Clim. Past, 15, 1913–1937, https://doi.org/10.5194/cp-15-1913-2019, 2019.
Shi, X., Cauquoin, A., Lohmann, G., Jonkers, L., Wang, Q., Yang, H., Sun, Y., and Werner, M.: Simulated stable water isotopes during the mid-Holocene and pre-industrial periods using AWI-ESM-2.1-wiso, Geosci. Model Dev., 16, 5153–5178, https://doi.org/10.5194/gmd-16-5153-2023, 2023.
Citation: https://doi.org/10.5194/egusphere-2025-2459-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 270 | 38 | 15 | 323 | 14 | 25 |
- HTML: 270
- PDF: 38
- XML: 15
- Total: 323
- BibTeX: 14
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1