Review of Karancz et al., “Glacial-interglacial contrasts in the marine inorganic carbon chemistry of the Benguela Upwelling System,” submitted to Climate of the Past, by Jesse Farmer

Karancz and colleagues present a reconstruction of ocean carbonate chemistry and its proximal drivers in the Benguela Upwelling System (BUS) covering the last ~27 kyr based on a combination box-piston core from the Walvis Ridge. I commend the authors for applying a wide range of analytically challenging proxies in their attempt to understand past carbonate chemistry in the BUS. The manuscript is overall well written, especially the detailed and extensive methods, and the top-line interpretation/conclusion of increased intermediate ocean carbon storage based on the difference between alkenone ε_P and boron-based pCO₂ proxies is quite novel.

For the bad news: I found several deficiencies that temper my enthusiasm for publication. Given that I think some major rewrites (and possibly additional data) will be needed, I recommend major revisions that allow for sufficient time to address this. I want to emphasize that, given the quality and quantity of data here already, I fully expect a revised paper would be worthy of publication in Climate of the Past.

Major comments.

1. Core stratigraphy. I found the combined stratigraphy from BC6 and PC8 to be lacking. If I read this correctly, the authors took the length of BC6 (41.5 cm) and appended PC8 to the bottom to create a composite depth scale of 141.5 cm. While this is probably in the general ballpark given the radiocarbon dates, this approach is far too simplified. The authors would have to be incredibly lucky to have it such that the total recovery of the box core (41.5 cm) just happened to align with the total amount of lost material from the top of the piston core! (In my experience, this has never once happened).

To address this, the authors should use physical properties measurements on the two cores to align them in depth space and create a composite depth scale. This can be any measurable property – bulk density, reflectance/color, XRF, etc. Once they have this data, they can then align the two cores based on these properties and create a single depth-age model in rBacon using the ¹⁴C dates vs. composite depth.

I don’t ask for new data lightly in what is already an impressive multiproxy work. But there is very little other than ¹⁴C (and then, only six of the eight shown dates) to benchmark this chronology. What else exists does not inspire much confidence – the benthic δ¹⁸O record (Figure 4b) is effectively missing the entire deglacial section and so is of minimal utility. And I do not see at all the assignment of millennial-scale deglacial events in Figure 6 (it should be noted that G. bulloides δ¹⁸O does not look like EDC). To this end, significant age model refinement will be needed if the authors wish to discuss millennial-scale features in their records during the deglaciation.

1. Presentation of alkenone ε_P-derived pCO₂. Keeping in mind my affinity toward boron isotope approaches, the presentation of alkenone ε_P-derived pCO₂ stuck me as rather outdated. Namely, the approach outlined in the methods Section 3.8 does not acknowledge any of the vigorous discussion around the feasibility of this approach or its potential limitations. (Note that the authors do start to tackle this much further down on L651-678). I think the authors need to be more up front about what is known regarding the limitations of this proxy and how they sought to address these. Do haptophyte CCMs matter in an upwelling region? Is Ba/Ca actually a functional phosphate proxy in an upwelling region, where high OM remineralization rates can lead to BaSO₄ precipitation? Even if that problem is overcome, is Ba/Ca-derived PO₄ by itself enough of a constraint on the physiology of the alkenone producing community to address these issues?

Ultimately, it does matter that the authors’ ε_P-derived pCO₂ is in reasonable agreement with ice core pCO₂, and the authors could use this to argue that these recognized proxy complications may not be as significant in this particular setting. But the complications must be addressed head-on, and the reasoning for the authors’ choices on calculation of pCO₂ should be made explicit to the reader.

1. B isotope results, S/Mg and calculation of pCO₂. There are a few issues here:
   • G. bulloides size fractions and their δ¹¹B. Numerous studies indicate that different size fractions of planktic foraminifera possess different B isotope ratios (Hönisch and Hemming, 2004; Henehan et al., 2013, 2016). Although I do not think this has been demonstrated explicitly in G. bulloides, the community tends to work in quite limited size fractions when measuring B isotopes in G. bulloides: 300-355 µm (Martinez-Boti et al., 2015) or 315-355 µm (Raitzsch et al., 2018, cited in the manuscript). In this study, however, size fractions are not constrained; it appears the authors used specimens from 150 to 425 µm (Section 3.1). This adds an additional source of uncertainty that should be propagated into uncertainty on the reconstructed pH (L285), and it may be quite significant (in excess of 1‰, Henehan et al. 2013 Figure 6).
   • Poor precision in some G. bulloides δ¹¹B replicates. Looking at Figure 5a, there are a few datapoints where the replicate precision on the sample is > ±0.5‰. Boron is a tricky isotope system to measure, so I wonder if there was some plasma instability during these measurements. But regardless, the authors may wish to exclude these data as they don’t appear sufficiently well-constrained to be useful for calculating pH or pCO₂.
   • Using carbonate ion alongside pH to constrain the carbonate system. If I follow the authors’ approach correctly, they seek to employ G. bulloides S/Mg or B/Ca as a separate constraint on the carbonate system to address uncertainty in paleoalkalinity (L680 and supplement). Unfortunately, this does not work well for a reason not included in the study: carbonate ion and pH strongly covary in the modern ocean ranges of alkalinity and DIC (see Figure 10 in Rae et al., 2011). For this reason, error propagation alone is not sufficient; the true uncertainty in the carbon system parameters derived using pH from B isotopes and carbonate ion from S/Mg, B/Ca would need to account for covariance of these parameters. Given this, I think it would be better just to remove this text/approach.
2. δ¹³C gradients and estimation of BCP. In section 5.2, the authors do a pretty good job of laying out the complications to using δ¹³C gradients as a proxy for the biological carbon pump (see also section 4.4 in Farmer et al., 2021). After laying out these significant complications, they state on L611-615 “a larger difference between planktonic and benthic foraminiferal δ¹³C values during the LGM compared to the Holocene is evident (Fig. 6 c; and Supplementary Fig. S3), suggesting a more efficient BCP”. After re-reading this section, I do not agree with this conclusion. The data could more simply be explained as a change to lower δ¹³C in AAIW source waters due to inefficient air-sea CO₂ exchange in a seasonally sea-ice covered Southern Ocean. To remove this influence, the authors could difference their benthic δ¹³C record from one of similar depth in the South Atlantic not under the influence of Benguela Upwelling.
3. Impact of Benguela Upwelling on atmospheric pCO₂. The manuscript sort of kicks around this idea that changes in upwelling intensity and/or its carbon content might have altered atmospheric pCO₂. In the biological carbon pump/preformed nutrient content view of ocean CO₂ uptake/release (e.g., Sigman et al. 2010), though, whether or not the BUS was a CO₂ source or sink would have minimal impact on global atmospheric CO₂ This is because any excess upwelled nutrients in the region would then be advected to regions where they would be consumed. That is, even a high rate of local CO₂ outgassing at the core site due to upwelling > productivity would be offset by adjacent regions, where productivity > upwelling and CO₂ uptake would occur. Put another way, the only places where upwelling has the capacity to alter atmospheric CO₂ is in regions where that upwelling adds nutrients to newly formed deep water, either around Antarctica or in the high latitude Northern Hemisphere. In all other regions, local imbalances are evened out spatially.

Minor comment (just one for now). Suggest changing the title to “Contrasts in the marine inorganic carbon chemistry of the Benguela Upwelling System since the Last Glacial Maximum”. “Glacial-interglacial” implies that there are multiple data realizations of glacial intervals and interglacial intervals, so I found myself surprised when the data only went back to 27 ka.

References:

Badger, M.P., Chalk, T.B., Foster, G.L., Bown, P.R., Gibbs, S.J., Sexton, P.F., Schmidt, D.N., Pälike, H., Mackensen, A. and Pancost, R.D., 2019. Insensitivity of alkenone carbon isotopes to atmospheric CO2 at low to moderate CO2 levels. Climate of the Past, 15(2), 539-554.

Cenozoic CO2 Proxy Integration Project (CenCO2PIP) Consortium*†, 2023. Toward a Cenozoic history of atmospheric CO₂. Science, 382(6675), eadi5177.

Farmer, J. R., Hertzberg, J. E., Cardinal, D., Fietz, S., Hendry, K., Jaccard, S. L., et al. (2021). Assessment of C, N, and Si isotopes as tracers of past ocean nutrient and carbon cycling. Global Biogeochemical Cycles, 35, e2020GB006775.

Henehan, M.J., Rae, J.W., Foster, G.L., Erez, J., Prentice, K.C., Kucera, M., Bostock, H.C., Martínez-Botí, M.A., Milton, J.A., Wilson, P.A. and Marshall, B.J., 2013. Calibration of the boron isotope proxy in the planktonic foraminifera Globigerinoides ruber for use in palaeo-CO2 reconstruction. Earth and Planetary Science Letters, 364, 111-122.

Henehan, M.J., Foster, G.L., Bostock, H.C., Greenop, R., Marshall, B.J. and Wilson, P.A., 2016. A new boron isotope-pH calibration for Orbulina universa, with implications for understanding and accounting for ‘vital effects’. Earth and Planetary Science Letters, 454, 282-292.

Hönisch, B. and Hemming, N.G., 2004. Ground‐truthing the boron isotope‐paleo‐pH proxy in planktonic foraminifera shells: Partial dissolution and shell size effects. Paleoceanography, 19(4).

Martínez-Botí, M.A., Marino, G., Foster, G.L., Ziveri, P., Henehan, M.J., Rae, J.W., Mortyn, P.G. and Vance, D., 2015. Boron isotope evidence for oceanic carbon dioxide leakage during the last deglaciation. Nature, 518(7538), 219-222.

Phelps, S.R., Hennon, G.M., Dyhrman, S.T., Hernández Limón, M.D., Williamson, O.M. and Polissar, P.J., 2021. Carbon isotope fractionation in Noelaerhabdaceae algae in culture and a critical evaluation of the alkenone paleobarometer. Geochemistry, Geophysics, Geosystems, 22(7), e2021GC009657.

Rae, J.W., Foster, G.L., Schmidt, D.N. and Elliott, T., 2011. Boron isotopes and B/Ca in benthic foraminifera: Proxies for the deep ocean carbonate system. Earth and Planetary Science Letters, 302(3-4), 403-413.

Sigman, D.M., Hain, M.P. and Haug, G.H., 2010. The polar ocean and glacial cycles in atmospheric CO2 concentration. Nature, 466(7302), 47-55.

Karancz et al. reconstruct past ocean conditions for the past ~27 cal ky BP using a suite of geochemical proxies from a box and piston core collected from the south flank of Walvis Ridge in the South Atlantic. Multiple techniques are applied to reconstruct seawater temperature and pCO2, representing a major analytical effort by the author team.

There are some issues with the manuscript that should be addressed prior to publication, described in more detail below.

1. There is an age reversal in the box core that complicates interpretation of the Holocene portions of all the geochemical records (~5-10 cal ky BP). The authors do not discuss the age reversal, keeping it in the age model (Fig. 3a), and suggesting a sedimentation rate of 0.01 cm/yr, or “Alternatively, the upper 10 cm bsf have constant ages due to bioturbation.” This is not a clear or satisfying discussion of the observed radiocarbon ages in the box core. On the figure, there is also a note “winnowing (?)” with no discussion of winnowing in the main text. In addition, if the aim of the study is to gain insight into millennial scale events using the geochemical records from this core, then it could be worth investing in tighter age control through more radiocarbon dates per depth interval (e.g., at least 1 age control point per 2 ky; Guilderson et al., 2021). At a minimum, the box core’s chronology is uncertain and the basis for splicing the box and piston core has not been clearly described (as noted by reviewer 1 in the previous comment).
2. After going through the major effort of generating several complementary geochemical records, it is surprising that the authors do not synthesize the data to the full extent possible. The most prominent missed opportunity is that, if I understand correctly, the authors apply a uniform temperature correction across the whole G. bulloides carbon isotope record, from ~5 to 27 cal ky BP. Why not use the available Mg/Ca derived temperatures and the established temperature dependence of d13C for G. bulloides to get the most accurate values possible? There is a temperature-corrected curve in the supplement (Fig. S3), but it looks the same as the gray curve in main text Fig. 6C, which was created by applying a constant offset, so it appears not to have been created using the variable, reconstructed temperature estimates (it would be nice if the Fig. S3 caption could be updated to clarify that). There is also a carbonate-ion corrected curve in the supplement (Fig. S3), but it is not clear how the paleo carbonate ion values were determined or why this correction was not also applied to the record that was ultimately compared with benthic isotopes to assess changes in the soft tissue pump in the main text.
3. In the discussion, the use of “intermediate” is confusing, sometimes being applied to G. bulloides records (which can live anywhere from the shallow surface down to a couple hundred meters) and sometimes being applied to benthic foraminiferal records. The cores were collected at 1375m water depth, which lies within the modern extent of AAIW in the region, so discussing the benthic foraminiferal records in the context of AAIW is reasonable and is also supported by prior studies. However, use of “intermediate” to describe the G. bulloides records is misleading, since there is no evidence that this species inhabits AAIW. Instead, G. bulloides tends to live very shallowly during the active upwelling season and extend more deeply in other seasons (e.g., Peeters & Brummer, 2002). Karancz et al. suggest, based on Mg/Ca-derived temperatures, that their Holocene bulloides are living at ~100-150m (~line 515), and this is shallower than the uppermost extent of AAIW. I’d strongly recommend using another word (like “subsurface”) to describe the G. bulloides record, and reserve use of “intermediate” for AAIW alone.
4. Considering the wide range of possible habitat depths (and seasonal variability) of G. bulloides, comparing G. bulloides d11B-derived pCO2 and alkenone-based pCO2 is not a robust approach for reconstructing an intermediate-to-surface pCO2 gradient.
5. The discussion would be strengthened by providing specific references and paleorecords when comparing their results with prior work. There are multiple instances (line 530 “changes known for the Southern Hemisphere”; line 500 “… general Southern Hemisphere temperature record”) where paleoclimate patterns are alluded to but not cited or described specifically. Changes were not uniform across the SH during the last deglaciation and therefore it is essential that the authors are very explicit about what they are referring to.

The data presented here represent an enormous amount of work, and they do have the potential to make a useful contribution to our understanding of paleoceanography in the South Atlantic. I hope that these comments are useful to the authors.

References:

Guilderson, T. P., Allen, K., Landers, J. P., Ettwein, V. J., & Cook, M. S. (2021). Can We Better Constrain the Timing of GNAIW/UNADW Variability in the Western Equatorial Atlantic and Its Relationship to Climate Change During the Last Deglaciation? Paleoceanography and Paleoclimatology, 36(8). https://doi.org/10.1029/2020pa004187

Peeters, F. J. C., & Brummer, G.-J. A. (2002). The seasonal and vertical distribution of living planktic foraminifera in the NW Arabian Sea. Geological Society, London, Special Publications, 195(1), 463–497. https://doi.org/10.1144/gsl.sp.2002.195.01.26