Review of the paper « Bottom water oxygenation changes in the Southwestern Indian Ocean as an indicator for enhanced respired carbon storage since the last glacial inception» by Helen Eri Amsler, Lena M. Thöle, Ingrid Stimac, Walter Geibert, Minoru Ikehara, Gerhard Kuhn, Oliver Esper and Samuel L. Jaccard, submitted to Climate of the Past.

The authors present new records of redox-sensitive elements to reconstruct bottom water oxygenation changes from the last glacial inception to the Holocene. These records were obtained on a North South transect of marine sediment cores in the western Indian sector of the Southern Ocean. The authors provide records of exported biogenic silica in the same cores to determine whether the bottom water oxygen changes are linked to increased organic carbon sedimentation or circulation changes. These data therefore provide important information concerning the mechanisms involved in the air-sea partitioning since the last glacial inception.

The paper is thus within the scope of “Climate of the Past” and could be of great interest for the community.

However part of the methods needs to be clearly explained.

The weakest part of the paper concerns the age models. For the sub-Antarctic core DCR-1PC the age model has been established in a previous paper (Crosta et al., 2020). However, it is necessary for the reader to see figures with

• the depth of the 14C dates and the tie-points on the aU, Mn/Ti and opal records to see where are the chronological constraints (could be added on figure 4 but it would be nice to see also the records that have been tuned)
• a depth/age plot.

In fact, the 14C dates presented in Crosta et al. 2020 for this core probably indicates a hiatus of ~5kyr between 33 and 41cm, that would roughly correspond to isotopic stage 2. This possibility should be discussed when considering this time period.

For the other cores, the dating strategy is not explained. Why correlating the core signals to benthic LR04-stack, while the sub-Antarctic core age model has been established by tuning with EPICA Dome C deuterium record? If there is a scientific reason to link the magnetic susceptibility records and the LR04-stack that have been aligned together, it has not been explained. Comparing the same/similar records of two neighbouring marine cores does not need a long explanation but any other tuning between various records requires at least a short explanation of the underlying assumptions.

The introduction is well written and the lines 59 to 69 clearly present the goal of this study. However it is disappointing to have a very simplified presentation of the role of iron in the Southern Ocean. This study concerns the Indian sector of the Southern Ocean, not the Atlantic sector and dust is probably not the major source of iron at the cores locations (Tagliabue et al., 2017, 2014 and reference therein) at any time of the last glacial cycle.

For bottom water oxygenation proxies, the authors indicate that they considered two different ²³⁸U/²³²Th ratio, 0.5 for cores within CDW with a large NADW component and 0.27 for the deeper and southern core PS2603-3 influenced by AABW and thus Antarctic continental crust. Within the discussion, the authors consider changes in the deep Southern Ocean circulation during the last climatic cycle, with shoaling of the NADW influence (Govin et al., 2009 should be cited for the circulation changes within the Indian sector of the Southern Ocean during the glacial inception). The authors should thus consider a possible decrease of the ²³⁸U/²³²Th ratio for the shallower cores during the glacial stage. It might not change significantly their results but it would be nice that they indicate the corresponding uncertainty.

Other questions and minor corrections are indicated with the manuscript line numbers in the following part.

Change Sigman et al., 2020 to Sigman et al., 2021

All the figures have a 2 before their number, to be suppressed.

Line 210 to 214, aU do not peaks at peak glacial conditions but at the transition to termination 1

Line 254 to 270: the authors could also consider the possible hiatus in the core with a missing isotopic stage 2.

Line 286: I do not understand the sentence: in the Polar frontal zone the nutrient availability was reduced compared to interglacial period but the nutrient availability is always higher in the Polar frontal zone than closer to the Subantarctic front. Again consider also a possible hiatus, as indicated by 14C data.

Line 306 “alternative” , n missing

Line 315 Is it the sampling resolution or the uncertainty of the age models that precludes to assess the potential time lag between cores?

Line 316 to 325 the increase in aU seems to be at the beginning of the Holocene not during the deglaciation, as well as the opal peak in the PS2603-3. Do the authors consider a possible 5kyr error on the age scale at that time? We really need to see the records that were tuned to benthic LR04 record or EPICA Dome C deuterium and the tie points considered.

This manuscript presents authigenic uranium (aU) concentrations, biogenic silica (bSi) concentrations and Mn/Ti elemental ratios in bulk sediments obtained with various analytical methods for five cores between 46ºS and 59ºS in the Southwest Indian Ocean. Mn/Ti profile was not obtained for the southernmost core. Three of the five cores provide 120 ka records, one core extends only 40 ka, and another one covers the past 180 ka.

The major conclusion is the important role of physical processes to oceanic carbon storage during cold periods due to reduced ventilation. The biological productivity is considered as a second factor. The link is proposed between Southern Ocean carbon storage and atmospheric CO2 concentration changes on glacial/interglacial timescales.

The strong points of the present study are i) the reconstruction of both oxygenation state and biological production inferred from sedimentary opal content and ii) latitudinal transect covering different frontal zones in the Indian sector of the Southern Ocean. The authors discussed various possibility affecting the aU, bSi and Mn/Ti records. They are careful but they did not explicitly provide their specific objectives and working hypothesis. Consequently, the present manuscript gives impression “just confirming the previous studies”. I will develop my major concerns below.

1. Too general objective and poor description of original finding

The major role of ventilation changes to oceanic carbon storage on glacial/interglacial timescales has been already reported by number of studies. What is the focus of the present study? Why are the authors interested in changes since the last glacial inception? Why the transect in the Southwest Indian? If the role of the Indian sector of the Southern Ocean is the primary motivation of the present work, introduction should be focused on state-of-art of the study region.

The discussion is qualitative and only confirms the observation of the previous studies. The authors are careful to interpret the obtained records considering different possibilities (ex. diagenetic burn-down that could modify aU records). But it is difficult to identify robust reconstruction and new insight supported by their own results. Also, there is no figure comparing the new results with previously obtained records except dD and pCO2 to discuss the processes that the authors proposed.

Reorganization of the manuscript with addition of discussion figures will be useful to identify targeted objective and working hypothesis to emphasize original aspect of the present study.

2. Lack of demonstration about age model quality

The authors described the age model in section 2.2 and Table 1 but the provided information is insufficient. Since sedimentation rate is a factor affecting the accumulation of authigenic U (Figure 4), more extended explanation is necessary with figures. For instance, it is helpful to show 14C dating levels and tie points of each core. Which size of reservoir age was applied? Which 14C calibration equation was used? The magnetic susceptibility (MagSus) records of PS2609-1 and PS260606 were tuned to LR04-stack. What is the hypothesis to relate MagSus to the benthic d18O stack? Were XRF data (Fe, Si, Ti, Ca) used to correlate between PS2609-1 and PS260606? The authors also used alignment of XRF Ti intensity and Ca/Ti intensity ratio of PS2606-6 with the EPICA Dome C dust record. What is the size of age offset based on the tuning to LR04 and to EPICA Dome C dust record? Concerning core PS2603-3, MagSus, XRF data (which elements?) and biogenic silica were graphically aligned to the LR04 reference curve. Did the authors assume that the changes are synchronous? Why? Overall, what is the size of uncertainty of age model of each core?

3. Estimation of authigenic uranium (aU) concentration

aU is estimated assuming a constant 238U/232Th that is variable with sites. Even if generally consistent aU trend is observed for the study cores on glacial/interglacial timescale, absolute aU is relatively small, often less than 3ppm except core DCR-1PC. Moreover, detrital U contribution might have changed on glacial/interglacial timescales. It will be useful to present figures comparing 238U/232Th activity ratio with aU concentration profile of each core to demonstrate potential influence of detrital 238U/232Th activity ratio on aU variability.

At last, this study used different analytical procedures to obtain the same parameter (aU, Mn/Ti and bSi) for the different cores. The consistency of the results is mentioned but it is not shown how the comparison was realized: some selected common samples were analyzed or common standards were regularly measured? Some more detail will strengthen the manuscript.

I recommend to accept this manuscript after major revision.

Minor / specific comments

Abstract last sentence (lines 23-24), “These records highlight… insufficiently documented role the southern Indian Ocean played in the air-sea partitioning of CO2 on glacial-interglacial timescales”. It is unclear how this statement is extracted from the results obtained in this study.

Line 25, “exogenic carbon cycle”. Please define this term.

Lines 26 and 41, “Sigman and Boyle, 2000”. The reference is missing in the reference list.

Line 67, “underrepresented Indian sector of the Southern Ocean”. It will be helpful to add the state of art about bottom water oxygenation state in the Indian sector to clarify unsolved issues. Such description will better define the objective of the present study.

Lines 71-84, “2.1 Core locations and material”. Add the description of the present-day water masses occupying the core locations.

Line 88, “neogloboquadrina” should be “Neogloboquadrina”.

Line 135, about Mn and Ti measurements. To avoid any confusion, indicate from the beginning, XRF scanning or ICP-MS measurement realized for different cores. Also, it is necessary to mention that Mn/Ti record was not obtained for core PS2603-3.

Line 184, “millennial-scale oscillations’. What is the temporal resolution of aU record? Considering the possibility of aU remobilization, is it appropriate to treat millennial-scale variability, in particular for the interval of low sedimentation rate such as MIS 5 (Figure 4a)?

Lines 189-191, “detritic values”. Mn/Ti variability of core DCR-1PC is estimated by XRF intensity ratios that are not converted to concentration. How did the authors know the background level corresponds to detrital values?

Lines 194-197, about glacial-interglacial trend of bSi for core DCR-1PC. Caution should be paid because the expected glacial high bSi value is not observed for MIS 2.

Line 207. Add “aU” between “Sedimentary” and “concentrations”.

Line 209, “a pronounced increase in sedimentary aU concentration during MIS 4”. This sentence should be revised because the description is true for PS2609-1 but not for PS2606-6 that shows a modest increase (Figure 3b).

Lines 213-214, “The highest aU…a gradual increase from about 30 ka, peaking during the LGM”. Core PS2603-3 does not show the described trend because no clear peak is identified (Figure 3d). Please revise the text.

Line 228. Delete “which seems to higher CO2 levels during MIS 5”. This is result section, thus premature to compare with pCO2 record.

Line 233. Add “inside of the sediments” after “at the sediment-water interface”.

Line 237, “the proxies broadly agree”. What does this sentence mean? The proxies follow an expected trend? If so, what is the hypothesis to expect some trend?

Line 240. Add “of core DCR-1PC” after “values”. It is unclear why the observed Mn/Ti trend can be treated as “a regional increase in carbon export and sequestration”.

Lines 292-293, “broadly similar to the SAZ record”. I don’t see the similarity because the SAZ core (DCR-1PC) is characterized by aU maximum during MIS 3 that is totally absent for the SAZ cores.

Line 293, “noisy Mn/Ti signal”. In general, the authors did not provide temporal resolution of different parameters for different cores. The mentioned “noisy signal” of COR-1bPC was possibly related to high-resolution XRF scanning.

Lines 302-303, “COR-1bPC was closest to the most vigorous upwelling location”. Is this statement enough robust? The bSi concentration of COR-1bPC is high but comparable with bSi at PS2606-6 considering different temporal resolution.

Line 306, “alterative” should be “alternative”.

Lines 327-333, about the deep or bottom water masses. This part should be placed in section 2.1. The present-day water masses (AABW, upper CDW and lower CDW) should be shown in Figure 1b. How did the author distinguish the water masses? Using a T-S plot?

Line 340. Add “and in pore water” after “interface”.

Line 372. Delete “XRF peak” since some Mn/Ti data were obtained using ICP-MS.

Numbering of the figures and the table should be corrected since the number always contains “2”.

Figure 1. (a) right panel. “AAZ” should be replaced by “AZ”. Show the position of transect indicated Figure 1b. (b) Indicate the present-day water masses.

Figures 2 and 3. Combine the two figures like Figure 4 to facilitate comparison between all study cores and avoid presenting atmospheric CO2 and dD twice. Indicate the latitude and water depth of each core.