Overview

The manuscript titled, “Low Cobalt Inventories in the Amundsen and Ross Seas Driven by High Demand for Labile Cobalt Uptake Among Native Phytoplankton Communities” written by Rebecca J. Chmiel and coauthors, describes how dissolved cobalt concentrations in the Ross Sea were much lower in the 2017-2018 season compared to two previous expeditions a decade earlier. The differences in these observations were explored by examining dissolved cobalt (dCo), zinc and cadmium uptake rates, as well as dCo vs. phosphate relationships to gain insights into processes acting on the dCo pool. Overall, I really enjoyed reading this manuscript and thought it was very thought-provoking and thorough.

Most of my comments below are very minor and driven primarily by interest. My only more general question for the authors was whether they had explored potential differences in the water masses sampled during the CICLOPS and CORSACS expeditions, which may impact the deep dCo concentrations. For example, if the expeditions both appeared to sample Circumpolar Deep Water (CDW) and Antarctic Bottom Water (AABW) equally well, then that would strengthen the argument that the dCo inventory differences are primarily driven by differences in Co uptake by the phytoplankton community rather than changes in the water masses over the Amundsen shelf or in the Ross Sea. Perhaps related to this, I was also wondering if the authors have any evidence that the potential for Mn-oxidation might have changed over the 2007-2018 time period, perhaps due to a change in temperature? It seems like Mn-oxidation is low to non-existent in this region, but perhaps it would be something to think about for future work in this area and might have a significant impact on Co scavenging.

In general, I thought this was an excellent paper and it will be an exciting contribution to the field. Below are some additional very minor more specific comments. 

Specific comments

Figure 1: Can you perhaps note broadly on this figure where the CORSACS cruises were? I realize it might clutter the figure to have all of the stations, but maybe just an outline of the regions that those cruises sampled?

Figure 4 and 7: How were the regression outliers selected?

Figure 6: I thought it was interesting that the CICLOPS expedition shows more of a scavenging signal for dCo compared to the CORSACS expedition. Any thoughts on why there might be those differences?

Figure 9: I really like this figure, the trends are very clear and it is really interesting.

Line 731-736: Perhaps split this into multiple sentences.

Figure 12: Is it possible to also plot an average of the dCo in the deep and surface box over time on top of the evolution of the pools? I thought it would be interesting to see if this dCo loss is a steady decrease or not, based on this model. It appears to be steady based on the trends in the deep and surface, but seeing the average plotted on top of this might be interesting.

Rebecca J. Chmiel and coauthors present an original, high-quality investigation of why the dissolved cobalt inventory of specific coastal Antarctic seas has decreased between 2005/6 and 2017/18. They achieve this by comprehensively exploring cobalt speciation, stoichiometry, and uptake rates alongside that of zinc and cadmium, trace metals with inter-related ocean biogeochemical cycles. With climate induced changes in the cycling of trace metals related to that of carbon, this decadal-scale study is both timely and thoughtful. I thank the authors for bringing this thought-provoking work together and congratulate them on the quality of the paper. Overall, I believe this study aligns well with the aims/scope of BG and will be of clear interest to the marine biogeochemical community.

My specific comments are relatively minor but reflect aspects of the paper I thought could be improved/clarified.

Specific Comments

Whilst this is an intriguing investigation of the coastal Antarctic Co cycle, the important big picture insight I took from this paper was that climate driven changes in the Antarctic coastal Fe cycle may be responsible for dramatic regional shifts in micronutrient biogeochemistry and limitation patterns. This point could be emphasized more in the introduction, without the need for drastic changes to the text. At present there is one sentence at the end, broadly describing shifts in Zn/B12 limitation as a result of a warming climate, but with no mention of the influence of Fe supply.

Was dZn measured during CORSACS? If Zn limitation is in part driving the dCo inventory differential between CORSACS and CICLOPS, it would be interesting to read whether there has been a change in dZn concentration over this same time period?

Could you say something about the broad pattern of sampling during CICLOPS at the start of the methods section? Where samples taken sequentially from Amundsen Bay to Ross Sea to Terra Nova Bay? On line 364 it is stated that surface dissolved metal concentrations decreased in this order, perhaps due to seasonal drawdown but I am unsure of the sampling timeline between the locations.   

Sections 2.2 and 2.3: Can you please clarify which samples were analyzed at sea versus the land-based laboratory? It is a little difficult to decipher at present. Importantly, were any seawater samples (and/or GSC / internal standard) analyzed both at sea and on land for intercalibration purposes?

Line 244: Shipboard incubator? Was this on deck with flow-through water and natural day/night cycle or temperature controlled with constant screened lighting? Please expand slightly.

Lines 359-365: This section briefly describes the pattern of decreasing surface trace metal concentrations between sampling locations and the possible mechanisms for this. Has any statistical test been performed to resolve this decreasing trend or is this assumption based on trace metal mean concentration values?

Figure 7 clearly shows dCo:PO4 slopes are being calculated below the mixed layer. Are underlying waters consistent between the geographic sampling locations, and/or could seasonal changes in subsurface water impact slope-based stoichiometry? 

The metal uptake values presented in this paper are especially interesting. However, throughout the text I found inconsistent expressions of the metal uptake term (ρ) that made me stop and question my understanding of the text and figures. For example, the use of ρCo and non-italicized ρCo for cobalt uptake rates, whilst pCo is also used to define particulate Co. The same is true for Zn and Cd. In this instance, I suggest the italicized variant for element specific uptake rates for the text, figures, and tables. Further, the use of ρM, non-italicized ρM, and ρTM for trace metal uptake rates. With pM also used extensively in the paper as a unit of concentration, I would suggest the italicized ρM for defining trace metal uptake rates.

Lines 521-524: Whilst there appears to be a clear concentration difference between CORSACS and CICLOPS mean Co profiles (Figures 8 and 9), there also appears to be considerable variability within the CICLOPS profiles. Do you have a feel for how sampling site variability within and between the cruises may have impacted the deep water dCo and labile dCo inventories shown in figure 9?

Lines 1015-1018: As your box model nicely shows, increased uptake, mixing and burial of DOM are all part of enhanced dCo removal from the water column, yet burial is excluded here.

Technical corrections

Line 42: DOM does not necessarily need to be defined here as no further use in the abstract, but rather at first mention in the main text.

Line 128: Please provide the location of the Saito Laboratory.

Line 131: Is this a clean-air van?

Line 132: Please use the term ‘macronutrients’ as used elsewhere in this paper.

Line 234: I think the reader will be interested in what resin you used for preconcentration.

Line 235: What is the elution matrix, 1 M nitric acid?

Line 242: What container/volume was the seawater used for the uptake experiments collected into from the TM rosette?

Line 252: Please clarify that pigment samples were collected from the trace metal CTD, as per macronutrients?

Line 299: No need to repeat (≥100 m depth) here as explained in the paragraph prior. Can just use ‘deep’

Line 303: Move the depths (770 m and 780 m) to line 302; i.e., elevated near-seafloor signal at Station 41 (770 m and 780 m)’…

Line 329: dFe has not yet been defined in the text.

Line 359: I think the Zn concentration should be in units of nM rather than pM.

Line 366-367: I would perhaps remove the term surface seawater, or replace with upper ocean, as the surface is generally defined as 10 m within this paper and <100 m as deep water.

Line 379: Remove the word ‘spikes’.

Line 408: Depths at which an uptake rate is below detection…….

Line 414: dCo and labile dCo vertical profiles…….

Lines 507-519: Would this paragraph on the nepheloid layer be better placed within its own sub-section rather than with stoichiometry?

Line 577: I think it may read better as; ‘In contrast, the 2006 CORSACS-2 expedition reported the presence of labile dCo at five stations with concentrations of 17 ± 7 pM at 6 m depth and 14 ± 9 pM at 16 m depth, with reported labile : total dCo ratios of 0.37 ± 0.13 and 0.28 ± 0.17, respectively.’

Line 677: Perhaps consider moving the final paragraph on depth thresholds up to line 671. It appears to me the reader would be better served understanding how the inflection points were chosen prior to the differences between metal:PO4 slopes.

Line 825: Do you mean 1% lability, as per line 767?

Line 937: 3550 pmol dCo m^-2 y^-1

Line 1064: Do you mean panels d-f?

Table 1: The surface and deep sections of this table have different labelling for n<DL column.

Figure 1: The stars and circles are small and quite difficult to differentiate on paper and even more so on screen. Could these be enlarged or made clearer?

Figure 6a: This figure would benefit from X and Y-axis lines.

Figure 11, Line 940: Please define the Co cycle as black arrows, as the pCo cycle has black text. Same with the pCo cycle (orange arrows).

Table 6: Is the burial flux here calculated as dCo or pCo? Line 960 suggests pCo yet this term falls under the dCo loss parameters.