Microbial Fe(III) reduction as a potential iron source from Holocene sediments beneath Larsen Ice Shelf

Recent recession of the Larsen Ice Shelf C has revealed microbial alterations of illite in marine sediments, a process typically thought to occur during low-grade metamorphism. In situ breakdown of illite provides a previously-unobserved pathway for the release of dissolved Fe2+ to porewaters, thus enhancing clay-rich Antarctic sub-ice shelf sediments as an important source of Fe to Fe-limited surface Southern Ocean waters during ice shelf retreat after the Last Glacial Maximum. When sediments are underneath the ice shelf, Fe2+ from microbial reductive dissolution of illite/Fe-oxides may be exported to the water column. However, the initiation of an oxygenated, bioturbated sediment under receding ice shelves may oxidize Fe within surface porewaters, decreasing dissolved Fe2+ export to the ocean. Thus, we identify another ice-sheet feedback intimately tied to iron biogeochemistry during climate transitions. Further constraints on the geographical extent of this process will impact our understanding of iron-carbon feedbacks during major deglaciations.

I realize that the authors might rebut this conclusion by pointing out the correlation between putative (see final comment below) DIRB abundance and illite crystallinity shown in Fig. 4. Here again, however, while I don't doubt that the correlation is real, this does not prove that illite was the source of Fe(III) for sustenance of DIRB populations. It seems entirely possible that Fe(III) oxides (as well as perhaps smectite, which is generally more readily reduced by bacteria than illite) also served as electron acceptors for DIRB. In other words, even if illite was undergoing reductive transformation by DIRB, this doesn't prove that illite would be the dominant Fe(III) source for Fe(II)aq generation. Along analogous lines, without further information on the Fe mineralogy, the incubation experiment with psychrophilic DIRB (two Shewanella species isolated from the Antarctic Peninsula) shown in Fig. S2 cannot be used to definitively infer that illite reduction was the main source of Fe(II)aq release, as some or all of the Fe(II)aq could have come from Fe(III) oxide reduction. In summary, the authors need to provide substantial additional analysis and interpretation to properly address their hypothesis that illite reduction was a major source of Fe(II)aq release from Larsen Ice Shelf sediments during the Holocene.
Regarding the putative DIRB populations inferred from 16S rRNA gene amplicon sequencing (Fig.  4), none of the three families (Comamonadaceae, Dehalococcoidetes, and Desulfobacteraceae) identified are canonical DIRB taxa. Although some organisms within the Desulfobacteraceae have been shown to be capable of Fe(III) reduction (Lovley et al., Mar. Geol., 113:41-53, 1994), this is by no means an established idea. And to my knowledge, neither the Dehalococcoidetes or Desulfobacteraceae families have been shown to contain organisms capable of Fe(III) reduction. So unless I'm missing something, the correlation between the abundance of these taxa and illite crystallinity could be completely spurious, having nothing to do with the role of these organisms as DIRB. In fact, given that the Desulfobacteraceae are well-known as sulfate-reducing bacteria, it seems possible that the correlation reflects the impact of hydrogen sulfide attack on illite as opposed to direct microbial reduction.
Reviewer #3 (Remarks to the Author): In this manuscript, Jung and colleagues examine a marine sediment core on the northwestern part of the Larsen Ice Shelf C (LIS-C) embayment. The authors argued that the microbe-mineral interactions may release ferrous iron (Fe2+) to porewaters from illite. Although this paper can be of potential interest for readers of Nature communication, I cannot recommend the manuscript for publication in its present form. This is an important investigation of the possible sources of Fe2+ in the Southern Ocean. However, the conclusion is not fully supported by the results presented in this paper. The key statement that "The biogeochemical data described here indicates a clear link between microbial Fe respiration, alteration of illite crystalinity and an increase in the Fe(II) content of illite under anoxic conditions." (Line 156-158) is vague. The biogeochemical data presented include X-ray diffraction of clay fraction, TEM characterization of illite particles and abundance of Fe reducing bacteria. In my opinion, the link extracted from the correlations from the depth profiles is weak due to following reasons: 1) The depth profiles of illite crystallinity might reflect the variation of smectite contents in illite/smectite mixed-layer components. The authors mentioned the presence of smectite in the core materials (Line 104), however, was not able to discuss the nature of the smectite and how it may impact the measurement of illite crystallinity. The existence of illite/smectite mixed-layer minerals in surface sediments in the Antarctic Ocean has been extensively documented in literature. The amount of smectite is likely to have a greater impact on the illite crystallinity rather than possible microbial modification. As shown in Fig. 1 in the manuscript, the smectite peaks at 14 Å are higher at depth where the illite crystallinity is higher (e.g. 136-190 cm).
2) The depth profiles of abundance of Fe reducing bacteria might reflect the variation of environmental factors such as decreasing concentrations of dissolved oxygen (DO) with depth.
3) The link between abundance of Fe reducing bacteria and illite crystallinity. Illite is not the only Fe-bearing minerals in the core. Even we are convinced that the microbial Fe reduction occurs, how do we now the bacteria utilize illite but not chlorite or Fe oxyhydroxide? Are there any data showing the depth profiles of chlorite and Fe oxyhydroxide? More specific comments: Line 57-59: Be more specific about the alteration. I think in metamorphic the alteration usually means smectite-illite transition. Do you mean the same? Line 72: there are several papers about illite reduction by microorganisms (Dong et al., EST, 2003;Zhang et al., Chemical Geology, 2012). Line 103-105: It is surprising that no quantitative XRD data (on clay fraction) were presented here. Are XRD data on ethylene-glycolated samples available? Line 193: There is little information in Supplementary. I didn't follow how the experiment was conducted and why smectite was mentioned here? Very unclear. Fig. 3: not sure how the diffraction pattern was collected? Please explain the dashed, white line and short white bar.
 We agree that the precise mechanism is not well constrained, although there is evidence from culture work. In general, the precise mechanisms by which iron-reducing microbes transfer electrons to insoluble minerals are unknown in nature, and require further study beyond the scope of this manuscript. Based on results from laboratory studies with pure cultures, Geobacter produces nanowires (i.e. pili) containing cytochromes which facilitate electron transfer to the surface of iron oxide minerals. Shewanella either produces similar electrically conducive appendages or uses riboflavins as electron shuttles, depending on culture conditions. Depending on the optimum temperature, microbes such as psychrophiles (<15 °C), mesophiles (20-45 °C), and thermophiles (>60 °C) showed different activities (Bozal et al., 2009;Jaisi et al., 2011;Lovley and Phillips, 1988;Zhang et al., 2007). To address this in the manuscript, we have added a clause to the sentence to make this clearer: (Line163) "….although the exact mechanism of electron transfer by which this occurs is not yet constrained (Bozal et al., 2009;Jaisi et al., 2011;Lovley and Phillips, 1988)." However, despite the exact mechanism being unconstrained, we would argue that the data presented here provide compelling evidence for the influence of microbial activity as reviewer commented.
I think the hook that microbes solve the problem in the cold -abiotic processes might be/would be too slow in these environments -is clever, but my gut feeling is that there is more to this story than is told at the moment. Am I going to recommend publication? Of course. Great data set, provocative idea, calls for clarification in how Fe loss from illite in the spectrum of marine environments really occurs. Go for it.
 We agree with the reviewer, and anticipate that our provocative study will spur on more work to identify the exact mechanism and systematics for the illite breakdown process, now that it has been identified in Antarctic sediments at low temperatures and pressures. This modification is conventionally thought to be affected by temperature and pressure in diagenetic environment (Eberl and Hower, 1976;Freed and Peacor, 1992).

Reviewer #2 (Remarks to the Author):
This paper deals with the potential for mobilization and delivery of dissolved (aqueous) ferrous iron (Fe(II)aq) derived from clay (illite) in the Larsen Ice Shelf to the Southern Ocean during the Holocene, and the implications of glacial retreat on this process. Input of Fe from ice shelves to what are generally thought to be Fe-limited Southern Ocean waters could have an important influence primary productivity, and possibly modulate the climate response to glaciation as ice shelves advance or retreat. To address this question, the authors obtained and characterized a marine sediment core from the northwestern part of the Larsen Ice Shelf C (LIS-C). The core encompassed four major lithological units, where the upper unit (U1) and the upper portion of the second unit (U2) are bioturbated, reflecting the presence of relatively oxidized bottom water conditions that arose as a result of ice shelf retreat something like 5000 years ago (5 ka). In contrast, the lower portion of U2 and the third unit (U3) are laminated, reflecting reducing bottom water conditions that were present as a result of relatively stagnant bottom water during the waning phase of the last glacial maximum ca. 5-12 ka. Among the various (solid-phase only; see below) measurements performed on the core, key for this paper was the analysis of down core changes in the crystallinity of the Fe-bearing clay mineral illite, which apparently comprises approximately 50% of the clay mineral content of Antarctic sub-ice shelf sediments. The authors put forward the hypothesis that alteration of illite coupled to the activity of dissimilatory iron-reducing bacteria (DIRB) could be responsible for release of Fe(II)aq from sub-ice shelf sediments during periods of prominent glaciation.
While this basic hypothesis generally makes sense, I am perplexed by the singular focus on illite in this paper.
Although I do not doubt the validity of the downcore illite crystallinity analyses, in and of themselves these data do not prove that microbial illite transformation would have been the main source of Fe(II)aq released to Southern Ocean waters. It is well-known that Fe(III)-bearing clays such as smectite and illite, though subject to microbial reduction, generally do not undergo anything like quantitative dissolution during this process. In contrast, Fe(III) oxides generally dissolve during microbial reduction (not completely, but much more so than clays), and it thus seems entirely possible that Fe(III) oxide reduction -rather than clay reduction -would be the more likely contributor to Fe(II)aq export from subglacial shelf sediments. Along these lines, I was surprised that the authors did not conduct standard wet-chemical extraction procedures to estimate Fe(III) oxide content through the downcore redox transition. Another set of analytical data that is missing is pore water Fe(II)aq concentrations; is seems to me that in order to argue convincingly for the past release of Fe(II)aq from unit U3 and the lower portion of U2, the authors should demonstrate that porewater in these units contain substantial amounts of Fe(II)aq. By analogy, porewater should be very low in Fe(II)aq in unit U1 and the upper portion of unit U2. To my opinion, without such data, together with information on other possible Fe(III)-bearing mineral phases that may contribute to Fe(II)aq generation, it is not valid to infer that microbial illite reduction would be the main source of Fe(II)aq release to the Southern Ocean.
 We thank the reviewer for noting the validity of the illite changes and the logic of our hypothesis and for giving us the opportunity to clarify our thinking and focus on illite, compared to other Fe-bearing minerals. As the reviewer points out, the possible source of Fe induced by microbial Fe(III) reduction would be a range of Fe-oxides, and Fe-bearing clay minerals including smectite, illite, and chlorite . Indeed, several papers have shown that Fe-oxides are the major source of Fe-liberation associated with Fe-respiration . This is well known. However, the novel aspect of our paper is instead the identification of illite breakdown as a new potential and important Fe source, which all of the reviewers recognized as novel and worthy of publication. We did not intend to say that it was the only or perhaps even the major Fe-source mineral in sedimentary environments, and we apologize if this was unclear. We would also argue that inclusion of other minerals would only enhance the iron source, and would likely be influenced by ice sheet changes in similar ways to illite. We do comment on other iron minerals in the paper. We have also added more interpretation of XRD data showing 4 possible Fe sources such as lepidocrocite, smectite, chlorite, and illite. We modified in the abstract (Line 18): "….illite and other Fe-bearing minerals…..". We added new data in the text (Line102-109): "X-ray diffraction profiles show that the major mineral composition for the clay size sediments throughout the core is smectite (S), chlorite (C), kaolinite (K), and illite (I) and lepidocrocite (L) ( Fig. 2 and Supplementary Fig. S3). Depth profiles of clay minerals ( Supplementary Fig. S4) throughout the core shows that illite is dominant (50-60 %) compared with smectite (~10 %), chlorite (~20 %), and kaolinite (~15 %).
There is a clear separation of chlorite (14 Å) and smectite (17.5 Å) for the glycolated samples and no XRD peak for illite/smectite interstratified layer (9.84 degree 2-theta) was observed. (Supplementary Fig. S3)". We also added in the text (Line 200): "…..clay structures……." Previous work Urrutia et al., 1998), has shown that the amount of Fe-release depends on mineralogy, surface charge, particle size, and crystal chemistry. It is well known that Fe-oxides are the major source of Fe , while clay minerals also showed Ferelease associated with microbial Fe-reduction. In the text we also added other Fe sources from Fe-oxides (Line 66-68): "Whereas a range of iron minerals are known to be sources of dissolved Fe upon breakdown by iron-reducing or -oxidizing bacteria , adding illite to this group, and at low temperatures, …….." We also added in the text (Line 144-147): "Other clay minerals may also undergo microbial-induced changes that involve release of reduced iron, however illite is the only clay mineral for which we can currently measure the crystallinity responding to alteration of crystal structure in various redox conditions ." It is practically impossible to quantify the amount of Fe-release from the mixture of Fe-bearing minerals (Fe-oxides and clay minerals), particularly for the natural sediments, and we do not attempt to do that here. In the paper by , crystalline magnetite showed less microbial Fe-reduction compared to the smectite. Nonetheless, goethite and amorphous Fe showed a large extent of Fe(III) reduction . Again, we did not show that illite is the major source of Fe(II) release to Southern Ocean water. We suggest that IC could be an indicator of depositional conditions under retreat and advance of Ice Shelf. Based on the IC, we can infer that Fe-bearing minerals including illite should undergo the same redox-reaction with microbes, releasing Fe(II), that responds to the movement of iceshelf. Please note that we addressed the objective of our paper in the text: "To address the possibility of illite crystallinity changes sourcing Fe to the water column beneath an ice shelf,…(Line 71)" We also addressed the importance of Fe-oxides minerals as a Fe-source in the text: "However, dissimilatory reductive dissolution of sediment releases isotopically light Fe 2+ is typically thought to involve Fe(III) minerals such as goethite, hematite and ferrihydrite [58][59][60] . Here, we propose that illite may also provide a substrate for microbial reduction that was previously thought to be largely inaccessible. This may be especially important in the Antarctic, since illite, and clays more generally, appear so prominently on the continental shelf around Antarctica 14 …(Line 191-196)".
Regarding the reviewer's surprise that we did not analyze porewaters, we agree with the reviewer's comments on the strength that porewater analysis would have added to this study. Unfortunately, the cores were not originally collected for clay mineralogy analysis, and, as Reviewer 1 pointed out, the results are so novel as to be unanticipated by the authors at the onset of this work. Thus, porewater samples were not taken for iron analysis. Now that the cores have been in the repository for several years, we do not feel like this type of analysis would reveal anything meaningful -it is simply too late to resample and obtain results which are representative. Our results, however, will ensure that future cores taken from beneath Antarctic ice shelves and elsewhere on the continental margin will likely be sampled in this way -by our group and others working on similar problems. This will be the impact of this paper once published.
I realize that the authors might rebut this conclusion by pointing out the correlation between putative (see final comment below) DIRB abundance and illite crystallinity shown in Fig. 4. Here again, however, while I don't doubt that the correlation is real, this does not prove that illite was the source of Fe(III) for sustenance of DIRB populations.
 Again, we did not suggest that illite is the only source of Fe-release, and we apologize if this was not made clear enough in our original submission. By quantifying oxidation states of Fe in illite by EELS acquisitions, IC corresponds to the redox conditions and microbial activity (Of course, other Fe-bearing minerals such as lepidocrocite, smectite, and chlorite undergo the same conditions as illite). However, illite is the only mineral that we can measure the crystallinity responding to alteration of crystal structure in various redox conditions. In the present study area, neither pressure nor heat can be the factor that modify the illite structure. Furthermore, the population of FeRB is inversely related with the value of IC. We addressed in the text the possible factors to control IC in the conventional diagenetic settings : "Previous observations of such diagenetic mineral alteration generally are reported in high temperature and pressure environments 34 , far from the conditions observed in sub-ice shelf sediments. Thus, here, instead of high temperature and pressure, we suggest that biogeochemical redox-sensitive reactions,…(Line 140)" It seems entirely possible that Fe(III) oxides (as well as perhaps smectite, which is generally more readily reduced by bacteria than illite) also served as electron acceptors for DIRB. In other words, even if illite was undergoing reductive transformation by DIRB, this doesn't prove that illite would be the dominant Fe(III) source for Fe(II)aq generation. Along analogous lines, without further information on the Fe mineralogy, the incubation experiment with psychrophilic DIRB (two Shewanella species isolated from the Antarctic Peninsula) shown in Fig. S2 cannot be used to definitively infer that illite reduction was the main source of Fe(II)aq release, as some or all of the Fe(II)aq could have come from Fe(III) oxide reduction. In summary, the authors need to provide substantial additional analysis and interpretation to properly address their hypothesis that illite reduction was a major source of Fe(II)aq release from Larsen Ice Shelf sediments during the Holocene.
 We did measure smectite, chlorite, illite and lepidocrocite for Fe-bearing minerals in our natural sample, and this information has been added to the paper. As we cited previous studies , and remarked above, we did not state that illite is the major or only source of Fe-release in our study area. What is new about this study is that we show that illite could be another additional source, complementing other Fe sources. Furthermore, illite crystallinity can be related to depositional conditions and thus potentially act as proxy for other sedimentary sources of Fe.
Regarding the putative DIRB populations inferred from 16S rRNA gene amplicon sequencing (Fig. 4), none of the three families Comamonadaceae, Dehalococcoidetes, and Desulfobacteraceae) identified are canonical DIRB taxa. Although some organisms within the Desulfobacteraceae have been shown to be capable of Fe(III) reduction (Lovley et al., Mar. Geol., 113:41-53, 1994), this is by no means an established idea. And to my knowledge, neither the Dehalococcoidetes or Desulfobacteraceae families have been shown to contain organisms capable of Fe(III) reduction. So unless I'm missing something, the correlation between the abundance of these taxa and illite crystallinity could be completely spurious, having nothing to do with the role of these organisms as DIRB. In fact, given that the Desulfobacteraceae are well-known as sulfate-reducing bacteria, it seems possible that the correlation reflects the impact of hydrogen sulfide attack on illite as opposed to direct microbial reduction.
 Fe-reducing bacteria are phylogenetically diverse across phyla. Conventionally the characteristic of Fe reduction has been confirmed for pure bacterial cultures like Geobacter and Shewanella. However, taking into account that cultivatable bacteria generally comprise less than 0.1-1% of natural bacterial community, previously known Fe-reducing bacteria found by cultivation-dependent approaches are underrepresented so far. Recently a combination of geochemical measurements and cultivation-independent molecular techniques (e.g. next-generation sequencing) revealed the presence of new potential Fereducing bacterial groups in sediment environments. The family Comamonadaceae (Betaproteobacteria) and Desulfobacter which is a type genus of the family Desulfobacteraceae (Deltaproteobacteria) were presumed to be putative Fe-reducing bacteria in recent studies . We have added the literature for referring to those groups as putative Fe-reducing bacteria in the revised manuscript. The distribution of Dehalococcoidetes reflects a tight-coupling with Fe-reducing bacteria, however they are not a direct driver to change IC . We In this manuscript, Jung and colleagues examine a marine sediment core on the northwestern part of the Larsen Ice Shelf C (LIS-C) embayment. The authors argued that the microbe-mineral interactions may release ferrous iron (Fe2+) to porewaters from illite. Although this paper can be of potential interest for readers of Nature communication, I cannot recommend the manuscript for publication in its present form. This is an important investigation of the possible sources of Fe 2+ in the Southern Ocean. However, the conclusion is not fully supported by the results presented in this paper. The key statement that "The biogeochemical data described here indicates a clear link between microbial Fe respiration, alteration of illite crystalinity and an increase in the Fe(II) content of illite under anoxic conditions." (Line 156-158) is vague. The biogeochemical data presented include X-ray diffraction of clay fraction, TEM characterization of illite particles and abundance of Fe reducing bacteria. In my opinion, the link extracted from the correlations from the depth profiles is weak due to following reasons:  We appreciate the reviewer's appraisal of the importance of our investigation and the constructive criticism that will make this manuscript a driver of iron biogeochemistry research in the Southern Ocean. We address the reviewer's criticisms point-by-point below.
1) The depth profiles of illite crystallinity might reflect the variation of smectite contents in illite/smectite mixed-layer components. The authors mentioned the presence of smectite in the core materials (Line 104), however, was not able to discuss the nature of the smectite and how it may impact the measurement of illite crystallinity. The existence of illite/smectite mixed-layer minerals in surface sediments in the Antarctic Ocean has been extensively documented in literature. The amount of smectite is likely to have a greater impact on the illite crystallinity rather than possible microbial modification. As shown in Fig. 1 in the manuscript, the smectite peaks at 14 Å are higher at depth where the illite crystallinity is higher (e.g. 136-190 cm).
2) The depth profiles of abundance of Fe reducing bacteria might reflect the variation of environmental factors such as decreasing concentrations of dissolved oxygen (DO) with depth.
 Although a depth profile of DO was not determined, oxygen penetration depth (OPD) likely would not exceed 50-60 cm in our sediment core (Sachs et al., 2009). In our results, a variation of Fe-reducing bacteria was relatively small in the upper oxic layer (indicated sedimentologically as bioturbated sediment), while the variation was large in the lower anoxic layer (Fig. 4, indicated sedimentologically as laminated, undisturbed sediment). Thus, DO is not likely to be an environmental factor for explaining the abundance of Fe-reducing bacteria in our study.
3) The link between abundance of Fe reducing bacteria and illite crystallinity. Illite is not the only Febearing minerals in the core. Even we are convinced that the microbial Fe reduction occurs, how do we now the bacteria utilize illite but not chlorite or Fe oxyhydroxide? Are there any data showing the depth profiles of chlorite and Fe oxyhydroxide?
 As we discussed for the previous reviewer's comments about the possible source of Fe, in the present sediment we measured Fe-bearing minerals including smectite, illite, chlorite, and lepidocrosite by XRD. Of course, we did not intend to say illite is the major source of Fe, associated with microbial activity. However, we would say variations in IC can be recorded by microbial Fe-reduction in illite structure implying the sediment depositional conditions associated with the retreat and advance of ice shelf. Our EELS data acquisition indicated that variation in Fe-oxidation states in illite is closely linked to the microbial activity under retreat and advance of Ice Shelf. Again, please note our responses to other possible Fe source above (See our responses to Reviewer 2).
More specific comments: Line 57-59: Be more specific about the alteration. I think in metamorphic the alteration usually means smectite-illite transition. Do you mean the same?
 As we described (Line 140), IC can determine the degree of diagenetic states of pelitic rock for the low-grade metamorphism, by measuring half height width of illite XRD peak that corresponds to the illite alteration with high temperature and pressure. It is surprising to detect the illite alteration given the variation in the redox conditions and microbial activity in the glacial-interglacial periods. REE data and crystal size distribution indicate that illite clay minerals characterized with IC in the core are from the same source. Therefore, alteration in our study must be associated with factors other than conventional ones (temperature and pressure). Microbial activity showed linear correlation with IC with depth, indicating that Fereduction in illite structure is related to the modification of illite structure that responds in IC.
 Yes, we included those published papers (Line 69) that discussed the microbial Fe-reduction in various clay minerals. Because we have illite, smectite, chlorite and lepodocrosite we discussed the possible source of Fe from clay minerals as well as Fe-oxides. Nonetheless, crystalline phase of magnetite showed less Fe-reduction rate than clay minerals, suggesting that there are several factors such as particle size, crystal chemistry, structure, and concentration of clays or Fe-oxides that control the Fe-release from the minerals. It is not easy to say which mineral can contribute more Fe-release as pointed out by Dong et al (2003). Again, we did not intend to say illite is the only source of Fe-release associated with microbial activity in the present sediment. Previous papers have shown a strong relationship between microbial Fe-reduction of clay minerals and Fe-release . Furthermore, we showed microbially induced Fe-release from biogenic smectite at low temperature, suggesting the feasibility of alteration of clay minerals ( Supplementary Fig.S2) along with Fe-oxides in Antarctic area.
Line 103-105: It is surprising that no quantitative XRD data (on clay fraction) were presented here. Are XRD data on ethylene-glycolated samples available?
 We added XRD profiles showing relative contents of clay minerals with increasing depth (Supplementary Figure S3 and S4). As we discussed above, illite is the dominant phase while smectite is minor content in the sediments. We added in the text (Line 102-109): "X-ray diffraction profiles show that the major mineral composition for the clay size sediments throughout the core is smectite (S), chlorite (C), kaolinite (K), and illite (I) and lepidocrocite (L) ( Fig. 2 and Supplementary Fig. S3). Depth profiles of clay minerals ( Supplementary Fig. S4) throughout the core shows that illite is dominant (50-60 %) compared with smectite (~10 %), chlorite (~20 %), and kaolinite (~15 %). There is a clear separation of chlorite (14 Å) and smectite (17.5 Å) for the glycolated samples and no XRD peak for Illite/smectite interstratified layer (9.84 degree 2-theta) was observed. (Supplementary Fig. S3)".
Line 193: There is little information in Supplementary. I didn't follow how the experiment was conducted and why smectite was mentioned here? Very unclear.
 In response to the reviews of this manuscript, we performed more batch experimentation with clays and microbes at low temperature. As we discussed above, previous papers showed the Fe-reduction in various clay minerals such as illite, chlorite and smectite indicating that Ferelease can be from various origins. In this manuscript, we showed one example of Ferelease from clay minerals associated with microbial Fe-reduction at low temperature for the first time, suggesting the feasibility of microbial role in Fe-release from clay minerals at low temperature. We provided the experimental setting in the method (Bacterial culture and experimental procedure) of the supplementary information file. Our experimental data (Supplementary Figure S2) suggests ~4 % of Fe may be released by psychrophilic microbial reduction of the clay mineral smectite (nontronite). Of course, this is single mineral experiment to understand the microbial Fe-reduction in clay minerals at low temperature. Therefore, we used nontronite which measured a large amount of structural Fe (23.4 % total Fe content by weight, where 99.4 % of the total Fe is Fe(III)  for the most optimum conditions of the microbial Fe-reduction at low temperature.  First, we verified the illite packet by measuring EDS and layer spacings with 10 Å. Then, illite layers were confirmed by Selected Area Electron Diffraction (SAED) patterns with the strongest Bragg reflections of 1.0, 0.5, and 0.33 nm (Fig. 3 inset). Dashed white line separated each illite packet. "A short white bar" is referred to the illite (I). We added in the text (Line 261-264): "…. 10 Å illite packets from the collapsed hydrous clay minerals, such as smectite, displaying the same spacings under the high-energy TEM beam. Illite layers were confirmed by Selected Area Electron Diffraction (SAED) patterns with the strongest Bragg reflections of 1.0, 0.5, and 0.33 nm."  In addition to the major changes, we changed or modified the words in the text to make it clear: Line111 added "more"; Line 132 corrected "for" to "within"; Line 187-199 deleted "Furthermore", added the "possibility"; Line 241 added "directly"

Reviewers' comments:
Reviewer #2 (Remarks to the Author): I think that this is good, provocative science that will lead to further high profile work. I remain very positive in my support.
 Thanks again for the positive comments. We strongly agree with R2, because that is the main purpose of publication in the high impact journal like Nature Communications.

Reviewer #3 (Remarks to the Author):
I must acknowledge that the authors did their best to respond to my comments on the original manuscript. However, on balance the changes/additions did very little to alter my take on the unwarranted focus on microbial illite transformation on Fe(II)aq release to Southern Ocean waters during Holocene glaciation, as well as other issues related to the microbial community analyses. As the authors readily admit, Fe(III) oxides could have been much more important sources of Fe(II) mobilization in these sediments. Unfortunately they chose not to conduct standard wet-chemical extraction procedures on the core materials to estimate Fe(III) oxide abundance so as to put some constraints their potential importance (or non-importance) as substrates for dissimilatory microbial reduction, e.g. in relation to the estimated abundance of illite and other clay minerals. I understand that the authors (as well as some of the reviewers) are excited about the idea of illite as a substrate for microbial reduction and Fe(II) mobilization, but this idea alone does not seem sufficient to justify publication of the paper in Nature Communications. I another words, I am not comfortable with the take home message being "microbial illite reduction might have been a source of Fe(II)aq mobilization in Holocene Southern Ocean sediments". Without further direct proof or constraints on this idea, the paper simply does not live up to the message conveyed by its title. In light of these weaknesses, in my opinion the only option to make the paper more realistic and honest would be to change the title to something like "Microbial Fe(III) reduction as a potential Fe source responding to depositional environments under the Larsen Ice Shelf C during the Holocene".
 We understand the concern raised by R3, and we have modified the title as suggested.
As R2 commented, we anticipate that this study will stimulate future high profile work, including the biotic mineral diagenesis under the ice shelves, Fe isotopic fractionation associated with biotic/abiotic reduction in various minerals including both clay minerals and Fe-oxides. We feel that the role of journals like Nature Communications are exactly why R2 is excited by the publication of these findings -it is to communicate potentially impactful multidisciplinary new ideas that are of significant interest to specialists in each field. Publication of this article will certainly lead to the type of wet-chemical techniques that R2 (and the coauthors) deem important, but that were not the original focus of research on the sediment cores collected.