The goal of the manuscript “Holocene sea ice and paleoenvironment conditions in the Beaufort Sea (Canadian Arctic) reconstructed with lipid biomarkers” by Santos et al. is to fill a spatial gap in knowledge about ocean surface conditions, including sea ice, primary production, temperature, and terrestrial input, spanning the Holocene in the Beaufort Sea. The study aims to fill this gap by developing age-depth models and analyzing elemental composition, foraminifera abundance, and biomarker abundance in cores from two sites, one on the shelf (shallow), and one on the shelf slope (deeper). The study concludes that the early Holocene was warm and productive with minimal sea ice and larger inputs of terrestrial material (including organic matter and freshwater) than the late Holocene. The study also compares new and published time series and finds that these patterns are generally similar to those reconstructed elsewhere around the margins of the Arctic Ocean during the Holocene.

The central goal of this paper is important, in that quantifying the response of sea surface conditions to past periods of warmth will provide useful context for ongoing and near-future changes in the Arctic Ocean. The two study sites fill a spatial and temporal gap in data, and are based on good age constraints, especially considering the challenges with developing good age-depth models in Arctic Ocean sediments. In general appropriate methods are used, although I have a few suggestions for the authors to more clearly state the uncertainties inherent to these proxies, detailed below. The discussion sections could also be more clearly written, detailed suggestions below. Overall, the data presented here do support the conclusions. My suggestions are minor to moderate, and do not require further analysis. With some modifications to the text and figures, I recommend this manuscript for publication, as it will represent a strong and useful contribution to the literature.

Suggestions that will require moderate modifications:

Throughout: there is some uncertainty on the ages of the time series discussed throughout the paper. It seems important to list that uncertainty when describing the timing of events. There are many examples throughout the paper, here is one: (line 361) “The concentration of brGDGTs and terrestrial sterols in the shelf slope location during the Early Holocene peaked at 11.3 and 8.2 ka”. Add ± uncertainty to these ages, throughout the manuscript.

Section 4.1 and 4.2: I’m having a hard time following whether the changes mentioned/inferred here are based on new data presented in this study or in other studies. I think most information from other studies is well cited, but there are a few spots without citations or figure callouts. I think these spots are based on data presented in this study. Can the authors add references to specific figure panels wherever data from this study are mentioned? Adding interpretive arrows to Figs 2, 3 and 4 (see suggestion below) will also help the reader follow more easily, as some of the inferences about the conditions are difficult to follow for people unfamiliar with the details of the many proxies presented here.

Section 4.3: I’m also having a hard time seeing in Fig. 5 some of the changes that are mentioned in the text. For example, the text states “Norther Greenland (Detlef et al., 2023) and the Laptev Sea (Fahl & Stein, 2012; Hörner et al., 2016) are the first regions to record permanent sea-ice cover after the Early Holocene minimum, around 9 ka.” I think I see the pattern described here in the PIP25 time series for two of the three Laptev Sea sites (the authors could mention here that it’s only the deeper Laptev Sea sites that show this pattern), but I don’t see this pattern in the Northern Greenland site (in fact this site seems to have the opposite trends?). Can the authors clarify the descriptions throughout this section, so this section is easier for a reader to follow? I think adding information about the interpretations of the PIP25 ranges to Fig 5 (see Fig 5 comment) will also help.

Section 4.3: I think an important takeaway from this Arctic-wide comparison is the fact that there are a few regions that respond differently than others. This has implications for Arctic Ocean response to modern change. The authors allude to this a little bit, but a few more sentences about this conclusion would be interesting and a useful contribution. Can the authors clarify this important takeaway?

Two suggestions about inferred Salinity:

Line 209-210: If I’m reading this sentence correctly, the ±7 psu uncertainty stems from an isotope measurement uncertainty of 4‰ and is based only on that one source of uncertainty. This estimate of uncertainty seems small, given the scatter in data points in Fig. S4b. The uncertainty on the inferred salinity measurements should also incorporate the calibration uncertainty, i.e. the uncertainty on the regression between salinity and palmitic acid isotope values. The total uncertainty reported should include both analytical and calibration uncertainty, and be propagated appropriately (i.e., typically the total uncertainty is the square root of the sum of the squares of all individual sources of uncertainty).

Line 289-295: Somewhere in this section, or in the discussion, it should be noted that the uncertainty in reconstructed salinity is larger than the magnitude of salinity change in the reconstruction. The authors should also address whether it is still okay to interpret the reconstructed salinity values (I think it is, as long as the caveats are made clear, and the interpretations are well supported by multiple lines of evidence)

Suggestions that will require minor modifications:

Line 52: lipid biomarker records where? Climate model simulations of where?

Line 54: rephrase to clarify: which single offshore location (or are there several studies, each of which focuses on a different offshore location)? It’d be helpful to show existing studies in a map, eg as dots on fig 1?

Line 240: the ages provided in this sentence seem very precise, given the uncertainties in the age control points. Radiocarbon labs have some information about rounding conventions for radiocarbon ages. It seems as if the authors could apply these rounding conventions to age-depth model-derived maximum core ages (e.g., https://www2.whoi.edu/site/nosams/radiocarbon-data-and-calculations/)

Line 273-274: I don’t understand this sentence, it doesn’t describe the trends in PIP25 in either core. Remove?

Line 289: Should this be referring to Fig S4?

Line 205-210 and line 289-295: Can the authors provide some more details and citations about which data points went into this updated isotope-salinity calibration?

Line 295: The salinity range quoted here (31 to 33 psu) is smaller than the range for the Baffin Bay samples shown in Fig S4B. Clarify why that’s the case, or perhaps fix the quoted salinity range?

Lines 304-306: seems like this could say that both cores have stable values in the middle/late Holocene?

Lines 301-305 and Figs S5a, S4d: the time series for PCB09 look different between these two figures. Perhaps this is because the PCB09-MC is plotted with the same color in S4d? Can this be fixed?

Line 304: it’s hard to see the data that support the statement that the inferred temperature approaches modern values toward present. It looks to me like the inferred temperature is highly variable in the past couple hundred years. Can this be illustrated more clearly and/or discussed differently?

Lines 309 to 312: I’m having a hard time following the explanation about the high BIT value at 1 ka in PCB09, I think perhaps because some of the ‘increase/decrease’ values are backwards, and because I don’t see any obvious changes in the cren or brGDGT concentrations in this core at this time. Can this description be rewritten for correctness and clarity?

Lines 331-332 and Fig 2: can stratigraphic log be added to clarify the intervals that are more rich in mud vs sand? This will be useful in general, not simply for understanding the foraminifera data.

Line 345: I’m confused by this statement that the HBI implies there is some sea ice, but the ‘interpretation shading’ in Fig 4a shows this time period is within the range of ‘no sea ice’. Can this be clarified/explained in the text, or the shading in Fig 4a be modified?

Line 357: can this statement about ammonia oxidizers be tied to data from the paper? If not, it sort of appears out of the blue, so should perhaps be moved or removed.

Line 361-362: the peaks described here (and earlier in the results) are based on single data points. Can the authors provide more justification for interpreting these peaks as real?

Line 362-363: Additionally, given the interpretation of these peaks as indicating terrestrial input due to Laurentide melt, I’d expect to see the salinity decrease in the same samples. Is this the case? If not, why not?

Line 364-365: it’d be helpful to see the foraminifera abundance plotted vs age for direct comparison with the other data discussed in this paragraph.

Line 384-385: this interpretation is really interesting and exciting, in that it leans on modern observations and the difference between the two locations and time series. I think it’d be helpful to state more clearly that this is an interpretation (i.e. use more hedge words, such as ‘may have been’), but one that is supported by multiple lines of evidence.

Can the authors describe what a ‘flaw lead’ is?

Line 450: remove ‘during the Little Ice Age’, as it is redundant with the ‘during the late Holocene’ statement earlier in the sentence.

Suggestions for figures:

Figure 1:

• define ‘modern’, provide citation for sea ice data source
• could zoom in on inset map of whole arctic ocean and show dots of previous publications examining past sea ice cover, which are referred to in the introduction

Figs 2, 3, and Fig 4: It’d help to add interpretive arrows on each panel, e.g. panel 3h would have an arrow pointing up labeled “increased terrestrial contribution”, or something like that. Can the authors add an interpretive arrow to each panel in these figures?

Fig 4 a-f: I think it’d be easier to see how these various time series align if they’re arranged in a single stack plot, perhaps with some dashed vertical lines every 1 or 2 kyr.

Fig 4d: what is the uncertainty in inferred values using this calibration? It’d be helpful to show a vertical line that’s the uncertainty, or some shading around the datapoints indicating the uncertainty.

Fig 4f: It’s most appropriate to compare with peak annual insolation, as this is the forcing that the climate system responds to. 21 June insolation is in phase with peak annual insolation (see Clemens et al 2010 Fig. 6 doi.org/10.1029/2010PA001926 for an explanation about this), so I’d suggest modifying this to plot 21 June insolation instead of mean June and July insolation, as the most appropriate point of comparison for the time series.

Fig 5. Are the interpretation cutoffs displayed using shading in Fig 4a applicable to all of the PIP25 time series shown in Fig 5? If so, it could be helpful to display those shaded regions in these figures as well. If not, it seems important to explain that they are not, and why they are not.

Fig S6: can some arrow annotations be added to this figure to highlight the different features of interest in these images?

Table S1 should also include information about the material dated.

Santos et al. present two new multi-proxy paleoceanographic records from the eastern Beaufort Sea. The unique sampling sites and wide range of proxies analyzed have the potential to enhance our understanding of Holocene sea-ice regime in this underexplored region. However, the current version suffers from methodological inconsistencies and a lack of focus in presentation (e.g., trying to do both analysis of site-specific and pan-Arctic trends), which limit its suitability for publication at this stage. While many proxies are analyzed, few are discussed in sufficient depth. The authors are encouraged to narrow their focus to the regional paleoceanography of the Beaufort Sea and to compare their results more thoroughly with nearby cores - especially Wu et al., 2020, which employed a very similar set of proxies. It is also important to critically evaluate the suitability of each proxy for this region. Indices such as RI-OH′ are still relatively new, and their environmental interpretations should be treated with appropriate caution until further validation is available. Below are more detailed suggestions to help strengthen the manuscript.

INTRODUCTION

Overall, this introduction is brief, glances through many proxies without properly explaining their mechanism, limitations, and justifying their applicability to your specific study area. It also doesn’t establish well the theoretical link between numerous oceanographic conditions and sea ice. The authors fail to identify knowledge gaps in the existing Holocene sea ice literature. If Holocene sea ice records have no regional heterogeneity among them, there is then no need for another paleo sea ice reconstruction. I understand the purpose of the statement in lines 55-56, but this present study does not offer more cores or a more coastal location than other average paleo sea ice studies. Regarding line 68-69 identified a valid gap in regional calibration for GDGT proxies, and this led naturally to the need for calibration from surface sediment samples, which was briefly mentioned in line 74-75, but this connection is textually hard to see. The authors should put more effort into highlighting this connection and use citations to support their claim.

line 47-51: As one of your cores covers the end of the last deglacial, discussing some pre-/early Holocene warming events that you expect to see in your record in chronological order might help streamline your narrative.

line 52-54: The summary of Holocene sea ice trend in line 52-54 might be too simplistic and overlook the regional disparity among sea ice records. The narrow age constraint on the Holocene thermal maxima is questionable. If your claim only describes the Beaufort Sea, please specify that, as it has not been made clear.

line 55-56: Which studies? Need citation or need to reformulate the sentence.

line 57-69: The authors presented the current arctic paleoceanography proxy toolbox in a confusing order, while leaving out the key HBI-based proxies and index (IP25, PIP25), which definitely deserve some discussion. There are numerous GDGT-based paleothermometers and indices; the authors should name them directly in the paragraph to avoid confusion. Please consider adjusting the use of “e.g.” from line 60 onward, especially for line 63. Regarding line 69 needs citations to support the claim.

line 70-79: The objectives 2 and 3 are bold statements; some reformulation while keeping the limitation of your proxies in mind is needed.

MATERIALS AND METHODS

Overall, the biomarker workflow is unconventional. The authors attempt to analyze too many biomarker classes from a single extract, which may compromise the analytical quality of each individual proxy. The solvent system, choice of standards, and selected m/z values raise concerns about data robustness. Methodologically, the section reads as a workflow appears fragmented and could benefit from clearer structure with uneven detail across proxies. While I appreciate the challenge of condensing complex workflows into a limited space, a clearer structure, such as a summarized workflow diagram or table, would greatly help readers follow the analytical sequence and evaluate reproducibility.

Figure 1: Where are the surface sediment samples? Please consider using a legend for the labeling of sea ice extent. Country names are not necessary; authors should use a brighter color to indicate the study area in the global map, or else the small map will not be very useful.

line 156: Could you clarify what is the resolution of your biomarker analysis?

line 158: Could you clarify what is the concentration and the composition of the alkaline solution used for saponification?

line 163: Could you explain why internal standards added post-extraction?

line 164: The choice of C22 5,16-diol as an internal standard for sterols is unconventional. As a long-chain diol, its polarity, structure, and chromatographic behavior may differ substantially from sterols, and it is unlikely to mimic sterol recovery or derivatization efficiency.

line 170: HBI IV is an isomer of HBI III with the same degree of unsaturation. Why is it monitored at 348? The authors should include ion monitoring values for the standards as well, since previously there have been a few different fragments monitored for the same standard (For example, 7-HND can be monitored with m/z 99 and 266).

line 172: Please provide comparison data in the supplementary material.

line 173: Saponification is not a viable strategy for GDGTs work. Assuming your post-saponification liquid-liquid extraction is between a methanolic KOH and hexane:DCM mix, there is a possibility that quite some GDGTs will be lost in the aqueous phase.

line 177: This raises concerns and warrants clarification if the sterols would end up in F3, as their polarity is different than GDGT and according to this protocol, would probably mostly elute in F2 (DCM) and potentially partially elute in F1. Also, how is the F3 used both for GDGTs and phytosterols? Is it a split of the fraction? This point please clarify.

line 178: What are the m/z ratios, and which sterols are you quantifying?

line 203: c factor should be reported here.

line 202: The statement of not being able to detect dinosterol in the samples is concerning, since dinosterol should be a regionally abundant biomarker. In Wu et al. 2020, the core, which is nearby, the PIP25 index was calculated solely based on dinosterol. The complete absence of dinosterol raises questions about the sterol recovery in this study under review.

RESULTS

Figure 3: The practice of including brassicasterol as a terrestrial sterol is potentially problematic and warrants reconsideration, even regionally, brassicasterol is primarily of terrestrial origin, but it’s still a sterol that has a mixed source.

Figure 3&4: The authors should consider grouping proxies that are reconstructing the same information in the same figure. (such as PIP25 with the HBIs, BIT with terrestrial sterols)

DISCUSSION

line 335: The authors claim that the comparison between the slope core and outer shelf core is the focus of this study, but the comparison is weak throughout the discussion. For most of the discussion, the authors seem to treat both cores as one single record without emphasizing the difference in their depositional environments. In the introduction (line 74) and results (line 289), the authors claim that they conducted further calibration work with surface sediment, but these samples weren’t included in the map, nor do we see the data presented or discussed anywhere else.

line 337: The use of HBI II as an arctic paleo sea ice proxy is far less common than HBI I; it does not add anything to the author's narrative. It is redundant and adds to the confusion. The same goes for HBI IV, since this paper isn’t exploring the HBI TR25 index, there is no real reason to present the HBI IV as a separate record.

line 345: Please consider reformulating “some sea ice coverage”.

line 348: Could you clarify what is the reasoning behind this heterotrophic production claim? Citations are needed.

line 399-400: This claim needs some further support.

Figure 5: All records should have a shared time axis. If authors insist on presenting both IP25 and PIP25, they should separate them into two clear columns.

line 450: The resolution of biomarker analysis in this study doesn’t allow the authors to make claims about centennial events like the Little Ice Age.

In this study, the authors report a multiproxy dataset analyzed on two sediment cores from the shelf slope and outer shelf of the Beaufort Sea. These records span the Holocene, and indicate that—similar to other proxy-based reconstructions in other sectors of the Arctic Ocean—the sea-ice extent in the Beaufort Sea has been increasing since the early Holocene.

The manuscript is well-written, and the dataset is a welcome addition to the literature. I applaud the authors for producing such a diverse dataset. The main conclusion about the Holocene trend in sea ice is supported by the IP25 and PIP25 data. However, I am not fully convinced yet by some of the discussion based on individual or few data points, and by some interpretations of the biomarker data. In addition, I feel that the organization of the figures and parts of the discussion can be further improved. I hope my suggestions and comments below will be helpful in revising the manuscript to improve its clarity and accessibility. Overall, I recommend moderate to major revisions.

General comments

[1] Interpretation of biomarker abundance

The biomarker data are presented as concentration (ng/g sediment extracted). In a region where large shifts in sedimentation regime occur, the concentration may reflect not only biomarker production but also dilution by clastic material and oxic degradation. For example, is it possible that the downcore decrease in almost all biomarkers in Figure 3 reflects oxic degradation post burial? TOC also shows such a downcore decrease; normalizing the biomarker data using TOC, as some workers prefer, would probably eliminate the Holocene “increase” in other biomarker data (e.g., HBIs, isoGDGTs, and OH GDGTs). I understand that it is not always possible to fully tease apart these factors, but I encourage the authors to discuss this possibility and, if applicable, caveat their discussion to acknowledge the uncertainty of these data.

[2] BrGDGTs as terrestrial indicator

The authors interpret brGDGT concentration as terrestrial input. However, recent studies have proposed in situ production of brGDGTs (e.g., Xiao et al., 2016, Biogeosciences). In particular, based on the IIIa/IIa ratios, Singh et al. (2025, OG) suggested that brGDGTs in the central Arctic are of marine origin and transported from the shelf. Might this explain the differing temporal patterns in the concentrations of brGDGTs and sterols? I suggest that the authors first discuss the origin of brGDGTs in the Beaufort Sea before interpreting brGDGT concentration as terrestrial input.

[3] Discussion not supported by the temporal resolution and precision of records

The temporal resolution of proxy records is relatively low, and there are large fluctuations in the data. In some parts of the manuscript, a single data point is interpreted as representing a climate event (e.g., “A sharp decrease in PIP25 at 1.5 ka…”). Given the temporal resolution of the proxy records (e.g., 15 HBI III data points over 9 ka, corresponding to ~600 years between data points) and the overall noise in the data, I am not convinced that the change between two data points (1.5 ka to 0.4 ka) can be robustly linked to the Little Ice Age—especially considering age-model uncertainty.

More generally, proxy records are subject to non-climatic noise from sedimentation and analytical error; unless demonstrated otherwise, interpreting absolute point-by-point values seems overconfident. It would be more robust to interpret overall trends or time-slice means instead of individual fluctuations. This can be achieved by applying a running average, smoothed series, or trend lines for defined intervals (e.g., Early, Middle, and Late Holocene).

[4] Overall clarity and accessibility

At several points, it was challenging to follow the authors’ reasoning, as it was unclear how some conclusions were reached. I’ve provided line-specific examples below, but the authors may want to check for this throughout. Readability could also improve if the authors indicate biomarker interpretations directly in Figures 3 and 4 (e.g., label “Terrestrial input” next to sterols).

In addition, the age unit is given in years in all figures but in ka in the text—please make this consistent. It would also aid comparison if Figure 4 panels were stacked vertically rather than side-by-side, and if time intervals were shaded in color to highlight different Holocene intervals.

Specific comments

L20: “…IP25, and other HBIs,…” → suggested phrasing “HBIs including IP25.”

L42: “…renewed…” needs context—did the interest lapse previously?

L56: Add reference.

L58: Use “Lipid biomarkers” for consistency with the title and throughout.

L62–63: “Several proxies for sea temperature exist using microfossils…” reads awkward; please rephrase.

L65–66: The phrase “usually include hydroxylated…” is too strong; OH-GDGT proxies are not yet widely used enough for “usually.”

L80: “Material” → “Materials.”

L83: Include and label Mackenzie Trough in the map for non-expert readers.

L94: “…an open water flaw leads occur…” — if singular, use “occurs”; if plural, remove “an.”

L99–102: Clarify whether multicores were used to generate proxy records. If not, remove.

L103: “…as in (Matsuoka et al., 2012).” → “…as in Matsuoka et al. (2012).” Check and correct similar formatting errors throughout.

L110: “extend” → “extent.”

L113: “…scanned shipboard on a Geotek…” → reword for clarity.

L121–123: Add more explanation of the assumptions underlying these interpretations.

L132–133: Define “well-preserved”; consider adding SEM images of foraminifera.

L156–157: Rephrase to clarify that 5 g refers to each depth/sample, not the whole study.

L165: C46 GTGT

L170: “Concentration of IP25 were…” → “was.”

L173: The PTFE filter should likely be 0.45 µm, not 45 µm. Please check.

L175: “according to Hopmans et al. 2016 and following Lattaud et al. 2021” → clarify which method was actually used.

L176: C46 GTGT (check notation).

L177: “sililated” → “silylated.”

L178: “C22 5,16” → add “-diol.”

L180: Define “IRMS” at first mention.

L208–209: Use consistent terminology for “salinity” vs. “S.”

L211: Define “SST” here, not later (L216).

L218–219: “SST” is italicized in L218 but not L219—be consistent.

L227: Only N. pachyderma counts are presented, yet the methods mention all species. Please clarify.

L241–243: Plot sedimentation rate in Figure 2 for clarity.

L248–254, 257–259: These sentences read more like discussion than results—consider moving or rephrasing.

L260–262: Add figure citations so readers can locate corresponding results.

L273–274: See General Comment [3]. It’s difficult to locate data at 1 ka; consider adding minor ticks and grid lines.

L285: Consider moving Fig. S3 to the main text (only 5 figures currently).

L301–308: Same as above. Alternatively, briefly state in Methods that RI-OH′ was not used because it yielded unrealistic SSTs.

L349–351: The results do not clearly show that SST follows summer insolation. To me, SST declines at the end of the Early Holocene while PIP25 remains low, suggesting weak coupling.

L363–364: This interpretation assumes Ti is relatively constant—please justify or provide supporting evidence.

L365: “slop” → “slope.”

L372: “have” → “has” (subject is singular).

L377–378: The conclusion “stable sea ice-edge or polynya conditions” is not well-supported—please elaborate.

L386–400: See General Comment [3].

L413: “close by” → “close-by”

L417: “Norther” → “Northern”

L440: “Conclusion” → “Conclusions”

L451–453: This claim is not supported by the lack of correlation between the PIP25 and SST records.

Overall assessment:

This is a valuable contribution to Arctic paleoceanography, and I commend the authors for assembling such a comprehensive dataset and transparency reporting reporting all results, including proxies that did not yield clear signals. Addressing the points above—particularly regarding biomarker interpretation, temporal resolution, and figure clarity—will greatly improve the robustness and accessibility of the paper.