This paper entitled “Biomarker characterization of the North Water Polynya, Baffin Bay: Implications for local sea ice and temperature proxies” submitted by Harning et al. presents the distribution of lipid biomarkers in marine surface sediments taken in the area in and surrounding the largest arctic polynya (North Water Polynya). They assess some of the most commonly evaluated lipid classes in organic geochemistry: highly branched isoprenoids, sterols, and GDGTs. Combining their results with data generated from previous studies in the area, these authors evaluate the use of highly branched isoprenoids and sterols as biomarkers for sea ice productivity. The authors also compare the estimated temperatures generated by biomarker (GDGT) proxies with measured data from the World Ocean Atlas Data from 2007 – 2017. The authors make several astute observations about the environmental controls that may govern the distribution of GDGTs involved in the temperature proxies in the Arctic. They have a reasonable argument for the power of regional calibrations in reducing unceratintity and errors in temperature estimates, but I feel that they must also acknowledge that the sample size of these observations is small. Overall, I am satisfied with their discussion of GDGT-based temperature proxies in the Arctic and sea ice cover/productivity and recommend this manuscript for publication. I do have a few comments for minor revisions, nevertheless:

Results

Line 257 onwards. It is not clear to me which sediment set you are discussing in terms of the GDGT and OH-GDGT data. Is it the 13 “Baffin Bay” samples collected for your study or are you including the 70 previously published samples? It would be good if you make the sample number clear in the text and in the figures. It is also confusing when the authors switch on line 283 to talking about “northern Baffin Bay”. I assume that you mean just the 8 NOW sediments, but this needs clarity. Including number of samples with the correlation coefficients would make it clearer for the reader when you are examining 13 sediments, or just the 8 in the NOW sample subset and if you are also ever including previously published data. Indeed, if all your GDGT and OH-GDGT data analysis is for the 13 samples then you aren’t saying anything with the GDGTs about the NOW (as you did do for the HBIs and sterols) but rather about the entire Baffin Bay area. This seems to deviate from the title and aim of the manuscript.  

It is a pity that the authors have not presented more of the GDGT dataset. For example, the GDGT & OH-GDGT fractional abundance for each of the 13 stations in a table or in a supplementary figure (perhaps similar as to that done for the sterols and HBI concentrations; Fig. S1 - S3). While Figure 6 provides an overview of the entire data set (although I’m still unclear as to whether this is 13 samples or not), I think it would be useful to know whether the GDGT and OH-GDGT distribution varies spatially inside and outside the NOW. Perhaps something similar to the way Spencer-Jones et al. (2021) presented the distribution of archaeal lipid (headgroup)s in different polar water masses. I expect that spatial differences in GDGT & OH-GDGT (core) distributions would not be great in surface sediments, but as you have the data it is a pity not to present it, or at least include a line about any spatial homogeneity or heterogeneity in the results.

Discussion

Overall, the discussion of GDGTs and OH-GDGTs is clear but I have a few suggestions. The sentence on lines 413 – 415 about the cren isomer reminded me of the analysis in Bale et al. (2019) of the cren’ ratio in 58 thaumarchaeotal cultures. They reported that growth temperature has a larger effect than thaumarchaeotal phylogeny on the proportion of the cren isomer.

I found this BGD preprint and discussion https://cp.copernicus.org/preprints/cp-2022-19/ useful when thinking about the relationship between hydroxy-GDGTs and shifting archaeal species composition and/or salinity. Perhaps if it is in print in time, the authors could reference it in relation to the statement on lines 378 – 380 and perhaps in relation to the statement on line 456.

A note throughout the manuscript. The authors refer to the cren isomer as a regioisomer, whereas this has been proven not to be the case and is more likely a stereoisomer (Liu et al., 2018; Sinninghe Damsté et al., 2018).

One final suggestion I would like to put forward is based on a number of recent articles about decolonization of geosciences, for example Liboiron (2021). Would the authors be prepared to acknowledge the indigenous people that traditionally occupy the regions they sampled? I believe that Inuit Nunangat occupy the West (Canada) and Kalaallit the East (in Greenland). More information about the people in the sample areas and how to acknowledge them can be found at https://native-land.ca/

Minor edits

Line 124 – change to “depth”

Line 135 – change to “planktonic”

Line 189 – remove extra )

Line 260 – the second half of this sentence is confusing.

Lines 262 – 263 – This is a confusing sentence as figure 7 only presents temperature correlations and the other variables listed are shown in the supplement. Perhaps easiest here to expand the figure range (e.g., Figs 7, S4 – S6).

Line 265 – change to “their correlation”

Line 269 – is this line missing the word temperature, SST perhaps?

Line 272 – change to either “autumn subsurface temperatures between 60 and 80 m bsl also feature similarly significant correlations” or “autumn subsurface temperature between 60 and 80 m bsl also features similarly significant correlations”. I’m not sure which of these you mean but you currently have a mix of both.

Lines 283, 285 – This should be a reference to Figure 8 not 9.

Line 380 – Could you add the actual samples size (i.e. n = x) and temperature range here?

Line 425 – this should be Fig 8c?

References

Bale, N. J., Palatinszky, M., Rijpstra, W. I. C., Herbold, C. W., Wagner, M., and Sinninghe Damsté, J. S. (2019). Membrane Lipid Composition of the Moderately Thermophilic Ammonia-Oxidizing Archaeon “Candidatus Nitrosotenuis uzonensis” at Different Growth Temperatures. Appl. Environ. Microbiol. 85. doi: 10.1128/AEM.01332-19.

Liboiron, M. (2021). Decolonizing geoscience requires more than equity and inclusion. Nat. Geosci. 14, 876–877. doi: 10.1038/s41561-021-00861-7.

Liu, X.-L., Lipp, J. S., Birgel, D., Summons, R. E., and Hinrichs, K.-U. (2018). Predominance of parallel glycerol arrangement in archaeal tetraethers from marine sediments: Structural features revealed from degradation products. Organic Geochemistry 115, 12–23. doi: 10.1016/j.orggeochem.2017.09.009.

Sinninghe Damsté, J. S., Rijpstra, W. I. C., Hopmans, E. C., den Uijl, M. J., Weijers, J. W. H., and Schouten, S. (2018). The enigmatic structure of the crenarchaeol isomer. Organic Geochemistry 124, 22–28. doi: 10.1016/j.orggeochem.2018.06.005.

Spencer-Jones, C. L., McClymont, E. L., Bale, N. J., Hopmans, E. C., Schouten, S., Müller, J., et al. (2021). Archaeal intact polar lipids in polar waters: a comparison between the Amundsen and Scotia seas. Biogeosciences 18, 3485–3504. doi: 10.5194/bg-18-3485-2021.

The manuscript by Harning et al. presents the results obtained from 13 surface sediments collected in the largest Arctic polynya (North Water Polynya, NWO). The authors analysed HBIs, sterols, and GDGTs which are used to calculate sea ice- and temperature-related indices. Based on their data, the authors discussed the utility of the paleoproxies and introduced two local calibrations for TEX86-L and RI-OH. Although their attempt sounds reasonable, the dataset is very small with the very narrow temperature range of 2°C and the correlations between indices and temperatures are moderate (R2 <0.5). Nonetheless, the data are valuable since there were no GDGT data from the study area in the global dataset published before. Some issues are listed below, which should be better addressed before the manuscript is accepted.

Major comments:

The authors suggest that all HBIs are derived from sea ice diatoms in Baffin Bay and thus cannot be used to distinguish sea ice and open water conditions. Although they all might be produced by sea ice diatoms, their concentrations and distribution patterns are different. Potentially, they might be derived from different diatom species. Discussion about potential biological sources of individual HBIs can be added based on the literature, although there are no direct information on specific species from Baffin Bay.

The TEX86-L calibration was based on 0-90 m water temperatures. But the R2 values are similar to those in 40-90 m water depth as shown in Fig. 7. Although the p values are >0.05 below 90 m water depth, this might be due to insufficient instrumental data. So it will be interesting to show how the calibration based on 0-200 m water temperatures does look like as well.

I see that there are no GDGT data previously published in the study area. However, there are HBIs data previously published. I feel that the discussion about HBIs is in general based on 13 samples, not well integrating the previous data from 70 sites. These data are not even incorporated in Fig. S1 to S3. It is not clear what might be the reason.

Other comments:

Line 177: an Agilent DB-1MS GC column (60 m x 250 μm x 250 μm)? Is the column information correct?

Line 182-185: Concerning to the response factors for HBIs quantifications, it is not clear how the approach used in this paper is comparable to that used in the paper by Belt et al., 2012.

Line 194-196: Similarly, concerning to the response factors for sterols quantifications, it is not clear why cholesterol is used instead of target sterols directly. The standard samples for ß-sitosterol, brassicasterol, and campesterol are available in the markets, except for dinosterol.

HBI III and HBI IV: It would be good to show HBIs chemical structures as a supplementary figure.

Line 236-237 & Fig. 4a: Looking at Fig. 4A, the standard deviations of mean concentrations of all HBI compounds appear to be overlapping. To better demonstrate the difference, some additional statistics should be done.

Line 240-246 & Fig. 4c: The sample number can be added. In addition, some statistical analyses should be done to better demonstrate whether the datasets between NOW and non-NOW are different. Although it is written in the text like “Although the standard deviations of mean dinosterol and campesterol concentrations overlaps between NOW and non-NOW sites, the standard deviations of ß-sitosterol and brassicasterol for the two regions is statistically different (Fig. 4c and S2).”, Figure 4c rather shows that ß-sitosteroal and dinosterol are different between NOW and non-NOW while the standard deviations of mean brassicasterol and campesterol overlap.

Line 249: The balance factors were obtained based on a combination of previous and current datasets. However, later on, the major conclusions related to the PIP25 indices were based on the current study (i.e. n=13). How are the balance factors if they are calculated only based on the current study? Are the resulting PIP25 values similar?

Line 259: “….the standard deviation of mean values between these regions is not statistically different (Fig. 4e and S3).” – What kind of statistics were done? The statistical results were not shown to compare both NOW and non-NOW datasets.

Line 266-280: The authors show the R2 values but it is not clear on how many samples these are based. Please provide the number of samples.

Line 281-286: There is no Fig. 9a and 9b.

Line 320-322: In the Kolling’s dataset, HBI III is correlated with dinosterol and brassicasterol which is not observed in the current study. What would be the reason?

Line 325-326 & Fig. 4: There are also some differences in sterols between two datasets. Is there any possibility that this is due to the different quantification methods applied?

Line 345-347: It is somewhat confusing to see the difference for PBIP25 and PDIP25 between two datasets. How does it look like if the data from the sites in front of fjords in the Kolling’s dataset are removed? If so, is the difference less then?

Line 401: Besides the regression analysis, other statistical analyses, such as PCA and RDA would be helpful to better illustrate the impact of the main environmental factors on the GDGT distributions.

Fig. 5: It is a little bit confusing to see the correlations between the same compounds. It is obvious that they have the value of 1. It would be better to remove them.

Fig. S1, S2, and S3: The color bar scale is not visible.

Table: It would be beneficial for other researchers to present data of individual HBIs, sterols and GDGTs as an Exel file or tables in Supplementary Information, although they can be deposited in a website later on.

Harning and co-authors present a study of biomarker distribution in surface sediment samples from Baffin Bay, with focus on identifying biomarker signatures for the North Water Polynya that may inform paleo-environmental reconstructions in the region. This is both a well-justified and significant effort. On the one hand, the North Water is a globally important ecosystem that is particularly vulnerable to climate change. On the other hand, there is considerable uncertainty in the behavior, source producers (HBIs and sterols), and fidelity of biomarker proxies and thus local and regional data are essential to refine their usability.   

The study includes a broad range of biomarkers and this is, in my opinion, the main strength of the work. It is also very well-written, and the figures are generally well prepared and adequate (see below for some suggestions for improvement).

My comments will focus on the HBIs and sterol work, as GDGTs are outside my realm of expertise.

While the study provides some useful insights (adding, for example, to the discussion on whether HBI III and HBI IV might in fact be produced by sympagic taxa), I found some of the data analyses and conclusions problematic and cannot recommend publication of the paper in its present form. I encourage the authors to consider a few points and revise the manuscript accordingly.

General comments:

- Limited number of samples and novelty

The study includes a surprisingly low number of samples (n=13, n=9 analyzed for bulk geochemistry) to represent such a large area. I found there are some overstatements throughout the manuscript that should be corrected having in mind the small sample size, and what indeed it adds in terms of novelty (or not) to the large Kolling et al. 2020 study. For example, in the Introduction it is mentioned that samples were collected in 2008 and 2017 but the total number is not given. From the table I assume 2008 (n=3), 2017 (n=10). It is also written in the introduction that biomarker proxies were assessed against modern instrumental data but this was only presented for GDGTs.  Another overstatement, although minor, in the discussion (line 299) is saying that “several additional sterols” were added compared to the work of Kolling et al. 2020 when in fact only two more were analysed.

-  Lack of information on surface sediment samples

In order to be able to assess the new findings, it is important that the authors provide additional information on the samples. Table 1 can be expanded to include at least “year of collection” (might not be obvious to all from the cruise code), “coring device”, and data on bulk geochemistry. Figure 3 shows 13C and C/N data for the 9 samples, but it is not clear which ones these are. Also, TOC values should be given in the table, as these are important when considering biomarker concentrations (see comment below also). Is there any age control on the surface samples? Were any of these cores analysed for 137Cs/210Pb? This would give some confidence at least that they might indeed represent recent deposition.

- Comparison with Kolling et al 2020

I found some of the comparisons with the Kolling et al. dataset quite confusing. In the materials and methods, it is written that the new dataset (n=13) will be compared with a subset of samples (n=70) from Kolling et al and that samples collected within fjords and bays will be excluded. However, later in the discussion, it is argued that the difference in brassicasterol and dinosterol trends between the two are likely due to the fact that some of the sites are in the vicinity of large fjords. If there is uncertainty whether some samples might be skewing the response, the authors could run the comparison with a different subset and evaluate if this is the case. Given that the Kolling et al dataset includes many more samples, the ranges of water depths, sedimentation rates, and likely sediment composition are likely larger than for the n=13 dataset. Simply comparing biomarker concentrations per volume of sediment across the two datasets and the now vs. non-now sites is not adequate, in my opinion.

- Influence of TOC contents on the biomarker signals

Given that the NOW is a highly productive area, one can expect that TOC values for the NOW sites will be generally higher than for the non-NOW sites. It has been recommended, and is common practice in paleo sea ice reconstructions using HBIs, to normalize the data by TOC. This way, we account for down-core changes in sediment composition. The same would be important for a dataset like this one, where large changes in sediment composition and organic matter content can be expected across the region. I strongly recommend that the authors plot all biomarker data in ng.gTOC-1 and revise the discussion and conclusions accordingly. Figures S1 and S2 might also show quite different spatial trends if TOC values are accounted for.

- Recommendations for paleoenvironmental reconstructions

This study highlights the complexity of biomarker signals in the highly dynamic Baffin Bay region, and our limited knowledge of their mechanistic behavior and applicability. Given my previous comments, and the uncertainty linked to comparing NOW vs. non-NOW sites based on concentrations of biomarkers per volume of sediment without accounting for sediment composition and in the absence of any form of age control, I cannot agree with the recommendation of proposing one type of biomarker (sterols) as a “more appropriate tool” (rather than HBIs) to characterize the NOW in the recent past. On the contrary, I think this study is a perfect example of why we need to pursue a multiproxy approach and not rely on single proxy lines of evidence. I would like to mention here that previous Holocene records from the NOW have mostly followed a multiproxy approach including microfossils, biomarkers, and biogeochemical proxies and I would be concerned if the community, based on this study, would go ahead and attempt to reconstruct the NOW based on sterols alone.

Detailed comments:

Lines 7 and 38 - Greenlandic Inuit is not a language. The correct term is West Greenlandic or Kalaallisut.

Lines 33,34 – The instability of the NOW (e.g. Ribeiro et al. 2021) has been shown by a combination of multiple proxies, including lipid biomarkers, microfossils and bulk biogeochemistry (not just lipid biomarkers).

Lines 44-45 – human occupation timelines are incomplete and outdated, please correct. See:

- Ribeiro et al. 2021 Fig 5 (already cited in this manuscript) and references therein, mainly: 1) Raghavan, M. et al. The genetic prehistory of the New World Arctic. Science 345, 1255832 (2014). And 2) Grønnow, B. & Sørensen, M. Palaeo-Eskimo migrations into Greenland: The Canadian Connection. In Dynamics of Northern Societies. Proceedings of the SILA/NABO Conference on Arctic and North Atlantic Archaeology, Copenhagen (eds Arneborg, J. & Grønnow, B.) 59–74 (National Museum, Studies in Archaeology & History, 2006).

Line 47 – This study should be mentioned here as well: Vincent, R. F. A study of the North Water Polynya ice arch using four decades of satellite data. Sci. Rep. 9, 20278 (2019).

Line 146 – specify coring device in the table per sample

Line 149 – here, add any information on age control for the core tops if possible.

Line 226 – did you mean to write “shoulder season months”? I am not familiar with this expression.

Line 229 – TOC data should be added here. Specify in the table which of the 9 samples were analysed for bulk geochemistry.

Line 307-308 – One could argue for the opposite, given that polynyas are characterized by intense sea ice formation, and the polynya area is under the influence of Arctic sea ice export. I suggest revising this section.

Line 227 – Would be useful to add more information here on other potential sources for campesterol and b-sitosterol (besides marine diatoms). Also note that the Detleft et al 2021 study is in a fjord setting, while the datasets in this study excludes such settings. I don’t completely follow the reasoning that correlation (of sterols) is supportive of a common source.

Line 350 – replace “all” with “partly” – biomarkers may partly originate from sea ice diatoms

Lines 365-366 – please revise this conclusion. Sterols alone are very unlikely to help us characterize the presence/absence of the NOW in the recent geological past, and other tools are available, such as “true” open water indicators.

Lines 472-473 – It is important to verify if this holds when accounting for TOC contents.

Line 479 – I suggest some caution here since sterols are not unequivocal open water indicators.

Figure 4 – I suggest replotting with TOC normalized values

Figure 5 – It took me a while to make sense of this figure. I suggest ordering the biomarkers in the same way as Fig 4 so they are easily comparable. IP25, HBI II, HBI III, HBI IV, Dinosterol, Brassicasterol, Campesterol, b-sitosterol. Also specify if the plotted data in b) are all Kolling et al or a sub-set as indicated in the text? Sample sizes (n=) should be given for all figures.