Dear Sebastian Naeher and Ashley Abrook,

This manuscript presents measurements of iso- and br-GDGTs in three Holocene varved sediment sequences in northern Europe, as well as at higher resolution for the last 300 years for two of the sites, allowing comparison with instrumental records. An impressive amount of data is presented, and the manuscript is well written with good structure. By presenting three new Holocene records, this manuscript is a valuable contribution to paleoclimate knowledge in northern Europe, and the investigation of GDGTs in varved lake sediments is of clear relevance to the proxy community. However, some interpretations, particularly for sparsely sampled late Holocene intervals, are not robustly supported by the data. A deeper discussion of discrepancies between GDGT based reconstructions and instrumental records would also improve the manuscript. Provided that these points and the more detailed comments below are adequately addressed, I see no issues with publication of this manuscript.

General comments:

• Why are the last 300-year data not included in the Holocene reconstructions presented in figs 5, 6, 7 and 9? It would be very useful to extend the long records towards the present if possible. If suitable, I would suggest adding it, or clarify in the manuscript why it is not suitable.
• The high resolution recent (last 300 year) data that is compared to instrumental data is very interesting. Since one of the cores follow the instrumental data, and the other does not, I think this section deserves more attention and some further discussion would be useful, as mentioned below. Can the findings from the core-tops be further used to indicate which records are more robust down-core?
• Section 4.5.1 discussed the temperature trends over the Holocene based on the three new records. Conclusions are made based on the changes in trends over the later Holocene, but the records only contain 2-4 data points over this period, leading to very high weight being given to single datapoints. Given that there can be large scatter in GDGT data, this is not robust, and the section should be reworked. Potentially this can be improved by adding the last 300 year data as mentioned above.
• Is TOC data available from the studied cores? Comparing the GDGT indexes with TOC may be very useful, since organic contents may be an important variable that covaries with GDGT distribution changes. Sudden shifts in TOC contents are strong indicators of environmental and limnological shifts, and are therefore likely to influence also GDGT distributions, see for example Hällberg et al., 2023 (https://doi.org/10.1016/j.orggeochem.2023.104702).
• 7 methyl and GMGT data from these sites would be interesting to see from these sites, if that data is available. It may provide clues to the provenance of the GDGTs as well as potentially providing an additional indication of temperature variability (Baxter et al., 2019, https://doi.org/10.1016/j.gca.2019.05.039). Of course, this manuscript is already long, and 7-methyl brGDGTs and GMGTs should only be included if the authors deem this to have large explanatory power.

Specific comments:

• Fig 1:
  • in the B inset, lake depth isobars are not numbered as in a) and c) insets. This would be good to add.
  • The brown symbols in depositional model for a) is not indicated in the legend what they are.
  • It is also not clear in the figure text why Diss Mere has three panels for the depositional model, so please add that explanation there. What does the light brown/green shading indicate?
  • Letters in the lake insets for a) and b) are not explained in the figure text, and what are yellow markers in b)? If they are important in the manuscript, please explain them in the figure text. If not important, perhaps remove from figure?
• Site description sections. Please provide references and explanation for the meromictic conditions at the lakes already here.
• How does the 0.5 cm resolution compare with the varve thickness? It would be interesting to know if it is technically possible to reach sub-annual resolution, and if some samples represent that in this study.
• The manuscript refers to GDGT-0 vs crenarchaeol as %GDGT-0. I would expect that to refer to the fractional abundance of GDGT-0 (relative to all isoGDGTs). Instead, why not refer to it simply as GDGT-0/cren, as frequently done previously, for instance by Baxter et al., 2021 https://doi.org/10.1016/j.quascirev.2021.107263 ? This would make it much clearer what is meant.
• Refer to Hopmans et al., 2004 https://doi.org/10.1016/j.epsl.2004.05.012. Since that original publication is used in marine settings only, I would suggest also referring to a study showing that it can also be used as an aquatic/land signal for terrestrial sites.
• Is this ‘wrong’ classification correlated with an increased TOC from that core? Since TOC has previously been found to strongly correlate with GDGT distributions elsewhere (see for example fig 3 in Hällberg et al., 2023), I think it would be valuable to show this data, as mentioned earlier, if available.
• “Across all three sites the BIT index is very high and is > 0.92”. A BIT value of 0.92 is a significant change from >0.99 which is normally found in soils/terrestrial sites, and I would therefore suggest that a value of 0.92 likely represents a significantly different GDGT distribution, likely resulting from some environmental change. However, looking at figures 5-7, I do not see any such low values, so perhaps this is a typo?
• Fig 4. What are the “modern” samples in a) and b)? Surface samples?
• Fig 4. What are the plot labels in the RDA plots? Ages? In that case, I would suggest adding a fill color gradient based on that value, to increase readability.
• Fig 5e. The arrow with “marine provenance” needs to be relabeled, since i assume you mean aquatic rather than marine?
• Is the value 0.9 correct? It looks like it’s higher based on the figures.
• 375 section, on source attribution. Additional evidence for soil input can be derived from degree of cyclization, IR6-methyl and 7 methyl GDGTs as done by Martin et al., 2019. GMGTs and their isomers may also provide further clues and indicate bacterial community shifts, see Hällberg et al., 2023.
• Please clarify how it is reflected in fig 4: “Nautajärvi are classified as ‘peat-type’ (this is also reflected in Fig. 4)”
• Reference needed.
• It is probably useful to also mention the results of Baxter et al., 2021 (https://doi.org/10.1016/j.quascirev.2021.107263) from meromictic Lake Challa here.
• Section 4.3.2, and in particular lines 468-475. CBT’ calculation includes 6-methyl brGDGTs, and is therefore a mix of cyclization and isomerization, despite the (perhaps misleading) name cyclization of branched tetraethers. It would therefore be better to compare degree of cyclization (Sinninghe Damsté et al., 2009 https://doi.org/10.1016/j.gca.2009.04.022) and IR6me instead of CBT’ and IR6me, to disentangle these compound influences.
• Fig 8. The figure text is quite messy.
  • The panels should be a, b, c, d, since it’s four panels.
  • Currently, panel b) is not specified in the text, only a).
  • I would propose labelling the panels for easier readability, such as “MAAT”, “Tmay-Nov” or similar.
  • Better to use common era (CE) instead of BC/AD?
• It would be interesting with a deeper discussion of your results from the short core presented in fig 8. Nautajärvi has a very good agreement with the instrumental record, but Diss Mere shows the opposite trend. Please elaborate on this, and potential causes for it. The sudden offsets in temperature around 1940 in Diss Mere may be useful in investigating this. What happens in the GDGT distributions (or other data) to cause this offset? After the offset, the reconstructed temperature is lower than before, contrary to instrumental data which show warming. Any clues to why?

• Fig 9.
  • I would propose to show all records with the same y axis spacing, so that it is possible to see which sites show very little change versus larger change.
  • The arrows indicating trends appear quite arbitrary and require some better explanation.
• Specify if you mean the trend here rather than amplitude.
• 541: “Peak warmth occurs at Diss Mere at ~ 8 ka BP” this seems to be based on only a single datapoint, and is only true for the MBT based calibration, but at odds with the Raberg calibration. The way I read that graph (9h), the Diss Mere reconstruction shows slightly lower temperatures at 10-9 ka BP and after ~3 ka BP, but this is based on very few datapoints. The period 8-4 ka BP has very slightly elevated but variable temperatures. The temperature at ~5.9 ka BP is for example the coolest of the full record. I therefore find the discussion of the Diss Mere temperature evolution to be lacking in robustness, and higher sampling resolution would likely be needed to draw these conclusions. That said, like mentioned earlier, if the near surface data are added to the full reconstruction, the trends for the Late Holocene may be clearer.
• The statements about peak temperatures around 5.6-4.3ka BP at Meerfelder Maar also doesn’t appear robust, with at least the datapoint at around 3k showing comparable temperature, with only one datapoint after that showing a slight cooling.
• Specify what is meant by “this” at the start of the paragraph.
• But it is highly uncertain how much of this 2-3 degree temperature variability stems from methodological uncertainty/scatter versus actual climate shifts. This needs to be mentioned in the text.

Technical or minor corrections:

• The original reference for tex86 is Schouten et al., 2002 https://doi.org/10.1016/S0012-821X(02)00979-2
• This sentence comes a little out of the blue and requires a little more clarification. Is this threshold applied in this study?
• Equation 9. Typo. As written now, it mathematically makes no sense. It should be GDGT0/(GDGT0+cren).
• Why capitalization of crenarchaeol?
• Section 3.1. Please refer to figure numbers when presenting index results.
• Fig 4. To improve readability of this figure, I would suggest removing “brGDGT” in front of each compound. It is unambiguous that Ia, IIa etc. are brGDGTs. This can also be done for figure 2 axis labels.
• Fig s1. Add in text that it is based on TEX86.
• Specify that you mean in reconstructed LST?
• Figures s4, s6, s8. Please make the order of the elements the same in all figures. Currently, the s6 figure has other order than s4 and s8, which reduces readability.
• “vs” doesn’t need to be in italics. But if you decide to still do that, be consistent throughout. For example, it is not in italics on line 362.
• Reference should be in parentheses.
• Replace imperfect with moderate, or similar.

In this manuscript Abrook et al. present a compilation of three european lake records from Holocene-aged varved sediments. These records are analized for their capability to preserve GDGT signals within the varves and whether the conditions that generate the varves may override the temperature signals of the GDGTs. Within this work the authors show that unique signals to the different lakes are captured, which can be related to specific conditions of the water chemistry, nevertheless, taken as a whole, the records suggest that the GDGT temperature signal is preserved, suggesting that varved sediments provide a useful tool to generate high resolution GDGT records.

I find this manuscript very interesting and it presents a very exciting dataset, additionally I consider that the main takeaways from the study are adequate and follow the data presented. The manuscript is also generally well written. I do however have some comments which I hope could help improve the manuscript and I will be happy to support the publication of his work once these have been addressed.

Main comments

I agree with Reviewer #1 that further comparison between the modern and Holocene records could be done for Nautajärvi and Diss Mere. In addition to the points made by Reviewer #1 I would suggest discussing further the separation observed for the Holocene and modern samples in Nautajärvi. Did you considered further PCs in the Diss Mere PCA that may show a similar spread (that may indicate a common feature explaining this)? Particularly since the first two PCs for Diss Mere account only for ~60% of the variance.

Additionally, given the dataset presented here, I think this could be an excellent opportunity to compare these samples and the changes in mixing regimes with the proportion of cren/cren’, as was proposed by Baxter et al., 2021 (10.1016/j.quascirev.2021.107263).

Finally, throughout the text, I find that the Figures and figure captions could be modified to be clearer and easier to read. See detailed comments below.

Specific comments

-Figure 1: The labelling of the different panels is confusing and starts over in the diagrams of he couplettes. Additionally the maps contain features that are not explained in the text.

-Line 219 and throughout the text: capitalize brGDGT and isoGDGT when at the beginning of a sentence (i.e., BrGDGT and IsoGDGT).

-Figure 3. Is b) a zoom in on triplot on a). Please explain in legend. 

-Figure 4. Label each plot with a letter for clarity.

-Figure 8. Legend is very hard to follow, should be rephrased.