In this study, authors presented results of a Thaumarchaeotal culture enrichment, surface sediments, and two composite sediment cores from the Baltic Sea to assess the use of OH-GDGT-based indices, OH-GDGT%, RI-OH and RI-OH’, as sea surface temperature (SST) proxies. Based on their enrichment and surface sediments, those indices (RI-OH’ in particular) indeed respond to temperature changes, although it appears that quantitative estimate of temperature remains challenging. However, in their sediment core records, those indices become problematic when the Baltic Sea experienced from the fresh Ancylus Lake to the brackish Littorina Sea phases. The RI-OH’ values are exceptionally high during the fresh Ancylus Lake phase. Authors conclude that a reduced salinity could increase the RI-OH’ values substantially, and thus this effect should be considered when past salinity changed dramatically.

OH-GDGT-based proxies have been proposed to reconstruct past SST, however, confounding factors for these indices have not been fully evaluated, limiting their use as SST proxies. This study presents solid evidence that other factors could also affect those indices. The manuscript is well written and the scientific issue is well articulated and addressed.

I have only one comment. Authors believe that either thaumarchaea adjust their membrane lipid composition differently in freshwater vs. brackish conditions or different salinity could result in a shift in the Thaumarchaeotal species composition and suggest that the salinity effect should be considered when waters became fresh in the past. I’m not sure whether authors could recommend a GDGT-based index, ideally to be retrieved from the same OH-GDGT analysis, to signal this effect, without knowing its salinity history. Based on authors’ present and previous study (Warden et al., 2018), it appears that the BIT and MBT’5ME indices also changed correspondingly, associated with the changes in sedimentary environment, which appears to me that the archaea community may have changed substantially in response to the salinity changes. Thus, the BIT index could be used to signal this effect (but I understand that the BIT index does not indicate specific changes in the Thaumarchaeotal species composition)? Just a thought.

Line 25: "are omitted", incomplete phrase

Line 205:  change "as for" to "than"?

This paper presents OH-GDGT distributions in surface sediments and two sediment cores from the Baltic sea and investigated the potential impact of salinity on the OH-GDGT distribution in these samples. The sea surface temperature (SST) was found to be the major factor controlling the RI-OH’, and a higher culture temperature results in a higher RI-OH’ value in a thaumarchaeal enrichment culture experiment. However, several outliers were found to deviate from the linear correlation between RI-OH’ and SST in surface sediments of the Baltic sea, suggesting that a lower water salinity may increase the RI-OH’ value. In addition, the RI-OH´ index in a sediment core shows a good correlation with the TEX₈₆^L in the brackish phase of the Baltic sea whereas this index shows abnormally higher values during the freshwater phase. These pieces of evidence collectively suggest salinity should be considered when applying RI-O’ to the reconstruction of SST in sediment cores. This paper is generally well written and easy to read. The finding of salinity impact on the distribution of OH-GDGTs is interesting. I have some concerns that need to be addressed.

The authors may need more evidence to support the argument that salinity influences the OH-GDGT distribution. Two points in the paper indicate there might be a salinity effect: abnormally high RI-OH’ values occur in surface sediments with potentially low water salinity as well as in the freshwater phase of a sediment core in the Baltic sea. These unusually high RI-OH’ values were attributed to the low water salinity. However, it remains to be determined whether there is a causal relationship between higher RI-OH’ and lower salinity. It would be interesting to see whether there is a close relationship between RI-OH’ and salinity in the thaumarchaeal enrichment culture. Also, the previously published global data set may provide some clues for the impact of salinity on the RI-OH’ because those samples were collected from seas with different surface water salinity.     

L20-22  The inconsistency of OH-GDGTs between the thaumarchaeal enrichment culture and core top sediments might be because the dominant thaumarchaeota producing OH-GDGTs in sediments differ from that cultured in this study.

L144-145  Why do you use base hydrolysis here? The base hydrolysis can only transform a part of intact polar lipid (IPL) to core lipids (CL) while those polar lipids with glucose headgroup remain intact.

L146  grounded should be ground

L176  Lu should be Lü

L186  Why were the thaumarchaeota cultured only at 4 and 22 °C? Culturing this thaumarchaea along a temperature gradient would be better to test the relationship between isoGDGT distribution and temperature.

L217 and 250  - ~  keep one

L294 ‘grown at’  at which temperature?

L350 than that of …..

Figure 1  I am confused about the legend and the illustration of the figure. I did not see green squares but only see red squares, which represent sediment cores in the legend.

Also, a scale bar and the north direction should be presented.

Figure 2 and 3  I do not know what the grey lines indicate. If they are not used to highlight anything important, I suggest the authors remove these grey lines to make a clearer figure.

Figure 4   Liu et al. 2015  should be Lü et al.,2015

I have noticed the sediment cores used in this study have been dated and described in Warden et al. (2016). However, core depths, instead of the ages, were presented in the figures of this study. That is not helpful for interpretation of the SST change if the time frame is not well constrained.   

Sinninghe Damsté et al. describe the distribution of OH-GDGT lipids in Baltic Sea surface sediments, sediment cores and a thaumarchaeal culture grown at different temperatures. The manuscript is thorough and well-written and many significant, open questions on OH-GDGT proxies are addressed by the authors. However, there are some issues with the methodology that could complicate interpretation of the results. First, the authors used base hydrolysis to degrade intact polar OH-GDGTs into core lipids, presumably to facilitate analysis. While this method preserves the hydroxyl groups compared to acid hydrolysis, it does not quantitatively remove the headgroups of thaumarchaeal IPLs (Schouten et al., 2008, AEM) and may selectively degrade IPLs according to their headgroup composition. This will lead to biases in the resulting OH-GDGT quantifications and indices because it is known that OH-GDGTs mostly occur in the form of glycosidic IPLs rather than phosphatidic IPLs in Thaumarchaeota (Schouten et al., 2008, AEM; Pitcher et al. 2011, AEM; Elling et al., 2014, GCA), which are not significantly degraded to core lipids by base hydrolysis. Further, the proportion of core structures is different for different IPL types and markedly different between core OH-GDGTs and OH-GDGTs (predominance of OH-GDGT-0) and IPL GDGTs (predominance of OH-GDGT-2) in thaumarchaeal cultures including the strain most closely related to the organism used in this study, N. limnia (Elling et al., 2014 GCA; Elling et al. 2017 GCA; Pitcher et al., 2011 AEM). It is unclear why the lipids were not analyzed as IPLs, which would have circumvented these issues. If aliquots of the lipid extracts are still available, I would recommend re-analyzing the samples using a method appropriate for detection of IPLs. Second, the authors quantified the fragmentation products of OH-GDGTs at m/z 1300, 1298, 1296 and did not include the 1318,1316, 1314 ions, which could explain some of the variation compared to previous studies and during replicate analysis due to variations in ionization efficiencies due to. This approach was used by some previous studies (e.g., Lü et al., 2015, OG), i.e., in contradiction to what is stated in lines 159-161 and 321-322. Third, the authors do not provide sufficient information on the growth of the cultures such as detailed growth conditions (e.g., volume of medium, volume and type of container and closure, substrate concentration, medium composition) or growth data (e.g. growth curves) to allow assessment of the influence culture conditions on lipid profiles. For instance, it is probable that growth at 4 °C was much slower than at 22 °C. It is known that growth limitation and growth rate influence GDGT cyclization (Qin et al., 2015, PNAS; Hurley et al., 2016, PNAS) and thus it is probable that these factors influenced the presented OH-GDGT data in addition to the influence of temperature. These issues need to be discussed in the manuscript and their potential impact on the results from the culture sample needs to be quantified or acknowledged.

Line comments:

Line 18: Missing word? OH-GDGT [indices] for SST reconstruction?

Line 145: Please add the original citation of the extraction method, not just a derivative.

Line 225-227: This is confusing. If the concentrations of OH-GDGTs could not be determined, then how was %OH-GDGT calculated?

Figure 1: There are several issues with the legend and caption. The caption states that surface sediments are green squares, but the map and legend show blue circles. The caption further states that sediment cores are blue circles but they are red squares in the map and legend. The red text on brown background in the map are hard to read. Consider coloring land areas grey.

Figures 2-5 are very low resolution and hard to read. Adjust?

Figure 4: The colors for the “Baltic Sea (Kaiser and Arz, 2016)” and “global set” should be changed. They are indistinguishable for red/green colorblind people.

Figure 5b: This panel needs a legend displayed on the figure. The colors of the triangles are indistinguishable for red/green colorblind people.