Referee comments are in italics and our responses are in bold.

This is a very well written paper that details the pros and cons of using GDGTs to reconstruct temperature in the paleorecord from Lake Chala. The study highlights the complexities of the modern lake environment, and how variations in the system through time make application of GDGTs to temperature reconstruction only appropriate in some intervals.

In general, I find the discussion very thorough, using other types of data to support their interpretations of GDGT data. The discussion of both isoprenoid and branched GDGTs is welcome.

We are very thankful to the referee for their positive assessment of our work. 

I find a few places in the discussion where it seems that the authors are speculating. For example, is there other evidence for the development of the strong chemical gradient over time? This is certainly possible, but I didn’t see any solid evidence for that. I don’t think it is a problem to speculate in this type of paper, but they should be clear when they are doing so.

We see the point of the reviewer and plan to strengthen the arguments for the development of the chemical gradient, partly, by referring to data on fossil diatom assemblages (Tanttu, 2022) and partly by more explicit description of i) the evolution in lake hydrology over time due to gradual infilling of the crater basin with sediments, and ii) the expected effects of major lake-level changes on the water column’s dissolved-ion content.  

The authors used the Pearson 2011 calibration to estimate temperature from brGDGTs, ultimately concluding that this “MST” reconstruction that combines the 5 and 6 Me GDGTs appears to be more appropriate to use in Lake Chala than MBT’_5ME. This is somewhat surprising as there is substantial evidence that separating the two groups improves temperature reconstruction, though the logic they use is clearly laid out. I am very curious to see if either the Zhao et al 2023 (QSR), O’Beirne et al 2023 (GCA), or Wang et al 2024 (EPSL) calibrations provide more reasonable results. I fully understand that these calibrations were not yet published when this work was being performed, but now that they are out there I wonder if they do a better job that the Pearson calibration, which has issues of its own that compromise its application.

It is true that separating the 5- and 6-me brGDGTs has improved the statistical relationship between brGDGT distribution and temperature in calibration data sets, i.e. brGDGT distribution in recently deposited sediments has become a better ‘predictor’ of temperature measured today at the sites included in those calibration data sets. However, there is no concrete evidence (for example, through comparison with time series of instrumental temperature data or with parallel reconstructions based on independent temperature proxies) that this improvement in modern-day calibration leads to more accurate reconstructions in down-core applications. Instead, as discussed in our paper, our results highlight that optimizing the calibration does not guarantee a credible long-term reconstruction. Further, considering the multitude of data presented in this paper, the already lengthy discussion (which we deem necessary to drive home the point made above), and the fact that the brGDGT palaeotemperature record is not the main topic of the current manuscript, we choose to not incorporate and discuss temperature reconstructions based on alternative calibrations into the present manuscript. Instead we refer to the detailed justification provided in Baxter et al. (2023, Nature) for our selection of the Pearson et al. (2011) calibration over other calibrations, including three that are based on MBT’_5Me (Russell et al., 2018; Martínez-Sosa et al., 2021; Raberg et al., 2021). We showed there that any calibration based on the ‘classically’ defined MBT’_5Me will result in reconstructed temperature records that are likely incorrect. As to the possible merit of the three newest calibrations cited by the referee, we plan to discuss potential application of these new calibrations in a forthcoming publication in preparation which focusses primarily on the Lake Chala temperature record and its climatic implications. 

A few minor typos:

            Line 164 needs closing parenthesis after johnson et al., 2016

            Line 351: change to “…Similar to brGDGTs, isoGDGT-0 is also produced”

            Lines 361 and 372: change to “Table 1”

            Line 461: Change to “…The degree of cyclisation (DC) of brGDGT was also calculated…”

            Line 497: change to “…within…”

            Lines 652-653: sentence is unclear, use of “respectively” twice in once sentence

            Line 676: change to “confirms”

            Line 722: change to “Stage”

            Line 796: change to “from”

            Line 886-887: used “inferred” twice

            Line 923: change to “insolation”

            Line 924: change to “from around”

We will change all outlined minor corrections as suggested by the reviewer in the revised manuscript. 

You use both “GDGT” and “GDGTs” when referring to them as plural. Choose one and use it consistently

We reviewed this inconsistency and will use “GDGTs” for the plural form in the revised manuscript. 

Referee comments are in italics and our responses are in bold.

This is a very well written paper that details the pros and cons of using GDGTs to reconstruct temperature in the paleorecord from Lake Chala. The study highlights the complexities of the modern lake environment, and how variations in the system through time make application of GDGTs to temperature reconstruction only appropriate in some intervals.

In general, I find the discussion very thorough, using other types of data to support their interpretations of GDGT data. The discussion of both isoprenoid and branched GDGTs is welcome.

We are very thankful to the referee for their positive assessment of our work. 

I find a few places in the discussion where it seems that the authors are speculating. For example, is there other evidence for the development of the strong chemical gradient over time? This is certainly possible, but I didn’t see any solid evidence for that. I don’t think it is a problem to speculate in this type of paper, but they should be clear when they are doing so.

We see the point of the reviewer and plan to strengthen the arguments for the development of the chemical gradient, partly, by referring to data on fossil diatom assemblages (Tanttu, 2022) and partly by more explicit description of i) the evolution in lake hydrology over time due to gradual infilling of the crater basin with sediments, and ii) the expected effects of major lake-level changes on the water column’s dissolved-ion content.  

The authors used the Pearson 2011 calibration to estimate temperature from brGDGTs, ultimately concluding that this “MST” reconstruction that combines the 5 and 6 Me GDGTs appears to be more appropriate to use in Lake Chala than MBT’_5ME. This is somewhat surprising as there is substantial evidence that separating the two groups improves temperature reconstruction, though the logic they use is clearly laid out. I am very curious to see if either the Zhao et al 2023 (QSR), O’Beirne et al 2023 (GCA), or Wang et al 2024 (EPSL) calibrations provide more reasonable results. I fully understand that these calibrations were not yet published when this work was being performed, but now that they are out there I wonder if they do a better job that the Pearson calibration, which has issues of its own that compromise its application.

It is true that separating the 5- and 6-me brGDGTs has improved the statistical relationship between brGDGT distribution and temperature in calibration data sets, i.e. brGDGT distribution in recently deposited sediments has become a better ‘predictor’ of temperature measured today at the sites included in those calibration data sets. However, there is no concrete evidence (for example, through comparison with time series of instrumental temperature data or with parallel reconstructions based on independent temperature proxies) that this improvement in modern-day calibration leads to more accurate reconstructions in down-core applications. Instead, as discussed in our paper, our results highlight that optimizing the calibration does not guarantee a credible long-term reconstruction. Further, considering the multitude of data presented in this paper, the already lengthy discussion (which we deem necessary to drive home the point made above), and the fact that the brGDGT palaeotemperature record is not the main topic of the current manuscript, we choose to not incorporate and discuss temperature reconstructions based on alternative calibrations into the present manuscript. Instead we refer to the detailed justification provided in Baxter et al. (2023, Nature) for our selection of the Pearson et al. (2011) calibration over other calibrations, including three that are based on MBT’_5Me (Russell et al., 2018; Martínez-Sosa et al., 2021; Raberg et al., 2021). We showed there that any calibration based on the ‘classically’ defined MBT’_5Me will result in reconstructed temperature records that are likely incorrect. As to the possible merit of the three newest calibrations cited by the referee, we plan to discuss potential application of these new calibrations in a forthcoming publication in preparation which focusses primarily on the Lake Chala temperature record and its climatic implications. 

A few minor typos:

            Line 164 needs closing parenthesis after johnson et al., 2016

            Line 351: change to “…Similar to brGDGTs, isoGDGT-0 is also produced”

            Lines 361 and 372: change to “Table 1”

            Line 461: Change to “…The degree of cyclisation (DC) of brGDGT was also calculated…”

            Line 497: change to “…within…”

            Lines 652-653: sentence is unclear, use of “respectively” twice in once sentence

            Line 676: change to “confirms”

            Line 722: change to “Stage”

            Line 796: change to “from”

            Line 886-887: used “inferred” twice

            Line 923: change to “insolation”

            Line 924: change to “from around”

We will change all outlined minor corrections as suggested by the reviewer in the revised manuscript. 

You use both “GDGT” and “GDGTs” when referring to them as plural. Choose one and use it consistently

We reviewed this inconsistency and will use “GDGTs” for the plural form in the revised manuscript. 

Referee comments are in italics and our response is in bold.

This paper assessed the reliability of GDGT-based proxies for paleoenvironmental reconstruction in a long sediment core from Lake Chala, a deep meromictic lake in equatorial East Africa covering the past 250 kyr. The GDGT-based proxies, including BIT, IR6ME, MBT’5ME, and TEX86 over the past 250 kyrs, were interpreted based on previous modern system studies and compared with the evolution of the lake basin. They found that paleoclimate and lake evolution exert joint effects on these GDGT-based proxies and thus the use of these proxies in this lake appears to be complicated. This paper incorporates a wealth of data and very detailed data interpretation. Finishing such a paper would be a tough task. The paper is generally well-written. However, reading a paper like this would be a challenge to a reader because it is too long and there is too much information that needs to be considered. I do not have comments on the scientific issues. Instead, I strongly suggest the authors remove some unnecessary parts of the paper (moved to the supplementary material) and make the paper succinct and easy to read. That would attract a much wider interest to read through the paper.

We appreciate the compliments made by the referee about the quality of our manuscript. We would like to stress that, although focused on Lake Chala, our demonstration that lake evolution influences GDGT-based records of climate proxies is likely widely applicable to other lake systems. 

Crucially, this demonstration required thorough analysis of multiple GDGT compounds and proxies in relation to multiple independent sources of information on lake-system variation through time. Although we acknowledge that this may have rendered the length of the paper challenging for some readers, we feel that most sections (in particular the discussion) would lose in clarity or persuasiveness if shortened considerably. See also our reply to referee #2 for question relating to L175 below. 

Table 1 ‘%GDGT-2’  A bracket is missing.  Two ‘iso-GDGT-2’ were in the denominator.

We will correct this. 

L110  ‘only few’ is not accurate. There are many downcore applications of brGDGTs in lakes.

We will change this sentence as follows: “The strong correlation between MBT′ _5Me in lacustrine surface sediments and temperature (Russell et al., 2018; Martínez-Sosa et al., 2021; Raberg et al., 2021) has led to successful downcore applications of lake-based temperature calibrations since discovery of the 5-Me and 6-Me isomers (Feakins et al., 2019; Stockhecke et al., 2021; Zhao et al., 2021; Zhang et al., 2021; Garelick et al., 2021; Ramos-Roman et al., 2022; Parish et al., 2023), although uncertainties about the exact source(s) of brGDGTs in lakes still complicate interpretation of these records”.

Although there may now be several applications of brGDGT temperature calibrations to lake sediments, we would like to again stress that often these studies do not settle on the classical MBT′ _5Me  model, but used modified calibrations they deem more suitable for their lake records (e.g., Ramos-Roman et al., 2022, Parish et al 2023, Bittner et al., 2022).

L175 I suggest that you delete the detailed descriptions of all previous results or make a summary of them. Such a detailed description of previous GDGT work in Lake Chala makes the part look like a review. The paper is too long, which eliminates the interest of careful reading. This detailed description can be moved to the supplementary files 

We agree with the referee that the paper is long but it should be realized that this is the first GDGT lake record where so many sediment horizons have been analyzed, resulting in an unprecedented resolution. In addition, the present-day lake has been studied extensively in many aspects including extensive studies of GDGT production and settling over various time scales. Hence, it is likely fair to say that Lake Chala is the best studied lake in the world in this respect. The >15 years of work has resulted in an enormous amount of background information that actually allowed the detailed interpretation of the 250-kyr GDGT record. We feel that the reader of the paper will require a summarized form of this previous work to understand the interpretation of the record. Putting section 2 in a supplemental would make it much harder for the reader to follow the argumentation in the manuscript. Well-informed readers can simply pass over section 2, hence our choice to put this description in a separate section and not in the introduction.

L260 delete ‘then’

L439 following the method of Hopmans et al.(2016)

L459 calculated according to De Jonge et al. (2015).

Figure 10  in e) Stage VI should be stage IV

L796 from 

L924 from around

L932 where these lipids 

L951 Data on this chapter?  Which chapter?

L955 ‘review and edited the?’

L1033 Chen et al…. This is a preprint. The paper has been published in GCA.

Citation: https://doi.org/10.5194/egusphere-2023-2486-RC2  

We will correct all these minor changes in the final submission.