Please see the attached document.

The authors investigated sections of a drill core from Lake Magadi (Hominin Sites and Paleolakes Drilling Project, HSPDP), a soda lake in Kenya, to reconstruct the microbial methane cycle of the lake system over the last 456 ka. The study is focused on molecular biomarker analysis, especially isoGDGTs, representing archaeal core lipids. Together with accompanying (organic-) geochemical data and published information, the authors interpret periodical shifts in microbial methane cycling (and consequently the archaeal community) to be associated with changes in the hydrothermal input at Lake Magadi. It is indicated that phases of low hydrothermal activity show increased microbial methane cycling as compared to phases of high hydrothermal activity.

Soda lakes are important habitats for life. Their investigation, including the microbial methane cycle over time, provides valuable information on these extreme environments and potential early Earth habitats. A detailed reconstruction of the microbial methane cycle of Lake Magadi over time does not exist so far. The findings of the study by Collins et al. are new, complement existing data, and improve our understanding of the Magadi system. The used core samples from the HSPDP are unique, and represent excellent material to study archaeal communities/the microbial methane cycle of Lake Magadi over time. However, the manuscript needs to be substantially improved in some areas before publication:

(i) The sampling strategy is not optimal (cf., l. 129–133). The authors focused on samples that were expected to have high total organic carbon contents (data not presented in the manuscript), which was only assessed by visual inspection (dark brown to black silty clay). The authors argue that those samples would yield the best results. This may have created a biased data set (also samples with low organic carbon contents may show a great molecular diversity). Additionally, the sampling scheme is not consistent. Between the defined intervals #1–6, several meters of core are not covered (3­­–15 m between single intervals), while within an interval the sampling steps are in parts as close as a few centimeters. It would be interesting to see, if microbial methane cycling was also active during the deposition of sediments with low organic carbon content.

(ii) The study lacks bulk geochemical data of the samples, which would be important to contextualize the presented biomarker and isotope data (e.g., total organic and inorganic carbon carbon contents, total sulfur content, bulk ¹³C_carb). Especially stable carbon isotope data of the carbonate phase (δ¹³C_carb) would improve the discussion of shifts in methane cycling (it seems that at least some samples contain carbonate, as the samples were acid-leached before ¹³C_org analysis; l. 174–175).

(iii) The presented bulk δ¹³C_org data lack context. In lake systems primary production and/or terrestrial input usually govern the carbon cycle. The presented data do not allow the assessment of the role of microbial methane cycling in the lake’s carbon cycle over time. It would help to present total abundances of compounds in relation to the total organic carbon content (amount per g TOC). In addition, the ¹³C_org data should be discussed together with the leaf wax data to evaluate the influence of terrestrial input on the ¹³C_org values. In the presented data set, only three values in interval #2 indicate methanotrophy (−48.1‰, −64.2‰, −89.4‰; Table 1), the rest of the ¹³C_org values could also be explained by variations in primary production and/or terrestrial input.

(iv) The discussion of microbial sulfate reduction in the system (e.g., l. 366–379) is not based on a solid data set. In the current version, only C_17:0 FAME and the appearance of pyrite are used to track microbial sulfate reduction. The C_17:0 FAME, however, is not only produced by sulfate reducers and represents a weak biomarker. Furthermore, it seems that only few samples contain C_17:0 FAME, and it does not necessarily co-occur with pyrite (cf., Table 1). The authors also speculate on the sulfate availability without presenting any robust indication on sulfate levels. Without further data (e.g., sulfur content, stable sulfur isotope composition of the pyrites) this part of the discussion needs to be significantly reduced. 

(v) The interpretation of increased microbial methane cycling at times of low hydrothermal input (and vice versa) is mainly based on the correlation of MI with REE data and Ca/Na-ratio. These data sets, however, do not always match (cf., figure 4). The authors should discuss the discrepancies in more detail, and present some explanations for the major discrepancies (e.g., low Ca/Na at the end of interval #2, high Ca/Na together with low REE abundance in interval #5, high REE abundance together with low Ca/Na at the end of interval #6). The MI data set seems to be much more consistent.

More specific comments:

• The title is misleading, as the manuscript is focusing on the reconstruction of the microbial methane cycle in Lake Magadi over time, driven by archaea, and not on the reconstruction of the entire microbial community and its change over time. Please replace “microbial” in the title by “archaeal”.
• The errors for the δ¹³C_org analyses should be presented (results section and Table 1).
• I suggest to include more details on the statistical evaluation (Fig. 5; PCA and correlation matrix) into the methods section.
• Why do the authors think the fatty acids <C_16:0 are degraded in the samples? I do not see any indication why this should be the case. The compounds were likely never present or below detection limit.
• In section 4.1.1 the authors discuss missing pyrite in some intervals and explain this by too small pyrite aggregates that could not be seen by the naked eye and/or sulfur incorporation into kerogen (l. 406–408). This is pure speculation. The authors could have easily checked the samples for small pyrite aggregates by using thin section microscopy and could have measured the total sulfur content.
• The headline of section 4.2 should be changed to something like “The influence of hydrothermal activity on the microbial methane cycle”.
• The REE data should be discussed in more detail in section 4.2.
• Figures 3 and 4 should be turned 90° and stretched (differences e.g. in interval #1 are barely visible in the current version), with age/depth on the y-axis. It would also be important to include the stratigraphic units and different lithologies.
• Please carefully check the color coding of the symbols in figure 6. Shouldn’t the cross at ca. 67% crenarchaeol be green or is the cross incorrect? What about the triangle at ca. 6% GDGT-2 (maybe blue or incorrect symbol)?
• Please add some representative GC chromatograms for each interval to the supplement.

Minor comments:

L. 30: Please list some studies that have investigated soda lake sediments/sedimentary rocks over geologic time scales here (some are already mentioned in the manuscript, incl. those from Lake Magadi).

L. 30/31: Delete space before full stop.

L. 87: Delete bracket in front of [2].

L. 92: Replace “microbial” by “archaeal” (please also do so in other relevant areas of the manuscript not mentioned here).

L. 95: The “n” of n-alkanes should be written in italics.

L. 117: Insert space in front of “Although”; delete comma behind “it”.

L. 188–190: Please check for correct phrasing (verb missing?).

L. 282: Please also calculate a mean value without the three outliers.

L. 324–326: Please check for correct phrasing (verb missing?).

L. 336: Replace “microbial” by “archaeal”.

L. 349: Change “biomarkers” to “a potential biomarker”.

L. 350: Change “(FAMEs) were identified” to “(C_17:0 FAME) was identified”.

L. 456: What do the authors mean by the “green checkered pattern”?

L. 606: Replace “predominantly microbial inputs” by “archaeal communities”.

L. 607: Delete “archaeal”.

Figure caption of figure 6: High MI is shown in green, not yellow.

Table 1: It would be great, if the color for “MI on” periods in the table would match the color used in the figures (green).