Summary

In this manuscript the authors reconstruct east Asian monsoon dynamics over the last 130 kyrs using biomarkers preserved in loess. Newly generated records of changes in the distribution of brGDGTs are combined with published records of plant wax d2H from the same section. The authors develop a quantitative method to reconstruct changes in MAP using brGDGTs preserved in loess and demonstrate that precipitation in the Chinese loess plateau varies at the precession and obliquity scale, the former indicating the northern hemisphere as a main driver of monsoon precipitation.

Main assessment

The manuscript provides a large amount of new data that are used to support novel insights into our understanding of the east Asian monsoon. This type of manuscript will be of interest to the readers of CoP. In addition, the newly proposed method to quantify precipitation using brGDGTs in loess will be of interest to organic geochemists. The manuscript was pleasant to read and the figures clear and informative.

            However, my main criticism is that the manuscript and main conclusions rely on a limited set of brGDGT-based indices: DC and, to some extent, IR. However, other brGDGT indices are influenced by pH (and thus precipitation), for example the well-established CBT index for brGDGTs. In addition, other GDGTs like crenarchaeaol and the BIT index can be used to infer changes in hydrology in terrestrial sections, as highlighted in the introduction of this manuscript. However, these complementary methods are not used here. Rather, the manuscript relies on the application of less often used indices like DC. There is no explanation why these other GDGT-based indices are not used, while they are measured. I assume they are excluded because they show different results? But these other proxies could provide additional insights into changes in hydrology in this region.

I therefore recommend moderate revisions. In the revised manuscript I would like to see an expanded discussion on the other GDGT based proxies (e.g., CBT, BIT, %cren) and justification for why they are not used here to assess changes in hydrology. Or better, they are included to obtain a more holistic reconstruction of EASM dynamics across the late Quaternary.

David Naafs

Minor comments:

Line 1: The method proposed here to quantify precipitation is explorative and needs to be verified at other sections. Remove “quantitative” from title to reflect the uncertainty surrounding this method.

Line 14: state here that both the speleothem and plant wax d2H records are already published

Line 30-31: I am not an expert, but NH summer insolation also has an obliquity component, especially when we look at 65 oN and higher. In this manuscript the focus is on 35 oN insolation (e.g. figure 2), but this nuance of low versus high-latitude NH summer insolation needs to be explained here and elsewhere in the manuscript. Also, the spectra of NH summer insolation (as shown in figure 2, so 35 oN) should be added to figure 3.

Line 32-34: this sentence seems to be crucial for the later interpretation of the data, but the reasoning behind this conclusion is not very clear for non-experts (like myself). The importance of this lag and why this argues against a NH insolation control needs to be explained a bit more here. This will help clarifying the discussion and conclusion later on.

Lines 46-48: Similarly, expand here to explain why a strong 23 kyr cycle is indicative for NHSI.

Line 68-onwards: somewhere in this section of the introduction of the manuscript explain where (and when) the biomarkers that are found in loess are produced. Do this for both the GDGTs and the plant waxes. For example, are the plant waxes produced in situ or transported with the loess? This nuance is important for the later discussion.

Line 83: this is a bit of a NIOZ/UU centred list of papers. Lots of other groups have worked on this, I suggest diversifying the reference list here.

Line 86-89: I was surprised that CBT was not discussed at all here (and not used at all in the entire manuscript). CBT is one of the most common methods to reconstruct soil pH. It needs to be introduced here. In this context, I wonder whether changes in the accumulation rates of brGDGTs hold any paleoclimatic information. The GDGTs were quantified using the C46 std, so this data is available.

Line 88: methylation can also occur at C7, see for example (Ding et al., 2016)

Line 90: we also discussed this in (Naafs et al., 2017)

Line 113: and is this benthic d18O record tuned to astronomical cycles like the LR04 stack is?

Line 116: change to “…corresponding to a sedimentation…”

Line 144: Explain here why IIc and IIIb-IIIc are not used in the DC index. Is their abundance too low?

Line 153: what is this assumed standard deviation based on? Can repeat analysis of for example a lab standard provide a data-supported value? If not, what is the impact of selecting a slightly different value? How does this impact the MAP reconstructions?

Line 177-179: show cross plots of DC versus GS and d2H to provide a statistical basis for this “match”

Line 179: also show cross plot for NH insolation and the IR record for comparison (and for other proxies used, see main comment)

Line 182: this grey is hard to see, add arrow to figure of where this splicing occurs

Figure 3:  How do you get 100 kyr cyclicity in a 130 kyr long record?

Line 195: but besides precession, the DC record also shows a strong 41 kyr signal

Line 199: to fully determine whether the DC and IR records are different, show a spectral analysis of the IR record to highlight that it lack precession and obliquity cycles as seen in the DC record.

Line 207: could the brGDGTs be used to quantify these variations in hot and cold conditions?

Line 211: doesn’t Ca2+ affect pH and that influences brGDGT production? Is there clear evidence that it is Ca2+ and not the resulting change in pH?

Line 215: I don’t understand how Ca2+ affects brGDGT production. The direct impact of Ca2+ on brGDGT producers needs explanation. If Ca2+ drives pH and that impacts brGDGT producers, explain that here

Line 222: doesn’t a r2 of 0.06 indicate no correlation?

Line 225: although the overall community might change, this doesn’t mean that the brGDGT producing community changes. The next sentence should state that this is speculation.

Figure 4: do I understand correction from this figure that pH is not correlated to MAP because IR is strongly correlated with pH, but not with MAP? Does that not undermine some of the earlier text of this manuscript where MAP and pH are suggested to be linked?

Line 243: cite reference for modern soil CLP GDGT data here

Line 245: how does this uncertainty of ±125 mm compare to other quantitative proxies used for the CLP? Is this correlation much better, worse, or similar to other methods? This context would be good for the non-expert.

Line 266: and how does this gradient compare to the modern gradient?

Line 268: is the difference in reconstructed MAP between Holocene optimum and MIS 5e statistically not different? State statistically proof for this statement.

Line 283: the manuscript states a “close resemblance”, but the DC has a strong 41 kyr signal that is lacking in the d2H record.

Lines 283-308: For this comparison with the d2H record, it is important to in the introduction explain where and when the different biomarkers are produced. Is there a possibility for a spatial and/or temporal offset between production of the plant waxes and bacterial membrane lipids?

Figure 6: why is the IR data not included here? Does that not have a clear precession forcing?

Line 343: ensure that the individual GDGT data is available for future usage

References:

Ding, S., V.F. Schwab, N. Ueberschaar, V.-N. Roth, M. Lange, Y. Xu, G. Gleixner, and G. Pohnert, 2016. Identification of novel 7-methyl and cyclopentanyl branched glycerol dialkyl glycerol tetraethers in lake sediments. Organic Geochemistry 102, 52-58. doi: 10.1016/j.orggeochem.2016.09.009

Naafs, B.D.A., A.V. Gallego-Sala, G.N. Inglis, and R.D. Pancost, 2017. Refining the global branched glycerol dialkyl glycerol tetraether (brGDGT) soil temperature calibration. Organic Geochemistry 106, 48-56. doi: 10.1016/j.orggeochem.2017.01.009

Guo and co-authors present a novel dataset, and apply the DC index as a proxy for precipitation, in alkaline soils from a loess sequence. The paper is in general very well written, the figures are clear and the paper is well structured. The authors did a very good job explaining the result, that at first sight seem to contradict the expected GDGT responses. I do have a few comments that could broaden the interest of the manuscript to users of GDGT proxy ratios.

 The discussion of the results is very to-the-point, to a degree where I wonder whether more GDGT ratios (Ri/b and BIT seem obvious choices (see introduction), possibly together with GDGT concentrations or accumulation rates) could have been included.

I would invite the authors to think a bit further about the local paleo-environmental conditions, and whether any of their assumptions can be confirmed with independent measurements. For instance, the authors surmise a link with a change in DC and available Ca in the soil profile. Could this be substantiated by the analysis of the sedimentology of this sequence? I also wonder whether the short-lived hydrological shift that resulted in a long impact on the 6-methyl branched GDGTs, could have been caused by ponding (i.e. the creation of a lake)? Is there any evidence for stagnant water (and associated anoxia) based on the sediments? The different signal between pit and outcrop is also interesting, and points to the large impact local (hydroclimate/ vegetation) changes have on the GDGT signal.

Guo et al. generated new records based on branched glycerol dialkyl glycerol tetraethers (brGDGTs) from the Chinese Loess Plateau (CLP) over the last 130 kyrs. The authors found that two pH-sensitive brGDGT-based indices, DC and IR, showed contrasting temporal changes at the same site. After a comparison of the new brGDGT-based records with several published records from the same site, such as another biomarker-based record (ice-corrected δ²H_wax), the authors found that DC showed promise as a mean annual precipitation (MAP) proxy. Then, the authors investigated the relationships between brGDGT-based indices (DC and IR) and soil pH and MAP using a global modern soil dataset, as well as a CLP modern soil dataset. The authors found that in alkaline soils, including within the CLP, DC showed a strong correlation with MAP, which enabled the development of a DC-MAP calibration for quantitative MAP reconstructions. The authors then applied their DC-MAP calibration at three sites within the CLP, including the study site, and investigated spatial differences in MAP within the CLP and their changes through time. The authors also did spectral and cross-spectral analyses and found that (i) their DC record showed precession and obliquity signals, contrary to δ²H_wax and δ¹⁸O_speleorecords which only showed the precession signal, and (ii) their DC record was in phase with δ²H_wax and δ¹⁸O_speleorecords at the precession scale. The authors thus concluded that Northern Hemisphere summer insolation was a direct forcing of precipitation amount rather than the result from confounding factors on precipitation records based on isotopes.

As a paleoclimatologist with expertise on GDGT-based proxies, I have read this manuscript with interest for several reasons. First, the authors tackled a topical and important subject, namely the understanding of the East Asian Monsoon which also has its own controversies and subjects of debate, as the authors stated in the Introduction. Second, this manuscript presents an interesting use of a brGDGT-based proxy, namely DC as a (quantitative) MAP proxy, as brGDGTs are classically used for quantitative reconstructions of land temperature and soil pH. Third, the authors reconstructed past MAP changes using a proxy that does not involve hydrogen or oxygen isotopes, contrary to δ²H_wax and δ¹⁸O_speleo, which strengthens the independence of DC relative to δ²H_wax and δ¹⁸O_speleo. Furthermore, I found the manuscript easy to read and well-organized. Overall, this piece of work would be a great addition to the literature and is worth publishing in Climate of the Past.

However, I have several comments, questions, and suggestions for revision, which I detail below.

General comments:

1) Recently, Zhao et al. (2020) introduced a precipitation index (PI) as follows: PI = (Ia + Ib)/(Ia + Ib + IIIa + IIa′ + IIIa′). Like the redefined DC, the PI takes advantage of the improved separation of 5- and 6-methyl brGDGT isomers. Importantly, Zhao et al. (2020) and Zhang et al. (2024) proposed PI-MAP calibrations in cancellous bones and soils, respectively, for brGDGT-based quantitative MAP reconstructions. Accordingly, I would like to know the authors’ thoughts concerning the PI. Specifically, the authors may check how well the PI would behave as a (quantitative) MAP proxy compared with e.g., DC and IR in the CLP and, in case of similar trends between DC and PI, which MAP reconstructions the PI would yield in the CLP. However, the authors do not need to switch to the PI, especially if DC has a better motivation and/or shows more meaningful results compared with the PI in the authors’ view.

Zhang, T., Han, W., Tian, Q., Zhang, J., Kemp, D. B., Wang, Z., Yan, X., Mai, L., Fang, X., and Ogg, J.: Tectonically controlled establishment of modern-like precipitation patterns in East and central Asia during the early late Miocene, Journal of Geophysical Research: Atmospheres, 129, e2024JD041025, https://doi.org/10.1029/2024JD041025, 2024.

Zhao, J., Huang, Y., Yao, Y., An, Z., Zhu, Y., Lu, H., and Wang, Z.: Calibrating branched GDGTs in bones to temperature and precipitation: application to Alaska chronological sequences, Quaternary Science Reviews, 240, 106371, https://doi.org/10.1016/j.quascirev.2020.106371, 2020.

2) Even more recently, De Jonge et al. (2024a) proposed another GDGT-based proxy which may track precipitation changes, specifically mean monthly precipitation (MMP) changes: MMP = (isoGDGT-1 + isoGDGT-3)/(isoGDGT-1 + cren). I reckognize that the involved GDGTs are isoGDGTs rather than brGDGTs and that the alternative GDGT-based index would likely yield uncertain reconstructions as well (see the [Eq. 14] versus MMP plot in Supp. Fig. 8 in De Jonge et al., 2024a). Nevertheless, provided that isoGDGTs are abundant enough to yield peak areas above quantification limit, I feel that the authors may consider this isoGDGT-based ratio as well.

De Jonge, C., Guo, J., Hällberg, P., Griepentrog, M., Rifai, H., Richter, A., Ramirez, E., Zhang, X., Smittenberg, R. H., Peterse, F., Boeckx, P., and Dercon, G.: The impact of soil chemistry, moisture and temperature on branched and isoprenoid GDGTs in soils: a study using six globally distributed elevation transects, Organic Geochemistry, 187, 104706, https://doi.org/10.1016/j.orggeochem.2023.104706, 2024a.

Detailed comments:

Main text

Line 11: In “Chinese Loess Plateau”, “loess” is not capitalized in the abstract, but is in line 40 in the Introduction.

Line 48: Which paper by Guo et al. (2022) is cited here? The one published in Geology (Guo et al., 2022a)?

Line 98: Which paper by Guo et al. (2022) is cited here? The one published in Organic Geochemistry (Guo et al., 2022b)?

Fig. 1 (also Fig. S3): Readers may find it hard to read the coordinate labels for the inset which shows the relevant wind patterns: the authors should consider changing the color from black to white, as they did for other labels within the larger map, and/or increasing the font size.

Lines 142–143: Replace “(De Jonge et al., 2014a)” with “De Jonge et al. (2014a)”.

Fig. 2: For panel A, the authors should pick a color pair different from the current green-orange one for the sake of accessibility to color-blind readers. For panel B, the authors could consider picking a color pair with a stronger contrast in terms of hue and/or lightness for the sake of readability.

Line 204: Replace “this event only last” with “this event only lasts”.

Fig. 4: For panels A–D, the authors should consider revising the colors to avoid the green-orange confusion for color-blind readers. Alternatively, the authors should distinguish the CLP datapoints from the other ones using a different symbol type, for instance with squares, diamonds, or triangles rather than circles. If the authors pick the second option, then the change in symbol type for CLP should be reflected in panels E and F as well.

Lines 247–248: The r² value represents the percentage of variance explained by the regression, not the correlation strength which is represented by the r value.

Lines 255–258: This is an important and welcome remark.

Line 294: “(i.e., δ²H_wax (Fuchs et al., 2023), speleothem δ¹⁸O (Cheng et al., 2016))”: A few parentheses should be removed so that only a pair of parentheses remains.

References: Could the authors recheck their reference list? The formatting appears a bit suboptimal at places, for instance in lines 367–368 (Baxter et al., 2019) where I spotted a “ScienceDirect” which appears out of place there, as well as in lines 537–539 (Wang et al., 2001) where I spotted an unexpected “(80-. ).” just after the journal name.

Supplementary Figures

Fig. S1: It would be great if the authors could write the m/z values of [M+H]⁺ ions with at least one decimal place rather than as integer values. Otherwise, researchers who would examine GDGTs for the first time may fail to do optimal GDGT analyses for the reasons discussed by Davtian et al. (2018) and partly reminded by De Jonge et al. (2024b).

Davtian, N., Bard, E., Ménot, G., and Fagault, Y.: The importance of mass accuracy in selected ion monitoring analysis of branched and isoprenoid tetraethers, Organic Geochemistry, 118, 58–62, https://doi.org/10.1016/j.orggeochem.2018.01.007, 2018.

De Jonge, C., Peterse, F., Nierop, K. G. J., Blattmann, T. M., Alexandre, M., Ansanay-Alex, S., Austin, T., Babin, M., Bard, E., Bauersachs, T., Blewett, J., Boehman, B., Castañeda, I. S., Chen, J., Conti, M. L. G., Contreras, S., Cordes, J., Davtian, N., van Dongen, B., Duncan, B., Elling, F. J., Galy, V., Gao, S., Hefter, J., Hinrichs, K.-U., Helling, M. R., Hoorweg, M., Hopmans, E., Hou, J., Huang, Y., Huguet, A., Jia, G., Karger, C., Keely, B. J., Kusch, S., Li, H., Liang, J., Lipp, J. S., Liu, W., Lu, H., Mangelsdorf, K., Manners, H., Martinez Garcia, A., Menot, G., Mollenhauer, G., Naafs, B. D. A., Naeher, S., O’Connor, L. K., Pearce, E. M., Pearson, A., Rao, Z., Rodrigo-Gámiz, M., Rosendahl, C., Rostek, F., Bao, R., Sanyal, P., Schubotz, F., Scott, W., Sen, R., Sluijs, A., Smittenberg, R., Stefanescu, I., Sun, J., Sutton, P., Tierney, J., Tejos, E., Villanueva, J., Wang, H., Werne, J., Yamamoto, M., Yang, H., and Zhou, A.: Interlaboratory comparison of branched GDGT temperature and pH proxies using soils and lipid extracts, Geochemistry, Geophysics, Geosystems, 25, e2024GC011583, https://doi.org/10.1029/2024GC011583, 2024b.