In this manuscript, Zhang et al propose a new proxy to reconstruct fluvial organic matter inputs to coastal marine settings. They suggest that brGMGTs are produced in-situ in rivers and estuaries and that the distribution of brGMGTs is principally controlled by salinity. Based on these facts they generate a new Riverine Index (RIX) using the fractional abundances of H1020c and H1034b versus H1020a and H1020b. To validate the RIX in deep time they compare RIX values to the BIT index and terrestrial pollen/spore deposits deposited during the PETM from IODP Expedition 302 Hole 4A. They report a closer relationship between RIX and terrestrial pollen abundance than BIT and terrestrial pollen abundance, indicating that at least in this site RIX outperforms BIT in accurately reconstructing riverine inputs. In all, this is an interesting study that will likely be of interest to BG readers. I have a number of comments that aim to strengthen the manuscript. 

General comments:

In some sections (see specific comments) the use of English in the paper is poor and obfuscates the meaning of the text. I suggest that the authors carefully read through the manuscript to catch all typos and grammatical errors. Likewise figure quality varies considerably. In some cross plots, it is impossible to see the data as the marker size is so small (see specific comments). Characters that should be superscripted/subscripted are left as regular text (see specific comments). Lines of best fit are drawn through data but there is no information as to how these lines were constructed (see specific comments). As such, this manuscript would benefit greatly from more attention to detail from the authors. 

Additionally, as the authors are proposing a new GDGT salinity index, I would like them to calculate and report previously formulated salinity indices from their samples. Specifically, the ACE index (Turich and Freeman 2011) and the IR6+7me (Wang et al 2021) both have been calibrated against water salinity in marine saline ponds and hypersaline lakes respectively. I know the author's brief touched on comparing IR6me from this study to values from Wang et al (2021) in the text but a more thorough examination of prior GDGT-derived water salinity reconstructions would strengthen the manuscript. Readers will be interested to see how these indices compare against RIX in reconstructing salinity from an estuarine environment.

Additionally, the evidence for in situ production of brGDGTs and brGMGTs in downstream estuary sites is pretty weak. This is demonstrated by Fig 2 where we see that distributions of d13Corg and d15N in soils and downstream estuary samples are very similar in addition to Fig 5 where your PCA on sample brGDGT and brGMGT distributions cannot separate out soils from downstream estuary samples. Yes you see (on average) higher concentrations of brGDGTs and brGMGTs in downstream estuary samples than in soils but the actual distributions of brGDGT and brGMGT abundance in soils are pretty large, indicating that some soils have pretty substantial quantities of these compounds. A great way to add more clarity to this sourcing issue is to train a random forest model using a similar method to Martinez-Sosa et al (2023) on your brGDGT and brGMGT samples (and isoGDGTs as these should be available to you). If the random forest model can accurately separate out soils from downstream estuary samples then you can be pretty sure that your downstream estuary samples were produced in situ. This won’t require much additional work and can be implemented easily using python (https://scikit-learn.org/) or another language of your choice. 

Specific comments

Line 35: This complicates paleoenvironmental interpretations in SOME aquatic settings not ALL aquatic settings

Line 37: “all along this basin, from land to sea” awkward phrasing 

Line 40: “Redundancy analysis further shows that both salinity and nitrogen loadings dominantly control the brGDGT distributions.” No, the loadings indicate that SALINITY (not salinity loadings) controls the brGDGT distribution.

Line 40-43: “Furthermore, the relative abundance of 6- methyl vs. 5-methyl brGDGTs (IR6Me ratio), Total Nitrogen (TN), δ 15N and chlorophyll a concentration co-vary in a specific zone with low salinity” Is this zone geographical, in your redundancy analysis, or something else?

Line 44-45: “Salinity is positively correlated with homologs H1020a and H1020b, 45 and negatively correlated with compounds H1020c and H1034b.” Is this in soils, sediments or SPM?

Line 45: “This suggests that bacteria thriving…” thriving is not the correct word (carries implications of a value judgment) replace with “living”. 

Line 45-47: It seems like you aren’t mentioning results from soils and sediments, only from SPM? Or maybe all your sediment samples are exclusively from rivers? The reader is unclear on this. 

Line 51-52: “a paleorecord across the upper Paleocene and lower Eocene,” You should name this record and say where it is.

Lines 51: “showing its potential applicability in both modern samples and in paleorecords.” Perhaps you could evaluate its usage in both these cases e.g. - we successfully/unsuccessfully applied RIX in …

Line  55: “, although some of them were attributed to the phylum Acidobacteria” Imprecise language.

Line 57-58: “The distribution of brGDGTs (number of cyclopentane moieties and methyl groups; cf. structures in Fig. S1) was empirically linked with pH and Mean Annual Air Temperature”  Again, imprecise language. The phrase “empirically linked” doesn’t convey much useful information. 

Line 60: Should really cite some earlier lake GDGT papers in addition to Martinez-Sosa et al., 2021.

Lines 60-61: “The brGDGT-based proxies (i.e. MBT’5ME and CBT’) have been largely applied to reconstruct MAAT and pH from sedimentary archives (Coffinet et al., 2018; Harning et al., 2020; Wang et al., 2020).” Not quite true - Martinez-Sosa et al (2021) and Dearing Crampton-Flood et al (2020) generated Bayesian linear regressions between the Mean temperature of months above freezing and MBT’5me. These BayMBT models have been used widely in the community since their publication.

Line 62-63: “In aquatic settings, brGDGTs were initially suggested to be predominantly derived from watershed soils and transported by erosion in the sediments (Hopmans et al., 2004).” Maybe you mean “transported by erosion to the sediments”?

Lines 63-78: The use of English throughout this paragraph is poor and hard to follow. Needs copyediting.  

Lines 63-78: You should read and cite Martinez-Sosa et al (2023) here for their work on a Random Forest approach to classify GDGT sources (i.e. Marine, Soil, Lake etc).

Line 80-83: “The improvement of analytical methods allowed the separation and quantification of 5-, 6- and 7-methyl brGDGTs (methyl groups at the fifth, sixth, and seventh positions; Fig. S1), that in previous chromatographic protocols co-eluted (De Jonge et al., 2013, 2014; Ding et al., 2016).” No real link between the previous paragraph and this one. Also, which methods? How did they improve? 

Lines 86-87: “In addition to temperature and pH, other environmental factors may influence brGDGT distributions in terrestrial and aquatic settings and hence the application and interpretation of brGDGT-derived proxies” This is a repetition from earlier in the introduction.

Lines 91-99: You should mention that brGMGTs have previously been called H-brGDGTs in the literature.

Lines 91-111: This paragraph was very well written and is an example of the standard the entire manuscript should meet. 

Lines 117-123: You go from talking about the hypothesis you aim to test in the paper to talking about the aims of the paper. Surely your aim is to test the hypothesis you have just laid out - why do we need to talk about more aims here? 

Line 125-126: “by high population density”. High population density of what?

Line 127: “macrotidal”. Please define this term.

Figure 1: I really like this figure - it nicely summarizes your water sampling campaign.

Line 167: “Both decarbonated and non-decarbonated samples (~6 mg for SPM and ~20 mg for soils) were enclosed in a tin capsule” You should mention that you will split your samples and decarbonate one aliquot. Otherwise, the reader is confused as to where your non-decarbonated samples are coming from.

Line 172-174: “The isotopic composition (δ 13C or δ 15N) was expressed as the relative difference between isotopic ratios in samples and in standards (Vienna Pee Dee Belemnite for carbon or atmospheric N2 for nitrogen)” Should be “and atmospheric N2…”. 

Line 176: What were these “additional…analyses”? Do you mean the same analyses aforementioned but on different samples, or different analyses on different samples?

Line 177: “(4-20g, n=51)” Looks to me like you’ve used the minus sign, not the en dash here. If so use the en dash.

Line 180-183: “The total lipid extracts were then separated into fractions of increasing polarity on an activated silica gel column, using (i) 30 mL of heptane, (ii) 30 mL of heptane:DCM (1/4, v/v), and (iii) 30 mL of DCM/MeOH (1/1, v/v) as eluents.” That seems like a nonstandard amount of solvent. Are you using very large columns here? If so state how many g of silica gel were used.

Line 233: Vegan should be vegan. No capital V. 

Lines 240-243: I don’t think you effectively explain how your hierarchical partitioning method actually works here. As some readers won’t be familiar with this method, more details are needed.

Figure 2. I really don’t like how the axis of these plots has been extended to include chart labels. The top left panel scale is completely distorted by the addition of these labels. You should also define the features of your “boxes” in your box plot. These comments apply to all boxplots in the manuscript.

Line 268: “The different brGDGTs were detected in all studied samples” Which brGDGTs?

Line 275: “The relative abundances of the brGDGTs were determined all along the Seine River basin” I feel like this sentence should be at the start of this section not in the second paragraph.

Line 290: “which explained 40.9% of the variance in two dimensions” Which two dimensions are these?

Line 291: “Samples from the downstream estuary clustered well” Colloquial language, you should describe the data using words that don’t convey a value judgment. 

Line 314-315: “It allowed to explain 39.79% of the variability through two dimensions.” Doesn’t make sense - please proofread your manuscript. 

I feel like you have just randomly placed the figures in the text. You should line up the first in-text citation of a figure with the location of the figure in the manuscript. Currently, the text and the figures are out of sequence which makes reading this document a challenge. 

Figure 5: Visually this figure is quite busy. I don’t think having the brGDGT names in blue (the same colour used for the downstream bubble) helps. I would use black for these names and also the arrows.

Line 336: “The brGMGTs identified in previous studies” Which brGMGTs and which studies? This lack of precise usage of language is present throughout the text.

Line 343-345: “In SPM and river channel sediments, the total brGMGT concentration was observed to be slightly higher in the riverine part (0.26 ± 0.24 μg/g Corg) than in downstream (0.20 ± 0.13 μg/g Corg) and upstream estuary samples (0.17 ± 0.18 μg/g Corg; Fig. S4b).” Slightly higher but not significantly higher. If it’s not significant you should say so.

Line 346: “The PCA analysis based on the brGMGT relative abundances (Fig. 5b) explained 70 % of the variance”. I’m unsure what the authors are trying to say here but I think they mean that the first two PCs sum to 70%. The second half of the sentence “which allows to observe that samples from the different parts of the basin clustered well apart from each other.” doesn’t make sense and I’m unsure what the authors are trying to say.

Line 357: “allows to explain” This phrase doesn’t make sense in this context - please remove all uses of it from the manuscript. 

Lines 406-408: “The similarity in distributions between soils and downstream samples may be due to the overrepresentation of downstream soil samples, as 82% of the soils were collected downstream (Fig. 1a and Table 1).” I don’t understand your point here. Are you saying that the similarity between downstream estuary brGDGT distributions and soil brGDGTs is because the downstream estuary predominantly receives brGDGTs from downstream soils? 

Lines 409-412: “Nevertheless, the soil-derived brGDGT contribution to the downstream samples is expected to be much lower than the autochthonous one, as the average brGDGT concentration in soils was ca. 3 times lower than the one in downstream (i.e. SPM and river channel sediment) samples (Fig. S4a).” Right, but it’s curious that the distributions are so similar between brGDGTs in soils and downstream estuaries. To bring more clarity to this point it would be interesting to see you attempt a machine learning approach (see Martinez-Sosa et al 2023, PP) to investigate whether (or not) a random forest model can distinguish soil samples from downstream estuary samples. 

Lines 426-429: It would be great to see you calculate and report IR6+7me following Wang et al (2021) to determine if these indices correlate to salinity in an estuarine location. 

433-436: “The distinct behavior of 6-methyl brGDGTs between lakes and the Seine river-sea continuum might be due to the lower salinity range in the Seine River basin (0-32 psu) vs. the lakes (0-376 psu) 435 investigated by Wang et al. (2021). This suggests that the limited range of salinity variation in the Seine River basin might be insufficient to trigger significant 6-methyl brGDGT production, as observed in hypersaline lakes.” This is actually incorrect. Wang et al 2021 report that IR6me is sensitive to salinity in the range of 5-1000 (mg/L) but relatively insensitive beyond this range. 

458-460: “As the nutrient concentration is higher in the upstream part of the Seine estuary (Wei et al., 2022), the substantial 6-methyl brGDGT production observed in the aforementioned zone (260 460 < KP < 340, Fig. 8)” Right but why would the nutrient runoff be higher for this specific section of the basin? Do we see more agricultural activity here or something? It would be good to try and flesh out this point. 

Figure 8 and throughout: Make sure to superscript 15 in d15N and subscript 6 in IR6me.

509-510: “The current knowledge on the parameters controlling the brGMGT distributions in the terrestrial and marine realm is still limited.” Why is it limited? Be specific.

Fig 9: Almost impossible to see the data points on some of the figure panels (e.g. panel e). Make the points bigger. Also, keep a consistent label text size to make the figure look neater. Also, you should say in the caption how you constructed the straight lines drawn through the data in some panels (e for instance). I’m assuming this is a linear regression but you have to inform the reader of your methods. 

557-558: “ However, the average concentrations of brGMGTs are an order of magnitude lower in the soils than in the river channel sediments and SPM samples of the Seine basin (Fig. S4b).” Maybe it is, but visually it doesn’t look like that, so include the numbers in this sentence. You can also argue that the brGMGT abundance within soils varies by an order of magnitude. Do you know what is driving such a large variance in the soil brGMGT abundance?

589: Missing the word “index” after BIT 

You need a map showing the location of IODP 302 Hole 4A

Lines 605-607: “This core is considered proximal to the coast and has considerable changes in terrestrial inputs (i.e. continental spores and pollen) over time (Sluijs et al., 2009, 2006), making it a suitable paleorecord for testing runoff proxies.” Again would be great to have some specifics. The readers will be interested in how close this core site was to the coast around the PETM. You should also say why there was a considerable change in terrestrial inputs (I’m assuming large changes in sea level are responsible). 

Lines 616-617: “Such decreased runoff during the PETM body was previously attributed to a local sea level rise” Ah here is the explanation - this should have been in the previous paragraph. Also, be specific, are you saying there was decreased runoff during the PETM, OR did your sediment core record decreased runoff due to a change in sea level? These are two different things.

References:

Martínez‐Sosa, P., Tierney, J. E., Pérez‐Angel, L. C., Stefanescu, I. C., Guo, J., Kirkels, F., ... & Reyes, A. V. (2023). Development and application of the Branched and Isoprenoid GDGT Machine learning Classification algorithm (BIGMaC) for paleoenvironmental reconstruction. Paleoceanography and Paleoclimatology, 38(7), e2023PA004611.

Wang, H., Liu, W., He, Y., Zhou, A., Zhao, H., Liu, H., Cao, Y., Hu, J., Meng, B., Jiang, J., Kolpakova, M., Krivonogov, S., and Liu, Z.: Salinity-controlled isomerization of lacustrine brGDGTs impacts the associated MBT5ME’ terrestrial temperature index, Geochimica et Cosmochimica Acta, 305, 33–48, https://doi.org/10.1016/j.gca.2021.05.004, 2021.