General comments:

“Sourcing and Long-Range Transport of Particulate Organic Matter in River Bedload: Rio Bermejo, Argentina” by Dosch et al. have analysed a range of chemical signatures of organic material carried in the bedload of the Bermejo river which has no tributary inputs in its lowland reach. They have shown that bedload OM primarily reflects recently eroded OM, and that it’s rapidly transported to the mouth and likely comminuted during  this transport. This is a thorough and well carried out study on a unique system, providing new and valuable insights into bedload OM dynamics. It is within the scope of ESurf, references relevant literature well, and describes the context, methods, results, and discussion clearly. The paper is also well written and presented, with some relatively minor technical suggestions for improvement below. The only substantial issue in terms of discussion/interpretation is the choice of the mixing model, as discussed below. I think overall this warrants minor revisions that can be assessed by the editors.

Specific comments:

• The efficient conversion of POMbed to suspended POM via comminution is one of the most interesting findings of this paper, in my opinion. Would there be value in discussing the potential implications of this more? For example, would it be possible to calculate a rough estimate of the rate of POMbed comminution into POMspm, from the “missing flux”? Acknowledging the difficulties in bedload sampling, of course. Yes, it is likely to vary with tectonics and climate, but probably not by a huge amount, and more likely to be strongly controlled by the type of POC (eg woody debris vs leaf litter) and river hydrodynamics, ie. turbulence + bedload grainsize (sand vs gravel) acting as the "comminuter"? Are there any experimental or field data showing similar comminution of POM? Could this be extrapolated to other river systems, acting as an “invisible” supply of addition POM to the suspended load? Overall, the transfer of POMbed into the total POM flux (dominantly SPM here, and likely in other river systems?), and the wider implications seem a bit lost in the long paper, and (to me, at least) seem worthy a bit more discussion/emphasis – but this is just a suggestion and the authors are free to disagree or to refrain from too much speculation.
• Is there any correlation of near-bed flow velocity and POMbed concentration? Seems like not – worth a mention that this is the case, either way. Also, it seems ADCP data was not available at all sites, so average velocity from other sites was used to then estimate the velocity in the ones with missing data – is this correct? This was not very clear, if so. Table 2 should include near-bed flow velocities where available, and the values used (and explanation of how they were estimated), where not available.
• I think a more consistent sampling site naming convention would benefit the reader. I felt in places it was difficult to follow and to remember all the different locale names and to cross-reference them constantly with river names, headwater vs downstream, etc etc. For example, HWnorth and HWsouth is intuitive and clear, so I would suggest using those more consistently throughout, instead of referring to “Pichanal” etc.
• In terms of interpretation / discussion, my biggest questions are related to the mixing model used:
  • A bit more explanation of the modelling approach would be useful, even if it’s replicating the Smith et al method, as it is not the most standard EMMA / monte carlo mixing approach, as far as I can tell. For example, were the endmember compositions resampled from a normal distribution, or a uniform one? How does the area optimisation work? What is the meaning of the final variance values? Etc.
  • Are you sure this method does not under-represent the potential mixing polygons? Given the large scatter in the samples comprising the different endmembers, and the quite large SD of the endmembers, the contours seem to plot very close together and to be very strongly constrained by the endmember mean/median value. I have not used this mixing modelling approach, but my guess is it has to do with the fact you are minimizing the convex hull area each time? What would the result be if you used a simpler monte-carlo resampling of the different endmembers from their distributions, without area optimization? If you think the Smith et al approach is better here, then I would suggest explaining how and why in a short paragraph.
  • On L364 you say the modelling is to “more quantitatively determine the sources (sic) areas” but there does not seem to be any quantitative discussion of the model results, as far as I can tell in Section 4.2.3. It’s fine if the model does not provide quantitative insights in the end, since most samples fall outside of it. But I wonder if alternative modelling approaches could be tried in this case. For example, could you not represent the floodplain endmember as the 13c-2H trend line, instead of the mean±SD value? Combined with ACL, it seems like this could yield at least semi-quantiative insights into the fraction of POMbed derived from the three different areas. Even visually, it is quite clear that HWnorth contributes very little while floodplain dominates. This seems to be obliquely alluded to in L451-459 but not actually said. And then it appears that the flux estimates are derived solely from sample concentrations in Section 5 (as far as I can tell).
  • Overall, the choice of symbols and colours, and especially how the legend is shown in Fig 6, makes it very hard to interpret. The legend in panles a and b implies the symbol shapes and colours are showing different things, but then looking at the actual figure, it seems there is only 3 types of shape+colour types over all? The colours should be more distinct rather than different shades of the same colour. Same applies to c and d, but there shapes should also be distinct, not just circles. Is there a reason the contour lines change colour between the panels? The linear fit is not explained in the caption, and even from the text it is not clear exactly which samples are included in the fit and which are not. The grey contours and the gery circles in panels c and d are too pale and hard to see.

Technical comments

• L103: “is a braided”, word missing
• 163: “The sampling in these earlier campaigns were performed under qualitative aspects” – unclear what you mean here, perhaps rephrase
• Fig1: it is hard to distinguish the lightest colours, especially when printed, esp soil and suspended sediment samples. The colourbar of panel b does not seem to match the map at the lightest end, ie. 0 looks much darker in the colour bar than on the map? Potrentially this has to do with the hillshading. I would suggest removing the hillshading in panels b and c, or making it much lighter. Panel labels a, b, c, are too small and hard to see here, and in a number of the other figures.
• Section 3 title “Analysis and preliminary data” is somewhat confusing. I would think “Methods” or “Analysis and data treatment” or similar would be suitable and more intuitive.
• L198: please provide more detail on the ADCP deployment from bridges. Was it floating on a boyant board, being dragged by a rope? How did you manage to do this upstream of the bridge, rather than downstream, without it being pulled under the bridge? I think useful for the readers interested in this.
• Fig 2 panel a: there is some “colouring outside the lines” going on
• There is a number of issues in the bibliography, with some references not formatted, or weirdly formatted. See L670, 674 for examples. It is also very difficult to tell where one ends and another begins as there’s no spacing between them (not an issue in the final paper but important for reviewers).
• L245 – should paragraph break take place at the end of the sentence here, rather than at the next sentence? Makes reading this part a bit confusing.
• Fig 3 / L262-263 – the R2 values don’t match the figure. Or if it’s a regression of a different set of samples, that is not clear from the text. Also some figure panel references are wrong.
• Fig 4: the open symbols are difficult to see, especially the paler ones. Consider thicker lines, stronger colours.
• L325-328: “POMBed CPI25-33 values (average: 7.4±3.0, n = 39) were not significantly different from… soils” followed by, “on average POMBed CPI25-33 values were lower than soils…”
• Fig 5: panels c and d mixed up in the caption
• Section 4.2.1 – it would seem more logical to place this after what is now section 4.2.2 – this is also because Fig 6 is really properly explained only in section 4.2.3
• I would encourage supplying the code that was used for the endmember mixing model in the supplementary material.
• L378 – what is a “resolution of 500” in this case?
• L434, the linear trend equation and R2 value does not match the one shown in Fig 6.
• L436-437: this sentence is unclear. Floodplain leaf litter is an important source of POMbed but is not mixed into the bedload? seems contradictory
• L439 – “downstream trend” – please rephrase for clarity. Is there an actual trend going downstream from site to site, or do you mean a trend AT the downstream sites.
• Fig 7 – has the same issue with the legend as Fig 6
• L449 – should be “land use” not “land consumption”
• L467 – “flow structures” perhaps, rather than “flow motions”
• L498 – should it say “values ONLY as high as 7.8”? because it seems the point you’re making is that these values are low, compared to POMbed
• Equation 3 – why are you using the funnel width, as opposed to cross sectional area? Presumably the implied assumption is that POMbed transport occurs in a tin layer near bed that is less than 8cm, ie the vertical dimension of the opening of your sampler? If so, it would be good to explicitly say this. Could you ponentially be underestimating the flux if the POMbed is transported in a thicker layer, saltating, etc?
• L518 – could you not estimate %OC of your own samples from your analyses? And/or explain briefly what the “van Bemmelen factor” is
• L522 – it would be more meaningful to give the length of the wet season in days rather than seconds
• L524 – sentence unclear, potentially some words missing
• Table 2 – caption is very difficult to comprehend and does not seem to reflect the table contents. For a number of samples “n” is the overwhelmingly more common convention rather than “i”. Is “total bedload (g) the average of the samples, or total sum of all samples? Why is this meaningful to show, compared to g/s? Water flow rates should be included here.
• L564 – “ad this eye-catching feature” – I would suggest rephrasing, it is not clear if you’re referring to POMbed transport or something else here.

General Comments:

Sourcing and long-range transport of POM in river bedload: Rio Bermejo, Argentina

I have read over the other reviewers’ comments, so I have not added some of those in this review but do encourage the authors to address his thorough comments. This is a very interesting paper, with a lot of samples ranging from various years, and previous studies, taken in a remote location. A continuation of engaging and interesting studies that comes from this region of the world and from this research group.

Interesting method to sample the near bottom of the river (POM_bed), and an area of the river that is important but neglected. Sampling focused on smaller particles, would large woody debris be an element to consider, as this often flows fast but along the riverbed (i.e. Peruvian foothills). Regardless, since this type of measurement is new, I recommend explaining clearly what it actually is measuring (i.e. the material flowing along the riverbed, rather than buried in the riverbed).

Well explained mechanisms of river POM recruitment and transport. The results and discussion are well written, and easy to read and follow.

You carried out this study during a very challenging time (Covid-19 Pandemic), which is commendable.

Specific comments:

• Section 4.1.1 Flux of g/min appears in the text, but how are you calculating a POM_bed flux from a grab sample? Explain please.
• In Section 5, you do a good job to estimate fluxes, but do you think it would be good to include more on limitations (i.e. some sites only had one sample), how might the sampling have been improved. It is understandable that you were limited because of Covid. It is amazing that you were able to do any sampling at all during this time.

Technical comments:

• L66: check spacing
• L106-107: check sentence structure
• L112: SS and bedload sediment make this yield?
• Table 1 caption is very long, do you need to include all the variables in the caption?
• L151: roughly what was a sufficient sample you collected in the field? If only 1 gram of OM isolated, would this be an area for improvement in the future? Or a study limitation?
• L166: spacing
• L165: I would make it clear what you mean by POM_bed, because I would have thought it was POM in the riverbed, rather than 8cm above the bottom.
• L178: are there any study limitations by not measuring POM_float in the northern tributary? Do you expect it to be consistent with the south tributary?
• L281: Section 4 appears, is there any distinction between materials and methods and results? Check journal specifics.
• L297: is there a correlation with velocity near the riverbed?
• Figure 4 a-c, really nice visuals of the samples
• Figure 6, POM_float in the legend needs fixing, too much space.
• L482: spacing needed
• L494: What about average particle size within the samples?
• Table 2 caption is very long. Check that you’re happy with this, otherwise maybe it can be simplified, or some content can move to a footnote.