This paper looks at the effects of tree plantations on erosion in the coastal range in Chile by comparing suspended sediment concentration to in situ 10Be-derived erosion rates in the context of hillslope connectivity and detailed land use land cover change mapping. Although the authors do not have particularly conclusive results and the methods are not that new, I really like this paper. The data are well explained and the results are interesting. It's a bit puzzling to see so little difference between the two different metrics of erosion and the authors do a good job thinking about why that might be. 

I do have a few very minor points that I would like to see the authors clarify. 

1) end of page 14 (around line 307), I got to wondering if storage in the system, like in floodplains or alluvial fans, could be part of the reason for the depressed sediment concentration. That is brought up later, but forecasting it earlier on would make things clearer. 

2) The last word on line 307 ("This") is a pronoun that is unclear. I am not sure what "this" is. 

3) Is it possible that with the high rates of chemical weathering and likely deep regolith, the 10Be is an underestimate of long-term denudation rates? (see: Campbell, M. K., P. R. Bierman, A. H. Schmidt, R. S. Hernandez, A. Garcia-Moya, L. B. Corbett, A. J. Hidy, H. C. Aguila, A. G. Arruebarrena, G. Balco, D. Dethier, M. Caffee (2022). Cosmogenic nuclide and solute flux data from central Cuba emphasize the importance of both physical and chemical denudation in highly weathered landscapes. Geochronology. ) 

4) Along the same lines of questioning the 10Be data, is it possible that you have stripped so much soil that these are artificially elevated and don't actually reflect the long-term pre-people rates? (like Hewawasam et al found in Sri Lanka [Hewawasam, Tilak, et al. "Increase of human over natural erosion rates in tropical highlands constrained by cosmogenic nuclides." Geology 31.7 (2003): 597-600.] or Schmidt et al (ESurf) found in China [Schmidt, A. H., Neilson, T. B., Bierman, P. R., Rood, D. H., Ouimet, W. B., and Sosa Gonzalez, V., 2016, Influence of topography and human activity on erosion in Yunnan: Earth Surface Dynamics , v. 4, n. 4, p. 819-830. http://www.earth-surf-dynam.net/4/819/2016/])

5) Is it possible that the hillslopes have been so disturbed that the sediment is totally stripped from them, leaving the hillslopes in a detachment limited system even while the valleys have stored sediment that is transport limited? Given the magnitude of erosion you are talking about, it seems like this could be possible. I could be entirely out to lunch though, not knowing the area, and I do think that intermediate hillslope storage is a reasonable explanation. 

6) The first two sentences of the conclusions are printed twice. 

This is a really neat study and I look forward to seeing the final version published. 

The authors present an interesting study on the development of stream flow and suspended sediment transport in Chilean headwater systems that are conditioned by changes of multiple drivers (drought, wild wire, tree plantations). This study is relevant and valuable to be published in the journal ESurf. The aim of the study, to extract the effect of tree plantations and wild fires on sediment transport is very challenging. In the end, the discussion of the results is very general and I have the feeling that the authors missed the chance to analyse the data in more detail (see general and detailed comments) and learn more about the specific controls. I made some suggestions to more specific approaches, which might shed more light to the discussion. Overall, I suggest major revisions following my general and detailed comments before publication.

Kind regards, Thomas

General comments:

Unravelling causes/drivers of changing suspended sediment transport is a challenging task give the multiple interdepended drivers of suspended sediment loads. The authors are correct stating that an unambiguous attribution of cause and effect is difficult to assess. In this study the authors rely their statements mainly on the trend analysis. However, more statistical approaches are available to learn more about the driving factors.

The major issue is that SSD (load) is directly related to Q, since it is part of the calculation of the load. SSC (suspended sediment concentration) is strongly conditions by Q, but is not directly related to its estimation. Changes in sediment supply should therefore be discussed by changes in SSC. To see if changes in SSC are related to changes in Q or by changing sediment supply from hillslope sediment rating analysis could be performed, e.g. SSC = a(Q/Qm)^b) à changes in coefficient “a”, which is the suspended sediment supply at Q=Qm should be related to changing supply conditions that are not related to changes in Q (see for instance Warrick 2015, WRR or Hoffmann et al. 2023, ESurf). I suggest to add a trend analysis of the rating parameters, to learn more about changing supplies.

Additionally, the observed patterns is superimposed by a declining trend. To extract the effect of single events (wild fire, Earthquake etc.) the authors should focus on the residuals of single years (with and without events) with repsect to these long-term trends.

Detailed comments:

Line 31: rephrase „Mediterranean section (35-37.5o S) of the Chilean Coastal Range (CCR)”

Line 33: “slow denudation rates” à typically Mediterranean regions are characterized by high rates of soil erosion and gullying.

Line 34: “secondary native forest” à should be explained

Line 52: “the CCR ranks amongst the highest…” rephrase to “CCR ranks amongst regions with highest forest loss and gains worldwide”.

Line 55ff: are the statements limited to the “storm and yearly scale”? Furthermore, no other scales are mentioned later on à I suggest to remove the reference to the time scale.

Line 87: “transport-limited conditions” à this strongly depends on the dominant grain size. Only streams in the humid south of the CCR show transport limited conditions à see and refer to Terweh et al (2019, Geomorphology)

Line 90: The two time-scales fall out off the box here: Why 10^4 years? à You should motivate the use of the two time-scale. Since 10^4 Yr is related to 10Be, we should explain here why you use 10Be!

Line 135: is there any chance to identify gaps due to ceased stream flow and gaps due to measurement errors/mistakes, defect measurement devices?

Line 149: the use of 10Be in the context of LULCC impacts should be motivated in the introduction

Line 213: The calculation of RC needs more explanation. How did you calculate IC_rs and IC_s? What was the reference point of both calculations? Does IC_s is the connectivity of hillslopes with streams and IC_rs the connectivity to either streams or roads, meaning that you treat roads as streams? What is the effect of subtracting both and what are the major assumptions in this approach? Please give more details on your approach. If I understood it correctly, calculating IC_rs assumes that the connectivity of roads is 100%, meaning that very sediment that enters a road will immediately enter the stream. Am I correct and is this assumption valid? If yes, what about roads that are not connected to streams or that first go downhill and uphill afterwards? Could you show these effects of using IC_s and IC_rs based on a simplified graphic?

Line 240ff: Not clear what exact criteria you used to define alpha=0.7. I would be helpful to use ranges of alpha to show confidence intervals of the analysis. The presented results of the main topic (changing suspended sediment loads) is very limited. More detailed description of the results should be given.

Line 255ff: This is in line with the comment in Fig. 5: It seems that you calculate the percentiles of annual estimates (as suggested by the units per year). Percentiles are derived from probability distributions, but as far as I understand you only have a single estimate of the annual load of each year. If you use the daily measurements as annual distributions you should highlight this using the correct units (e.g. t/day). Please clarify.

Line 257: “Only the lower percentiles of SSD revert the decreasing trend at the end of the time series”?????? I don’t see that!!!!!

Line 260: How do you see if data are homoscedastic? Homoscedastic data show a normal distribution of residuals with respect to predicted values (e.g. using a linear regression). Why is homoscedasticity important in this context?

Line 263: post-wild fire SSC/SSD is very low (due to low discharges) and changes of SSC are mainly related to changing discharges. Again, a rating analysis of pre and post-fire SSC~Q relationships would be very helpful.

Line 276-277: you should also indicate whether there are places that show reduced connectivity, as the legend in Fig.8c indicates that there are areas with negative RC.

Line 280: The threshold is very arbitrary and likely depends on the input data and their resolution. Therefore, it is questionable if the threshold from another study can be used here. If the definition of a “threshold” is necessary, why don’t you use the distribution of IC to identify breakpoints.

Line 328: In Figure 6 no mean stream flow is visible. The question is whether means stream flow is a good proxy of the impact of wildfire. Streamflow is strongly related to rainfall. Thus, stream flow should be related to rainfall to see the effects of the wild fire.

Line 335: Whether sediment transport is transport-limited or not depends on the grain size. This should be discussed.

Line 338f: “a lack of minimum rainfall intensity required to trigger runoff and soil erosion on hillslopes (Mohr et al., 2013) and/or an increase in the residence time of sediments stored within the valleys is plausible.” Not sure what you want to say?

Line 341-354: The discussion is very general and only weakly related to the results of the study. You should discuss the result from the IC here. The IC does not explain any changes of the transport capacity, but only regarding the connectivity between hillslopes and channels! How are the increased connectivity values related to the decreasing SSC?

Line 355ff: Again, limited links are drawn to the results of the study.

Figure 4: Stream flow and base flow show very similar pattern and similar tau values. Given the difficulties of defining the correct alpha value for stream flow separation and the similarity of the trends, I suggest to use total discharge as a more robust estimate of the discharge.

Figure 5: Please give more details on the percentiles. Percentiles of what? It seems that you calculate the percentiles of annual estimates. Percentiles are derived from probability distributions, but as far as I understand you only have a single estimate of the annual load of each year. I assume that you used the daily data here. This should somehow represent in the use of the unit. Please clarify.

Figure 6: Unfortunately there are major data gaps in SSC/SSD after the fires.

Figure 7: I suggest to represent burned areas as bar and not as “*”. The “*” are difficult to see.