The Authors present a numerical thought experiment, based on the soil erosion model MMMF, to show the importance of including soil truncation processes in soil erosion models. The main thesis of the Authors is that, in the long term, soil truncation will result in a change in the soil properties, and thus in its vulnerability to erosion, leading to a change in the erosion rate. I want to commend the Authors for the style of writing (very easy to read) and for the clarity of images and equations shown. The results are well documented and discussed. The study is interesting, sort of an "opinion paper" backed by modelling experiments, helping raising an issue on something which is very interesting: which erosion processes that we disregard at short time scales may be relevant at long time scales?

However, I have some issues with what comes before the results: in my opinion, the article would benefit from a rewriting or restructuring of the abstract, introduction, and materials and methods sections, and from a more-focused title. While reading the manuscript, I had to "read between the lines" for assumptions and work-flow, piecing the real objective of the study together just as in a detective story, just to find the culprit confessing in the results and discussion (I do love detective stories, and I enjoyed doing this reading quite a bit, but I still think it is not the best way to write about science).

I want to stress that I believe this was an interesting study, and that my (extensive) comments are only aimed at improving the manuscript.

* The title of the article, even if alluring, is not very informative: the reader is left in darkness about what the article is about, and what those feedback system is. I'd suggest to try to use the title to inform the reader about, at the very least, the keywords of the manuscript, namely: long-term erosion, UK (more on this later), numerical thought experiment.

* If the title of the manuscript may be more "catchy" than informative (depending on the style of the Author), the abstract really needs to be informative, especially filling in what is not in the title. The abstract should be clear about (a) the model used, (b) the type of approach to modelling (if "numerical thought experiment" phrase is used in the abstract, that would be enough to wet the appetite of a theorist), (c) the scope of the research - the fact that the result of the analysis may be limited to conditions similar to those studied (i.e. UK), with possible usefulness to other areas where "saturation excess" is the dominant overland flow mechanism.

* The Introduction section should include a paragraph about modelling, since this article is based on it: which models are the most used, how and why do they keep soil erodibility fixed throughout the soil profile, what is MMMF model and why it was selected above the others. The Authors should also make it clear to the readers what has been observed in the field and what is not known yet (observations part), what was done already in research for long-term soil erosion predictions, etc... The introduction should state clearly the scope of the study: in this case, the results may be applicable only to the UK, even though the problem raised (soil truncation) is certainly important in all areas of the world. Finally, the Introduction should state clearly the objective of the study; even better would be to state a research question that will be directly answered in the conclusions; even-even better would be to state a null hypothesis to be tested statistically through the numerical experiment (but this may be very limiting). As a general comment, the Authors should try to create a clear connection between the introduction and the conclusions they reached.

* Finally the Materials and Methods: in my opinion, the work-flow followed is not very clear, I suggest to include a figure in the article detailing what was done step by step, e.g.: 1) MMMF model (already described) was modified to include topsoil truncation (state assumptions made); 2) soil data from UK was used to derive pedotransfer functions; 3) soil data and pedotransfer functions were used to create N instances of the model with soil parameters as in the 265 soil profiles; 4) the climatic and LUC information from the 265 experiments were used to set mean and variance for the respective parameters in MMMF, and this normal distribution was used to run a Monte Carlo analysis for each of the N instances of the model; 5) the results were analysed by (describe). Note: I may have missed or misinterpreted something in the example given, and this should show that the work-flow was a bit confusing. If not with a figure, the workflow should nevertheless be made more clear by restructuring the section 2.4.

* Bonus point: Discussion. Even though the discussion section is very nice, I noticed that it does not discuss the limitation of considering 500 yr soil truncation when assuming no sedimentation (no input of soil material to the modelled profile). Moreover, I think that it would be nice to organize the discussion by dividing clearly the limitations of the study depending on: inherent limitations/assumptions of the MMMF model; limitations/assumptions in the modifications introduced on the model; limitations/assumptions in the dataset and simulations performed. However, the assumptions could be also introduced in the Materials and Methods section (in the respective sections 2.2, 2.3 and 2.4).

A major question: why did the Authors did not perform a Global Sensitivity Analysis of the modified MMMF model, using as parameter space the edaphoclimatic conditions of the UK sites, after deriving their probability density functions?

Minor correction: line 283, "with the variation in soi losses over" is missing an "l" in "soil".

In this manuscript the authors study the effect of soil truncation/thinning in the evolution of the annual soil erosion rates over 500 years. For this purpose, they used a parsimonious soil erosion model and 265 profiles in different locations in UK. They considered in the simulations the change in soil properties as subsurface layers get exposed as a consequence of the removal of the soil surface by soil erosion. In order to isolate the effect of soil truncation, factors usually considered as variables, such as climate, land cover and topography, are considered as constant. First of all, I would like to congratulate the authors for a very interesting paper and for developing a simple but effective method to consider long term soil truncation in soil erosion modelling. In general, I think that the manuscript is well written and structured.

- My one major concern is that one of the equations applied is wrong. The error is Eq. 6, this equation should be multiplied by LD, in other words, the KE of LD should be proportional to the annual LD, the same way as the KE of DT is proportional to the annual DT (Eq. 4). This error was already present in the original publication of RMMF model (Morgan, 2001) and has propagated to the MMMF model, equation 7 in Morgan and Duzant (2008) and other models including PSYCHIC (Davison et al., 2008) and SERT (López-Vicente et al., 2013). These KE equations were originally proposed by Brandt (1990) where they present two equations of kinetic energy per mm (J m−2 mm−1) of LD:

E = 8.95 + 8.44 log I

and of DT:

E = 15.8 PH0.5 – 5.87

So, they need to be multiplied by the volume of rainfall (mm), LD and DT respectively, to obtain the KE (J m−2). Hopefully this correction won’t change much the main results and conclusions. This error was previously corrected in some studies (Choi et al, 2017; Peñuela et al., 2018; Sterk 2021) however this error in the formulation of the RMMF and MMMF models was only pointed out by Peñuela et al (2018). To avoid further propagation of this error, I also encourage the authors to highlight in the manuscript the need to correct it when applying either RMMF or MMMF models and when developing new models based on them.

- How is the soil truncation calculated from SL? Can you further explain this? I think that while the formulation of the MMMF model is well described, it would be very helpful for other researchers interested in applying this method in their models to show explicitly the equations used to consider the effect of soil truncation.

- Something that I missed in this study is an analysis of temporal evolution of the influence of soil truncation, in particular I would be very interested in knowing more about when this influence starts to be significant. It would be very useful for modellers to have an idea of under what circumstances, in particular number of years simulated, soil truncation should be taken into account or not. For example, if I do a 100-year simulation, should I include the effect soil truncation? While the results of this study cannot be generalized to other regions, I think that this can provide a first attempt to set recommendations, at least in UK, of when soil truncation should be considered depending on the number of years simulated.

- In the Discussion, please include a paragraph evaluating the performance of the model and the SL values simulate. Please include references of studies of measured annual soil loss in agricultural fields, in particular winter cereal, in UK, for instance Evans et al 2016 and Boardman 2013, and compare them to the simulated soil loss, are they similar or in the same order of magnitude?

Minor comments:

- Please justify in the manuscript the 20cm plough/mixing depth considered for the simulations.

- Can you please further develop the justification of using the MMMF model, for example, why is it important “its ability to simulate multiple erosion subprocesses” for this study? And that it is parsimonious?

References

Boardman J. 2013. Soil erosion in Britain: updating the record. Agriculture 3: 418.

Brandt CJ. 1990. Simulation of the size distribution and erosivity of raindrops and throughfall drops. Earth Surface Processes and Landforms 15: 687–698.

Choi K, Arnhold S, Huwe B, Reineking B. 2017. Daily based Morgan– Morgan–Finney (DMMF) model: a spatially distributed conceptual soil erosion model to simulate complex soil surface configurations. Water 9: 278.

Davison PS, Withers PJA, Lord EI, Betson MJ, Strömqvist J. 2008. Psychic – A process-based model of phosphorus and sediment mobilisation and delivery within agricultural catchments. Part 1: model description and parameterisation. Journal of Hydrology 350: 290–302.

Evans R, Collins AL, Foster IDL, Rickson RJ, Anthony SG, Brewer T, Deeks L, Newell-Price JP, Truckell IG, Zhang Y. 2016. Extent, frequency and rate of water erosion of arable land in Britain – benefits and challenges for modelling. Soil Use and Management 32: 149–161.

López-Vicente M, Navas A, Gaspar L, Machín J. 2013. Advanced modelling of runoff and soil redistribution for agricultural systems: the SERT model. Agricultural Water Management 125: 1–12.

Morgan RPC. 2001. A simple approach to soil loss prediction: a revised Morgan–Morgan–Finney model. Catena 44: 305–322.

Sterk, G. A hillslope version of the revised Morgan, Morgan and Finney water erosion model. Int. Soil Water Conserv. Res. 2021, 9, 319–332.

The manuscript describes a model experiment on how erosion induced changes in soil properties (by soil thinning) affect soil erosion rates at large temporal scales. The authors applied the modified MMF model to 265 soil profiles in the UK over a 500-year time period. The results show that only 39% of the profiles were sensitive to the changes in soil properties by soil thinning and most of these profiles showed decelerating erosion rates. 

In general, the manuscript is very well written and accompanied by clear figures. I think that the model experiment is very useful and gives some interesting insights on the feedback between soil thinning and soil erosion rates. However, I have the feeling that the implications for soil erosion modelling may be a bit overstated. The authors apply a 500-year simulation period, which generally speaking is much longer than normally applied in soil erosion modelling studies, which mostly apply a decadal time period. The results show that over this 500-year simulation period there are some notable changes in soil erosion rates, however, when focusing on the first few decades, the changes are negligible. Of course, as the authors point out in the Discussion section, there are some processes that are not accounted for by the MMF model, which may affect the results, also in the first few decades. But still, I doubt if the results of this study have significant implications for soil erosion modelling. The authors may agree or disagree with this, but I welcome to the authors to discuss this point somewhere in the manuscript.

Below I have provided specific comments to the text.

Specific comments

Line 205: Table 2 shows the parameter values used in the Monte Carlo simulation. I’m not sure if this is really important, but if you would calculate the ground cover occupied by the number of stems (i.e. stem area times number of plants per unit area = (0.025**2 * pi) * 250) then I arrive at a value of around 0.49, while the ground cover is assumed 0.3.