This is an interesting paper that explores the interactions between land use change on soils with two contrasting parent materials in Congo.  There is a significant amount of work here and the team are to be congratulated on collecting a significant data set for the region. The results point to the importance of understanding the soil quality and, in particular, the availability of aluminium as the lower soil horizons are brought closer to the soil surface due to soil erosion.  The focus on soil depth and its interaction with soil quality is often overlooked in erosion studies so this is good to see.

I have made a few of mostly minor comments below:

Page 5 L123. Results are being introduced here. Suggest that they are placed in the results.  The same data also needs to be presented for the mafic area.

P7 L155 change to ‘difficulties in finding’

Page 11 L225.  The calculation of soil erosion depths and then the calculation of the land use change age are key to the paper, but the explanation here is hard to follow.  Given the importance to the overall paper I suggest adding a figure and using it to help explain the concept of matching SOC contents to calculate soil depth change.

Page 14 Figure 4 Depth in forest profile is a rather confusing term.  Do you mean depth of forest profile? See my previous content about adding a diagram to help make this term clearer. Also label the axis the same as in the caption ‘equivalent depth in forest profile’.  Better, consider using the term ‘ calculated soil depth change’.

Page 25 Line 501.  While there is a nice bit of symmetry with the introduction  here. It seems to me that the requirement for more sustainable soil management is not restricted to just kalongo soils.

This is an interesting study in which the authors focus on erosional processes as an explanation for soil degradation in two regions that contrast in soil mineralogy in the Albertine Rift Valley in Africa. Whereas most studies on soil degradation in tropical regions avoid steep slopes (and erosional processes), this study is different by focusing on areas that have experienced substantial erosion. This makes it both interesting but also challenging. Another interesting part of the study is that the authors compared a region where soils are formed on mafic rocks with a region where soils are dominated by felsic rocks, which affect the secondary minerals in these regions. The felsic area also experience influence from volcanic ashes in the Holocene.

Writing. While generally well written, the manuscript suffers in some places from long and winding sentences. The most extreme example is between lines 103 and 109 which all appears to be one sentence. Also, sometimes you try to put too much information in one sentence, which makes them difficult to understand. For example, I did not understand what you were trying to explain in lines 166-167. There are more examples like these and I would like to encourage the authors to critically look through their manuscript for long or complicated sentences because they make your manuscript in parts difficult to read.

Soil analyses. I understand from the manuscript that a substantial part of the soil analyses were conducted using soil infrared spectroscopy which is an indirect method (you write it was used to ‘support’ the quantification of key soil variables).  I can understand why it was done, probably given a lack of reliable soil analytical equipment and laboratory capacity. Although I think this is OK for variables such as SOC and TN, I remain quite skeptical when it is used for other variables such as pH (CaCl2),  ECEC, and  pyrophosphate and oxalate extractable Fe and Al. My main reason is that I don’t see how these variables can affect the infrared spectroscopic information from soil samples. It remains a black box and if you have unusual soil samples or outliers that are not included in your calibration samples you may work with erroneous data, so you add another level of uncertainty to the quality of your data and thus your analysis.

Assumptions.The most important and critical part of the interpretation of the data is the comparison between forested are eroded hillslopes. This is described in 2.5 and I had to read it three times to understand what was done. I appreciate the approach that you chose, but it is also my task as a reviewer to question how solid your assumptions are.  If I understand it correctly, it was assumed (1) that originally (before clearing) SOC profiles and SOC stabilization mechanisms in the cleared hillslopes were similar to the SOC of the forested hillslopes. Then, it was assumed that (2) the SOC levels of the topsoil from the cleared hillslope sites were correctly matched to the comparable levels of SOC of the forest profiles, which was somewhere below the surface of the forest profiles. Finally, (3) it was assumed that the soil above this level, which was still present in the forest profiles, had been eroded from the cleared hillslopes. That is quite a lot of assumptions and it would be really good if you had some independent parameter to test these assumptions, which I don’t think you have provided.

Let’s first look at (1). You write that the reference data from undisturbed sites were retrieved from the TropSOC database v.1.1.  If we do a study in my group where sites are compared, we try to select our own reference sites. We make sure that soil types are comparable and we make sure that soil texture is comparable and that the distance between sites in relatively small.  I was wondering if you can provide the minimal, maximum and mean distance between reference sites and cleared sites? Are they having similar parent material? Are you comparing similar soil types, soil textures? Did you test this? In the discussion you mention for the felsic region that croplands were likely established on more fertile soils, which you use to explain why soils on cropland actually had higher SOC contents than forests. But it also put into question your approach. Can you exclude that you are also comparing different soil types in the mafic region?

Assumption (2) is also critical, because with each centimeter deviation from a correct match you introduce more errors. Therefore, it would be really good if you had some independent variable in your soil profiles to check how good the match is between forest and cleared sites. Would it be possible to use soil texture to test your match? Another option might be the 13C values of SOC, since they typically change systematically with soil depth, it should be possible to check with such independent data whether your match between forest and cleared sites is correct. I am asking for this because later in the manuscript you argue that there is almost no replenishment of new organic matter (e.g. from the Eucalyptus monocultures), but this could also the result from wrong ‘matching’ of the forest and cleared profiles. An idea might be not to match the topsoil of the cleared area but focus on the subsoil of the cleared sites to match with the forest profiles.

Assumption (3) is also not supported by independent data. Although you mention in the text that you are careful with the interpretation of SOC losses (l 226) in the discussion you mainly focus on the erosion as major process of SOC losses. If this assumption is correct, you should be able to find some of the eroded soil material at the footslopes of these hillslopes. Did you find colluvium and are the amounts comparable to what you would expect based on your analysis in this manuscript?

I found the explanation of volcanic activity in the felsic region vague. It made me wonder how representable this region is for soils derived from felsic parent material. First, carbonate vulcanism is very rare and mostly consists of carbonates instead of silicates. I am not sure how this information fits with the occurrence of allophanes, as mentioned in the manuscript, which are silicates and not carbonates. Also, in the felsic region the sand content is much higher than in the mafic region. A higher sand content may have affected water infiltration rates which may have affected soil erosion. Could that be one of the explanations why you were not able to quantify erosion in the felsic region? Climates are also quite different, but not included in any of the explanations of the observed differences between the two regions.

Overall this study is very interesting because it clearly goes beyond typical studies on land use change effect of soil carbon stocks. However, the weakness is in the many assumptions that are currently not checked independently. Maybe the solution is to conduct an uncertainty analysis and error propagation of your calculations, because you are introducing quite a lot of additional errors by your methodology of soil analyses and your assumptions.  For guidance see: https://seismo.berkeley.edu/~kirchner/Toolkits/Toolkit_05.pdf 

There were a few other minor issues, but I have already spent a considerable amount of time on this review and hope you won't mind me leaving it at that. I hope you find my review helpful, and that it will enable you to improve your manuscript.