Dear authors,
thanks a lot for responding to the reviewer comments. Both reviewer found that this manuscript presents an interesting data set and makes a significant contribution in exploring factors of SOC storage in grasslands. However, both reviewer noted that there are several formal aspects, which can be clarified as outlined in your responses. However, the overall writinng has to be improved considerably.
In addition to the reviewer, I have the following comments:
1. In the Abstract make clear on how many sites your study is based on and to which depth samples have been taken.
2. Take care that all abbreviations are explained properly. E.G. BRT might be known by statisticians but not by all soil scientists.
3. Tables. Explain all abbreviations (especially the not well known ones, e.g. TSIS). Tables and Figures should be self-explanatory
4. Final model. Is it meaningful to include soil N? More than 90% of N are bound to soil organic matter and consequently N must correlate with C, at least their concentrations. Due to this inherent relationship, I would argue that N is a controlling factor. In the model you have used to explain stocks (whose estimate includes bulk density), but you have to provide the relation between C and N concentrations and then the reader may judge how menaingful soil N is.
Could you so provide support that the final model indeed allows that many factors (e.g. by showing AIC)?
5. How did you consider that factors are autocorrelated e.g. pH and Mg?
6. Table 1. Is it really meaningful to show what has been measured in which study but not showing the actual outcome of these studies? For instance SOC stocks, main controlling factors, etc.
7. Figure 2: Clarify that the relative importance is related to explained variance.
8. The explanation of interaction seems very speculative.
9. Language should be carefully checked. It can be smoothed at many places.

Dear authors

Thanks a lot for carefully revising the manuscript. It reads much better now! I find it a great exploration of driving factors of SOC storage although I still find some sections rather speculative. For instance at the first view, I have found TSIS a potentially great variable tackling seasonality but at a closer look the observed range seems close to the detection limit and thus somewhat ambiguous. Well, this is part of the discussion and identifying or presenting new controlling factors may stimulate other researchers to test them in their data set with a greater data span.
In a similar sense at
L. 554 ‘contrast with most other’ – this is not correct, the other studies have simply not included TSIS in their set of parameters – I would rather argue that your data suggest to include it in other studies which may also test it for stronger gradients than in this study.

Looking at Figure 3, the range of TSIS seems very small (1°C, >70% of the values ranging less than 0.5°C), almost being in the range of measurement uncertainty, especially because TSIS is derived from two measurements and the relationship primarily be driven by three sites with low TSIS. Consequently, I find your conclusions that TSIS is a key driver somewhat too strong.

Please carefully check the use of upper and lower case throughout the whole manuscript (see examples below).

L. 426 change 88.31% to 88%
L. 427 change Clay to clay
L. 435 change Combined to combined
L. 436 add empty space before (Fig…)
L. 439 Geophysical Model
L. 503 the sentence reads incorrect, especially the last half.
L. 561 Cold Sites

All comments have carefully been addressed and the revised version of the manuscript is a great contribution on drivers of SOC stocks.