|Review Luo et al. Biogeosciences (Revision 1)|
I have reviewed the revised version of Luo et al. with the title “Similar effects of edaphic and climatic factors on soil organic carbon stocks of the world” submitted to Biogeosciences. I was also reviewer for the original submission and I have reviewed both the response letter as well as the revised submitted manuscript files. In general the authors did a good job in revising the manuscript. I don’t have many remaining comments. Most of them are of minor size and should be easy to address. I find the discussion in parts a bit too long and speculative but this can be easily addressed. What I was confused about was the large amount of supplementary material that is not referred to in the text, even though it might be important. Please find my comments below. Line references refer to the revised manuscript file.
1. The results section is very short. And I think you need to explain the loading of the different PCs in a better way. You could also create a table for that shows the correlation of independent predictors with each PC, cumulative variance explained, Eigenvalue of PC etc. To me it seems a bit unclear from the fig 2 and description what really the most important soil variables were and how they cross-correlated with other variables.
2. The uncertainty section regarding model testing etc. mentioned in the response letter might be well suited to be a separate results section.
3. Also a small section on the model performance in different biomes or climate zones might be interesting in this regard. (Fig 3 shows some weird bias in performance, see comment there.
1. Discussion: 4.1. this section is a bit speculative. You describe in the results section that predominantly texture variables and pH were the most important explanatory factors. Yet you begin the discussion with mediation of decomposer communities by these variables, stabilization mechanisms etc. I would think you better start this section with what you wrote in l. 187 and so on before you go into all the stabilization mechanism and microbio stuff and be very careful to state that your variables actually mean indicate that. It is good that you try to interpret what mechanisms are behind the variables that the model selects but you should add a sentence here that interpretation has to be done with great caution or that we need to address/measure certain aspects of the linkages of these soil variables to mechanism carefully (with examples of changing relationships of soil variables across climate or land use zones from the literature ). To stay with the example of pH and texture: this could mean all of what you write here is true, or maybe nothing as these variables integrate so many effects at the same time.
2. Discussion section 4.3 (l. 218 .227). This is super interesting. I read something similar in a discussion paper by Fromm et al. currently reviewed in SOIL. You could check that paper and also the references cited in there (which you should refer to regarding that this section of the discussion is not well underpinned with literature yet). I do share your interpretation that most soil systems, on the long term show SOC stocks not to be limited by annual C input but by how well C can be stabilized in soil. Make sure this is understood and maybe add something on that in this discussion section 4.3.
3. Discussion section 4.4 (l.233-249). I would add to this section that for many areas we simple lag behind with gathering soil data. Just look at the bias that your dataset has towards the temperate zone. Many soil types are simple not well constrained and we don’t know much about whether some of the relationships we find between C stocks and predictor variables are universal or maybe fundamentally different in less studied soils (dryland, tropical soils, melting permafrost soils, developing or degraded soils, to name just a few examples).
1. Figure 2. I think you need to add a legend that explains the abbreviations directly to the figure and not refer to table 1. It is simply too many variables.
2. Figure 3. Fig 3 shows some weird underperformance bias for high observed log SOC ,consistent across all four depths. Explanation? Is this discussed? Maybe some part of the data consists of soil data points that are not covered by the environmental variables used here ? for example peat or wetland soils that are not well predicted (= constrain to model performance)?
I think the authors must have missed to revise parts of the supplement. Fig 3-5 in the main manuscript appear in a similar form in the supplement, similar for one of the tables. In addition, none of the figures in the supplement are referred to anywhere in the text. There is also additional analyses shown (partial dependence plots) that are not reflected on in the manuscript, even though they are important and interesting for the reader. Can you explain and address that?
L 42. Space missing in the reference list.
l. 55. I would replace “cultivation” with land cover or land use.
l. 84-85. What you write within  I would put as a separate sentence following the one you have it currently included in.
l. 101 delete “the” at: at the strict steady state
l. 105 delete “also”
l. 138-139. Remove the two “is”.
l.143: “land use” instead of “cultivation” ?
l. 165: “,” after “respectively.
l. 173. Revise wording. Make two sentences, one for the decrease if climate, one for the increase of soil influence
l. 181. Spaces missing between references. I stopped checking here as the problem seems to be also in many other places).
l. 198 grammar.