The paper by Laub et al presents a newly developed model called SAMM. The model aims at representing the effect of aggregation on soil organic matter (SOM) dynamic. The paper is well written and the subject fits perfectly with the GMD scopes. I appreciated the approach proposed by the authors and in particular the comparison between SAMMnoagg and SAMM. It is fair to recognize that SAMMnoagg performed as well as SAMM when properly calibrated. I think this paper deserves publication after some corrections:

1.    The authors claimed that this is the first model considering aggregation. It is not totally true, the MIMICS model published by Wieder et al., (2014) consider a physically protected pool that could be similar to the Agg pools presented here though not as detailed as what presented by the authors.

2.    The authors evaluated the SAMM model using microbial biomass obtained after fumigation extraction. This is problematic because the chloroform extraction method extracts the full biomass including the dormant one and, in your model, you represent the active one. You can't directly compare both since 90 % of the biomass is dormant (Lennon and Jones, 2011).

3.    The initialization procedure of the model is not detailed enough, does the simulation showed here started after a spinup? What are the consequences of the initialization procedure on the results?

4.    I don’t understand what is the rational behind the ProtLAB pools, how the presence of structural litter can protect the labile pool. It needs to be more justified.

5.    It is not clear what the time step of the model is, please clarify.

6.    In the main text, the information on the boundary’s conditions is not clear. A sentence refereeing the appendix would help the reader to find the information.

7.    Since the model is newly developed a mass balance calculation showing that the mass balance is closed is necessary to trust the model behavior.

8.    From Fig. 7 it is not clear whether the prediction of SAMMnoAgg recalibrated are different from SAMM. It might be interesting to test through a statistical analysis if the two models give predictions that are significantly different.

9.    L46-49. You should write “One of the important processes…” not only CUE matters

10.    L95: The data should be show in supp mat because you may have no significant changes for 2 mains reasons :1. There is indeed no or a very limited effect or 2. The variance between plots is so high that the statistical power of your setup is not strong enough to detect any change.

11.    L307: This comparison is not totally fair because you are comparing with 1st order kinetics models, you should compare with Millenial and MIMICS.

References cited

Lennon, J. T. and Jones, S. E.: Microbial seed banks: the ecological and evolutionary implications of dormancy., Nat. Rev. Microbiol., 9, 119–130, https://doi.org/10.1038/nrmicro2504, 2011.

Wieder, W. R., Grandy, A. S., Kallenbach, C. M., and Bonan, G. B.: Integrating microbial physiology and physio-chemical principles in soils with the MIcrobial-MIneral Carbon Stabilization (MIMICS) model, Biogeosciences, 11, 3899–3917, https://doi.org/10.5194/bg-11-3899-2014, 2014.

The manuscript by Laub et al. presents a new, advanced soil carbon dynamics model featuring a measurable-pools structure, which includes an explicit aggregate formation process and its connection with microbial growth. The model parameters were calibrated against measurements in a long-term experiment at a tropical site, showing low parameter correlations that indicate a parsimonious model structure. Their model results could reasonably reproduce the observed microbial biomass and soil carbon changes after litter addition, and highlighted the role of aggregate protection which accounted for about half of soil carbon stabilization at the tested site. Overall, the manuscript is well written and logically organized. Model limitations are also well discussed. I have only some minor comments that need to be addressed/clarified.  

For the Bayesian calibration of the parameters, it is not clear what data from the observations were used for the optimization. Do you use all the observations including the time series of carbon changes for different pools after each litter addition? If so, the model evaluation metrics actually represent the potentially highest level that the model can reach, which would be expected to degrade when applied to other sites.

It is not clear how the initial state of the model was derived. Was a spin-up process employed to reach equilibrium, or were initial values prescribed for each pool?

Line 222: Is there an explanation for the 1~2 months delay in the peak of MICc compared to the peak of LMWc?

Figure 2: It would be helpful to add a panel showing changes of the total SOC. Besides, line colors for the different pools are a bit difficult to distinguish, please consider using more distinct colors. “STRUc” in the legend should be “STRc”.

There are a few typos in the current manuscript, such as “MAOc” being written as “MOAc” in some places, “depolimerization”, “One the one hand”. Please check carefully throughout the text.

The current abstract is not a very concise and engaging summary of the study, please refine it.