|Most points that I raised regarding the initial submission of this manuscript have been addressed satisfactorily. However, I recommend addressing the following remaining points (referring to numbers in the author's response document):|
12. The point I was trying to make was that the 15N added is well mixed in the model in mineral N pools (irrespectively of its depth), whereas in the field, it may first be immobilised in the litter layer. I was trying to argue that in the model, natural 15N has the same probability as experimentally added 15N for being taken up by the plant (well-mixed mineral N pool), whereas in the field, it may not. The author’s response and added text does not fully address this point.
13. Added text addresses my point well. I didn’t understand, however, whether the entries ‘0’ in Table 1 (Amount of N fertilizer, g N m-2 yr-1) were wrong in the first submission and have been changed now to 0.5.
14. Greatly appreciated. No units are given in Table S1. Data that is of interest for use in other research should be made available following the FAIR principles (see https://libereurope.eu/wp-content/uploads/2017/12/LIBER-FAIR-Data.pdf). Here, it may be made available as a CSV file in a “tidy” format, i.e. one variable per column, one entry per row (no merged cells), and deposited on a public repository (e.g., Zenodo) with a DOI.
16. Added text under 4.1 (author response to comment 40) is helpful as a statement regarding how to improve the representation of N cycling in CLM.
18. I don’t agree with the author’s argument. Following the definition I proposed, then, if for two systems, Nin is the same (hence Nout is the same in steady-state) and NPP is the same, then their “openness” is the same, too (unless the C:N ratio of new production and/or the resorption efficiency). I am not sure if I understand correctly how the index was calculated here to estimate how often N cycles through plants before being lost fro the system. I assume it’s the residence time of N in at the ecosystem level (Nplant+Nsoil+Nmineral / Nin), divided by the residence time of N in the plant (Nplant / (Nuptake+Nfix) = Nplant / Nlitterfall). Is this the case? Please specify this more precisely in the manuscript.
20. This is a point that I have been thinking about and I may not understand what’s the best answer. The author’s response doesn’t solve my puzzle either. When doing an ANOVA, you consider the uncertainty of two samples. The uncertainty from the observations is both the result of measurement error and variability in the system. In the model, it’s only variability in the system, but has no error (it’s a deterministic model, not stochastic). Is it permissible to treat variability in the model as an uncertainty and do an ANOVA. I’m not an expert on this, but have doubts.