The manuscript investigates the combined effect of fertilization with different elements. The study analyzes data from the NutNet grassland experiment, encompassing 34 sites that are distributed globally. The sites are subjected to standardized treatment and all in arranged in a block design. The manuscript presents results on the effects of the addition of N, P and K and combinations on biomass, temporal stability of biomass as well as synchrony using mixed effects models. They find that effects of different nutrient additions are additive, interactions were not observed. The researchers also draw some conclusion about the effect of nutrient addition in general, e.g. showing that decrease in stability was driven by variation and not the mean.

Overall this is a great dataset with wide coverage that allows drawing general conclusions. The analysis is sound. While already some research has been conducted on fertilization on grassland ecosystem stability, the combined effect of different nutrients also on reducing variation adds a new insight. The manuscript is overall well written and organized and potentially a valuable contribution to the literature. My comments are mainly editorial.

The introduction deals very generally with the ecosystems in general. it only returns to grassland ecosystems very late, when the current design is introduced. The introductions very general related to the effect of resources limitation on ecosystems, and does not relate to grasslands at all until the very end where it introduces the experimental setup. Grasslands differ substantially from other ecosystem types in ways that may affect their reaction to resources supply, e.g, species turnover, species life cycles, access to deeper soil for water, herbivory, etc. The introduction should bring out the novelty of the research and why grasslands are relevant for drawing general conclusions.
The manuscript meanders a bit between the general topic of stability in general and the combined influence of addition of different nutrients. This is especially evident in the discussion. For example, the last part off the discussion „Was there evidence for mechanisms associated with destabilization following nutrient enrichment“ does barely touch on combined effects but makes up 40% of the discussion. This probably due to the fact that new patterns showed up in this analysis compared to a previous study by Hautier et al., 2016. that were worthwhile reporting. There are some parts in that specific section (L 459-462) could be extended upon to bring out the effect of individual nutrients.

Introduction

L 96-97 „important undertaking“ - this is very general, can you be more specific? You are looking at this in more detail below, but either say here „(see below)“ or find a very short way of expressing why it is important.

L 107 Reference is missing or at least additional statement required why and how this is true. This is the core part of the manuscript and I think it would be better to be very explicit here.

L 143-144: „define changes in invariability …“ Could you rephrase this to bring out more clearly what you mean?

L 145-147 „increased likelihood of crossing thresholds“ and L 150 „passes such thresholds“:
I personally find this reference to multiple steady states is rather far fetched for the example at hand which is grassland ecosystems receiving additional nutrition. But probably I am not thinking of the right scenario. Can you please underline those statements with one or two examples illustrating the „irreversible change“ caused by passing thresholds?

L 160: „mean and variability“: I propose making explicit that this is about the within treatment variation. This will help understanding the explanation in the next phrases without further reference.

L 163: better use „resources enrichment“ to avoid ambiguity with richness term, e.g. species richness

Methods

L xxx I generally like the introduction of the mixed effects models, but I find the notation in Eq. 1 and 2 confusing. I am not sure what beta really stands for? Currently the difference between the fixed effects, which are a product, and the random effects, which are fitted values for each category is not well represented in the equation. Especially in model 2 this is rather confusing. There, the biomass appears in the index of beta, thus rendering the index an odd mix of a discrete count of the treatments and the biomass which is a real number and an observation. Writing the fixed effects as factor would help alleviate this.

Results

L 285: I propose replacing the word „outpaced“ with „accompanied“ which is a bit more neutral.

L 317-319 (and above): Generally, it is expected that the variance increases as the mean increases. A real increase in the variation would only be detected if Sd shows this effect. Please report Sd values or remove this part. Also, it would help, if the notation was unambiguously applied throughout the manuscript and also into the supplement. For example the term „variation“ can mean a lot of things. Table 10c probably refers specifically to sigma_detr which is properly defined in the methods section and therefore unambiguous. Therefore sigma_detr should appear in the description of Table 10c. Similarly for mu, Sd, etc.

L 329: Maybe better „with N addition“ instead of „with N“

L 334: Please remove „changed“ to remove ambiguity

L 342: Maybe better „Species synchrony and Sd responses were negatively related“ or similar.

Discussion

Line 357-358: What is meant by „This signals“ ? Please rephrase.

L 437 - 438: Be aware that a difference in mean annual precipitation (MAP) does not necessarily relate to differences in aridity. Aridity is defined by the ratio of MAP and potential evaporation. Therefore, the conclusion is incomplete and may be due to neglecting the atmospheric demand. I think the limitation-question is interesting and would encourage the authors to calculate the ratio and include it to the analysis. Otherwise, at the very least, leave out this statement.

L 459-462: Can you mention on the underlying reason?

My comments have been appropriately addressed. The new introduction now provides background on grasslands specifically and also motivates better the discussion part on stability. The re-ordering of the Mixed-effects models has improved this section.