This manuscript clearly explains the method and performance of integrating the phosphorus cycle into the LPJ-GUESS Dynamic Global Vegetation Model. The writing is logically structured, and the model development process is well-reasoned. The authors also provide a thorough discussion of the results. I have only a few detailed comments, which are outlined below.

Introduction:

The introduction does a good job in establishing the need for incorporating the phosphorus cycle into dynamic vegetation models, highlighting its importance as a plant growth constraint. However, it would benefit from a more thorough discussion of past challenges in phosphorus modeling and recent advancements, especially regarding prior efforts in Earth system models, where phosphorus cycling, though less common than nitrogen, has been integrated in models like ORCHIDEE-CNP and CASA-CNP. Clarifying research gaps and this study’s unique contributions compared to these models would enhance its impact. Additionally, providing context for the four scientific questions would help readers understand their relevance, as they currently appear somewhat abruptly.

Methodology:

In Figure 1, consider adding some equations (or other elements) to illustrate the principles behind phosphorus limitation, such as how soil phosphorus cycling regulates available phosphorus levels and how soil phosphorus uptake limits plant photosynthesis. This would help readers better understand the model's configuration, especially as this is a central focus of the study. Additionally, is the leaching path missing from the figure?

I recommend moving the content related to P soil processes from the appendix into the main text, particularly sections on P sorption and P weathering. For readers less familiar with these processes, equations can facilitate understanding.

Integrating phosphorus limitation into the model is indeed a significant advancement. However, we know that plant growth may also be limited by other elements, such as potassium or calcium. Therefore, there is still the potential to overestimate plant growth in this model (at least from a nutrient supply perspective). I suggest adding a few sentences in the discussion to address this limitation.

Results:

The results are well-documented.

Discussion:

The discussion section provides a thorough evaluation of the model's performance and areas for improvement. I suggest more discussion of potential limitations. For example, the model lacks certain processes in the phosphorus cycle, such as the impact of plants and microbes on weathering, and there is uncertainty in data sources, as phosphorus content in rock is typically based on sample averages that may not fully represent the mineral composition across all regions. Highlighting these limitations would inspire future research to address these aspects.

My recommendation is minor revision or rather technical corrections. The manuscript is well written and structured, and most importantly, the transparency of the model assumptions and evaluation. The aim with the extended LPJ-GUESS-CN with the P cycle was to addresses the following research questions (1) Does including the P-cycle improve model agreement to biomass and gross primary production (GPP) observations? (2) What drives N and P limitation across climate zones? (3) How has that changed during the 20th and early 21st century? and (4) Which environmental factors are more relevant for N and P limitation change? Questions 1-3 are clearly answered, and the answers form also the major contribution of the novelty value of the manuscript.  Concerning research question (4), I recommend a specification which environmental factors are studied or in mind.  Do you mean drivers of plant growth such as N deposition, temperature, precipitation and ambient CO₂-concentrations or do you have other factors in mind?

Concerning the P cycle concept, I have one question or minor concern considering the effect of fire on P cycle. L 124 It says “we consider burnt P to be completely retained in the soil”. Do you mean burnt P in the soil or burnt P total plant? Earlier, the authors mention that in the tropical region, most of P is found in the biomass. Hence, I rather expected an addition of both burnt fine and burnt coarse woody debris. Adding specific pools and/or cohorts for burnt debris, would also allow to include the impact of burnt debris or charcoal on decomposition and sorption processes. I have the impression that disturbance by fire was not applied in this study, and thus the results shown are independent of the fire concept. Yet, fires are expected to become more frequent and intensive, and for future use of LPJ-GUESS-CPN, it maybe of importance.   

Detailed comments

L 65 Cramer,  W.,   Bondeau, A.,  Woodward, F.I.,  Prentice,  I.C.,  Betts, R.A.,  Brovkin,  V., Cox,    P.M.,    Fisher,  V.,   Foley,   J.,   Friend,    A.D.,   Kucharik,   C.,   Lomas,   M.R., Ramankutty,  N., Sitch, S., Smith, B., White, A., Young-Molling,  C., 2001.  Global response of terrestrial  ecosystem  structure  and function to CO2 and climate change: results  from six   dynamic  global vegetation  models.  Global Change Biology 7, 357e373. Is a relevant reference here.



L 69 Wrong reference for He et al. 2021.  Change to He et al. 2021a and add He H., P-E Jansson, A.I. Gärdenäs. 2021. CoupModel (v6.0): an ecosystem model for coupled phosphorus, nitrogen and carbon dynamics – evaluated against empirical data from a climatic and fertility gradient in Sweden. Geoscientific Model Development 14, 735–761, doi.org/10.5194/gmd-14-735-2021.

The later references to He et al. 2021 such be changed to He et al. 2021b

2.5 Driving data

L184 Add the ranges in N and P deposition, so that the reader get an impression of magnitude.

Other driving data?

Table 1 Heading, denote CN and CNP-versions same as in text.

L281 Please clarify is water runoff considered important for transport and losses of weathered P or for weathering process?

L356 Here He et al. 2021a is relevant. Please clarify why plant root-symbiosis maybe more important for plant P than for plant N acquisition.  Slight overlap with paragraph L425.

Figures A5 and A6 incomplete figure legend; dots, lines and area denote …

In this study, Dantas de Paula et al presented the phosphorus cycle into LPJ Guess model. Given the emerging importance in P on regulating the C cycle, the N cycle, and vice versa, several DGVMs are incorporating P into its model structure, this among one of its kind. The authors discuss mainly on the GPP and the stocks’ part and conducted a sensitivity analysis for climatic and edaphic drivers. The CNP model fits with the empirical stock data, showed expected P limitation geological distributions. Factor experiments suggest N being more temperature dependent and P to CO₂ changes. These results seem sound and logic. Overall, the paper is well written and clear, thus I would recommend publishing at GMD. The other reviewers have made many comments, here I would emphasize the following which I believe can further improve the manuscript to advance the P incorporation in ecosystem models.

In my view, the overall P model development is mostly centred on two questions, Frist, How N and P limitations are determined, and co-regulations/co-limitations are designed, further how the interaction with C, e.g. GPP calculation is structure and parameterised. These two questions essentially determine the output of the research questions the authors want to address here and beyond. I would rather suggest the authors compare the design of those in LPJ with existing models (although I do see many processes follows existing approaches), and how the difference design or representations make a difference or not? To me this can be an important implication to the community of this model development here.

I would like to raise the importance of parameterisation, since the outputs e.g. table 1, is essentially a product of parameterisation, thus it is crucial to present and discuss which is the crucial parameter, how those parameters are determined, how it compares with empirical process studies. For instance, the discussion of soil organic P and labile P stocks at Line 340 to me is again a parameterisation issue. A major drawback for P model developments is the scarcity of data for parameterize the major soil-plant fluxes, for instance, the weathering/deposition flux, particularly the sorption-desorption dynamics in the soil is the key in regulating the root-fungi-soil continuum in P availability. Discussion of the parameters would also add the implications of current modeling study to further measurements to improve our future understanding the coupled C-N-P cycle. 

What is the explanation for the increase of GPP with factoring P cycle (e.g Line 236), and why is so? Would this relate to feedback of the SOM decomposition? What processes in the model is suggesting this? Is this seeing from other models?

I would also suggest discussing further the drawbacks of current model configuration. e.g. The vegetation-soil P cycle is quite tightly coupled, meaning the mineralization and/or root- fungi uptake is the dominant flux in many cases. Given the soil components in LPJ is simplified, e.g. Line 121-124. How does that influence the P cycle? What’s the limitations of that.

Finally, another way to present the stocks and fluxes are present the stoichiometry, and compare with the empirical data.

The paper is overall well written, some minor suggestions below:

Line 62 what is large time and spatial scales, do you mean P limitation occurred in the geological scale? But DGVMs normally do not run in those scales, so what better be specific here.

Line 112 Precisely the reference should be Wang et al 2007 GBC?

Line 237 why is that, also see above

Table 1 be specific about the “Soil”, in the “Litter+Soil” term.

Line 246 Table 2, the two versions are rather the similar if you consider the uncertainties in your reference dataset, a slight change in r2 and RMSE, is surely non-significant particularly when you compare the global averages.  To me it would be more logical to present this as a benchmark, then discuss the model setup and parameterisation, rather than arguing an improved model performance..

Fig 4, legends a bit too small to see

Fig 5, the same what is soil P here?