Review of: PHOREAU v1.0

This paper introduces PHOREAU, a new individual-based, process-based forest dynamics model. The model is a significant development as it couples three existing models: a forest gap model (ForCEEPS), a plant hydraulics model (SurEau), and a phenology-based species distribution model (PHENOFIT). The primary goal of PHOREAU is to improve predictions of forest dynamics under climate change by integrating detailed, trait-based representations of climate-sensitive processes like water use, phenology, and competition for both light and water. The authors perform a comprehensive multi-scale validation, evaluating the model's performance on metrics ranging from daily tree-level hydraulic functioning (e.g., stem water potential) to long-term, landscape-level species composition (Potential Natural Vegetation). The results show that PHOREAU provides reliable predictions across these scales and generally outperforms its predecessor, ForCEEPS.

General Comments

First and foremost, I would like to congratulate the authors for this impressive and substantial piece of work! The manuscript, while long, is exceptionally thorough, providing a wealth of necessary detail for understanding and evaluating the model. It is well-structured and written with great clarity, making it a joy to read.

The presentation of the model is a particular strength. The authors guide the reader effectively from the foundational ForCEEPS model through the various layers of modification and integration. This, combined with high-quality figures (e.g., Figure 2), makes the complex structure of PHOREAU highly accessible. The integration of the sophisticated water modeling via its effects on growth and mortality is elegantly done and represents a significant conceptual advance. Furthermore, the authors should be commended for their transparency regarding model simplifications and the assumptions that underpin them.

The evaluation is very extensive, examining different processes across multiple scales. This approach is reminiscent of "pattern-oriented modelling," and the authors might consider framing it in this context and citing the relevant literature. I particularly appreciate the forward-looking perspective of establishing an evaluation framework that can be easily repeated, which will undoubtedly guide future model development.

The presentation of the evaluation results could, however, be improved. In several figures, font sizes are quite small, and some plots appear stretched. For long time series of daily data (e.g., Fig. 10), the fine temporal scale is lost, making them difficult to interpret. Presenting more of this data as scatter plots (as in Fig. 13) could enhance clarity. To improve readability and focus given the paper's length, the authors might also consider moving some detailed results to the supplementary materials, while retaining the key findings from each evaluation level in the main text.

A minor suggestion would be to more frequently remind the reader in the technical sections (e.g., 2.1.2) that additional details are available in the appendices. Given the manuscript's length, this would help reader navigation. Similarly, a clearer, earlier statement regarding the different time-steps (e.g., daily, hourly) and the multi-layered soil structure used in the water modeling would help orient the reader from the outset.

In summary, PHOREAU v1.0 represents a substantial and impressive advancement in forest modeling. The authors' forward-looking approach, designing the model and its evaluation for continuous improvement, signals a long-term commitment to advancing the field, which I highly appreciate!

Specific Comments

Introduction:

• The introduction provides a very detailed account of species mixture effects (e.g., lines 70-95), which could be shortened. On the other had, a broader context of other forest modeling approaches is a bit lacking.
• L105-110: The text mentions "identified two main shortcomings in forest models." It would be helpful to briefly elaborate on how these specific shortcomings were identified (e.g., through literature review, previous modeling experiments, etc.).

Model Description:

• L159: Typo: "plaform" should be "platform".
• L216 (Eq. 1): There appears to be a layout issue in the equation. It should likely read 2*H_max,s - b_s * e^(...) rather than having the allometric parameters in the denominator's exponent. Please verify the formula.
• L232: Suggest inserting the word "species": "shade intolerant species having...".
• L251-265: The concept of "crown ratio reversion" needs clarification. Does this mechanism allow the base of the living crown to move downwards again, effectively re-greening parts of the stem that were previously bare? This should be clarified here and in the appendix.
• L277: The symbol for the clumping factor appears to be missing from the parentheses: clumping factor ( ).
• L333ff: While "symplasm" is defined by contrast, a brief explanation of "apoplasm" (the continuum of cell walls and extracellular spaces) would be beneficial for non-specialists.
• L368: Typo: "depending on depends".
• L382: It would be useful to briefly explain what a "semi-implicit solver" is and give an indication of the runtime difference it makes compared to an explicit solver.
• L531 (Eq. 13): There may be a typo in the denominator. The text reads P_gSBB, which seems inconsistent with the parameter P_gs88 defined on L525. Please check.
• L545: The growth reduction factor GR_crown is used here but has not been fully explained. A brief definition is needed.
• L553: The citation Hammond et al., 2019 appears to be missing a closing parenthesis.
• L620: The rationale for weighting light availability by daily mean temperature needs more justification. This method heavily weights hot summer months when growth may be limited by other factors (like drought). A temperature response curve that is primarily limiting at the cool end of the spectrum might be more ecologically realistic.
• L651 (Eq. 19): Notation for leaf unfolding and coloration intervals is slightly inconsistent between the text (Uls, Cls) and the key below the equation (UI_s, CI_s).
• L695: A word appears to be missing in "the fine root area of a tree in a determines...".
• L749: There is an extraneous character (a hyphen) after the period at the end of the paragraph.

Results, Discussion & Figures:

• L765/Figure 4: This figure is difficult to interpret due to very small font sizes and hard-to-distinguish color-coding. Furthermore, the claim that it shows "acclimatization" over 1500 years seems more likely to reflect changes in stand structure and species composition rather than plastic adaptation within individual long-living trees. Please clarify.
• L880/Figure 6: There appears to be an inconsistency in the visualization. For instance, for the Puéchabon site, the caption states "3 patches of 100m²", but the grey grid lines on the ground seem to depict a different arrangement (e.g., 4x4 grid). This should be checked for all subplots.
• L914: The paper states that longer time-lapses "would have mechanically improved simulation results". This is counter-intuitive, as one might expect simulating longer periods to be more challenging and prone to error accumulation. Could the authors please clarify what is meant by "mechanically improved" results in this context?
• L1019: Typo: "crown Al ratio" should likely be "crown ratio" or similar.
• L1059: There is a minor date discrepancy for the Hesse site thinning. The main text mentions a cut in 2005, whereas Appendix Q lists thinnings in 2004 and 2009. This could be harmonized for clarity.
• L1250/Figure 17: This figure effectively illustrates the model's performance across ecological gradients. A very nice visualization.
• L1375 (and elsewhere): The citation 'Allen, Macalady, Chenchouni...' is very long. This format occurs multiple times (e.g., L54) and could be consistently shortened to 'Allen et al.' for readability.

References:

• L2231: The reference for Bréda, Soudan and Bergonzini is listed with "(no date)", which is unusual and could be clarified.

General comments

This manuscript describes the integration of three models that represents forest gap dynamics (ForCEEPS), a phenology model that predicts species distributions (PHENOFIT) and a ecohydrological model (SurEAU). The paper highlights the main characteristics of the integrated model (PHOREAU) and show comparisons with multiple data streams across Europe to appraise the model’s ability to represent multiple processes governing water and carbon cycles, as well as the structure, composition and dynamics of forest ecosystems. The model development and assessment presented in this manuscript is without a doubt very scientifically valuable and in scope for the journal. 

However, my impression is that the current manuscript is trying to cover too much material, which makes the manuscript difficult to follow at times. I also found that the model assessment could be elaborated further, because many results are only briefly described and discussed. I think the best option would be for this paper to be split into two parts, one focussing on the model description, and another that presents and discusses the model assessment in more detail. There are multiple examples of two-part contributions in Geoscientific Model Development, and I would encourage the authors to consider this alternative.

I have some additional suggestions on how to improve clarity and streamline the model description and model assessment, which I go into more detail in the points below.

Section 1, Introduction: I think the introduction is for the most part concise and accurate. My main suggestion is to try to present PHOREAU in a broader context of vegetation demography models. A few references that could be interesting as starting points are Fisher et al. (2018) (https://doi.org/10.1111/gcb.13910) and Bugmann and Seidl (2022) (https://doi.org/10.1111/1365-2745.13989).

Section 2, Presentation of the model: I think this section could benefit from some extensive restructuring. Currently, the sub-sections have several back and forth points (e.g., section 2.1.1 points to specific equations in section 2.4.5), and in many cases equations are shown in blocks (e.g., near lines 425-440) that do not really describe each and every term. In addition, some information is presented in sub-sections with titles that do not describe them (e.g., rain interception is presented in section 2.4.3 “leveraging leaf phenology and hydraulics to temporalize competition for light”). Part of this may have stemmed from the authors seeking to keep the sub-sections aligned with the contributing models (ForCEEPS, PHENOFIT, SurEAU). Whilst I appreciate this, I think it is more important to streamline the description of PHOREAU and organise sections according to processes, and refer to the original models for the specific modules as needed. I also think it would be better to start with a brief overview of the model (as the authors already do), describe the initial and boundary conditions needed by PHOREAU, present the fundamental equations in the first subsection (i.e., combine 2.1.1 and 2.4.5), then split the other sections by process. To link specific processes to each originating models, the authors could consider adding a table in section 2 that lists all processes, their time scales, the originating model and the sub-section where they are described. Finally, it would make it much easier to follow this section if the authors presented the equations and terms in the same order as they first mention them.

Section 2.1.1 and 2.4.5. I missed one equation that brings together establishment, growth and mortality to describe the change in forest characteristics (stem number density or basal area). Even though it may be a bit obvious, it would help visualise how PHOREAU resolves forest dynamics.

Sections 3 and 4. I truly appreciate that the authors acknowledged that PHOREAU is a process-rich model and provided a comprehensive assessment of the model that went beyond a basic comparison of a single metric. I also liked that the authors appraised the model performance under multiple climates across Europe, and pointed out where the model predictions work best and has issues. That said, in part because the paper is rather long, the results of this comparison are only briefly described, and some of the comparisons may need more explanation, because of the multiple assumptions on both the model and the observations. I think that if the authors keep this as a single paper, they should simplify many of the analyses. For example, their Figures 10 and 11 have several simulation details (e.g., individual contributions of transpiration and evaporation from soils, trunks and canopy interception), but section 4.1 mostly describers the comparison of total ET. If the individual components do not help explain differences, the authors could simplify the figures and only show what they can describe. Likewise, Figure 12 is only mentioned as a support to explain Figure 10, and little is said about PHOREAU’s ability to represent the seasonal cycle of stem water potentials, and how the model perfomance varies across the sites. More critically, the comparisons with observed forest inventories use potential vegetation simulations. This may be the comparison that is feasible, but most of the forests are managed, so it is unlikely that the discrepancies are entirely due to PHOREAU being unable to represent forest structure and composition.

Specific and minor points

Figure 1. Considering that PHOREAU is an integration of three models, it would be helpful to indicate which processes in the figure are primarily coming from ForCEEPS (e.g., add ForCEEPS beneath the “competition for light” bubble.

Section 2.1.2. I suggest citing each individual appendix near the specific process that they describe in detail.

Lines 161–163. The authors refer to the Capsis modelling platform multiple times throughout the text. Even though they provide references, it may be worth describing briefly what this platform is and does, for those readers unfamiliar with the platform. 

Eq.1. The denominator seems to have some formatting issue.

Line 261. This is a bit cryptic. When the authors refer to “crown reversion when light availability increases”, are they referring to increased light availability due to changes in forest structure, or something else?

Line 285. Is SLA the only variable controlling the diversity of drought resistant strategies in PHOREAU? If so, provide a high-level overview of how this works in the model (e.g., through empirical relationships with X, Y and Z).

Figure 2. Replace “density-dependant” with “density-dependent”

Line 314. The correct term seems to be meteorological data, unless PHOREAU expects long-term averages for each day of the year. In addition, no need to change the model, but from reading the paper I got the impression that PHOREAU takes ERA5-Land data aggregated by day, then uses some internal procedure to disaggregate the data to hourly. I suppose in the future it would be better to take hourly data directly from ERA5-Land.

Lines 321-326. This explanation appears multiple times throughout section 2 (e.g., 454-456, 465-468). I suggest adding this information only once, when providing the model overview early in the section, and dropping all the other instances.

Lines 328-334. The authors describe the main characteristics of the symplasm, but not the apoplasm.

Line 351. Replace “phenomenons” with “phenomena”.

Line 356. There seems to be a problem with the parentheses.

Line 383. Consider dropping the word “exactly”, as it implies bit-for-bit comparability, which is unlikely to be the case.

Line 483. Repetitive, consider dropping the sentence.

Line 501. How are the results disaggregated? Does the model assume that all trees had the same values for the predictors of growth and mortality, or does it assume some sort of distribution? Either way is fine, but a bit more detail would be helpful.

Figure 3. This figure currently has too many symbols and details that are not really described in the figure, the caption or the paper. Perhaps they are explained in the original papers. Additionally, the image quality is a bit poor, so it is hard to read. I suggest replacing this figure with a simplified version of it, where only the key processes are spelt out more clearly.

Lines 551-554. Does this assumption influence the seasonality of variables that may depend on PLC, such as GPP or ET?

Line 565. Consider replacing “random” with “spurious”, unless referring specifically to a process that is represented by a random variable in PHOREAU.

Lines 619-623. How does the weighted average by mean daily temperature works? Does PHOREAU use absolute temperature (Kelvin) or a relative scale (e.g., Celsius). This would give very different results… Also, if the latter, how does the averaging work when the temperature is at or below freezing?

Line 679. Unless I missed it, this is the first occurrence of SurEau-Eco, this needs a description.

Lines 689-690. This part is a bit confusing. What are the inputs the user is supposed to provide?

Line 705. Consider using the present tense instead of future tense.

Lines 713-716. Are there any assumptions on the vertical distribution of roots within the rooting zone (uniformly distributed, exponential decay).

Figure 4. The specific species and sizes are a bit hard to read due to the colours being similar to the background. Also, would it be possible to disaggregate the above-ground contributions of each species to the basal area and LAI, similarly to what is done in the below ground.

Line 782. Equation 26 does not strike me as a core model equation, at least not for PHOREAU.

Line 809 (Eq. 31). How is light tolerance defined?

Lines 829-831. Even though this is explained in more detail in Appendix M, this paragraph comes a bit out of nowhere. Perhaps add some more context of what the bootstrapping does.

Figure 5. Drop the word proposed? This is the implemented framework, “proposed” implies something to be developed in the future. In addition, the meaning of the arrows is a bit unclear.

Lines 844-847. I do not think this is really the case. For example, Maréchaux and Chave (2017) (https://doi.org/10.1002/ecm.1271) has an extensive assessment, using multiple observation metrics.

Table 1. Some rows are a bit unclear. For “Stand inventory”, perhaps replace with something more specific, like “Available stand inventory information”. Likewise, I was not sure about the meaning of “available soil water quantity”, is this the average, maximum, minimum or the initial value?

Line 891. The section title is “ICP II sites”, but the section describes the RENECOFOR network too.

Line 944. Missing description for SILVAE

Figure 8. Is there a reason to refer to the old INRA logo instead of INRAE (which is also shown)?

Line 986. Provide actual numbers instead of “a few dozens”.

Section 3.2.3 I did not follow this section. Is PHOREAU being compared with ForCEEPS, which is itself a component of PHOREAU? This struck me as rather circular….

Section 4.1. Again, the sub-section title does not correspond to what is described. The first paragraph is entirely about forest structure, not water balance and plant hydraulic functioning. 

Line 1055. I thought Figure W3 would help complement what is shown in Figure 9 and could be presented in the main text (likely omitting Barbeau in both figures as it does not have observations). I think this would strengthen the point that PHOREAU seems to be doing a very good job representing the basal area dynamics, though likely through compensating biases in both growth and mortality.

Line 1073. I may have missed it, but I think Figure 12 is mentioned before Figure 11.

Figure 10. The comparison for Hesse is fundamentally different from the other sites. There are many assumptions on how to scale sapflow measurements to total transpiration (which is not the same as ETR). Some context or some reference is needed for explaining how this scaling was made, and the comparison should focus on transpiration only, not ETR.

Figure 11. Why is the colour ramp scale stretched from 0-500 cm, if all the depths in the labels are within the top 150 cm only?

Figure 12. Units in the Y axes should be MPa

Figure 15. I think this plot is rather difficult to interpret. The comparison of basal area increments by species across multiple sites can go wrong for so many reasons, including the fact that the forest structure and composition predicted by PHOREAU does not match with the observations. I wonder if some assessment of emergent properties of the ecosystem would be more valuable in here. For example, the authors could compare how plot-scale basal area increment relates to total basal area in both PHOREAU and the observed inventories, and understand if the model has reasonable representation of canopy occupancy. Similarly, mortality rates as functions of total basal area could indicate the model ability to represent canopy thinning.

Section 4.5 and Figure 17. How are the “accurate prediction”, “partially accurate prediction” and “false prediction” are quantitatively defined? Without knowing these thresholds, I think there is little value showing this figure. In addition, panel (a) overlays too many colours, and it is rather difficult to identify the patterns of model performance. Perhaps the authors could use a white background in panel (a), and the same colour code and same symbols in both panels to make these figures a bit more comparable. 

Lines 1254-1256. I think similar assessments exist for other models, if not individually based, at least cohort-based models (which seems to be the approach used by PHOREAU in any case). For example Xu et al. (2021) (https://doi.org/10.1111/nph.17254) and Eller et al. (2020) (https://doi.org/10.1111/nph.16419) have to some extent assessed similar characteristics.

Lines 1271-1277. In principle, I agree with this discussion point, but there is a downside in the simplification too. Models that are too simplistic may lack mechanisms to represent how forests would respond to shifts ti no-analog conditions, such as changing climate or changing disturbance regimes.

Lines 1333-1342. This was not extensively assessed in this manuscript, so I think before implementing more complex approaches, it may be useful to test whether these new processes are indeed needed, or if the results as they are are already reasonable. 

Line 1362-1371. Use HEH instead of EHE?