The manuscript describes a new version of TROLL, which represents 3D variation in canopy heterogeneity and individual-level dynamics that can best match field observations. The main development is to include (1) water fluxes, plant water stress, and some degree of micro-environment variability, (2) random intra-species variation, and (3) more realistic crown shapes.

As a vegetation modeler, I read the manuscript and the companion paper with great interests because representing species-level dynamics and 3D heterogeneity are important yet challenging in predicting tropical forest dynamics. I appreciate the efforts by the authors to push in this direction. At the same time, I feel the organization and presentation can be improved to better highlight the novel development and some of the model assumptions warrant justification and further discussions.

Overall, I would recommend reporting more key intermediate variables in the new modules (beyond ecosystem-level outputs in Fig. 4), such as predawn water potential, stomatal conductance, evaporation/transpiration partitioning, LAI_opt, etc. These will help the readers understand the behavior of the model and better interpret the other companion paper. Given the unique 3D configuration, it would be cool to show the simulated vertical/horizontal variation of the new processes/variables.

In addition, the leaf phenology section is tough to read through. Some schematic figures to explain the phenological dynamics would be helpful.  

Finally, it can be useful to conduct/report some sensitivity tests of key model parameters (e.g. crown area allometry, rooting allometry, vertical temperature/VPD gradient etc.) so that we know how results might change with these parameters qualitatively.

Below I list my specific comments along the order of the manuscript

Line 52-65, there are some efforts to extend gap models regional to global scales. See the two reviews by H. Shugart

Shugart, H. H., Wang, B., Fischer, R., Ma, J., Fang, J., Yan, X., Huth, A., & Armstrong, A. H. (2018). Gap models and their individual-based relatives in the assessment of the consequences of global change. Environmental Research Letters, 13(3), 033001. https://doi.org/10.1088/1748-9326/aaaacc

Shugart, H. H., Foster, A., Wang, B., Druckenbrod, D., Ma, J., Lerdau, M., Saatchi, S., Yang, X., & Yan, X. (2020). Gap models across micro- to mega-scales of time and space: examples of Tansley’s ecosystem concept. Forest Ecosystems, 7(1), 14. https://doi.org/10.1186/s40663-020-00225-4

Line 168-176: Does this radiative transfer scheme consider solar geometry? It seems to me LAI(v) only considers the LAI right above the voxel, which means the model does not consider the diurnal and seasonal changes in solar zenith angle. This seems to be too simplistic to me, which can lead to biases in leaf physiology predictions. In addition, is *k* a fixed global parameter or species-specific?

Line 177-188: this LAI-based extrapolation of temperature and VPD seem problematic to me for at least three reasons. First, this scheme means both canopy and understory have the same diurnal temperature range based on eqn. 4 but understory will just be a few degrees cooler. In reality, canopy temperature has large diurnal variations (due to radiative heating) while understory has smaller variations (more similar to air temperature). One idea is to use radiation/air temperature to estimate crown-level temperature (e.g. following Rey-Sanchez et al. 2016). Second, such empirical extrapolation will likely violate the physical linkage between T and VPD. A better method is to interpolate relative humidity and temperature, then calculate VPD based on RH and T. Third, these empirical relationships might not hold when making future predictions under novel climate regimes. Ideally, energy balance should be tracked but I understand it might be too 'heavy' for a 3D model.

Rey-Sánchez, A. C., Slot, M., Posada, J. M., & Kitajima, K. (2016). Spatial and seasonal variation in leaf temperature within the canopy of a tropical forest. Climate Research, 71(1), 75–89. https://doi.org/10.3354/cr01427

Line 218-227. It might be better to expand equation 10 into 2, one for top layer one for other sub-surface layer.

Line 254: how is soil temperature calculated?

Line 274: it is good to see van Genuchten equation is used here!

Line 285: it is unclear to me how LA is used in the model. Is it used to calculate LAD of each voxel in the crown? If so, how?

Line 296: it seems that there is no light-associated vertical trait gradient (or, light-driven trait plasticity) in the model, but this plasticity has actually become a common feature in many vegetation models. Besides, our work with ED2 model (e.g. the method section in Xu et al. 2021 and a manuscript under review) has shown that light-driven plasticity is critical to model realistic LAD profile in tropical forests. And representing realistic environmental plasticity is recognized as a research priority (Fisher & Koven 2020; Xu & Trugman 2021)

Fisher, R. A., & Koven, C. D. (2020). Perspectives on the Future of Land Surface Models and the Challenges of Representing Complex Terrestrial Systems. Journal of Advances in Modeling Earth Systems, 12(4). https://doi.org/10.1029/2018ms001453

Xu, X., & Trugman, A. T. (2021). Trait-Based Modeling of Terrestrial Ecosystems: Advances and Challenges Under Global Change. Current Climate Change Reports, 7(1), 1–13. https://doi.org/10.1007/s40641-020-00168-6

Xu, X., Konings, A. G., Longo, M., Feldman, A., Xu, L., Saatchi, S., Wu, D., Wu, J., & Moorcroft, P. (2021). Leaf surface water, not plant water stress, drives diurnal variation in tropical forest canopy water content. New Phytologist, 231(1), 122–136. https://doi.org/10.1111/nph.17254

L 304: If allometry has a random term, how is growth in height tracked in simulations? Wouldn't it cause troubles in carbon conservation when converting NPP to woody and height growth? Or this is only used during initialization?

L 319: "the treetop grows quicker in height than the outer crown parts" reads confusing to me. How quicker? what functions are used to determine the difference? How is carbon/leaf area balance is maintained in the process? In addition, how is total leaf area calculated for each tree crown? Is any leaf area allometry used? If leaf area allometry is an emerging feature of the model, is it consistent with observations (e.g. the BAAD data base, Falster et al. 2015)

Falster, D. S., Duursma, R. A., Ishihara, M. I., Barneche, D. R., FitzJohn, R. G., Vårhammar, A., Aiba, M., Ando, M., Anten, N., Aspinwall, M. J., Baltzer, J. L., Baraloto, C., Battaglia, M., Battles, J. J., Lamberty, B. B., van Breugel, M., Camac, J., Claveau, Y., Coll, L., … York, R. A. (2015). BAAD: a Biomass And Allometry Database for woody plants. Ecology, 96(5), 1445. https://doi.org/10.1890/14-1889.1

L 357: I believe the horizontal extent of the root is usually bigger than crown. You might check Schenk et al. 2002 for some global meta-analysis of lateral root extent

Schenk, H. J., & Jackson, R. B. (2002). Rooting depths, lateral root spreads and below-ground/above-ground allometries of plants in water-limited ecosystems. Journal of Ecology, 90(3), 480–494. https://doi.org/10.1046/j.1365-2745.2002.00682.x

L. 391: Sounds like Ca has vertical gradient in the model? How is the gradient simulated in the model?

L. 420: internal leaf temperature reads confusing, does the model separates leaf surface temperature and internal temperature?

L. 426: Isn't photorespiration already accounted for in the Farquhar model? Or do you mean reduction of daytime respiration due to Kok effect?

L 479-480: for tropical forests, many recent work by M Slot has evaluated non-stomatal limitation....

L 508: In Xu et al. (2016) we applied this limitation at sub-daily scale instead of daily scale in order to better capture midday suppression in photosynthesis and stomata functioning. To be honest, I am not aware of solid physiological support for such subdaily non-stomatal effects. That being said, applying this at daily scale won't have much effects (psi_pd is usuallal much higher than psi_tlp)

L 531-582: The "leaf energy balance" from Penman-Monteith equation is interesting... I am not sure whether this is actually circular given gs as already been calculated. In my understanding, Tl and VPD are 'inferred' to match Medlyn gs with Penman-Monteith ET... This does not seem right to me. Again, figures of simulated diurnal/seasonal cycle of Tl and VPDs would be helpful.

L 610: what does "conversion factor" mean? convert to what?

L. 681: great to see woody branch turnover is included in the model! They are important!

L 702-704: interesting way to simulate increasing in new leaves and loss of old leaves in dry season but it still only assumes water is the main driver of phenology? Where does leaf age limitation comes into play in this scheme?

L 706-712: I have a hard time to interpret these parameterizations and how they influence phenology. It would be helpful to plot a diagram of the phenological cycle under different a/b/delta parameters. 

L 739-740: what is the fate of the senescenced NPP? Are they lost to respiration?

L 761: what is reproduction opportunity? established 1cm sapling?

Line 879 Fig.4 panel (a) Gross not Growth, panel (d) Psi_TLP is hard to read. In addition to these figures, it would be helpful to show carbon balance at individual tree to ecosystem level (I guess only carbon conservation is tracked?)

L 973: how many CPUs (what type of CPU) are used?

L 1034: I am surprised that the C++ codes are all in a single cpp file for TROLL. From a software engineering perspective, this is not the best practice to organize the codes of 10k+ lines. I would recommend modularization in the future to facilitate model development and sharing....

1. The introduction part from the beginning to Page 3 Line 90 is too long, while there is too short to introduce current research gaps and your model contributions (from Lines 96-100). I suggest the authors reorganize the paragraph structure to shorten the modeling history and stick out why the community need your model. In you paragraph, I learn that (from Lines 96-100) the finer model may able to address the spatial mismatch problems between model simulations and observations in some degrees. However, there is far away enough, add more to make your readers feel like your model is important.
2. Lines 109-111: I suggested the author can point out how much uncertainties of the previous TROLL in simulating some biochemical processes and structures so that they can let the readers know why their new developed model TROLL 4.0 really matters.
3. Lines 332-333: I disagree with this agreement that LAI has no relationship with crown extent.
4. Line 335: why ”total fine root biomass is equal to leaf biomass”? 
5. Line 485: TROLL calculated Vcmax and Jmax based on the LMA and N, have you considered the impact of water stress on LMA and N?
6. Fig 4. I think figures are not enough to explain TROLL model at all
1. Figure 4 should showed the results of these two tropical sites seperately and validate the model accuracy with the site level measurements 
2. For all process revised and new incorprated like hydrulics, soil water, photosynthesis, LAI, the authors should plot figures to let readers know what’s the TROLL model’s performance. That’s to say, all processes you listed in the model description should be visulized and validated. Also detailed analysis  these results are very necessary.
7. Lines 901-911: I agree but why you think current TROLL model development is appropriate? what scientific quesitons new TROLL can answer?
8. Line 963 What’ s the meaning of data-driven knowledge?
9. Lines 983-984: rephrased this sentence, too exeggerated

In this manuscript, the authors present the TROLL 4.0 model, which simulates forest dynamics and processes at the scale of individual trees. Improvements in TROLL 4.0 from the previous version of the model include representation of the water cycle and non-structural carbon storage. In addition, traits can now vary for individuals within a PFT/species – this is an exciting development!

The design of this model seems ideal for parameterizing based on field measurements and for comparing output to field and lidar data. It was a bit unclear to me how the model could help us to improve DGVMs that can be coupled to GCMs, and I would love to see this discussed more explicitly in the text! In addition, it would be helpful to include some figures showing parameter sensitivity.

Major comments:

Figure 4 – I think this figure should be presented in companion paper instead. In its place, it would be useful to see some figures showing the sensitivity of the model to variations in PFT/species input parameters.

Line  965 – “TROLL 4.0 can help inform the development of DGVMs” – can you explain more about how this model can act as a bridge between observations and DGVMs that can be coupled to GCMs?

Line 356-8 – “Tree roots are distributed across vertical layers, but do not spread across belowground voxels horizontally. As a result, trees only deplete the water content of the belowground voxels located below their trunk position” – does this mean that trees are not competing for water? Also, I was a bit confused by this in contrast to Figure 1 (right), where it looks like the roots of the large tree illustrated span multiple belowground voxels. Can you clarify if trees are competing for water in the model or if each tree has its own “bucket”?

Minor comments:

Line 92-93 – consider adding a reference for the FATES model:

Koven, C.D., Knox, R.G., Fisher, R.A., Chambers, J.Q., Christoffersen, B.O., Davies, S.J., Detto, M., Dietze, M.C., Faybishenko, B., Holm, J. and Huang, M., 2020. Benchmarking and parameter sensitivity of physiological and vegetation dynamics using the Functionally Assembled Terrestrial Ecosystem Simulator (FATES) at Barro Colorado Island, Panama. Biogeosciences, 17, 3017-3044.

Line 162 – do you mean that canopy density is recalculated once a day? Does the incident irradiance vary with solar angle throughout the day?

Line 294-5 – do leaf properties vary vertically throughout the canopy? E.g.

Lloyd, J., Patiño, S., Paiva, R.Q., Nardoto, G.B., Quesada, C.A., Santos, A.J.B., Baker, T.R., Brand, W.A., Hilke, I., Gielmann, H. and Raessler, M., 2010. Optimisation of photosynthetic carbon gain and within-canopy gradients of associated foliar traits for Amazon forest trees. Biogeosciences, 7, 1833-1859.

Lamour, J., Davidson, K.J., Ely, K.S., Le Moguédec, G., Anderson, J.A., Li, Q., Calderón, O., Koven, C.D., Wright, S.J., Walker, A.P. and Serbin, S.P., 2023. The effect of the vertical gradients of photosynthetic parameters on the CO2 assimilation and transpiration of a Panamanian tropical forest. New Phytologist, 238, 2345-2362.

Line 335 - add reference for this “common assumption”?

Line 722 - should the size of NSC storage vary with species? E.g.

Signori‐Müller, C., Oliveira, R.S., Valentim Tavares, J., Carvalho Diniz, F., Gilpin, M., de V. Barros, F., Marca Zevallos, M.J., Salas Yupayccana, C.A., Nina, A., Brum, M. and Baker, T.R., 2022. Variation of non‐structural carbohydrates across the fast–slow continuum in Amazon Forest canopy trees. Functional Ecology, 36, 341-355.

Figure 2 – can you use a thicker lineweight and larger font size for this figure?