General comments:

The authors of this study investigated the impact of harvesting on the fluxes of carbon in forests over a large gradient. Based on eddy covariance measurements and on modelling approach, the authors explored the hypothesis that below a certain value of LAI, any forest management action such as harvesting or pruning does not affect Net Ecosystem Productivity. On the basis of a non-linear relationship between gross primary production (GPP) and leaf area index (LAI) characterised by saturation above a threshold of 4-5 m2 m-2, they concluded that above this value, the reduction in leaf area (due to forest management) therefore has little effect on net CO2 uptake and that it remains constant after partial harvesting.

Overall, the study is well structured and of great interest. However, I would suggest some major revisions, detailed below.

Major comments:

With regard to LAI values, it is difficult to understand whether the threshold value indicated by the authors is relevant whatever the PFT. In fact, the definition of LAI varies between deciduous and coniferous stands, due in particular to a difference in clumping index. As a result, its impact on carbon fluxes can also be expected to be different. This point deserves to be discussed. In addition, the results based on the analysis of carbon fluxes measured by eddy covariance technique should be further discussed in the light of the ‘known’ uncertainties concerning the estimation of GPP and Reco during the day.

In general, it is difficult to assess the contribution of using the LPJ-GUESS model. This tool was mainly used to confirm the non-linear relationship between GPP and LAI and to confirm the LAI threshold value, but it could have been used to go further in analysing the weak impact of forest management (competition for light, for example).

Specific comments:

151-153: For the sites studied, are the age and forest management of the plot described, and how have these characteristics been taken into account in the analysis?

162-166: For estimating LAI based on remote sensing, is the spatial resolution of MODIS images sufficient, particularly in relation to the size of the plots (consistent with the comparison with carbon flux measurements), to detect differences in LAI between managed and unmanaged sites?

223-227: In general, it is difficult to assess the contribution of using the LPJ-GUESS model in this study because the description is not very detailed: how is competition for light taken into account, in particular as a function of tree density, the age of the tree stand, etc.? how do photosynthesis parameters vary as a function of age, as a function of PFT? how does a reduction in soil water impact photosynthesis and/or production?

228-230: Does this mean that carbon allocation is only calculated on an annual time step in the model? There are seasonal dynamics that affect the respiration rate associated with organ growth and therefore the NEE. This point needs to be clarified in relation to the conclusions of this study.

230-232: Does the SLA vary with position in the canopy (profile of SLA?) relative to leaf exposure to incoming radiation? This is an important point to take into account when considering light competition and its impact on NEP in relation to tree density.

237-238: How does clumping index vary between PFTs, stand age and tree density? Is this variation taken into account when analysing the results?

267-268: Is the threshold of 4.5 m²/m² the same regardless of the clumping effect? Is this value the same for coniferous stands? Generally speaking, there is no discussion of the definition of LAI for a deciduous stand and that for a stand of conifers (see lines 285 & 305-307).

273-275 & 394-396: This result is relatively expected because if the LAI value increases, we expect an increase in biomass (linked to an increase in canopy photosynthesis) which leads to an increase in growth respiration, one of the two components of autotrophic respiration. Why not use the model to deeply analyse the differences in partitioning of the two components of autotrophic respiration (respiration due to the energy cost of tissue maintenance and respiration due to the cost of tissue construction during the growth phase) between sites and forest management to confirm the hypotheses proposed by the authors? Can the model support the hypotheses mentioned, particularly with regard to the non-linear relationships found with GPP, the distribution of NEE between GPP and Reco, and even the distribution of Reco between growth respiration and maintenance respiration?

417-418: Yes, a discussion on the uncertainty of the GPP estimate could be added, as well as for Reco values during the day (see also lines 304-305). The impact of the age of the stands selected for this study on the growth respiration rate in terms of the amount of living tissue (not total above-ground biomass) should be discussed. An increase in growth respiration could also be expected if there is a stand management practice such as pruning.

420-421: Why didn't the authors try to validate the model's predictions of NEE, GPP and Reco on these two sites? Once this had been done, the model could have been used to validate the hypothesis of an equilibrium LAI and to confirm the threshold value of 4.5 m²/m², and to test the impact of a change in the clumping index due to forest management.

Fig 1: The GPP/LAI relationship is difficult to interpret due to the high variability of GPP values (e.g. for managed conifer/mixed). No point corresponds to the case of managed broadleaves (mentioned in the legend). For the Reco/LAI relationship, it would be interesting to indicate the uncertainties on the graph in the same way as for the GPP/LAI relationship.

Fig 3: as for figure 1, it would be interesting to identify coniferous sites from broadleaves sites.

Fig 4: Why not show the measured NEE in addition to the simulated NEE?

Bouriaud and others studied the impacts of forest management on carbon dioxide flux and found relatively straightforward relationships between LAI and flux that could serve as a valuable guide to sustainable management if the statistical analysis was improved. I was left confused by multiple passages and the referencing could have been stronger. I am also a bit confused as to why a number of North American studies weren’t included for example Andy Black’s transects in British Columbia, foundational BOREAS studies, slash pine in Florida, etc.

54: value should have associated uncertainty for management guidance. Is 4 m2/m2 a minimum?

57: note that this applies to temperate forests

67: 'counteracts climate change mitigation’ sounds like a bit of a double negative and was confusing to read to start of the manuscript. I get it, but had to pause. The next sentence discusses mitigation rather than counteracting mitigation so one’s mind is pulled in two directions.

71: who assumes this? I wasn’t aware that this was a common perception amongst scientists, at least forest scientists.

77: this isn’t always the case e.g. https://www.nature.com/articles/nature12914 and ‘very low’ is at a minimum qualitative. There’s a huge literature on this topic (https://www.nature.com/articles/nature07276) with lots of controversy (https://www.nature.com/articles/s41586-021-03266-z) as the authors are well aware and the statement as written discounts this rick literature. I’ve come to the opinion that people with forest management training think that old stands stop growing because monoculture forests basically do, but natural forests can keep taking up carbon even if at a slightly slower rate. Having been in temperate forests where the mid story is comprised of trees that we would think of as fully grown mature adults with overstory trees proper old growth giants, I’ve always been mystified at the idea that old growth forests don’t still take up carbon. Obviously there is some physical limit. I don’t disagree that older forests might take up less carbon, rather the assumption that they always do; for example take a look at the data points in Fig. 1 here instead of the curves that were hacked through them: https://assets-eu.researchsquare.com/files/rs-5183310/v1_covered_49c18487-5655-429a-a89d-b2e4d64fa22e.pdf?c=1735007962

88: I somewhat that wood provision is considered a disturbance if sustainably harvested to simulate natural forest processes. The passage could easily be re-written to emphasize what the paper is actually about: that harvesting can occur with minimal disruption to carbon sink strength.

95: disagree that selective harvesting is a disturbance or at a minimum that a disturbance is a bad thing; forest harvesting can simulate ‘natural' forest processes as noted above.

As a whole, the Introduction makes a number of valid points, but is weakened by assumptions and poorly-cited statements. It should be re-written to focus more strongly on the matter at hand, and can be guided around the LAI of 3.5 m2/m2 found by Schultze to expand this argument to carbon gain in addition to conductance.

151: couldn’t most of the Canadian sites from BOREAS be considered unmanaged conifers?

153: how was LAI estimated? I see some text on line 162 but this could be written in a much more systematic way for a methods section. Remote sensing and ground-based estimates should be compared to understand their differences (and both have substantial uncertainty that should be estimated if possible).

168: the flux is transport across an area, so yes by measuring transport across the sonic anemometer and gas analyzer the eddy covariance system is physically measuring a flux. Calculating a surface-atmosphere flux does require some assumptions, I agree. Most instruments really just measure voltage differences so one could also argue that nothing measures a flux.

172: low  ustar needn’t represent an error in measurement, it just seeks to represent a case of insufficient turbulence where the assumptions that underlie the eddy covariance technique are not good assumptions. For all we know the sensor measurements themselves can be of the highest quality.

Figure 1: surprised that red and green are being used at the same time. Please use different choices for our colorblind colleagues.

194: please cite the R package

210: this can vary widely based on tower height and environmental factors; Chu et al. (2019) have the most systematic study of footprints across multiple sites and citing this study here can help clarify quantitative aspects of flux footprint dimensions at the network scale.

221: ‘demonstrated’ instead of ‘proven’ is probably a better verb here.

263: in the results section define what is meant by ‘near’. The manuscript should be strengthened by including uncertainty estimates in multiple locations including here (also line 266, etc.).

267: yes, because it’s not a location, it’s a mean value with uncertainty. That could and should be quantified, either as a single value or a threshold that varies as a function of climate or forest type.

279: if it’s not significant, it didn’t tend to be higher. But it could be higher in future studies with more statistical power perhaps.

292: there are more significant digits reported than warranted for a study of dry matter at a plot scale.

321: this is a great rule of thumb but adding an improved statistical analysis could further improve it, or note that the analysis points toward a rule of thumb that could be valuable guidance for foresters with additional research.

331-332: add scientific names. Note also that where the authors are writing from there is a single type of Fagus, but another in the eastern Mediterranean (although the eastern European one is now recognized as a subspecies), and quite a few species in Asia such that simply stating ‘Fagus’ might cause unnecessary confusion for an international study. Note also that Fagus is italicized on line 420 but not in other places.

Fig. 1: are these data points from the eddy covariance data or the model? If the former, what partitioning method was used to infer GPP and Reco?

Figure 3: is <10 etc. the age of the forest since last stand replacement or the time since last management prescription?

Fig. 4: slightly larger font sizes would make this easier to read.

735: species name