This is a well designed stydy with original and long-term measurements of reactive gas fluxes. I have noticed a rigorous experimental set up, data is clearly presented with no language flaws. The originality is limited in the sense that no major finding were observed and typical range of existing BVOC were described. However, this is one of the longest time seriers currently existing, and this dataset could be extremly valuable for modelling and other scientific purposes, so I strongly belive this would can be published as it is. 

Dumont et al. provide a highly valuable dataset of multiyear observations of VOC fluxes and ozone fluxes. Knowing how challenging such measurements and their analysis are, I applaud the authors for not only performing them, but also for dealing with such a large dataset. It is also highly commendable that the data, which will be highly valuable for the atmospheric chemistry modelling community, are already publicly shared including an open source processing code, which, in turn, will be useful for researchers performing similar measurements. I find the paper/data description extremely thorough – which causes its length, but seems necessary if other researchers are supposed to use the dataset – and only have some minor comments that should be addressed before accepting it for publication in ESSD. Those comments relate mostly to treatment of fragments and water clusters in the PTRMS dataset, and the necessity to compare differences in criteria for “significant” fluxes between this study and others.

Specific comments:

2.1 Study site: with the available data it must be possible to give percentages of the different tree species in the footprint. This would be an important piece of information for the result interpretation.

l. 259: according to Ionicon, negative peaks in IDA only occur when overfitting (e.g. the total peak area is already 102% with the first 7 peaks, so the 8th peak is made negative to get to 100%), so this points at a general problem in the data processing. This should be discussed here.

l. 397: the wood factory is only outside the 80% footprint, maybe even within the 90% or 95% footprint?  - and definitely within the 99.99% footprint, so it is not a surprise that emissions from it can be measured

l. 399: at night, footprints are larger due to more stable conditions, so again it is not a surprise that influence from a larger area can be seen then

l. 491: relative to what? 

l. 569: what brand/model of HPLC cartridges were used?

l. 621: not just fragmentation, also water clusters

l. 628: is benzene really benzene or could it be a fragment e.g. of a phenolic plant emission? 

l. 636: The fact that some compounds were not detected by the algorithm despite exhibiting diurnal trends, and others were detected but were considered insignificant by the authors, raises questions about the algorithm. Why could such mis-filtering happen and is there any way to improve the algorithm in that regard? In the end, if manual work is needed anyway, one could wonder why an algorithm is used at all.

l. 649: I think the authors should compare sensitivities/LODs or filtering criteria to find out if these could explain why more or less fluxes were considered significant in other studies. 

l. 654: in what way were trunk flux dynamics more difficult to assess, can you give an example?

l. 654: why are acetic acid and its fragment listed separately? If acetic acid was calibrated with a gas standard on the parent mass, the fragment should be omitted to avoid double counting; otherwise why are the two not added up?

Fig. 6., Table 3, treatment of fragments of gas standard-calibrated compounds: for example, why are mass 81 and mass 137 considered separately? If the monoterpenes were calibrated on the parent mass, then including the mass 81 fragment leads to double counting of a certain fraction of the monoterpene flux and concentration. The same is true for the mass 41 isoprene fragment.

Table 3 (some of these points are also relevant for the supplementary list of detected ions): 

-    Please mark the compounds that were calibrated with a gas standard and thus have higher certainty than the others. 

-    When you report the seasonal mean mixing ratios, it would also be interesting to show the standard deviations. 

-    The mass 151.112 could also be an oxygenated monoterpene like thymol, carvol or similar, why is the fragment of pinonaldehyde considered as the likeliest in table 3? 

-    water clusters should not be ignored. I assume that “ethanediol” is more likely acetaldehyde + H2O, and “propanediol” is more likely acetone + H2O. Please check through correlations with the parent mass. 

l. 709: “xylene” and “trimethylbenzene” sound unlikely to be emitted from a forest. Mass 121.101 could be a fragment of sesquiterpenes, and mass 107.09 a fragment of limonene, see Kari et al.: PTR-ToF-MS product ion distributions and humidity-dependence of biogenic volatile organic compounds, International Journal of Mass Spectrometry, 430, 87–97, https://doi.org/10.1016/j.ijms.2018.05.003, 2018.

l. 781: It has been reported in other studies that the fraction of carbon allocated to VOCs increases under stress (e.g. Peñuelas, J. and Llusià, J.: BVOCs: plant defense against climate warming?, Trends in plant science, 8, 105–109, https://doi.org/10.1016/S1360-1385(03)00008-6, 2003.). How did this fraction change during the heatwave in 2022?

l. 830: I think the main point from Millet et al. was that deposition fluxes have lower individual contributions, that is why there are more species needed to explain e.g. 90% of the deposition flux.

l. 865: Was there any influence of rain events or soil moisture seen on the TRUNK emissions? Emission bursts of VOCs from soil after wetting have been reported in several studies, e.g. Bourtsoukidis, et al.: Strong sesquiterpene emissions from Amazonian soils, Nature communications, 9, 2226, https://doi.org/10.1038/s41467-018-04658-y, 2018.

l. 946: It looks like the O3 deposition was larger when ozone concentrations were larger. Is this relevant besides stomatal aperture?

l. 1026: it is not clear which kind of remote sensing measurements are available. They should be made available with a doi, like the rest of the data. 

Technical comments:

L. 8 abstract: „processing pipeline“ sounds to figurative and can be confused with an actual pipeline, I would just say software

l. 91: “same two levels” – which ones?

l. 93: I assume you mean vertical profiles of turbulence measured by anemometers (“profiles of anemometers” sounds wrong)

l. 145: “absolute and stable” needs to be explained. It became clearer to me later in the text when the other ozone instrument was introduced, but at this point it was not clear

l. 176: what material was the pump membrane? PFA? Otherwise the ozone might react with the membrane?

l. 220: you can make clear here already that the PAP software is available and cite it here

Fig. 7: Nice way of presenting the data, but way too small to make out anything useful from it. 

Fig. 8: the colors of mass 137 and of “other > 0” look the same in print. 

The investigation by Dumont et al. at the ICOS Vielsalm station presents a unique dataset of long-term observations of concentrations and eddy-covariance fluxes for BVOCs and ozone. The study is unique in the scale and duration of BVOC flux measurements, spanning several years with observations both above and below forest canopy, and also thorough its presentation of methods and analysis used for preparing the dataset. Combined, this study and the dataset collected is a significant contribution to BVOC research, providing valuable data to the modelling community, and should be accepted for publication following minor revisions. Of my comments, the main area where the manuscript would benefit perhaps most substantially concerns the presentation in the text of VOC-related signals vs identified VOCs (for both concentration and significant ecosystem exchange), particularly in Figure 6 and Table 3. The distinction between detected ions and identified VOCs needs to be clarified (in both the Figure, Table, and their captions) to prevent ambiguity and potentially misinterpretation of the presented results.