General comments

The paper by Lee et al., aims to characterize the biogeochemical and biophysical responses to wildfire smoke episodes from a forest and a wetland ecosystems by using eddy-covariance and meteorological datasets.

Although the paper addresses a relevant scientific question, in my opinion it lacks of substantial conclusions, and the reported ones are not sufficiently robust and supported by scientific evidence. The main limits of the current manuscript - as also stated by the authors – are all related and are:

• Few observations (in their current form): despite the wildfire episodes are sporadic and limited in time (there is not much to do for this) data can be treated and aggregated in a more sophisticated way than by the simple day-to-day comparison (e.g. creating layers of data using some representative ancillary variables)
• the attributions of ecosystem response: the very likely presence of confounding effects is not treated in an accurate way so that to exclude (or limit and identify) their effect on results.
• the selection of experimental and control datasets: the simple definition of "smoky" and "non-smoky" days, with the latter being the remaining day of the month, risk to confound the picture too much. This is a challenging task, but a more sophisticated method (e.g. a data driven approach) could be explored.

Some of the literature cited in the paper could provide good methodological examples (e.g. Hemes et al., 2020; Park et al., 2018).

Specific comments

• It would be important to have a proxy variable to scale the AOD values to the measurement sites (e.g. PAR measured at the Saturna Island) and use this to select (or verify) the smoking days (or hours).
• Why did you use different partitioning methods for the two sites? This adds further uncertainty and makes respective estimates harder to compare
• At line 284:285 you report, within brackets, that control days are "all the remaining days in the same month". This criteria must be better clarified and described reported in the method section
• 263-270: did you compare the estimated diffuse PAR fraction with measured data when available? (e.g. from 2017 at the Buckley Bay site)
• Given the small numbers of samples and their distributions, the differences between smoky and non-smoky days should be tested by non-parametric statistical methods.
• The difference in PAR between smoky and non-smoky days seems to be never significant (except for August 2017 in the wetland site).
• The same holds for the diffuse radiation
• The Section 3.1.1 should be drastically reduced, moving all the text describing Fig. 3 in the supplement and keep the most relevant info.
• 360 this is true for the smoke-free days too.
• Section 3.1.3: the described differences are never significant
• Section 3.1.2 Albedo and energy partitioning:
  • Avoid basing discussion on single days. As you actually mention and as it is visible from Fig S4, heat fluxes are extremely variables. It's hard to draw robust conclusions here
  • ll 360:362: No need to compare flux absolute values between the two sites.
• Section 3.2.1 Net ecosystem exchange
  • 392:394: you report that "Both sites became a stronger CO2 sink ... These increases were statistically significant in the first two years." It is actually true that the significance (assuming its estimate is appropriate) is verified for the first two years in the bog and just for 2017 in the forest.
  • I would not discuss single-day values (e.g. ll 403-406), or base a conclusion on 2 days (e.g. ll 430-421)

• The same consideration as for NEE hold for GPP and Reco discussion
• Section 3.2.3 Relationship between smoke and gross primary production
  • the title is not reflected the text. In addition, the relation between PARg and the diffuse fraction is not a result (and well known already) and should be moved in the supplement or in Methods
  • R2s reported in Fig. 6(b) do not seem to reflect the scatter of respective points, please check (however, it is not necessary to report them here)
• Discussions: it is good to compare results with similar studies, but it is better to limit the use of conclusions from other works

Technical comments

• Fig. 1: please increase the quality, e.g., type of maps, distinguishing between EC sites and areas interested by fire.
• 143: insert AOD500 acronym after the word wavelength
• 143: please recall why you select the 500 nm wavelength
• 145-147: please provide some reference examples for AOD value (e.g. during clear sky)
• 2: please improve the axis labels, and align the two panel according to dates (days)
• 240: inconsistent formula for albedo: should be switched, reflected/incident
• 343: "Fig. S3" should be Fig. S4
• 412: add "g" to –5.40 C m-2 day-1
• It would be better to move Fig S5 and S6 into the main text

Sung-Ching Lee et al. “Biogeochemical and biophysical responses to episodes of wildfire smoke from natural ecosystems in southwestern British Columbia, Canada”

GENERAL:

This paper explored the impacts of wildfire smoke on environmental variables and terrestrial productivity using long-term observations from both forest and wetland sites, and found moderate smoke promotes plant productivity via diffuse fertilization effects (DFE) but negative effects under dense-smoke conditions. Overall, this work is interesting and shows the significant impacts of diffuse radiation on ecosystem productivity. However, there are some important limitations need to be overcome before the publication in ACP.

First, this paper neglected necessary comparisons of DFE with other studies. Smoke aerosols can increase diffuse radiation and enhance gross primary productivity (GPP). The magnitude of such DFE should be compared with previous studies, such as Hemes et al. (2020) or Zhou et al. (2021). Such quantification can help explore the contributions of DFE to site-level GPP under different smoke episodes and for varied vegetation types.

Second, this paper had explored the changes of temperature, relative humidity, and other environmental factors caused by smoke aerosols, but did not consider the impacts of these changes of meteorology on GPP. I suggest that the authors consider impacts of other factors besides diffuse radiation using some statistical methods (e.g., simple linear regression used in Cheng et al. (2015)). The explorations of other environmental factors are important especially for temperature and humidity, because smoke aerosols can largely influence them as shown in Figure 3.

Third, this study needed better explanations of their findings. For example, why the changes of PM2.5 and AOD in Figure 2 are not corresponding to each other. The high AOD in 2017 was associated with low PM2.5 while the high PM2.5 in 2020 was associated with low AOD. How such inconsistency affected the impacts of fire smoke on plant photosynthesis? Furthermore, there are some differences in the responses of meteorology as shown in Fig. 3. Why the RH was higher during smoke days for forest site in 2020 but was lower in other years? Also, section 3.1.2 discussed the changes in energy fluxes but did not explain why the responses of H and LE were different in different years.

Finally, the authors needed to exclude other associated perturbations such as cloud and ozone. As shown in Figure S2, the site-level observations show large reductions in surface shortwave radiation during some non-smoke days. Such reductions are likely caused by cloud. Previous studies found that cloud could mask the positive effects of aerosol DFE (Yue and Unger, 2018). So it is not reasonable to compare the smoke and non-smoke days, which is defined as all the remaining days in the same month (Line 288), because the sky conditions between fire and non-fire days could be different. The authors need to redo their analyses based only on clear-sky days to exclude the cloud effects. Meanwhile, fires can cause large enhancement of ozone, which can decrease GPP. Such effects are not considered in this study and need to be discussed about their possible impacts on final conclusions.

SPECIFIC

Abstract: This section needs to explain which processes in this study are biogeochemical and which are biophysical.

Line 23: H and LE should be explained at the first appearance.

Line 267: Is this formula of diffuse fraction applicable for the two sites in this study?

Figure 6: Although the relationships between LUE, PAR and diffuse fraction had been explored in this figure, I suggest that the relationships between GPP and diffuse fraction should be shown and compared with previous studies.

Reference:

Cheng, S. J., Bohrer, G., Steiner, A. L., Hollinger, D. Y., Suyker, A., Phillips, R. P., and Nadelhoffer, K. J.: Variations in the influence of diffuse light on gross primary productivity in temperate ecosystems, Agricultural and Forest Meteorology, 201, 98-110, 10.1016/j.agrformet.2014.11.002, 2015.

Hemes, K. S., Verfaillie, J., and Baldocchi, D. D.: Wildfire-Smoke Aerosols Lead to Increased Light Use Efficiency Among Agricultural and Restored Wetland Land Uses in California's Central Valley, J Geophys Res-Biogeo, 125, 10.1029/2019jg005380, 2020.

Yue, X., and Unger, N.: Fire air pollution reduces global terrestrial productivity, Nature Communications, 9, 10.1038/s41467-018-07921-4, 2018.

Zhou, H., Yue, X., Lei, Y., Zhang, T., Tian, C., Ma, Y., and Cao, Y.: Responses of gross primary productivity to diffuse radiation at global FLUXNET sites, Atmos Environ, 244, 117905, 10.1016/j.atmosenv.2020.117905, 2021.