This manuscript presents a modeling investigation of particulate matter (PM_2.5) changes by the mid-21^st century driven by changes in wildfire smoke emissions across North America. Based on simulations with and without wildfire emissions in a coupled fire-climate-ecosystem model (RESFire-CESM), the authors reveal elevated summertime wildfire-induced PM_2.5 concentrations during 2050s over the entire North America, with most substantial increase in wildfire contribution to the total PM_2.5 across the eastern United States. The authors further attribute this remote PM_2.5 enhancement to smoke transport and positive climatic feedbacks on PM_2.5. While this study provides insights on a timely issue, it seems this study could be improved in terms of its completeness. More specifically, the proposed mechanism underlying the remote effects of wildfire emissions on PM_2.5 over southeastern US needs further verification or exploration. Therefore, I suggest to return the manuscript to the authors for major revision.

Main suggestions:

• While the authors investigate the thermodynamical feedbacks associated with absorbing aerosols, it looks like neither the dynamical feedbacks or circulation changes under climate change are explored. Intuitively, one would expect the dynamical/circulation features to be important for the long-range transport of smoke; without a thorough analysis on the dynamical aspects, it is hard to believe that the thermodynamical feedbacks are the only or dominant mechanism underlying the projected enhancement of the remote smoke-induced PM₅. I suggest to 1) analyze observational data to identify circulation features that are responsible for long-range transport of western North American smoke to the southeastern US region; 2) evaluate the historical representation of such circulation features in your model; 3) investigate changes in such circulation feature between 2050s and 2000s, with and without wildfire emissions.
• In terms of the currently analyzed climatic feedbacks to smoke, the current interpretation of Figure 5 does not seem to be complete or robust. For example, the authors argue that “lower-tropospheric stability is enhanced by wildfire aerosols in the future”, but the changes in AAOD (Figure 5A) or stability (Figure 5C) are only significant in very small and inconsistent areas in the eastern US. In addition, how do you explain the lack of stability changes in western US despite the significant changes in AAOD? Are the changes in stability sensitive to vertical distribution of absorbing aerosols? It is even harder to believe that the changes in AAOD (Figure 5A) solely cause the changes in precipitation (Figure 5D), given their very distinct spatial patterns.
• While the authors focus their analysis on decadal mean changes and spatial distribution of such changes, the temporal variations in PM₅ are worthy of investigation too. A natural question would arise whether we will experience elevated occurrence of extreme PM_2.5 days. For example, it would be interesting to see something similar to Figure 6 but presenting the spatio-temporal PDF of PM_2.5 from all grid cells and all days in both the historical and future simulations.
• Does your simulation include multiple ensemble members? Will an ensemble simulation enhance the robustness of your results?
• Finally, while the burnt area does not increase in the southeastern US, it is possible that smoke emission still increases. Does your model account for future biomass increase and possible smoke emission rate increase?

Minor suggestions:

• On line 217-218, the authors attribute the inconsistency between model and satellite-derived PM₅ to the bias in the pristine region, but the scatterplot does not seem to support such bias in the low-value region.
• On line 249, it is also useful to indicate sample size.
• On lines 315-316, please rephrase this sentence, it is not clear what this sentence means.
• On lines 396-398, you can actually test this hypothesis with your simulations, for example by compositing absorbing aerosols in the subsequent days following fire events in Canada.
• Figure 5, what statistical test did you use?
• Around line 450, it is also helpful to report the percentage of grids with seasonal mean PM₅ exceeding 10 μg m^-3.

The authors applied fire-climate-ecosystem model to study the effects on increasing wildfires on summer PM2.5 over the US. Although some good results obtained, I am more worried about the real accuracy of the model simulations, especially in the irregular wildfires occurred in the western United States, which needs to be well verified. In addition, additional analyses are needed to make the results more robust.

Major comments:

The authors are suggested to summarize the previous studies related to PM2.5 changes in the US, especially those focusing on wildfires in the Introduction.

My biggest concern is the accuracy of the model simulation results, especially for the western US, which is essential for the current study. The author simply compared the model results with a single satellite remote sensing product. Note that the sampling frequency of these two is different since there are a large number of missing values in satellite products under cloudy conditions. I suggest using ground-based observations to evaluate the model simulations from different spatiotemporal scales over North America since a rich ground-based observation network of PM2.5 concentrations and its components are available, e.g., EPA, and IMPROVE, etc.

Another concern is that compared with summer (JJA), wildfires in recent years mostly occurred in dry autumn, e.g., California fire in 2020, which has been burning for nearly two months (September and October).

Section 3.1: Suggest comparing different satellite PM2.5 and component datasets over the US since many open datasets are available. In addition, it is suggested to add a validation of a single wildfire year or month, e.g., a severe and continuous wildfire occurred in western US in September 2020.

Section 3.2: It will be very interesting to take a look at the Fire Burnt Area, PM2.5 changes, and related contributions caused by wildfires in the past decade (2010-2020). Wildfires in the west of US have been burning more frequently in recent years, much higher than in the first decade.

Minor comments:

Line 172: Please spell out the JJA and check such issue throughout the paper.

Line 193: Figure 1

Lines 216-217: Again, sampling frequency is also a potential reason resulting in the differences that should be discussed.

Figure 5: Confusing. Are these temporal trends? If not, what the significant confidence level here used and how to calculate?