The authors tended to enable DSSAT is able to simulate 1) how O3 affects photosynthesis and leaf senescence, and 2) how O3 effect interacts with CO2 and water stress by associated changes in stomatal conductivity for crops. My comments are as follows:

(1) All stress factors the authors newly incorporated into model are all for the effects of photosynthesis (like FO3) and lead senescence (like SLFO3). But I noticed that the model parameterizations are based on only relative yields (Fig. S1-S3). It is important to have photosynthesis and leaf observations as you tend to simulate their changes to validate model, rather than fitting yields by giving different combinations of parameters. Otherwise, it is very possible you have a “right” yield simulation but wrong parameters, which will give users a lot of trouble when they tend to project model to some unknown conditions.

(2) In line 260, “until the best fit was found for the phenology, growth, and relative yield loss for each cultivar across all O3 treatments.” Where are the “growth” observations? I can only see relative yield data.

(3) For their experiment, I noticed some odd observations in Fig. S2. In rice and soybean, I found yields of some cultivars increase with higher O3, which seems impossible for the new models except unrealistic parameters were fed into (like change some parameters from negative to positive to make it increase rather than decrease. But this is inconsistent with the theory the models were built). Please explain the odd observations in the main text.

(4) I appreciate for the modelers consider the interactions between O3, CO2 and water stress by stomatal conductivity, which I am really interested into. I wish the authors could add more observation points in Fig 4 and 5 to ensure the model can simulate the key interactions quantitively.

This is an impressive compilation of data for many crops from multiple sites to calibrate the data intensive crop models. This was an enormous amount of work.  Overall, this is a good paper that highlights the impacts of ozone exposure on crop yields along with other stressors and should be of interest to a wide audience.

The work focuses on the M7 (7-hour daily mean) ozone metric to alter daily photosynthesis and accelerate leaf senescence.  The authors claim that their approach is more representative than a generic annual damage function.  It would have been nice to see a comparison of a simpler damage function approach with the simulation models presented in this paper to compare the results of the two approaches.  The use of the models outlined in this paper may be difficult to apply because of the need for large amounts of detailed data for each site.  Using weighted seasonal metrics like AOT40, W126 or SUM06 to modify yield might produce robust results as well.  However, I realize the M7 metric for yield loss is the most readily available in the literature.   

The models are well thought-out and carefully constructed.  However, it is hard to know how the models will work in uncalibrated situations.  The only predictive modeling seems to have been done on the evaluation year, 2010, at the SoyFACE study in Illinois calibrated with the 2009 SoyFACE study.  More attention to investigating why the simulations overestimated biomass and yield in 2010 could have been presented.  For example, even though the rainfall at that site was similar between years, 2010 appears to have more of the rainfall at the beginning of the season compared to rainfall patterns in 2009 (Betzelberger et al., 2012).  And the mass of individual seeds were smaller at lower ozone exposures in 2010.

I am not an expert in the modeling field or the uses of this model.  I would have liked to have seen more discussion the implications of this model. How can it be used in the near-term? Do users of the model have to have access to detailed site data or can modeled parameters be used to drive the model?