I cannot find any reason not to publish this paper. Below I list five points that I think might improve the paper.  I recommend publication after some revising.

Suggested Improvements

(1)  At the top of page 3 the authors provide a graphical abstract, which is quite helpful to the paper.  But I wonder if it is possible to do something similar regarding the physical processes or (water) pathways that are emphasized or de-emphasized or changed from one (temporal + spatial + disturbance) regime to another? I hope this suggestion is clear.  As I understand this paper, the authors are describing a systems approach (or model) addressing tree mortality in the Western US. But it seems to me that the paper is largely descriptive of some of the environmental conditions and how they lead to different outcomes for a forest. I think it would be more insightful to discuss the ways in which pathways, by which water moves through the forest ecosystem, change in response to changes in the drivers. 

(2)  As a follow on from (1) above, I think it would be helpful to move the last paragraph (lines 616-629, page 29)  to the introduction.  It would help to put the modeling effort into context.  Otherwise I was left to wonder until the last paragraph what the application of the authors’ modeling effort might be and who might benefit from reading this paper. 

(3)  I have no doubt of the importance of aridity in their findings.  Current expectations are that aridity in the western US will continue to worsen as climate change progresses.  Nonetheless, I am having some difficulty with the term “long-term” aridity.  Yes, they do have a 38-year record.  But at present aridity is dynamic (and has the potential to get much worse in far less than 38 years). I think the authors need to recognize and mention in their work that the past record may not be as useful in trying to project into the future as their findings suggest.  Simply extrapolating from the past 38 years of data may bias their results somewhat, especially if aridity (as represented by the aridity index) is prone to rapid intensification.  The paper would benefit by including further discussion of this issue.

(4)  In the Introduction the discussion of sublimation should probably cite Frank et al (2019: Bayesian analyses of seventeen winters of water vapor fluxes show bark beetles reduce sublimation. Water Resources Research 55: doi:10.1029/2018WR023054).  The findings of Frank et al. (2019) are germane and relevant to the authors’ paper.  Furthermore, Frank et al. cite other works that the authors should probably cite when discussing the impact that beetles can have on the (canopy-atmosphere-environmental) processes involved in sublimation. Given the importance of snowpacks and sublimation to forest ecosystem water balances I would suggest that the authors provide further discussion of the details of the model’s performance regarding sublimation.  The model’s predictions regarding the change in sublimation (from the snowpack and from the foliage-intercepted snow) before and after the beetles have killed the trees would provide some further insights into how well the model captures these important sublimation-related processes and observations.  And although different observers/papers report somewhat different findings, I think comparing the model’s predictions of changes in sublimation amounts and stream flow to previous observations would be worthwhile, especially for a systems level model like the one the authors are using.  

(5)  Lines 344-354, Pages 16-17 – These equations do not make dimensional sense to me.  Either Q, P, E and Sublim are rate variables (i.e., mass/unit time) or the storage terms, ΔS, should not be divide by dt.  If they are rate variables, the authors should include the physical units in their definition.  If they are total amounts (i.e., mass) then they should say that and correct the storage terms. 

The authors contributed a very interesting manuscript that is within the scope of the journal and the scientific quality of the ms is very good. Much of my research is in ecohydrology and one recent ms showed the effect of beetle defoliation in dryland riparian corridors of the SW USA and how water use (ET) on these corridors (13 rivers and streams) changed before and after the introduction of the beetles (see Restoration Ecology 2018); therefore, the authors contribution is of great interest to me and certainly contributes something new to the field of hydrology. There are citations that could be added to the ms. background to further demonstrate changes in riparian corridor ET before and after beetle introductions, although their paper is unique in looking at mountain pine beetle infestation and adding in other types of woodlands may not be needed. I have really learned from their discussion and the long-term aridity index is an excellent contribution to water yield research. These results (key points) show that separating wet years and dry years may provide important knowledge that is useful in other systems. I am curious now to apply similar methods in riparian corridors to see if in fact the response to mortality level remains nonlinear and varies by location and year, as I suspect it would in other beetle-infested land covers. My findings suggest that in canopies that were monotypic with high density and extent had increased water yield could be wiped out entirely but then regrown. The re-greening post mountain pine beetle does not exist I presume and therefore this work may not be transferrable to other ecosystems, but I do believe this ms and its findings, especially the drought information, is of great interest to the readership. This conclusion was therefore of most interest: " in a dry year, low to medium MPB-caused vegetation mortality decreases water yield, and high mortality increases water yield; this response to mortality level is nonlinear and varies by location and year."