Dear Authors,

We have received two detailed and insightful referee letters. After my own assessment of the manuscript and looking at the referee’s comments, I agree with both reviewers that the paper presents an interesting study of scaling characteristics of flow regulation, and an application to the classification of large Amazonian sub-catchments. I also agree that some aspects of the analysis need further analysis, particularly in the discussion, before consideration for publication.

The main points raised by the reviewers that need to be incorporated in a revised version of the manuscript are:

1) As mentioned by reviewer #1 Decreased water storage in the atmosphere, produces decreased precipitation (P is decreasing, so the sum of ET+R is not constant). It is difficult to understand that runoff can increase, and this is the idea of why there could be a shift from a wet to a dry state at longer timescales as suggested in other studies of precipitation recycling. It is clear that runoff can increase as a result of decreased evapotranspiration, but perhaps you need to clearly identify differences in response times and consequences for short and long time scales. This is still unclear in your current responses to the referee’s comments. It is also important to consider that, as mentioned by reviewer #1, moisture recycling has characteristic spatial scales. Please revise figure #6 and its interpretation accordingly. Therefore, though the link between reduced evapotranspiration and increased discharge is clear, the link to precipitation recycling is not. This has important implications for the discussion, as it emphasizes precipitation recycling.

2) As mentioned by Reviewer #2, and based on the results presented in your supplementary material, it is unclear what is the advantage of using LA instead of area for the scaling relations. Please explain and include this aspect in the discussion. At the moment, I don’t see a clear discussion between differences, and the supplementary figures (that show the trends for A and LA, are not discussed). This is linked also to the last comment of this same reviewer that mentions that the role of LAI in the relationships proposed needs to be better explained. Please also consider including a discussion of the differences in results by accounting for LAI as opposed to using A (S=A instead of S=LA).

3) As mentioned by reviewer #2 the 60% threshold is not clear, as having just one “unregulated basin” is not enough to identify a threshold or transition. Note that the modified text, in the response to the comment, is still not addressing this issue.

Please also consider addressing all the other comments by the reviewer’s in the revised paper.

Dear authors,

I believe that the revised manuscript has greatly benefited from the review process. The paper will be ready for publication after answering and clarifying the discussion in response to the minor comment of reviewer #1.

Best Regards,
Patricia Saco

The authors have addressed all comments, and therefore the paper is now ready for publication.