Dear Dr. Hudiburg,
thank you for providing detailed responses to each reviewer comment.

Your response to reviewer 1:
1. I ask you to consider using existing AR4 or AR5 climate change scenarios to apply them to DayCent for your study region because these climate scenarios provide physically consistent climate variables for a 2-degree warming. Otherwise, the error propagation is too high and your results can be biased.
2. Your response, "With respect to climate forcing, again,
we are forcing the erosion events to occur regardless of precipitation, based on the
reconstructed fire history record." Please make sure that this approach is thoroughly explained in the methods section. Also explain, why you cannot derive such type of information from your climate forcing data.
3. Net ecosystem responses cannot be derived from simulating fire pattern alone. Please re-consider your response and revise your manuscript as demanded by reviewer 1.
4. Reviewer 1 has offered you two options for improving your manuscript. Please reconsider to take one of the options to allow this manuscript getting published.

Reviewer 2:
1. Provide the information demanded by the reviewer in the manuscript text, accordingly.
Cf. Reviewer2: Materials and Methods: L165 What exactly is the size of the simulated area? Are fires
spatially-explicit? Or just based on random selection of cells? Perhaps a few word on
this.
Your Response: This was a ‘point’ simulation (size is not explicitly modeled) for a single
study site. The simulation represents the watershed (c. 30 hectares) that would be
affected in a high-severity fire with erosion. The fire is spatially-explicit to the single
point, as there are no other points/grids. We have clarified that this is a point simulation
in the text.
2. Explore all available options for validating also vegetation composition or productivity as demanded by reviewer 2: "This removes the necessity to do the paleo-informed, but nevertheless paleodata comparison is necessary as a validation step" and describe it in the manuscript.

In addition to these changes that need to be taken into account in the revision of the current manuscript, all other changes demanded by the reviewers need to be considered. You have announced that these changes were or will be conducted in the revised manuscript. These changes will be essential.

Best wishes,
Kirsten Thonicke.

Dear Tara Hudiburg,
thank you for the in-depth revision of your manuscript. It can now be accepted with technical corrections. This includes:

It will be important to provide the link to the detailed model description of DayCent (cf. "DayCent has 280 excellent documentation online (powerpoints, step by step instructions, publication lists; 281 http://www.nrel.colostate.edu/projects/daycent-downloads.html)." as well as the link to the Dryad repository with acceptance of this manuscript. The respective link or information on the Dryad archive needs to be added.

Best wishes,
Kirsten Thonicke.