Dear Dr. Hildebrand,

I very much agree with both reviewers that your work presents a fresh view on root water uptake that is worth to be published within HESS. In line with both reviewers I think that the current manuscript needs to be thoroughly revised to fully explore the potential of the framework you proposed. The reviewer provided many points for improvement and your final response makes me confident that these will be thoroughly addressed in the next manuscript version.
Both reviewers particularly pointed out that your main conclusion: “that uniform root water uptake is most efficient as it minimizes dissipative losses” is misleading, because it is based on spatially uniform soil water retention properties in your four box model. In case the soil water retention properties are not constant, the energetic minimum will likely neither correspond to the spatially uniform extraction nor to a spatially uniform “root density”. This could and should be investigated using additional test simulations. Thereby you should as recommended by Uwe Ehret use a realistic set of 2 to 3 end members, preferably based on soil water retention properties that have been observed in real systems to assure realism of the underlying parameter sets.

The statement “we ignore soil water redistribution in our model, and thus changes in water content are due to root water uptake alone” suggest that you do not use a fully coupled model for soil water flow and root water uptake? This simplification is valid if root water uptake is faster than lateral redistribution in soil. This can and should be shown. In case of heterogeneous retention properties soil water redistribution crucially affects the ratio between dissipation within the soil and within the plant. I am not sure, whether it makes sense, to pool both dissipation terms into an overall uniform dissipation. Personally I think the plant has an advantage by minimizing the necessary work that has to be performed to extract water from the environment. This might even include that it takes advantage from lateral distribution of water towards the plant, which as the roots do not grow for nothing.

Last not least I think the model could be used to investigate the energetic tradeoffs in the vertical pattern of root water uptake, which is known to be non-homogeneous. I guess the tradeoff between the amount of work that has to be performed to extract water from an increasingly drier top soil or to from the wetter sub soil is the key to understand how forests extract their water from variable depths.

Minor point: Your notion of Darcy’s law is a little different to what is usually used in hydrology. Darcies law determines a flux (m/s) based on the potential gradient (m) and a hydraulic conductivity (m/s). You use a bulk formulation which combines the potential difference (m) with a hydraulic conductivity (m^2/s) to a flow (m^3/s). With this the flux becomes independent from the spatial separation between both boxes, which is unphysical.
Best regards,

Erwin Zehe

I do concur with the reviewers and I am assigning minor revisions simply because I would like to give the authors the opportunity to consider the very minor technical issues raised by the reviewers. As the manuscript stands, I will be in a position then to make a final decision without the need to involve additional reviewers.

I think the manuscript can be considered as ready for acceptance. I only have a very minor issue. The authors state "... (that is, where porosity and the soil water retention curve are defined)". I am not sure that in the context of heterogeneous system the definition of an REV is unambiguous. Limitations about the concept of REV in heterogeneous media have emerged and considered in the field of stochastic hydrology/hydrogeology. What is relevant is that a quantity be observable at some (measurement) scale. The latter does not need to be an REV, as long as it is well defined. This is a detail, but I suggest the authors to consider it.