Dear Authors,

There have been two reviews of your manuscript which were in general quite positive on your work. However, there have been two major points raised by the reviewers, which have not been addressed sufficiently in your response.
1. Temperature controls on growth (fAPAR)
Reviewer 1: “As mentioned in the description of several ecohydrological classes and sometimes in the interpretation of the results, spatial covariates like temperature play a role and will explain some of the patterns of correlations found, especially those with precipitation. So, why not remove at least the contribution of spatial gradients in temperature as a known important control on forest growth by partial correlations to narrow down the contributions of the hydrological controls? “
Clearly temperature affects growth significantly (mentioned by both reviewers!). Temperature also correlates strongly with the hydrological variables P and WTD as you show in the response to reviewer 1. Further, also fAPAR and T must be strongly correlated! Therefore it is important to control for the temperature effect when trying to identify the regions where hydrology limits growth.

2. Choice of significance level
The presented maps of the ecohydrological classes are based on significance tests: L123: “Based on the significance of the correlation analysis between WTD and fAPAR, and between P and fAPAR, we distinguish 9 ecohydrological classes.”
Testing for significance of the correlation analysis has two key assumptions (https://en.wikipedia.org/wiki/Pearson_correlation_coefficient#Testing_using_Student's_t-distribution): a) independence of the samples b) normal distribution of the samples.
Looking at the high-resolution maps of fAPAR, P and WTD one can see that there is a substantial spatial correlation which varies between the three variables. Therefore the samples samples you draw for the correlation test are not independent. This reduces the degrees of freedom for the t-test applied here and requires a larger threshold to assess if there is a significant correlation.
In your response to reviewer 1 you state that “Currently we are using a alpha threshold value of 5%, which yields a significant correlation value (for n=225) of 0.11. We believe this is the most straightforward way of interpreting the data, as we are not interested in the explanatory power of WTD and P but merely in its significance in driving vegetation growth.“
This answer is not satisfactory to me. The interpretation of the data can be biased when underlying assumptions of a test are not met. Please address the issue of independence of your samples for the test of significance. This is important because the definition of the 9 classes which you call ecohydrological classes relies on it.

Addressing these points has the potential to change the maps of the so-called eco-hydrological classes and to improve the paper towards the aim described by the title. Hence I decided to that revisions are required with further review.
In your revision please address all points raised by the reviewers and in this editorial review.

Best Regards,
Maik Renner
--

Dear Authors,

you addressed all reviewer comments and the revisions of the manuscript improved the paper! The revised paper received excellent reviews by both reviewers including myself. Congratulations.

Please consider the technical comments and corrections by the reviewer and my comments below when preparing the final version of the manuscript.
Note that I used line numbers of the track-changes version.


Comment on aridity Index:
I appreciate that the authors changed their methodology and used the aridity index instead of precipitation. This makes the analysis more general and probably more comparable.
The authors interpret the Aridity Index as
Abstract: “climate driven water availability”.
And in the introduction: “and (thermally controlled) evaporative demand is found.“
From my point of view this is not precise and should be modified, similar to Line L60 in the manuscript.
Explanation: The aridity index normalizes water and energy availability. To estimate energy availability Budyko used Net Radiation. Since Net Radiation measurements are relatively sparse, one often uses Potential radiation. The Penman formulation also uses Net Radiation as one input, but there are variants and approximations, which allow the use of more commonly available meteorological data, especially temperature. However, I would like to stress that this is an approximation and it is not temperature alone which drives energy availability.
Please revise this where appropriate in the manuscript.

L101ff: P/PET Please note that P/PET is often referred to as the humidity index, the aridity index is PET/P !