I found the manuscript to be well written and the analyses to be strong. My major concern is the third-order polynomial used to fit the NEE/microclimate relationships. This has little physical meaning. Could a non-rectangular or rectangular hyperbola be fit to the GPP data instead? A stronger focus should be placed on the responses of GEP and RE, rather than NEE, on climatic drivers. For example, on page 9 line 42, is the 'quadratic' relationship between NEE and temperature not due to the increase in respiration at high temperatures? The manuscript needs to be refocused toward an analysis of the processes that make up NEE - namely GEP and RE - rather than NEE itself. (Of course some discussion of environmental controls on NEE is warranted, and can help motivate analyses of GEP and RE).

An additional reference to consider when justifying the importance of dryland ecosystems is Poulter et al. (2014, doi:10.1038/nature13376) (http://www.nature.com/nature/journal/v509/n7502/abs/nature13376.html).

On page 3 line 3, how rapidly are dryland ecosystems expanding?

'organic matter' on page 3 line 19.

On line 36 (page 3...using continuous line numbers makes it easier to review manuscripts), it is true that this phenomena is widely known, but are there references to help quantify the magnitude of desertification in China?

On line 42, has vegetation been recovering or has it been given the opportunity to recover?

On page 4 line 24, the eddy covariance instruments are plural.

The hyphen between 10 (and 30) and cm on page 6 and elsewhere is unnecessary.

I feel that the authors have carefully considered all comments and have written an excellent manuscript with exemplary uncertainty reporting. I recommend that this version of the manuscript be published in Biogeosciences after copy-editing.

Sincerely,
Paul C. Stoy