See comments on pdf attached here!  I really liked reading this paper and think the deployment/data analysis/results were compelling.  I have comments on the narrative flow and writing elements that I think would greatly strengthen the message the authors are conveying; currently the narrative arc is not supporting what is very solid data and presentation of those data.

The authors present a description of a low-cost mesocosm CO2 flux and ET measurement system. The basic idea of the manuscript and the measurement system is good, as there is a large need for low-cost instrumentation for scientific studies in the developing world; the authors well point out this reasoning for their study. In its current form the manuscript is, however, not publishable without major revisions and further tests.

The level of technical detail within the manuscript is a bit too varying; on the one hand, the Mosfet (the meaning of which many researchers probably are not familiar with!) is described down to a component code and the precise ohm numbers of the resistors, but the manufacturer and model of the linear actuator or the data logging shield, of which there are many available, are not disclosed; neither are the propeties of the air-mixing and ventilation fans disclosed: what volume of air do they move per minute.

The schematics in Fig. 2 are of little use: at first sight, they appear detailed, but the small scale of the images makes deciphering the precise connections made difficult or impossible. A proper schematic drawing (describing which pins on the microcontroller are connected to which pins on the relay board, for example) should be made available along the Arduino microcode to enable readers to build systems of their own.

The design of the "coffin" is not well described. It's not clear whether it's a ready-made design by the Polish firm Romid (if, then what order code?), or constructed by the authors; and if it's self-constructed, how the door and sliding window are constructed (hinges, rails, etc?), how is a tight seal ensured when the window is closed, etc.

These inconsistencies make it unclear whether the manuscript is meant to be a general description of the principles of a measurement system or a blue print. The authors should decide which approach they want to use.

A smaller issue, on line 179, I think the Li-850 already corrects its readings for H2O interference? The authors should double-check this (and present the formula for H2O correction if they need to apply one!)

I find the flux calculation method somewhat strange. Using a variable-size moving window and discriminating against larger temperature changes would seem to prioritize moments when the sun is occluded (low temperature rise) or in the case of constant sunlight cases when the temperature difference between inside and outside is already high (higher outflux of heat lessens the T rise within the chamber), or short fitting times. Instinctively I'd prefer a more constant approach to the fitting, e.g. decide that the fitting time is 4 minutes, leaving 1 minute out from the start. This is not a critical issue here, but if the authors plan to use the system for some actual measurement campaign, they should further examine how gas fluxes are estimated in closed-loop setups in other studies.

The method for testing the sealing of the system is seriously lacking: a smoke bomb creates aerosols, which are multiple orders of magnitude larger than the CO2 and H2O molecules which are the object of measurement here. The proper way of estimating leak (which is inavoidable in a system like this!) is to create a large mixing ratio of CO2, such as 1000 or 2000 ppm, in the chamber and then monitor the development of the mixing ratio within the chamber compared to the surrounding mixing ratio (ppm s-1 delta_ppm-1, where ppm is the mixing ratio within the chamber and delta_ppm is the difference between the outside and inside). Thus a solid estimate of the proportion of air exchanged between the measurement system and the ambient atmospere can be made. The same method can be used to estimate the rate of leakage between the chambers in multi-chamber mode, without the need to use a semi-random factor such as plants in the process. The method of leak evaluation chosen by the authors is not proper for the job. A small difference between mixing ratios on the inside and outside makes leaks nearly undetectable.

Another thing missing is an estimation of how the enclosure affects air temperature: it's a rather well-closed system without any cooling aside the ventilation, so I suspect that the temperature inside can get quite high on a sunny day.

Currently there is an increasing interest in non-CO2 GHG:s (esp. N2O, CH4) emitted from and/or consumed by plants. These can be tricky to measure as the mixing ratios are low and this makes the estimation of leaks even more important. Another important thing is that the materials used for constructing measurement setups, such as rubbers, plastics, glues etc. can emit the GHGs themselves or volatile organic compounds that can mimic or mask these GHGs in the measurement devices. An estimation of the blank flux rate of other greenhouse gases than H2O and CO2 would be very interesting; or the authors should include mention of the need for such a test in their first enhancement proposal (ll. 343-345).