>The new version is a nice improvement over the previous one, particularly 
>with the inclusion of full MHD magnetically forced simulation. I have a 
>few minor comments that should be fixed up before acceptance. 

We thank the referee for a careful reading of our revised manuscript, and the
useful comments provided. We have now responded to all of them, and indicated
the changes with blue font in the manuscript.

>- Is equation 7 missing a bar{U}*b?

Thanks for spotting this. Corrected.

>- After equation 12 it is written "...the corresponding expressions 
>for the fluctuating terms the respective first version...". Is this 
>a typo? Or if not, please reword and explain in a different way as 
>it is confusing at the moment.

We agree with the referee that this sentence was overcomplicated
and confusing. We have now reformulated the text below Eq. (12).


>- Currently, the paragraph with equations 13-14 looks like it's part 
>of "The kinematic limit", which presumably shouldn't be the case. 
>Please add another heading (e.g., method of solution) to make this clearer.

We have removed the paragraph heading "The kinematic limit" and
mention this term in the text.

>- In section 2.6 or elsewhere, I couldn't see any mention of the 
>numerical resolution used (aside from 64^3 in the appendix). Is 
>this 64^3 for all simulations? Please add discussion of this and 
>justify the choices (since if it is 64^3 this seems very low).

We have now added to the caption of Table 1:

"The number of grid points in the A=1 models was 144^3, while the 
number of grid points in the vertical (z) direction was increased 
as the aspect ratio of the box increased, so that 288, 576, and 1152 
grid points in z were used for A=4,8, and 16, 
respectively."

>- Likewise for the Mach number. Although not thought to be 
>important to the dynamo, since the paper insinuates that some 
>of the differences between previous works arise due to incompressibility, 
>this is presumably an important parameter for any FMHD case. It 
>would be helpful to add to table 1.

We have now added to the caption of Table 1:

"The Mach number in the simulations was around 0.03 and 0.04, for low and 
high ReM runs, respectively."

>- Typo "SMHDwhere" above equation 21

Corrected.

>- It was nice to see the FKM case in figure 2, but now this is 
>the only case in figure 2 not also shown in figure 1. Please 
>include there for consistency and because it is important for 
>the reader to be able to compare these more quantitatively.

We have now added the energetics plots from FKM1bd to Figure 1,
lower panel.

>- I presume (from figure 1 FK1b) that the small-scale dynamo is 
>always stable in all simulations because of the low Re_M. But 
>this is potentially important, so would be nice to clarify somewhere 
>in the text.

We have now added to the caption of Table 1:

"Due to the low ReM in all the models investigated, no 
small–scale dynamo could get excited, as for low 
Mach number flows the critical ReM for this instability is 
around 30 (see, e.g., Haugen et al. 2004)."


>- I also noticed on this second reading that the main comparison 
>in figures 1 and 2 is given between SK1b, FK1b, but SKM1a - i.e., 
>they have different Pm. All of the other simulations are at the 
>lower Pr_M as I understand it. While I understand that the FK1a 
>case was stable, which is presumably why this was motivated, in 
>the spirit of comparing apples to apples, it is important to show 
>SK1a and/or FK1a. I was left wondering if the observed differences 
>are due to the magnetic fluctuations or the different Pm or 
>something else. Please comment and ideally show one or both of these cases. 

We thank the referee for pointing out the magnetic Prandtl 
number issue. We noticed an unfortunate typo in Table 1 caption,
where "1/3" was reading instead of "10/3". We have now corrected this error.
Hence, the Prandtl numbers are not varying by so much as was implied
by the erraneous table caption.

Plots from the kinetically forced "a" runs are very non-informative, 
as the decay/growth is very slow, and the butterfly diagrams do not
show any structure. What is evident is that, when a growing solution
is finally seen in SK1a, the structure and growing Fourier modes are the
same as in the corresponding SK1b run. Judging from that, the
Prandtl number dependence appears weak in the kinetically 
forced cases. The magneto-kinetically forced cases were only performed 
with the lower Prandtl number, hence the Pm dependence of those 
solutions cannot be investigated with the simulations at hand.

According to the study of Squire & Bhattacharjee 2015, Phys.Rev.E,
the Prandtl number has an influence on the magnitude of the
eta_{yx}, in their case always negative, in such a way that the
magnitude becomes smaller when the Prandtl number is increasing. 
Their system includes both rotation and shear, and in addition
they do not specify, whether the Rm and Re are changed at the same
time. Hence, the applicability of these results to our case is
uncertain. We have now mentioned their results in our conclusions:

"We note that we have not investigated the magnetic Prandtl 
number (PrM) dependence of the magnetokinetically forced cases 
although, according to the study of Squire & Bhattacharjee (2015a), 
the PrM has an influence on the magnitude of $eta_{yx}$ 
(in their case always negative) such that it decreases
when PrM is increasing. Their model includes both rotation and shear, 
and in addition they  do not specify, how Rm and Re changed when Pm was changed. 
Hence, the applicability of these results to our case is uncertain, 
but studying the PrM dependence is an important future direction. "

It is evident, however, that the level of magnetic fluctuations is
drastically different in the kinetically forced wrt the 
magnetokinetically forced cases: While the magnetic fluctuations
remain modest in all the kinetically forced cases (the magnetic
energy is mostly contained in the mean fields) independent of the
Prandtl number, the fluctuating energy is much stronger in the 
magnetokinetically forced cases. The fluctuations could be
somewhat stronger if the Pm was increased further, but the
main difference in the fluctuation levels would persist
between the two kinds of forcings. 

Therefore, we have now added, to the discussion of kinetically forced
runs

"We note that most of the magnetic energy is in the mean fields, 
while only a small fraction (less than 20%) in the fluctuations."

and 

"The fraction of the mean field energy to the energy in the fluctuations
remains unchanged w.r.t the lower Rm (and Pm) runs."

while for the discussion of the magnetokinetically forced runs

"We note that now the energy in the magnetic fluctuations is dominating over the
energy in the mean fields, with roughly 70\% of the total contained in the former.

>- I didn't understand what was the purpose or the claim of footnote 4. 
>What is an exclusively magnetic SC effect? And what does that have 
>to do with the MRI. Please clarify or remove. Also, the 
>simulations in Shi+ (2016) do not have a mean field, as 
>claimed in this footnote.

Shi+ (2016) do deal with mean fields (even in the xy-average sense),
see their Eq. (10), although their initial field has no such average.
Quite generally, our picture is that the  MRI needs a large-scale
background field to be launched. The MRI-driven *dynamo*
can be considered a turbulent (or mean-field) dynamo, fed on the MRI turbulence
and peculiarly evolving just in a way that its mean field takes over the role of the
initial background field. Hence, it is an essentially nonlinear dynamo.
Without this taking-over, the background field would need to be maintained to 
maintain in turn the turbulence, and the whole situation could not be called a dynamo.
We have modified the footnote for clarification.

>- In the models section 3.3, it says (second to last paragraph) 
>"but this could also be due to the SMHD simplifications." What 
>does this mean? Because the conclusions here come from studying 
>the 0-D model, the "simplifications" should be already contained 
>within the measurement of the coefficients, I would have thought. 
>Please clarify.

The sentence

"Dynamo excitation is easier than in the FMHD cases, which might 
indicate that the coherent SC effect assists dynamo action, 
but this could also be due to the SMHD simplifications."

refers to the DNS models, not the 0-D results, as with closely
matching parameters, kinetically forced FMHD does not give a dynamo,
while the corresponding SMHD does. This could either be due to the
small but negative (coherent) eta_yx present in the system, or then
this is due to the neglect of the pressure term, or both in combination.
The 0-D analysis does not provide here any means to separate these 
possibilities. We have now edited the sentence to read:

"In the SMHD models, dynamo excitation is easier than in the 
FMHD ones, which might indicate that the coherent SC effect 
assists dynamo action, the SMHD simplifications could also be the cause."