Proofs correction of April 14 (still valid):

Page 5. right column, in Ref.[32],
replace "(unpublished)" -> "arXiv:2304.06612"

Additional proofs corrections of May 1:

Lines 104+105: delete ", which here is the case for values
of $R$ that are comparable to or less than $\xiM$"

Replace Figure 1 by the corrected one: rspec_select_HoskM_1024a_mu10_k002b.eps
(Here, just a subscript 5 has been added in the lower inset.)

Replace Figure 2 by the corrected one: psaff_1024a_mu10_k002b_rr.eps

Replace the caption to Figure 2 by:

$I_{\rm H}(t)$ normalized by its initial value.
The inset shows ${\cal I}_{\rm H}(R,t)$ versus $R$ at different times $t$:
solid lines correspond to $t=70$, 200, 700, 2000, 7000, and 20,000,
which are also marked by selected colored symbols in the graph of $I_{\rm H}(t)$.
The adopted Hosking integral is evaluated as $I_{\rm H}(t)={\cal I}_{\rm H}(R_*,t)$.
The vertical dashed-dotted line marks the value $k_0 R_*=100$ where the
curves show a plateau.
The slopes $\propto R^2$ and $\propto R^3$ are also marked by dashed-dotted lines.

Replace the text between lines 182 and 213 by:

In \Fig{psaff_1024a_mu10_k002b}, we plot the adopted Hosking integral
$I_{\rm H}(t)$, normalized by its initial value.
It is evaluated as $I_{\rm H}(t)={\cal I}_{\rm H}(R_*,t)$ with
$k_0 R_*=100$, where ${\cal I}_{\rm H}(R,t)$ is shown in the inset
at different times as functions of $R$.
Note that $I_{\rm H}(t)$ is essentially flat and shows only toward the
end a slight decline $\propto t^{-0.12}$, which is similar to what
has been seen for other simulations at that resolution; see, e.g.,
Ref.~\cite{Bra23}.
Thus, the adopted Hosking integral appears to be well conserved --
even better so than the Hosking integral in ordinary MHD, studied in
Refs.~\citep{Hosking+Schekochihin21, Zhou+22}.
There is not even the slight uprise $I_{\rm H}(t)$ reported
in Ref.~\citep{Zhou+22}, which was there argued to be due to
strong non-Gaussian contributions to the field that emerged
during the nonlinear evolution of the system.
Note also that for $R\ll R_*$, we see
${\cal I}_{\rm H}(R,t)\propto R^2$, which is shallower
than the expected cubic scaling.
This might change at larger resolution, although an intermediate range
$\propto R^2$ is also seen in Fig.~4(d) of Ref.~\citep{Zhou+22}, before
cubic scaling emerged for $R/2\pi<10^{-4}$.

Remove Ref. [33], which is now no longer being quoted.
Because of this, and since the old [32] is not quoted later, we have
[32] --> [37]
[34]-[38] --> [32]-[36]
[39]-[42] --> [38]-[41]