Regarding to the question of basis dependency of decoherence, I can perhaps give some examples to clear things up. For a pure qubit state $|0\rangle$, when complete decoherence happens, the basis where the maximum coherence loss is witnessable is in the maximum coherent state
$$
\rho= H | 0 \rangle \langle 0 | H^{\dagger} =
\frac{1}{2}
\left(
\begin{matrix}
1 & 1 \\
1 & 1 \\
\end{matrix}
\right)
\rightarrow
\frac{1}{2}
\left(
\begin{matrix}
1 & 0 \\
0 & 1 \\
\end{matrix}
\right)
,
$$
where $H$ is a unitary transform by a Hadamard matrix, $\rho$ is turned into the maximally mixed state. No matter what unitary operation you perform on it, coherence cannot be recovered.
However, if $|0\rangle$ and the environment interact in a way where partial decoherence happens and a full decoherence is only viewable from a non-maximal coherence basis
$$
\rho = U | 0 \rangle \langle 0 | U^{\dagger} =
\frac{1}{3}
\left(
\begin{matrix}
1 & -\sqrt{2} \\
-\sqrt{2} & 2 \\
\end{matrix}
\right)
\rightarrow
\frac{1}{3}
\left(
\begin{matrix}
1 & 0 \\
0 & 2 \\
\end{matrix}
\right),
$$
where
$$
U=
\left(
\begin{matrix}
\sqrt{\frac{1}{3}} & \sqrt{\frac{2}{3}} \\
-\sqrt{\frac{2}{3}} & \sqrt{\frac{1}{3}} \\
\end{matrix}
\right).
$$
The sight from a Hadamard matrix point of view(where maximal coherence is exhibited) of the decohered state would be
$$
\frac{1}{6}
\left(
\begin{matrix}
3 & -1 \\
-1 & 3 \\
\end{matrix}
\right),
$$
which still has coherence.