[guided]We need to verify that the expression $\mathcal L_{\sigma,q}(dF_pZ,dF_pW)$ is legitimate. For that, $dF_pZ$ and $dF_pW$ must lie in $T_{1,0,q}N$.
Use the defining-function description of the holomorphic tangent space. Since $\rho$ defines $M$ near $p$,
\begin{align*}
T_{1,0,p}M=\{Z\in T_{1,0,p}\mathbb C^n:\partial\rho_p(Z)=0\}.
\end{align*}
Similarly, since $\sigma$ defines $N$ near $q$,
\begin{align*}
T_{1,0,q}N=\{\Xi\in T_{1,0,q}\mathbb C^n:\partial\sigma_q(\Xi)=0\}.
\end{align*}
Let $Z=(Z_1,\dots,Z_n)\in T_{1,0,p}M$. Because $F$ is holomorphic, its complex differential maps $(1,0)$ vectors to $(1,0)$ vectors, and in coordinates
\begin{align*}
(dF_pZ)_a=\sum_{j=1}^n \frac{\partial F_a}{\partial z_j}(p)Z_j.
\end{align*}
The chain rule for the first complex derivative of $\rho=\sigma\circ F$ gives
\begin{align*}
\partial\rho_p(Z)=\partial\sigma_q(dF_pZ).
\end{align*}
The left-hand side is zero because $Z$ is tangent to $M$ in the holomorphic sense. Hence
\begin{align*}
\partial\sigma_q(dF_pZ)=0.
\end{align*}
This is exactly the condition that $dF_pZ\in T_{1,0,q}N$. Applying the same argument to $W$ gives $dF_pW\in T_{1,0,q}N$.[/guided]