Let $k$ be a field, $X$, $Y$, $Z$ smooth geometrically connected curves, and $f: Z \to X$, $g : Z \to Y$ finite morphisms.
Suppose that $f$ is separable.
Then we have $f_* \circ g^* : \Gamma(Y, \Omega_Y) \to \Gamma(Z, \Omega_Z) \to \Gamma(X, \Omega_X)$.
By definition, the Hecke operator is this map for certain morphisms.
The definition of $f_*$ is as follows:
Now $\Omega_{Z, \zeta} = \Omega_{X, \xi} \otimes_{k(X)} k(Z)$.
Using this, $f_*$ is defined as $\Gamma(Z, \Omega_Z) \to \Omega_{Z, \zeta} = \Omega_{X, \xi} \otimes_{k(X)} k(Z) \to \Omega_{X, \xi}$,
where $\Omega_{X, \xi} \otimes_{k(X)} k(Z) \to \Omega_{X, \xi}$ is $1 \otimes \operatorname{tr}$.
But why this map factors through $\Gamma(X, \Omega_X)$? I've found a reference (Zannier - A note on traces of differential forms (MSN)), but its proof is too long. I wonder whether we can give a more elementary proof (using $\dim = 1$).
And I've heard that this $f_*$ is the Serre dual of $H^1(X, \mathscr{O}_X) \to H^1(Z, \mathscr{O}_Z)$. Is this true? And how can I show these two maps are the same?