I'm interested in the first Chern class of the cotangent bundle. I concretely work on the sphere $S^2$, but the reasoning below seems to work for any manifold.
I take the symplectic point of view considering the standard symplectic form $\omega$ on $T^*S^2$. We may also take a Riemannian metric on $T^*S^2$ in order to get an almost complex structure so that $TT^*S^2$ is a complex vector bundle. My advisor pointed out that now one can choose a splitting $TT^*S^2\cong H\oplus V$ (via a connection) in the horizontal and vertical bundles. These are then identified with $\pi^*(TS^2)$ and $\pi^*(T^*S^2)$ so that from the fact that $c_1(E^*)=-c_1(E)$ and $c_1(E\oplus F)=c_1(E)+ c_1(F)$ one concludes that $c_1(TT^*S^2)=0$. Here $\pi: T^*S^2 \to S^2$.
I would like to understand the details of this. In particular, I feel that we have to make sure that $TT^*S^2$ and $\pi^*(TS^2)\oplus\pi^*(T^*S^2)$ are isomorphic as complex vector bundles as $c_1$ depends on the complex structure. But I can't see what the complex structure on $\pi^*(TS^2)\oplus\pi^*(T^*S^2)$ should be.
Furthermore, I'm not sure about $c_1(E\oplus F)=c_1(E)+ c_1(F)$. Wiki says that the chern class of the direct sum is the cup product of the chern classes so I guess it should say $c_1(E\oplus F)=c_1(E)c_1(F)$, but many authors use addition.