He is writing the group element $$g = 1 - i T_a\ \Theta^a(x)$$ in a neighborhood of the identity, and then using... the chain rule? At least I though so, but then I can't understand why there is no $$i$$ factor in equation $$(3.11)$$.
For context: $$g$$ is an element of the Gauge group, $$F^a$$ are $$n$$ functionals (where $$n$$ is the dimention of the local gauge group), $$A_\mu$$ is the connection form and and $$A_\mu ^g$$ is the transformed $$A_\mu$$ under the action of $$g$$.