Joined: Mar. 2005
One could also point out that the validity of the theory doesn't depend on direct observation of speciation. As I understand it, that's usually a slow process, in which as two populations diverge the ability to interbreed declines, perhaps slowly.
But, the theory does imply patterns that should be consistently observed if it's correct (or at least a close approximation), but that would not be expected in combination otherwise. The main pattern here is the the matching nested hierarchies constructed from multiple traits or DNA segments. (Ironically, hybridization causes an exception to the nested hierarchy thing.)