There is no way how to do that. For this reason. Both are functions, where the input space is infinite (ok, its finite, because of the finite Damgard Merkle padding, but lets consider it infite. It practically is.).
The result space of both functions is 2^128, because both produce 128 bit digests.
So we have an almost unlimited domain and a limited codomain. So there is (almost) unlimited number of collisions for each target for both functions.
Hence you can't distinguish which function was used, because you are not even able to tell, what was the source for one of these functions...
There may be one exception: if some of the hash functions does not generate some part of its codomain (but this would almost certainly also mean that the function is terribly broken). Then you would be able to say for sure that it can't be that function....