There are three approaches to tracing technological evolution. The first one is to trace the evolution of implementation of a function. The second one is to trace the evolution of a system's modifications. The third one is to trace the evolution of applications of a system/function.
TRIZ employs the first approach only. The thing is that tracing evolution means its reconstruction. And reconstruction of all modifications that a technical system went through is THEORETICALLY impossible.
Here lies the difference with biological evolution. In biology it is assumed that THEORETICALLY it is possible to trace all modifications of species, that THEORETICALLY it is possible to reconstruct the complete evolutionary tree. Such a belief is based on the assumption that all modifications were preserved in fossils. (Even if not all then at least there are no big gaps.) Sooner or later the fossils containing the missing chains will be found and the tree will be completed.
Such an assumption is not valid in technology. The patent archives contain a tiny fraction of all modifications that systems went through. Most modifications took place in the brains of inventors as the intermediate steps toward their inventions. They usually do not put these intermediate steps/modifications in writing and no trace is usually left of them.
Thus, in case when intermediate modifications are not small, it is impossible to reconstruct a meaningful evolutionary tree of a technical system in the spirit of evolutionary trees of species. It is even impossible to identify two systems as evolutionary related.
Indeed, suppose that inventor started with system A1 and mentally modified it into A2, which he subsequently modified into A3, etc. Eventually he arrived at modification An which turned out to be a solution to the problem he worked on. He made An public by patenting and/or building it. He never disclosed his starting point A1 and the intermediate steps A2,..., An-1. How would an external observer determine that An and A1 are evolutionary related if the intermediate modifications A2, A3, etc. are unknown to him ? It is only possible if An is still close to A1. (As was the case with cornets for which Niles Eldredge managed to build a seemingly almost complete evolutionary tree.) But if the sum of the slight modifications A2 to An-1 is not slight, then it is impossible to recognize similarity of A1 and An and suspect that they are evolutionary related.
That is why studying technological evolution by building evolutionary trees did not take roots in TRIZ (and in the history of technology on the whole). Historians restrict themselves to studying either evolution of implementation of a function or evolution of application of a function. So far the latter has not been shown to be useful for TRIZ purposes, however. This explains why TRIZ restricts itself to studying evolution of implementation a function only.