This article summarizes recent research in which it was claimed that mitochondrial import proteins evolved separately in different lineages and therefore present an elegant example of convergent evolution. Intersting results though from Perry et al.
Abstract. Complex machinery has evolved to recognise and import nuclear-encoded proteins into mitochondria. Recent work now shows that the plant Tom20 mitochondrial protein import receptor has a similar tertiary structure to animal Tom20, although the proteins are evolutionarily distinct, representing an elegant example of convergent evolution.
First, the endosymbiotic theory is not proven, but more importantly, the transfer to the nucleus and the necessity of an import pathway represent an irreducibly complex system. They are interdependent and you can’t have a transfer to a nucleus, after which the cell evolved an import pathway. Cavalier-Smith understood this and proposed that an export system was converted, but this seems unlikely.
Mitochondria are generally accepted to have descended from a eubacterium that was engulfed by an archaebacterial host cell [1,2]. During the evolution of this endosymbiotic relationship, the vast majority of organellar genes were transferred to the nucleus, necessitating an efficient system to import nuclear-encoded mitochondrial proteins into the organelle . All extant mitochondria possess this protein import machinery, consisting of the translocase of the outer mitochondrial membrane (TOM).
It seems that the core complex has a shared ancestry, whereas some additional proteins can be the result of convergent evolution.
In contrast to the core TOM complex, the receptor proteins Tom20 and Tom70 are not conserved between highly divergent lineages. Despite high-quality genome sequences, proteins with significant sequence similarity to the animal and fungal Tom20 and Tom70 import receptors could not be identified in a wide range of eukaryotic organisms, including plants, alveolates, trypanosomatids, chromists and red algae [8–10]. However, antibodies raised to a 20 kDa protein that co-purifies with the plant TOM complex were found to inhibit mitochondrial protein import in vitro, suggesting the protein is the plant functional analogue of the animal/fungal Tom20 [11,12]. In a recent issue of Current Biology, Perry et al.  present a detailed investigation of the function and three-dimensional structure of the plant Tom20, concluding that functional and structural similarities between the Tom20 receptors found in plants and animals/fungi probably arose via convergent evolution.
The last paragraph shows again the simplicity of their thinking.
Perry et al.  present a elegant example of convergen evolution on a molecular scale, where different organisms adapted distinct proteins to fulfil a function demanded by a similar cellular environment. With only the core import pore present in the early stages of mitochondrial evolution, great selective pressures would have existed to develop a discriminating receptor protein to increase targeting fidelity and import efficiency.