These are excerpts from the Wiki mitochondria page. Non-mechanistic as usual: “somehow survived endocytosis and became incorporated into the cytoplasm”. Possible mechanistic scenarios would not show any fitness advantages. I think the evidence is rather weak, disguised in vague language. Also again the funny thing about the Tree of Life. Ribosomal RNA shows that they are early diverging, in line with their appearance, but when the evidence doesn’t suit the theory, one starts to fiddle with the many parameters used in those studies and suddenly they are highly derived.
As mitochondria contain ribosomes and DNA, and are only formed by the division of other mitochondria, it is generally accepted that they were originally derived from endosymbiotic prokaryotes. Studies of mitochondrial DNA, which is often circular and employs a variant genetic code, show their ancestor, the so-called proto-mitochondrion, was a member of the Proteobacteria. In particular, the pre-mitochondrion was probably related to the rickettsias, although the exact position of the ancestor of mitochondria among the alpha-proteobacteria remains controversial. The endosymbiotic hypothesis suggests that mitochondria descended from specialized bacteria (probably purple non-sulfur bacteria) that somehow survived endocytosis by another species of prokaryote or some other cell type, and became incorporated into the cytoplasm. The ability of symbiont bacteria to conduct cellular respiration in host cells that had relied on glycolysis and fermentation would have provided a considerable evolutionary advantage. Similarly, host cells with symbiotic bacteria capable of photosynthesis would also have an advantage. In both cases, the number of environments in which the cells could survive would have been greatly expanded.
This relationship developed at least 2 billion years ago and mitochondria still show some signs of their ancient origin. Mitochondrial ribosomes in mammals are the 70S (bacterial) type, in contrast to the 80S ribosomes found elsewhere in the cell.. One mitochondrion can contain 2-10 copies of its DNA. As in prokaryotes, there is a very high proportion of coding DNA, and an absence of repeats. Mitochondrial genes are transcribed as multigenic transcripts which are cleaved and polyadenylated to yield mature mRNAs. Unlike their nuclear cousins, mitochondrial genes are small, and many chromosomes are circular, conforming to the bacterial pattern. In humans, mitochondrial genes lack introns, yet other Eukaryotic mitochondrial DNA has 1-37 of them. Further, there are codon differences in mitochondria: in the mitochondria, the UGA codon specifies tryptophan; AGA and AGG are stop codons; and AUA, AUC, and AUU are each allowable start codons.
A few groups of unicellular eukaryotes lack mitochondria: the microsporidians, metamonads, and archamoebae. These groups appear as the most primitive eukaryotes on phylogenetic trees constructed using rRNA information, suggesting that they appeared before the origin of mitochondria. However, this is now known to be an artifact of long branch attraction — they are apparently derived groups and retain genes or organelles derived from mitochondria (e.g. mitosomes and hydrogenosomes). There are no primitively amitochondriate eukaryotes, and so the origin of mitochondria may have played a critical part in the development of eukaryotic cells.