Complete Mitochondrial Genomes and Eutherian Evolution

Recent large-scale nuclear DNA phylogenies have supported unconventional interordinal relationships among modern eutherians as well as divergence dates (100 mya) that substantially predate the first appearance of fossils from modern eutherians near the Cretaceous/Cenozoic (K/T) boundary (65-70 mya). For comparison to the nuclear data, I analyzed 12 complete mitochondrial DNA (mtDNA) protein-coding genes (10,677 bp) from 53 eutherian taxa, using maximum-likelihood methods to estimate model parameters (GTR + I + ) and to optimize topology and branch-length estimates. Although closely resembling the nuclear DNA trees, the mtDNA maximum-likelihood tree is just one of seven statistically indistinguishable ( lnL 1.747) trees, each suggesting different evolutionary relationships. This 53-taxon data set and another including 56 taxa provide no statistically significant support for a monophyletic afrotherian clade. In fact, these mitochondrial DNA sequences fail to support the monophyly of three putative eutherian divisions suggested by the nuclear data (Afrotheria, Laurasiatheria or Euarchontoglires). By comparison to well-supported branches describing relationships among families, those describing interordinal relationships are extremely short and only tenuously supported. Neither these sequences, nor sequences simulated under a known tree, fully resolve any interordinal relationship. Even simulated sequences that are twice as long (22kb) as mtDNA protein-coding genes are too short and too saturated to resolve the deepest and shortest interordinal relationships. Further, the mammalian mtDNA sequences appear to depart significantly from molecular-clock and quartet dating assumptions. Unlike recent nuclear DNA studies, I find that mtDNA genes, by themselves, are inadequate to describe relationships or divergence times at the base of the eutherian tree.