Mitochondrial portraits of human populations using median networks

Analysis of variation in the hypervariable region of mitochondrial DNA (mtDNA) has emerged as an
important tool for studying human evolution and migration. However, attempts to reconstruct optimal
intraspecific mtDNA phylogenies frequently fail because parallel mutation events partly obscure the true
evolutionary pathways. This makes it inadvisable to present a single phylogenetic tree at the expense of
neglecting equally acceptable ones. As an alternative, we propose a novel network approach for portraying
mtDNA relationships. For small sample sizes (< -50), an unmodified median network contains
all most parsimonious trees, displays graphically the full information content of the sequence data, and
can easily be generated by hand. For larger sample sizes, we reduce the complexity of the network by
identifjmg parallelisms. This reduction procedure is guided by a compatibility argument and an additional
source of phylogenetic information: the frequencies of the mitochondrial haplotypes. As a spinoff,
our approach can also assist in identifjmg sequencing errors, which manifest themselves in implausible
network substructures. We illustrate the advantages of our approach with several examples from
existing data sets.