Phylogenetic relationships displayed by gene trees on the same taxon
set often disagree with one another, and may not accurately show
evolutionary relationships at the species level. The multi-species
coalescent model, however, models one mechanism for the production of
such incongruent gene trees -- incomplete lineage sorting.
Since distances on gene trees are affected by mutation rates, which
may vary over a tree or across genes, in some circumstances it is
desirable to consider the relationship this model predicts between
species trees and gene tree topologies only, as studied by Degnan and
Salter (2005).
In this talk, we show that the rooted species tree topology is
identifiable from either the unrooted gene tree distribution or the
clade distribution. Branch lengths for the species tree are also
identifiable from unrooted gene trees. Thus statistically consistent
species tree inference is, in theory, possible using only such
information.
This work is joint work with James Degnan of the University of
Canterbury and John Rhodes of the University of Alaska Fairbanks.