Even human-chimp-gorilla hard to resolve without statistical methods. One approach either concatenation or supertree– doesn’t always work. Incomplete lineage sorting. She is working on “One thousand transcriptome project” and “Avian phylogenomics project”– lots of ILS.
Under the multiple-species coalescent, the species tree is defined by a probability distribution on the gene trees.
Standard methods: Consistent species tree methods: MP-EST, BUCKy-pop, star-BEAST. Inconsistent: MDC and greedy consensus. Consistency of concatenation is not known.
MP-EST paper: http://www.pnas.org/content/109/37/14942.short
Springer and Gatesy– are not happy about treating gene trees as observed.
Bin-and-conquer:
How to combine genes into bins? First effort: random binning. Today: statistical binning.
Input: estimated gene trees with bootstrap support, and choose a minimum support threshold t
Output: partitioning of the estimated gene trees into sets:
Make an incompatibility graph and find a coloring of the nodes such that two connected nodes don’t have the same clustering.
This is vertex coloring problem– hard.
They have a heuristic approach.
Simulations: binning decreases the gene tree error by 50%. Application to Avian species tree. Seems useful. Error in species tree branch lengths is a problem: if you overestimate branch lengths, then you overestimate ILS.