A species tree is a weighted tree-graph that represents the order and the magnitude of separation between a set of species. Species trees have significant applications in biology, microbial epidemiology and tumor phylogenetics. Statistical estimation of species trees is an integral part of the field.
Although various likelihood and Bayesian estimators of species trees are available, none of these methods are fast enough to estimate very large trees under a computationally expensive, but commonly used statistical model: the coalescent. Estimation under the coalescent model is especially computationally expensive when one needs to include multiple organisms per species.
Here I will present an approach of fast likelihood estimation of species trees, exploiting a certain special structure of the tree space. Using this approach, one will be able to estimate larger trees than previously possible in a reasonable time.