Estimation of Species Trees

Abstract

During the last decades, gene trees have been often interpreted as species phylogenies. However, the extensive gene tree discordance found in multi‐locus datasets has put into question this interpretation, and a variety of new methods that explicitly consider species trees have been proposed in recent years. Some of these explicitly consider evolutionary processes that can lead to true gene tree discordance, namely incomplete lineage sorting, gene duplication and loss and horizontal gene transfer. Choosing the most appropriate species tree method for the data at hand is not straightforward due to different data prerequisites, model assumptions, analytical strategies and computational implementations. Key Concepts: We could think of at least three different phylogenetic layers corresponding to species trees, locus trees and gene trees. These depict, respectively, the history of the sampled species, loci and genes copies. Traditional phylogenetic inference has focused on the reconstruction of gene trees, assumed to be accurate proxies for species history. True species, locus and gene trees can be incongruent due to the effect of evolutionary processes like incomplete lineage sorting, gene duplication and loss and horizontal gene transfer. This incongruence might appear larger due to estimation error. Extensive gene tree incongruence unveiled in multi‐locus datasets has encouraged the development of methods that explicitly reconstruct species trees. The supermatrix approach combines loci into a super‐alignment and estimates the corresponding supergene tree. Supertree methods combine gene trees to obtain an estimate of the species tree. Other methods co‐estimate gene and species trees in a single step using full probabilistic models. Different species tree methods require distinct data specifications, mainly related with the consideration of paralogs, number of sampled species, missing taxa and number of loci. Rooting and reconstruction uncertainty must be carefully considered before choosing a species tree method.

SimPhy: Phylogenomic Simulation of Gene, Locus, and Species Trees

“…Indeed, it is well known that gene and species phylogenies can be inconsistent due to evolutionary processes like incomplete lineage sorting (ILS), gene duplication and loss (GDL), and horizontal gene transfer (HGT) ( Goodman et al 1979 ; Pamilo and Nei 1988 ; Takahata 1989 ; Maddison 1997 ; Page and Charleston 1997 ). Not surprisingly, the gene tree/species tree “dilemma” has received a lot of attention in recent years, and consequently a plethora of species tree reconstruction methods have been published ( Chaudhary et al 2010 ; Heled and Drummond 2010 ; Liu et al 2010 ; Boussau et al 2013 ; De Oliveira Martins et al 2014 ; see Mallo et al 2014a for a review; Mirarab and Warnow 2015 ). Although several benchmarks of species tree methods have been carried out ( Leache and Rannala 2011 ; Yang and Warnow 2011 ; Bayzid and Warnow 2013 ; Mirarab et al 2014 ), they have generally focused on single causes of phylogenetic discordance.…”