Toward Extracting All Phylogenetic Information from Matrices of Evolutionary Distances

Abstract

Theoretically Speedy Trees Phylogenetic tree construction often uses distance matrix methods, which offer promising scalable methods for rapid building of very large phylogenies. It has generally been assumed that trees constructed from distance matrices are intrinsically less accurate than those inferred from maximum likelihood methods, but the latter are slower to make. Roch (p. 1376 ; see the Perspective by Allman and Rhodes ) theoretically examined this problem and showed that distance methods can achieve the same level of accuracy as likelihood methods. Thus, the generally assumed tradeoff between accuracy and speed in phylogenetic reconstruction can be overcome.