Estimating Phylogenies from Shape and Similar Multidimensional Data: Why It Is Not Reliable.

摘要: In recent years, there has been controversy whether multidimensional data such as geometric morphometric or information on gene expression can be used for estimating phylogenies. This study uses simulations of evolution in phenotype spaces to address this question and identify specific factors that are important answering it. Most the use phylogenies with four taxa, so just three possible unrooted trees effect different combinations branch lengths studied systematically. a comparison methods, squared-change parsimony performed similarly well maximum likelihood, both methods outperformed Wagner Euclidean parsimony, neighbor-joining UPGMA. Under an evolutionary model isotropic Brownian motion, phylogeny estimated reliably if dimensionality is high, even relatively unfavorable lengths. By contrast, phenotypic integration most variation concentrated one few dimensions, reliability phylogenetic estimates severely reduced. Evolutionary models stabilizing selection also produce highly unreliable estimates, which little better than picking tree at random. To examine how these results apply more we conducted further up eight indicated effects extend convergence among internal nodes may additional complications specifically greater numbers taxa. Overall, suggest data, under plausible biological do not reliable phylogeny. [Brownian motion; data; morphometrics; morphological integration; parsimony; phylogeny; shape; selection.].