Phylogeny through brain traits: trees generated by neural characters

Abstract

Phylogenetic trees were computed by the Wagner algorithm from data on up to 15 brain characters scored on 154 specimens of 134 mammalian species. Because the data were not complete on all specimens, only one tree, of 18 taxa, was generated on all 15 features; a tree of 99 species was computed from 10 characters, and trees of 38 species from 10 and 12. The 38-taxon trees were considered best because they preserved most completely the integrity of mammalian orders. All trees consistently separated the subclasses of mammals and suggested that rodents, insectivores, and the tree shrew were most derived on the basis of brain characters. The trees' shapes are sensitive to small alterations in character scorings, largely because of the relatively few characters available and small differences in the number of states among them.