Abstract
Plethodontids, a monophyletic group, comprise about 60% of living salamanders. Many have a tongue that may be projected rapidly and for a distance up to half the body length. Previous studies have focused on the functional morphology of the projectile tongue and the comparative morphology of the feeding system in the family. Here we present a working hypothesis for the phylogeny of plethodontids and a scenario for the evolution of tongue feeding based on the phylogenetic hypothesis. Primitive and derived states are identified and transformation series proposed for 30 characters. Four alternative phylogenetic hypotheses are presented. Our choice of a preferred phylogenetic hypothesis is based on consideration of conflicting characters and tongue projection mechanics. The preferred hypothesis is not the most parsimonius, in a strict technical sense, because several homoplasious characters appear (on functional grounds) to be closely associated with attainment of tongue projectility, which itself has evolved several times. In our hypothesis the bolitoglossines and plethodontines are sister taxa. They in turn are the sister taxon of the hemidactylines, and this grouping is the sister taxon of the desmognathines. Within the bolitoglossines, Batrachoseps is the sister taxon of the Neotropical genera (supergenus Bolitoglossa). In our scenario for the evolution of tongue feeding, the ancestral plethodontid had an attached protrusible tongue. From this condition attached projectile tongues have arisen at least three times within the family: in the ancestors of the hemidactylines; in the ancestors of Ensatina; and in the ancestors of the bolitoglossines. In turn, the attached projectile condition has given rise to free projectile tongues at least once within the hemidactylines and twice within the bolitoglossines. This scenario tends to maximize the number of times attached projectile and free projectile tongues have evolved in the family.