Abstract
The cladogram of Trichodectidae produced by Lyal (1985) is interpreted, with reservations, as a phylogeny and is compared to such host phylogenies as are available. While the predominant pattern of louse relationships is broadly congruent with host relationships, indicating either phyletic tracking or a mixture of phyletic tracking and host-group-limited resource tracking by the lice, a substantial proportion (20·7%) of speciation events of the lice could only be explained by invoking secondary infestation of distantly related hosts by the lice. The ancestor of the Trichodectidae almost certainly parasitized a eutherian, but the identity of the eutherian group is unclear. The primary host association of the Neotrichodectinae was with an American rodent, from which resource tracking onto Mephitinae and then Mustelinae, Procyonidae, Bradypodidae and Melinae took place. The Trichodectinae had an ancestral association with the ancestor of the Feloidea, with subsequent secondary infestations of Mustelidae, Canidae, Ursidae and Procyonidae. The parasitism of the feloid ancestor was probably derived by secondary infestation from a rodent or hyracoid. The ancestral host of the Dasyonyginae may have been a hyracoid, and two secondary infestations of primates are recorded in the subfamily. The erethizontid hosts of the Eutrichophilinae gained their lice from an unknown host. The ancestral host of the Bovicolinae is not known. Future work to resolve these problems must involve: (1) Further collection of lice, with particular emphasis on Artiodactyla, Perissodactyla and Carnivora. (2) Further investigation of the cladistic relationships of the Trichodectidae, particularly at the primary furcations of the cladogram proposed by Lyal (1985), where apomorphies have been very difficult to discern, and in the Bovicolinae. (3) Production of a more fully resolved cladogram of the Eutheria, and the identification of holophyletic mammalian groups at all levels.