Studies on the phylogenetic position of the Ctenodactylidae (Rodentia)

Abstract

More than half of all mammalian species belong to the order of the rodents. Although many phylogenetic relationships among rodents have been established beyond doubt, others are not yet established. The phylogenetic position of the Ctenodactylidae (gundis) was discussed in several contributions to a symposium in Paris in 1984 (Luckett and Hartenberger 1985). This is a small rodent family with only four species living in the border regions of the Sahara Desert in Africa. This taxon has been grouped either with the hystricognathous rodents or as a separate branch, as a very early o&hoot from the rodent stem: Luckett ( 1985), Bugge (1985), and George ( 1985) concluded, from different sets of data, that the Ctenodactylidae share with the hystricognathous rodents a number of derived traits and that the Ctenodactylidae should be considered as an early offshoot of the latter. Wood ( 1985) discussed the phylogenetic position of several sciurognathous hystricomorph families, including the Ctenodactylidae. These families are also restricted to Africa and have often been associated with the hystricognathous rodents. However, in Wood’s opinion there are no indications for these relationships, the similarities being the result of conservation of ancestral features or of convergence. Hartenberger ( 1985) concluded from the fossil evidence that the Ctenodactylidae are one of the oldest recognized rodent families, diverging from all other rodent taxa in the Lower Eocene. With additional molecular data, it may be possible to shed more light on this controversy and to learn more about the origin of the rodents. No molecular evidence is yet available on the phylogenetic position of the Ctenodactylidae. Therefore, several tissues and blood of gundis were collected, and a number of proteins were isolated and investigated. Gundi myoglobin was isolated, and its amino acid sequence was determined (Beintema et al. 1990). Only four other rodent myoglobin sequences have been determined so far: mouse (A4z.o musculus) (Harris et al. 1985), casiragua (Proechimys guairue) (Harris et al. 1985 ), viscacha (Lugostomus muximus) (Gumett et al. 1984)) and mole rate (Spulux ehrenbergi) (Gumett et al. 1984). The gundi amino acid myoglobin sequence was compared with the 73 other known vertebrate myoglobin amino acid sequences by using computer procedures, described by Czelusniak et al. ( 1990), that search for the tree or set of trees with shortest nucleotide substitution length. The shortest trees for the 74 myoglobin sequences found in this search failed to represent rodents, primates, and artiodactyls as monophyletic orders. In these shortest trees, the gundi joins the hystricognath (casiragua/viscacha) branch, which is rather widely separated from the mouse/mole rat branch. If the four other rodents are forced together, which requires four additional nucleotide substitutions, the branch to the gundi still joins the two hystricognaths, or, for the same score, the gundi branches off before the divergence of the other four rodent sequences. The amino acid sequences of the a and p hemoglobin chains of gundi were