Base-compositional biases and the bat problem. III. The question of microchiropteran monophyly

Using single-copy DNA hybridization, we carried out a whole genome study of 16 bats (from ten families) and ¢ve outgroups (two primates and one each dermopteran, scandentian, and marsupial). Three of the bat species represented as many families of Rhinolophoidea, and these always associated with the two representatives of Pteropodidae. All other microchiropterans, however, formed a monophyletic unit displaying interrelationships largely in accord with current opinion. Thus noctilionoids comprised one clade, while vespertilionids, emballonurids, and molossids comprised three others, successively more closely related in that sequence. The unexpected position of rhinolophoids may be due either to the high AT bias they share with pteropodids, or it may be phylogenetically authentic. Reanalysis of the data with varying combinations of the ¢ve outgroups does not indicate a rooting problem, and the inclusion of many bat lineages divided at varying levels similarly discounts long branch attraction as an explanation for the pteropodid^rhinolophoid association. If rhinolophoids are indeed specially related to pteropodids, many synapomorphies of Microchiroptera are called into question, not least the unitary evolution of echolocation (although this feature may simply have been lost in pteropodids). Further, a rhinolophoid^pteropodid relationshipöif trueöhas serious implications for the classi¢cation of bats. Finally, among the outgroups, an apparent sister-group relation of Dermoptera and Primates suggests that £ying lemurs do not represent the ancestors of some or all bats; yet, insofar as gliding of the type implemented in dermopterans is an appropriate model for the evolution of powered mammalian £ying, the position of Cynocephalus in our tree indirectly strengthens the argument that true £ight could have evolved more than once among bats