On the Phylogeny of the Glossophaginae and the Proper Use of Outgroup Analysis

Abstract

In two recent papers on the phylogeny of phyllostomid bats, Hood and Smith (1982) and Haiduk and Baker (1982) commented unfavorably on my recent cladistic analysis of the New World nectar-feeding bats (Griffiths, 1982). Both pairs of authors suggested that (1) my outgroup comparisons were incorrect, and (2) my character analysis was incorrectly performed. Haiduk and Baker (1982) went so far as to redo the character analysis and claimed that my hyoid/tongue morphology data better supported the branching sequence suggested by them than the one I presented (Griffiths, 1982). Despite the alleged errors in my analysis, both Hood and Smith (1982) and Haiduk and Baker (1982) admitted that at least parts of my conclusions seemed to be correct. Neither paper mentioned that for the first time in the 15 years the nectar-feeding bat problem has been intensively examined, two phylogenies-produced independently and based on unrelated data sets-are remarkably congruent (Baker et al. [1981] and Haiduk and Baker [1982] using biochemical and karyotypic data; Griffiths [1982] using hyoid/ tongue morphology data). Except for very minor differences in the exact relationship of Hylonycteris and Choeroniscus and a minor disagreement on the relationships of the three brachyphylline genera, the only difference in the two cladograms produced is in the treatment of the genera Lonchophylla and Lionycteris. Haiduk and Baker (1982) indicated that the two genera are relatively derived members of the subfamily Glossophaginae. On the other hand, I suggested (Griffiths, 1982) that the two genera evolved nectivory independently of the true glossophagines. I indicated further that no synapomorphous character state was known that united Lonchophylla and Lionycteris with the true glossophagines and, until such a character state was found, Lonchophylla and Lionycteris (along with Platalina which Haiduk and Baker did not karyotype) might best be placed in a separate subfamily, the Lonchophyllinae. In this note, I respond to the allegations that my outgroup comparisons were incorrect. In addition, I show that Haiduk and Baker (1982) are quite incorrect in their assertion that my hyoid/tongue data support the placement of Lonchophylla and Lionycteris as relatively derived glossophagine bat genera. Outgroup comparisons are made to determine correct character polarity within the group one is studying and, most especially, to determine which character states are synapomorphies, which are autapomorphies, and which are symplesiomorphies. When studying the relationships of the genera within the taxonomic group A, it is obviously important to study also outgroup representatives from all closely related sister groups (B and C), as well as representatives from more primitive groups (D) that are thought to be derived from the direct ancestor of A, B, and C. Without so doing, the systematist fails to get an accurate idea of the polarity of characters and of the uniqueness of the synapomorphies in the group (A) in question. Unitl recently, in a study of the relationships within the taxon Glossophaginae, there was no way of determining which other groups were closely-related sister groups (B and C) and which were modern representatives of ancestral groups (D). This was because no study had been done that showed the relationships of the Glossophaginae to other bat taxa. In the absence -of a definitive study, Haiduk and Baker (1982) are entitled to their opinion of the proper outgroups to use, as am I. Such a definitive study (Hood and Smith, 1982) has now been done,