Karyotypic megaevolution and phylogenetic analysis: new world nectar-feeding bats revisited

The phylogenetic relationships among the genera of the New World nectar-feeding bats, subfamilies Glossophaginae sensu lato and Brachyphyllinae of the Phyllostomidae (see Handley [1980] for justification of this family name), have been disputed for over 15 years (see review in Griffiths, 1982). Most recently, Haiduk and Baker (1982) published a hypothesis of "glossophagine" relationships based on a cladistic analysis of chromosomal G-band patterns. Their paper is clearly a reply to Griffiths' (1982) hypothesis based on an analysis of hyoid and lingual morphology of these nectar-feeding bats. The use of multiple data sets provides a powerful tool for developing robust phylogenetic hypotheses, but it also complicates the process as discordances often arise (Straney, 1980; Arnold et al., 1982). Resolution of discordances is a necessary adjunct of the process of hypothesis building. While Haiduk and Baker (1982) claimed a reanalysis of Griffiths' data indicates that the most "parsimonious" cladogram consistent with the morphological information is identical to their cladogram based on chromosomal data, this is not necessarily true, as I will document below. Additionally, I amplify the underlying assumptions of Haiduk and Bakers' analysis and discuss some of the alternative interpretations of the several sets of data now available which bear on the issue of "glossophagine" relationships. The basic premise of phylogenetic reconstructions based on chromosomal preparations is that homologous elements can be identified among karyotypes and, thus, synapomorphic rearrangements can be discerned; homology, in this case, is assessed by visual appraisal of similarity in banding pattern (including number), relative staining intensity of bands, and relative spacing between bands. This further assumes that chromosomal rearrangements are generally unique events (i.e., homoplasy is rare; I will presume this to be true although it is not necessarily so). An additional assumption relied on by Haiduk and Baker (1982) is that the primitive karyotype for the Phyllostomidae is like that of Macrotus. Although I have doubts about identifying a karyotype as plesiomorphic for the whole family, since outgroup comparison cannot be performed (see Baker and Bickham, 1980), one can tentatively allow this assumption. I am not convinced, however, that they, or anyone else, can unambiguously identify a sufficient number of chromosomal homologies in the karyotypes of "glossophagine"' bats to support the cladogram they present. For instance, the major difference between Haiduk and Bakers' (1982) and Griffiths' (1982) cladograms are the positions of Lonchophylla, Lionycteris, and Platalina (compare Figs. la and lb). There are only two chromosomal synapomorphies that would place these three genera within the glossophagine (sensu stricto) clade, a 13/2 fusion and the presence of arm E. Unfortunately, a karyotype indicating the numbering system for chromosomal arms they used was not presented in Haiduk and Baker (1982), nor was it referenced; this information is in Baker (1979) and Baker and Bass (1979). A 13/2 fusion would require a minimum of three rearrangements from the supposed primitive karyotype of the group, that of Glossophaga (Baker and Bass, 1979). These rearrangements are a 13/28 fission, a 2/18 fission, and then a 13/2 fusion (alternatively, a 13/2 chromosome could result from a reciprocal translocation). Based upon the chromosomal spreads presented in Baker and Bass (1979) and Haiduk and Baker (1982), I cannot see a clear basis for distinguishing between a "13/2" chromosome, a " 13/18" chromosome, or a "2/28"