Base-compositional biases and the bat problem. II. DNA-hybridization trees based on AT- and GC-enriched tracers.

We conducted a series of parallel DNA-hybridization experiments on a small group of bats (species of Pteropus, Rhinolophus, Noctilio and Pteronotus) and outgroups (Lemur, Cynocephalus, Didelphis), using whole-genome labels and tracers made from extracts enriched with AT and two levels of GC content. FITCH (additive phylogenetic trees) topologies were constructed from the four sets of comparisons, indexed as both delta Tmode and delta NPHs (normalized percentage of hybridization). Based on our previous work showing that the shared AT bias of pteropodids and some microchiropterans may affect the rank-ordering of taxa based on either AT- or GC-rich labels, our expectation was that the resulting trees would show differing topologies when generated from tracers made with the variously enriched DNA extracts. Whereas there was some variation among the trees, most of them grouped the bats together, and almost all paired the representative megachiropteran and rhinolophoid microchiropteran as sister-taxa in contrast to the other microchiropterans. As the pteropodid-rhinolophoid relationship is an unexpected and unlikely one, we attribute this association to an AT bias that was not obviated even by our most GC-rich labels, and suggest that such a bias may compromise the truth of some molecular trees. Accordingly, we believe the broader issue of bat monophyly remains unresolved by DNA-hybridization and probably also by gene-sequencing studies.