Base compositional bias and phylogenetic analyses: A test of the "flying DNA" hypothesis

Phylogenetic methods can produce biased estimates of phylogeny when base co mposition varies along different lineages. Pettigrew (1994, Curr. Biol. 4:2 77-280) has suggested that base composition bias is responsible for the app arent support for the monophyly of bats (Chiroptera: megabats and microbats ) from several different nuclear and mitochondrial genes. Pettigrew's "flyi ng DNA" hypothesis makes several predictions: (1) that metabolic constraint s associated with flying result in elevated levels of adenine and thymine t hroughout the genome of both megabats and microbats, (2) that the resulting base compositional bias in bats is sufficient to mislead phylogenetic meth ods and account for the support for bat monophyly from several nuclear and mitochondrial genes, and (3) that phylogenetic analysis using pairwise dist ances corrected for compositional bias should eliminate the support for bat monophyly. We tested these predictions by analyzing DNA sequences from two nuclear and three mitochondrial genes. The predicted base compositional bi as does not appear to exist in some of the genes, and in other genes the di fferences in AT content are very small. Analyses under a wide diversity of criteria and models of evolution, including analyses that take base composi tion into account (using log-determinant distances), all strongly support b at monophyly. Moreover, simulation analyses indicate that even extreme bias toward AT-base composition in bats would be insufficient to explain the ob served levels of support for bat monophyly. These analyses provide no suppo rt for the "flying DNA" hypothesis, whereas the monophyly of bats appears t o be well supported by the DNA sequence data.