Reader Comments

Post a new comment on this article

Reply to Steve Rossiter

Posted by stjpetti on 09 Mar 2008 at 11:38 GMT

Reply to Steve Rossiter:

Significant conflict between DNA-based and amino acid-based phylogenies from the same taxa and genes present a dilemma. Should one side with the DNA tree, as Steve Rossiter and colleagues have now done using FoxP2 on two occasions (constrained tree in original paper and “gene tree” in Commentary)? Or should one pay more attention to the amino acid tree?

There are problems associated with the DNA tree. In this case it supports paraphyly of microbats, an entrenched position that might be seen to be receiving biased support if the amino acid alternative is rejected summarily. Moreover there is evidence, from two different techniques, that DNA leads to anomalously close distances in bats, which lead to a different when appropriately corrected. There is no way to detect anomalous similarity between two sequences from the sequences alone, so recourse must be made to outside information. Kitazoe and colleagues use a vector technique in n-dimensional space that allows simultaneous treatment of many taxa, to detect and measure anomalous similarity (convergence) between DNA sequences. They found that the phyllostomid microbat, Tonatia, moved dramatically, across many nodes in the tree, out of the Laurasiatheria superorder and into the Afrotheria, when a correction is applied. In a formally similar method that takes taxa four at a time and can be solved using 2-dimensional geometry, I have shown that rhinolophoid-megabat distances are seriously underestimated, by an average factor of 2 in 15 genes. Such a large error, when corrected, moves the megabats completely from their association with microbats in the DNA tree.

In contrast to DNA trees, amino acid trees are known to be less affected by such convergent errors in DNA. There do not seem to any cases where the amino acid tree has a greater degree of anomaly than the DNA tree. A number of mechanisms have been suggested to explain how two DNA molecules could converge, such as isochore formation. Moreover, there are well-documented examples where the anomalous phylogeny created by DNA can be corrected by using amino acid sequence, such as the eucaryote Dictyostelium, which DNA says is a prokaryote. In the case of bats, controversy about phylogeny may not be resolved soon, so it is not possible to test whether the DNA or amino acid sequence produce a more correct phylogeny. This is all the more reason not to discount the amino acid tree.

In the case of FoxP2, the unique amino acid substitutions provide a remarkably consistent phylogeny of microbats that does not include the megabats, which share none of the many substitutions found in rhinolophoids. The clear familial microbat lines that are identified by these substitutions and the consistent timing information about the divergence of theses lineages are both corrupted by the DNA tree, which might explain why they make no sense to Steve Rossiter and colleagues.

Fig. FoxP2 Substitutions: Amino acid tree

RE: Reply to Steve Rossiter

PLOS_ONE_Group replied to stjpetti on 17 Mar 2008 at 11:52 GMT

The figure cited by Professor Pettigrew in the above comment (FoxP2 Substitutions: Amino acid tree) can be viewed and downloaded at: