Tree confidence

Considering the almost inevitable errors in phylogenetic inference, the "best" tree (e . g., ML tree) is not necessarily the true tree Thus, even if the host and symbiont trees differ from each other, the symbiont tree may not be a significantly worse explanation of the host data, and vice versa In the realm of ML, a set of methods has traditionally been available to test the significance of tree differences with respect to the underlying data The most popular tests are the Kishino-Hasegawa test (KH test) (Kishino and Hasegawa, 1989) and its later modifications Although the likelihood-based tests are extremely popular and were often used, their application is not always straightforward In the analyses of symbiotic associations, the competing topologies are trees derived directly from the data It has been convincingly shown that under such circumstances, the KH test is not applicable (Shimodaira and Hasegawa, 1999; Goldman et al , 2000b) A modified version, the SH test developed by Shimodaira and Hasegawa (1999), overcomes this obstacle, but pays a considerable cost in the form of increased conservativeness (Goldman et al , 2000a)

For insect-symbiont coevolution, this kind of test was used, for example, by Clark et al (2000) to assess the congruence between Buchnera and aphids of the genus Uroleucon. An important outcome of this study is the finding that an ML-based test can reject compatibility of the host and symbiont data even under high topological congruency The author concluded that in the observed aphid-Buchnera system, this might be due to the heterogeneity of evolutionary process in different regions of the aphid mitochondrial DNA Since this report, several other tests were proposed, most of them based on the ML method Shi et al (2005) tested several of these approaches in simulation studies and stressed the good performance of the SOWH test, a method based on parametric bootstrap (Goldman et al , 2000a) There is currently only limited experience on performance of this and other tests in the coevolutionary field To the best of our knowledge, the only application of the SOWH test to insect-symbiont coevolution was done in the study of Downie and Gullan (2005) . Comparing phylogenies of mealybugs and their P-symbionts, the authors were able to show that the incongruence between the two topologies is not caused by sampling error

0 0

Post a comment