Compatibility of host and symbiont data

The question of whether two or more partitions of the dataset come from the same phylog-eny is very common in phylogenetics Incongruence length difference (ILD) (Farris et al , 1995a; 1995b) and its variants are the most frequent tests used to address such problems Although in most cases these partitions are represented by sets of different genes collected from the same taxa, this method can be easily adopted to matrices combined from two different organisms Due to this property, the ILD test can be applied in coevolutionary studies (Johnson et al ., 2001; Lopez-Vaamonde et al ., 2001; Hughes et al ., 2007) .

For insect symbiont associations, the ILD test was used in a few studies and allowed for both the rejection and corroboration of a null hypothesis Results from analysis on psyl-lids and Carsonella (Thao et al , 2000b) as well as Cicadellidae vs Sulcia (Takiya et al , 2006), showed that despite topological incongruences, the null hypothesis of strict cospeciation could not be rejected when the data matrices were taken as the base A different case was reported by Downie and Gullan (2005); for a 75% topological correspondence between mealybugs and their symbiont, the overall congruence of the datasets was rejected by ILD Considering the similarities of all of these insect-symbiont associations and the different implications of the ILD tests, knowledge of the latter's properties and reliability becomes particularly important There have been many debates in the phylogenetic literature on what kind of information and sources of error should be expected from this test (Yoder et al ., 2001; Hipp et al ., 2004; Ramirez, 2006; De Vienne et al ., 2007; Quicke et al ., 2007) . With respect to the insect-symbiont studies, the most disquieting weakness of ILD is its sensitivity to unequally distributed homoplasies among data partitions (Dolphin et al , 2000) In a coevolutionary framework, such difference in noise content between the two partitions is very likely Most recently, a new modification of the test, arcsine ILD, has been suggested to suppress this tendency (Quicke et al , 2007) At the time of this chapter preparation, no study on insect-symbiont association had applied this new approach, and its power remains to be tested

Another problem with the ILD approach is that while assessing overall congruency, it is not able to identify taxa responsible for the incongruences A simple remedy for this shortcoming was proposed by Johnson et al. (2001). Their approach consists of removing taxa (e g , host-parasite pairs) from the dataset until significant congruence between the partitions is achieved This restricted set of taxa is then used to build a phylogenetic backbone common for both counterparts Because the removed taxa are supposed to be incongruent, their position is inferred by a subsequent phylogenetic analysis constrained with the backbone topology, and the resulting trees are finally compared by a tree-based method The efficiency of this technique to localize the source of incongruence has been recently demonstrated on a host-parasite system (Hughes et al , 2007) Due to its capacity to identify a maximal set of entirely congruent taxa, this approach might be particularly valuable for investigating insect-symbiont associations To the best of our knowledge, it has not yet been applied to any of the known insect symbionts

A different approach in the identification of incongruent taxa is offered by the method of partitioned Bremer support (PBS) (Baker and DeSalle, 1997) . Dividing the overall Bremer support of each node to individual partitions, this method can highlight the taxa minimally contributing or even contradicting the particular node In insect-bacteria coevolu-tion, the partitioned Bremer has recently been used to identify possible incongruences between mealybugs and their insect hosts (Downie and Gullan, 2005) As agreed by many authors, none of the incongruence testing methods can provide definitive proof for tree congruence/incongruence and should be combined with other sources of evidence .

0 0

Post a comment