Tuesday, August 29, 2006

Some comments on the ant phylogenetics symposium held at Washington D.C. Part II.

Ant phylogenetics: New molecular trees to address old problems in ant biology. XV Congress IUSSI. August 1, 2006. Washington D.C.

Part I dealt only with the first four talks of the symposium; those happened to be the talks addressing phylogenetic problems at the highest hierarchical levels within Formicidae. The rest concentrated mostly on single genera or less inclusive clades with an improvement on detail. As such, they also highlighted the reasons behind phylogenetic reconstruction: improve taxonomy (Schöning et al.; LaPolla and Schultz; Wild); test predictions about the evolution of behavior (Savolainen and Vepsäläinen; Peeters); and introduce a historical component to ecological questions (Crozier et al.; Schultz and Brady; Solomon and Mueller). Rather than reviewing each of the remaining talks here I want to comment on two of them.

Alex Wild (Tucson, Arizona) presented part of his work on Linepithema. The revision of this genus is important not only because the taxonomy of the included species was outdated, but also because the Argentine ant, Linepithema humile (Mayr), is a predominant invasive ant worldwide. What is worth noting is Wild’s integrative approach to species delimitation. He draws data from morphology, mitochondrial and nuclear molecular markers and combines the information looking for agreement. The project is a painful reminder of the current state in ant species-level taxonomy; while female workers are the most commonly collected and therefore used cast in classification studies, adult males seem to display the greatest morphological diversity useful at the species level. Incidentally, Wild also showed that pure molecular approaches may be insufficient for the task, as with the use of a short COI barcode where calibrating an adequate threshold to reflect species circumscribed by an integrative approach proved daunting. Wild states in print that he explicitly follows E. Mayr’s biological species concept (Phil Ward’s fault I suppose), however it seems to me that by searching for congruence of the different lines of evidence he is really reconstructing species as historical units of the sort advocated by the phylogenetic species concept, a result that is conveniently more in line with phylogenetic reconstruction in general.

Riitta Savolainen (Helsinki) talk presented some preliminary results on a phylogenetic study of the host-parasite association between Formica and Polyergus. Previous phylogenetic work using both morphology and molecular data support a close or sister relationship between these two genera. My understanding is that the possibility exists that Polyergus may be a derived subgroup within the large Formica genus. It was puzzling to see, therefore, that a phylogeny for each genus was reconstructed independently, even though the molecular markers used were the same. Even if both genera are monophyletic, the best reconstruction within each genus will be achieved by pooling together all the species into a single matrix and rooting the result at the branch between them or using the more distant outgroups originally included.

In comparison with the similar symposium held at the previous IUSSI meeting in Sapporo (2002), the number, scope and quality of the talks presented at Washington D.C. was far superior and reflects the long overdue incorporation of cladistic methods at all levels of ant taxonomy. We have to congratulate Sean Brady and Riitta Savolainen for organizing such a stimulating symposium.

Monday, August 21, 2006

Some comments on the ant phylogenetics symposium held at Washington D.C. Part I.

Ant phylogenetics: New molecular trees to address old problems in ant biology. XV Congress IUSSI. August 1, 2006. Washington D.C.

The general mood of the symposium was good. In part this had to do with the fact that the venue was adequate for the event; a large and well-lit conference room at its full capacity. Phil Ward (Davis, California) opened the symposium talking about their Ant-AToL project clearly feeling grand about the full attendance and with the advantage of presenting before Moreau et al.’s talk, yet with characteristic lusterless. Their result so far is the same as the one recently published by Moreau et al.’s in Science (and it better be, since they have practically the same taxon sampling and genes, except using EF1-alpha instead of CO-I), but, unlike the Science report, they voiced the concern that the ant tree may be rooting in the wrong place; an analysis including just the ant sequences results in a topology that cannot be rooted to achieve the same results as with outgroups included. Basically there cannot be a Poneroid clade independent of Leptanillinae (see Moreau et al. 2006). They proceeded to constrain the root at different places within the ingroup comparing the parsimony/likelihood/probability scores (Lumberg-rooting style) to argue that in any case placing the root inside the large Formicoid cluster produces highly suboptimal results. I fail to see the point of this exercise since the global optimal solution for these data (including the ingroup and outgroup sequences simultaneously) already showed that the root doesn’t end up inside that clade. At the end recovering a highly supported Formicoid clade seems to be what most excites the AtoL group.


Corrie Moreau (Harvard) followed Ward's talk. She gave a good and fluid talk and stood pleased even though it was a repetition of results from the previous talk (yet already published by her group), but including the part of dating and correlation with the diversification of Angiosperms. She also cast doubts on the root position, mentioning the fact that both Myrmecia and Amblyopone have been hypothesized as the most primitive ants even though you cannot have both, since they occur well nested inside unrelated groups. Both her and Ward said that it will probably be necessary to sequence the other nominal Leptanillinae taxa to break the assumed long-branch attraction artifact. However the question remains if the problem if confined to Leptanillinae alone or if the whole so-called Poneroid clade is acting as an “attractor” of the divergent outgroup sequences and thus the ant tree is getting rooted upside-down. After all this “clade” contains several long branches (e.g., what is the lone Paraponera clavata doing in the middle of it?). A more promising solution would be to balance the outgroup portion of the analysis, but we will have to wait for the HymATol team for this.

Chris Schmidt (Tucson, Arizona) talked next. He is working on a phylogeny of the Ponerinae sensu Bolton and not wasting any time since he downloaded many sequences from Moreau's et al. to include as many outgroups as ingroup terminals. This talk was much more sophisticated than the rest in the symposium in terms of phylogenetic methods (probably under the influence of D. Maddison), and talked about the use of mixed models instead of the traditional way of applying the same model across a given gene or a priori codon partition. The results on the tree topology are very different, with the exception of Platythyrea that always comes as a long branch sister to Ponerini. The use of mixed models seems to be a good trend among researches fond of model-base approaches, and I wonder how far will they go before realizing they came back full circle to the realm of parsimony. I only regret that C. Schmidt omitted a discussion of the taxonomic and nomenclatural problems of this group as advertised in his abstract, since sorting out taxa like Pachycondyla is going to be the real challenge for this subfamily.

After this came the second turn for the Ant-AtoL team with a talk by Sean Brady (Smithsonian), co-organizer of the symposium. In this talk the AtoL team addressed the issue of dating the molecular ant phylogeny. The results are, again, basically the same as the ones published and presented by Moreau et al. However Brady also showed that his team is exploring the sensibility of their estimations to different parameters like models, taxon sampling and tree topology. This is important given the apparent problems with the ant tree reconstruction itself discussed earlier.

I will end this already long first part noting that all of the above speakers in some way or another acknowledged the need to incorporate morphology into the picture. In part to ameliorate the issues of poor resolution and indecisiveness that the current molecular data is showing about the relationships among the ant subfamilies, but also to be able to incorporate the fossil information more precisely and achieve a more complete picture of the ant’s phylogenetic history. In all, this positive attitude towards morphology is the best thing coming out of a molecular phylogenetics symposium.

To Part II