Bloody Roots: GC-rich Genes Hide the Root of the Placental Tree

Despite the rapid increase of size in phylogenomic datasets, the rooting of the placental mammal tree is still a controversial issue. One difficulty lies in the pervasive phylogenetic conflicts among genes, with each one telling its own story, which may be reliable or not. Here we identified a simple criterion, i.e. the GC-content, which substantially helps in determining which gene trees best reflect the species tree. We assessed the ability of 13,111 coding sequence alignments to correctly reconstruct the placental phylogeny. We found that GC-rich genes induced a higher amount of conflict among gene trees, and performed worse than AT-rich genes in retrieving well-supported, consensual nodes on the placental tree. We interpret this GC-effect mainly as a consequence of genome-wide variations in recombination rate. Indeed, recombination is known to drive GC-content evolution through GC-biased gene conversion, and might be problematic for phylogenetic reconstruction, for instance in an incomplete lineage sorting context. When we focused on the AT-richest fraction of the dataset, the resolution level of the placental phylogeny was greatly increased, and a strong support was obtained in favor of an Afrotheria rooting, i.e. Afrotheria as the sister group of all other placentals. We show that in mammals most conflicts among gene trees, which have so far hampered the resolution of the placental tree, are concentrated in the GC-rich regions of the genome. We argue that the GC-content - since it is a reliable indicator of the long-term recombination rate - is an informative criterion that could help in identifying the most reliable molecular markers for species tree inference.

Reference: Romiguier J., Ranwez V., Delsuc F., Galtier N. & Douzery E.J.P. (2013). Less is more in mammalian phylogenomics: AT-rich genes minimize tree conflicts and unravel the root of placental mammals. Molecular Biology and Evolution 30: 2134-2144.

Commentary by Emma Teeling & Blair Hedges in Molecular Biology and Evolution