Harvesting Evolutionary Signals in a Forest of Prokaryotic Gene Trees

Phylogenomic studies produce increasingly large phylogenetic forests of trees with patchy taxonomical sampling. Typically, prokaryotic data generate thousands of gene trees of all sizes that are difficult, if not impossible, to root. Their topologies do not match the genealogy of lineages, as they are influenced not only by duplication, losses, and vertical descent but also by lateral gene transfer (LGT) and recombination. Because this complexity in part reflects the diversity of evolutionary processes, the study of phylogenetic forests is thus a great opportunity to improve our understanding of prokaryotic evolution. Here, we show how the rich evolutionary content of such novel phylogenetic objects can be exploited through the development of new approaches designed specifically for extracting the multiple evolutionary signals present in the forest of life, that is, by slicing up trees into remarkable bits and pieces: clans, slices, and clips. We harvested a forest of 6,901 unrooted gene trees comprising up to 100 prokaryotic genomes (41 archaea and 59 bacteria) to search for evolutionary events that a species tree would not account for. We identified 1) trees and partitions of trees that reflected the lifestyle of organisms rather than their taxonomy, 2) candidate lifestyle-specific genetic modules, used by distinct unrelated organisms to adapt to the same environment, 3) gene families, nonrandomly distributed in the functional space, that were frequently exchanged between archaea and bacteria, sometimes without major changes in their sequences. Finally, 4) we reconstructed polarized networks of genetic partnerships between archaea and bacteria to describe some of the rules affecting LGT between these two Domains.