Department of Molecular and Cell Biology, University of Connecticut, Storrs, Connecticut 06269-3125, USA. e-mails: gogarten@uconn.edu; Jeffrey.Townsend@ uconn.edu doi:10.1038/nrmicro1204 Published online 1 August 2005 TREE OF LIFE The tree-like representation of the history of all living and extinct organisms. MUTUALISM An association between two organisms, often from different species, that benefits both partners. RETICULATION A network that is formed through the fusion of independent branches. Bifurcating trees, in which evolutionary lineages split and evolve independently from each other, have a long history as tools to visualize the evolution of species: Lamarck introduced tree-like binary schemes for taxonomic classification 1 and Charles Darwin described the evolution of species as the TREE OF LIFE 2 . Darwin also noted that the ‘coral of life’ might be a more appropriate metaphor, because only the outer- most layer in the tree of life is actually alive, resting on a base of dead branches 3 . The tree of life became the standard imagery to depict species evolution, implying a common root of all life on Earth and a bifurcating evolutionary process. However, there are several clear exceptions to this standard view. For example, botanists found that many plant species violate a bifurcating model as they are allopolyploid, combining the genomes of different parental spe- cies. This process of a new line of descent originat- ing from the hybridization of two parent species has been termed reticulate evolution 4–8 . The fungal–algal symbiosis of lichens illustrates that symbiosis can lead to long-term partnerships with different prop- erties from those of either parent species, and some of the most dramatic breakthroughs in cellular evolution, that is, the mitochondria and plastids, are the result of endosymbiosis 9 . Throughout the decades, MUTUALISM and RETICULATION have often been considered the most important processes in species evolution 10,11 . However, for most branches of biology these processes were only exceptions, albeit important ones, in an otherwise steadily furcating process of species evolution. By introducing ribosomal RNA (rRNA) as a taxo- nomic marker molecule, Woese and Fox extended the tree paradigm to the realm of microorganisms 12,13 . However, the large-scale availability of sequence data provided information that effectively sundered the cambium of the tree of life metaphor. Different mol- ecules were shown to have different histories 14 , and members of the same species were found to differ dra- matically in gene content. For example, of the genes revealed by the sequencing of three Escherichia coli genomes, fewer than 40% were common to all three 15 . Furthermore, it has been suggested that extinct species have contributed genes to the extant layer of life, even though these contributors might not have been in the direct line of ancestry 16,17 . Reticulate models of evolu- tionary history that incorporate gene transfer might HORIZONTAL GENE TRANSFER, GENOME INNOVATION AND EVOLUTION J. Peter Gogarten and Jeffrey P. Townsend Abstract | To what extent is the tree of life the best representation of the evolutionary history of microorganisms? Recent work has shown that, among sets of prokaryotic genomes in which most homologous genes show extremely low sequence divergence, gene content can vary enormously, implying that those genes that are variably present or absent are frequently horizontally transferred. Traditionally, successful horizontal gene transfer was assumed to provide a selective advantage to either the host or the gene itself, but could horizontally transferred genes be neutral or nearly neutral? We suggest that for many prokaryotes, the boundaries between species are fuzzy, and therefore the principles of population genetics must be broadened so that they can be applied to higher taxonomic categories. NATURE REVIEWS | MICROBIOLOGY VOLUME 3 | SEPTEMBER 2005 | 679 REVIEWS FOCUS ON HORIZONTAL GENE TRANSFER