BIOLOGICAL SCIENCES / EVOLUTION
Genomics, biogeography, and the diversification of placental mammals
Derek E. Wildman,,,¶, Monica Uddin, Juan C. Opazo,||, Guozhen Liu, Vincent Lefort, Stephane Guindon, Olivier Gascuel, Lawrence I. Grossman, Roberto Romero,¶, and Morris Goodman,¶,
Perinatology Research Branch, National Institute of Child Health and Human Development/National Institutes of Health, Department of Health and Human Services, Bethesda, MD 20892; Center For Molecular Medicine and Genetics, and Departments of Obstetrics and Gynecology and Anatomy and Cell Biology, Wayne State University, Detroit, MI 48201; ||School of Biological Sciences, University of Nebraska, Lincoln, NE 68588; and Laboratory of Computer Science, Robotics, and Microelectronics, Centre National de la Recherche Scientifique, Université Montpellier II, 161 Rue Ada, 34392 Montpellier, France
Contributed by Morris Goodman, May 9, 2007 (received for review May 4, 2007)
Previous molecular analyses of mammalian evolutionary relationships involving a wide range of placental mammalian taxa have been restricted in size from one to two dozen gene loci and have not decisively resolved the basal branching order within Placentalia. Here, on extracting from thousands of gene loci both their coding nucleotide sequences and translated amino acid sequences, we attempt to resolve key uncertainties about the ancient branching pattern of crown placental mammals. Focusing on 1,700 conserved gene loci, those that have the more slowly evolving coding sequences, and using maximum-likelihood, Bayesian inference, maximum parsimony, and neighbor-joining (NJ) phylogenetic tree reconstruction methods, we find from almost all results that a clade (the southern Atlantogenata) composed of Afrotheria and Xenarthra is the sister group of all other (the northern Boreoeutheria) crown placental mammals, among boreoeutherians Rodentia groups with Lagomorpha, and the resultant Glires is close to Primates. Only the NJ tree for nucleotide sequences separates Rodentia (murids) first and then Lagomorpha (rabbit) from the other placental mammals. However, this nucleotide NJ tree still depicts Atlantogenata and Boreoeutheria but minus Rodentia and Lagomorpha. Moreover, the NJ tree for amino acid sequences does depict the basal separation to be between Atlantogenata and a Boreoeutheria that includes Rodentia and Lagomorpha. Crown placental mammalian diversification appears to be largely the result of ancient plate tectonic events that allowed time for convergent phenotypes to evolve in the descendant clades.
Atlantogenata | Eutheria | Notolegia | phylogeny | vicariance
PNAS, September 4, 2007, vol. 104, no. 36, 14395-14400. OPEN ACCESS ARTICLE.
Phylogenetic analyses can elucidate the history of diversification within a group of organisms such as placental mammals (i.e., Placentalia) (1, 2). However, if the analyses use too few characters or taxa, an inaccurate phylogenetic tree can be obtained because of sampling error (3). Here we seek to reduce such error by using abundant character information from mammalian and other vertebrate genomes that have been completely or nearly completely sequenced. On examining phylogenetically the coding nucleotide sequences and translated amino acid sequences for thousands of genes, we attempt to resolve key uncertainties about the ancient branching order of crown placental mammals. Our results are promising, but uncertainties remain concerning the basal diversification of Placentalia. The practice of phylogenomics is still in its infancy and has yet to produce an authoritative model that infallibly predicts all of the real patterns of nucleotidesubstitutions within evolved genomes.
Among the placental mammals, phylogenetic branching events have been inferred by using either nucleotide sequence data (4–11) or rare insertion/deletion patterns (12–15). Many of the results recognize four primary eutherian groups: Afrotheria, Xenarthra, Laurasiatheria, and Euarchontoglires. Afrotherians (e.g., elephants, hyraxes, manatees, aardvarks, tenrecs, and allies) are a clade of mammals that originated in Africa, and whose extant members still mostly remain on that continent with the exception of Asian elephants and sirenians such as the Florida manatee. The Xenarthra includes the sloths, armadillos, and anteaters that today are restricted to South and Central America (although some Xenarthra, such as the nine-banded armadillo, have recently dispersed to North America). The Laurasiatheria (e.g., bats, eulipotyphlans, pangolins, carnivores, perrisodactyls, and cetartiodactyls) is a diverse clade including extant lineages that originated in the ancient northern continent of Laurasia. The Euarchontoglires includes the species from five living mammalian orders (e.g., primates, treeshrews, flying lemurs, rabbits, and rodents). This last group remains the most controversial, and a number of recent studies have suggested it is not valid (4, 6, 7).
The studies that have identified and supported these primary groupings have been able to resolve the branching pattern within some of the clades, but a statistically robust determination of the branching order at the base of the placental clade continues to be elusive. Three primary hypotheses (Fig. 1 A–C) supporting the branching orders among the four major placental groups described above have been proposed. In the first scenario (9, 10, 16), Afrotherians split from the other three clades (Notolegia = Exafroplacentalia) at the base of the placental tree. In the second hypothesis (13), the Xenarthrans split from the other three clades (Epitheria). The third hypothesis (5, 11, 14, 17) proposes that Xenarthra and Afrotheria group together (Atlantogenata) to the exclusion of Laurasiatheria and Euarchontoglires (Boreoeutheria).
Some studies have suggested a fourth hypothesis, that one of the groups, Euarchontoglires, is polyphyletic, and that Glires (rodents and lagomorphs) is also not monophyletic (Fig. 1D). Based on a parsimony analysis of nuclear-encoded genes, it was reported (8) that murid rodents as represented by mouse and rat were the sister group to all other placental mammals, and that rabbits were the next branching clade. More recently, a number of large-scale genomic studies (4, 6) have reported that primates grouped more closely to laurasiatherians than to Glires.
To test the four alternative hypotheses (Fig. 1), we constructed phylogenetic trees and performed topology tests based on a data set that included nucleotide and amino acid sequence data from the complete genome sequences of 11 mammalian species representing the major placental clades depicted in Fig. 1, a marsupial opossum, and two nonmammalian outgroups. The ingroup taxa include three primates (human, chimpanzee, and rhesus macaque), two rodents (mouse and rat), rabbit, dog, cow, armadillo, African elephant, tenrec, and the South American Gray Short-tailed opossum. Chicken and frog were used as outgroups to root the tree. To avoid common pitfalls associated with phylogenomic studies, we focused our analyses on conserved protein coding genes to reduce saturation and long-branch attraction effects. Our alignment method also preserved the codon reading frame for all loci and all taxa, so we were able to remove potentially saturated third codon positions from the parsimony analyses. Because all tree reconstruction methods fail to recover the "true" tree given certain conditions, we used four different inferential tree reconstruction methods: maximum parsimony (MP), maximum likelihood (ML), Bayesian, and neighbor-joining (NJ). The probabilistic methods (ML and Bayesian) may be more robust than the others to model assumption violation, although NJ can also often recover the true tree when the wrong model of sequence evolution is used. MP minimizes the amount of evolution (i.e., nucleotide substitutions) and in some cases, outperforms the probabilistic methods (18, 19). Although we would be encouraged if several of these tree reconstruction methods converged on just one of the four hypotheses, we realize that all methods might converge on an incorrect branching pattern for the basal diversification of Placentalia.