LETTER doi:10.1038/nature15697 A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing Richard O. Prum1,2*, Jacob S. Berv3*, Alex Dornburg1,2,4, Daniel J. Field2,5, Jeffrey P. Townsend1,6, Emily Moriarty Lemmon7 & Alan R. Lemmon8 Although reconstruction of the phylogeny of living birds has pro- It has long been recognized that phylogenetic confidence depends gressed tremendously in the last decade, the evolutionary history of not only on the number of characters analysed and their rate of evolu- Neoaves—a clade that encompasses nearly all living bird species— tion, but also on the number and relationships of the taxa sampled remains the greatest unresolved challenge in dinosaur systematics. relative to the nodes of interest9–11. Theory predicts that sampling a Here we investigate avian phylogeny with an unprecedented scale single taxon that diverges close to a node of interest will have a far of data: .390,000 bases of genomic sequence data from each of greater effect on phylogenetic resolution than will adding more char- 198 species of living birds, representing all major avian lineages, acters11. Despite using an alignment of .40 million base pairs, sparse and two crocodilian outgroups. Sequence data were collected using sampling of 48 species in the recent avian genomic analysis may not anchored hybrid enrichment, yielding 259 nuclear loci with an have been sufficient to confidently resolve the deep divergences among average length of 1,523 bases for a total data set of over 7.8 3 107 major lineages of Neoaves. Thus, expanded taxon sampling is required bases. Bayesian and maximum likelihood analyses yielded highly to test the monophyly of neoavian clades, and to further resolve the supported and nearly identical phylogenetic trees for all major phylogenetic relationships within Neoaves. avian lineages. Five major clades form successive sister groups to Here, we present a phylogenetic analysis of 198 bird species and the rest of Neoaves: (1) a clade including nightjars, other caprimul- 2 crocodilians (Supplementary Table 1) based on loci captured using giforms, swifts, and hummingbirds; (2) a clade uniting cuckoos, anchored enrichment12. Our sample includes species of 122 avian bustards, and turacos with pigeons, mesites, and sandgrouse; (3) families in all 40 extant avian orders2, with denser representation of cranes and their relatives; (4) a comprehensive waterbird clade, non-oscine birds (108 families) than of oscine songbirds (14 families). including all diving, wading, and shorebirds; and (5) a compre- Effort was made to include taxa that would break up long phylogenetic hensive landbird clade with the enigmatic hoatzin (Opisthocomus branches, and provide the highest likelihood of resolving short inter- hoazin) as the sister group to the rest. Neither of the two main, nodes at the base of Neoaves11. We also sampled multiple species recently proposed Neoavian clades—Columbea and Passerea1— within groups whose monophyly or phylogenetic interrelationships were supported as monophyletic. The results of our divergence have been controversial—that is, tinamous, nightjars, hummingbirds, time analyses are congruent with the palaeontological record, sup- turacos, cuckoos, pigeons, sandgrouse, mesites, rails, storm petrels, porting a major radiation of crown birds in the wake of the petrels, storks, herons, hawks, hornbills, mousebirds, trogons, king- Cretaceous–Palaeogene (K–Pg) mass extinction. fishers, barbets, seriemas, falcons, parrots, and suboscine passerines. Birds (Aves) are the most diverse lineage of extant tetrapod verte- We targeted 394 loci centred on conserved anchor regions of the brates. They comprise over 10,000 living species2, and exhibit an extra- genome that are flanked by more variable regions12. We performed all ordinary diversity in morphology, ecology, and behaviour3. Substantial phylogenetic analyses on a data set of 259 genes with the highest progress has been made in resolving the phylogenetic history of birds. quality assemblies. The average locus was 1,524 bases in length Phylogenetic analyses of both molecular and morphological data sup- (361–2,316 base pairs (bp)), and the total percentage of missing data port the monophyletic Palaeognathae (the tinamous and flightless was 1.84%. The concatenated alignment contained 394,684 sites. To ratites) and Galloanserae (gamebirds and waterfowl) as successive, minimize overall model complexity while accurately accounting for monophyletic sister groups to the Neoaves—a diverse clade including substitution processes, we performed a partition model sensitivity all other living birds4. Resolving neoavian phylogeny has proven to be a analysis with PartitionFinder13,14, and compared a complex partition difficult challenge because this radiation was very rapid and deep in model (one partition per locus) to a heuristically optimized (rclust) time, resulting in very short internodes4. partition model. Phylogenetic informativeness (PI) approaches15,16 In the last decade, phylogenetic analyses of large, multilocus data provided strong evidence that the phylogenetic utility of our data set sets have resulted in the proposal of numerous, novel neoavian rela- was high, with low declines in PI profiles for individual loci, data set tionships. For example, a clade consisting of diving and wading birds partitions, and the concatenated matrix (Supplementary Fig. 4). We has been consistently recovered, as well as a large landbird clade in estimated concatenated trees in ExaBayes17 and RAxML18 using a 75 which falcons and parrots are successive sister groups to the perching partition model. Coalescent species trees were estimated with the gene birds4–8. Recently, phylogenetic analyses of 48 whole avian genomes tree summation methods in STAR19, NJst20, and ASTRAL21 from gene resulted in the proposal of a novel phylogenetic resolution of the initial trees estimated with RAxML (see Methods.) branching sequence within Neoaves1. Although this genomic study Our concatenated Bayesian analyses resulted in a completely provided much needed corroboration of many neoavian clades, the resolved, well supported phylogeny. All clades had a posterior prob- limited taxon sampling precluded further insights into the evolution- ability (PP) of 1, except for a single clade including shoebill ary history of birds. (Balaeniceps) and pelican (PP 5 0.54) (Fig. 1). The concatenated 1Department of Ecology & Evolutionary Biology, Yale University, New Haven, Connecticut 06520, USA. 2Peabody Museum of Natural History, Yale University, New Haven, Connecticut 06520, USA. 3Department of Ecology and Evolutionary Biology, Fuller Evolutionary Biology Program, Cornell University, and Cornell Laboratory of Ornithology, Ithaca, New York 14853, USA. 4North Carolina Museum of Natural Sciences, Raleigh, North Carolina 27601, USA. 5Department of Geology & Geophysics, Yale University, New Haven, Connecticut 06520, USA. 6Department of Biostatistics, and Program in Computational Biology and Bioinformatics, Yale University, New Haven, Connecticut 06520, USA. 7Department of Biological Science, Florida State University, Tallahassee, Florida 32306, USA. 8Department of Scientific Computing, Florida State University, Tallahassee, Florida 32306, USA. *These authors contributed equally to this work. 22 OCTOBER 2015 | VOL 526 | NATURE | 569 G2015 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER Aves 70 60 50 40 30 20 10 0 Ma Struthio Palaeognathae 190 Rhea 191 Apteryx 192 197 Casuarius Dromaius 193 Tinam. Tina 195 Eudromia 194 Nothoprocta am. 196 Crypturellus Tinamus Leipoa Galliformes 176 183 Ortalis Crax 177 Numida Galloanserae 178 182 Odontophorus Colinus 175 179 1 Rollulus 180 181 Bonasa Gallus Chauna Anseriform. 184 Anseranas 185 Dendrocygna 186 Oxyura 187 Anser 188 189 Anas Aythya Eurostopodus 173 174 Caprimulgus Chordeiles 163 Steatornis 172 Strisores Columbaves Gruiformes Aequorlitornithes 2 Nyctibius 164 Podargus 165 Aegotheles Hemiprocne Apodiform. 166 168 169 Streptoprocne 167 Chaetura 171 Topaza 170 Phaethornis Archilochus Otidimorph. 162 Tauraco Corythaeola 157 Ardeotis 158 Tapera 159 Centropus 160 161 Cuculus 3 148 Coccyzus 156 Monias Columbimorph. 154 Mesitornis 155 Syrrhaptes Pterocles 149 153 Ptilinopus Treron 150 Columbina 151 152 Leptotila Columba Heliornis 144 Neoaves Sarothrura 141 Rallus 142 143 Micropygia 140 Porphyrio Psophia 4 145 Aramus 146 147 Balearica Grus 139 Phoenicopterus Rollandia Burhinus 136 Charadrius 123 137 138 Haematopus Recurvirostra Pedionomus 124 132 133 Jacana 131 Rostratula Limosa 134 135 Arenaria 125 Tringa 98 Turnix 5 126 Glareola 127 Uria 128 Chroicocephalus 129 quorlitorn 130 Sterna Rynchops 122 Eurypyga Phaethon Gavia n nithes 99 Spheniscus 114 Phoebastria 115 121 Oceanites 100 Pelagodroma 116 Oceanodroma 117 Fulmarus 118 Puffinus 101 119 120 Pterodroma Pelecanoides 113 Ciconia 6 Leptoptilos Fregata 102 110 Morus 111 112 Anhinga 103 Phalacrocorax Theristicus Tigrisoma 104 106 107 Ardea 105 Ixobrychus Scopus 108 109 Balaeniceps Pelecanus Upper Palaeocene Eocene Oligocene Miocene Pli. Ple. Cretaceous Palaeogene Neogene Q. Figure 1 | Phylogeny of birds. Time-calibrated phylogeny of 198 species of sister clades are: Strisores (brown), Columbaves (purple), Gruiformes (yellow), birds inferred from a concatenated, Bayesian analysis of 259 anchored Aequorlitornithes (blue), and Inopinaves (green). Background colours mark phylogenomic loci using ExaBayes17. Figure continues on the opposite page geological periods. Ma, million years ago;