Tectonic collision and uplift of Wallacea triggered the global songbird radiation

Abstract

Songbirds (oscine passerines) are the most species-rich and cosmopolitan bird group, comprising almost half of global avian diversity. Songbirds originated in Australia, but the evolutionary trajectory from a single species in an isolated continent to worldwide proliferation is poorly understood. Here, we combine the first comprehensive genome-scale DNA sequence data set for songbirds, fossil-based time calibrations, and geologically informed biogeographic reconstructions to provide a well-supported evolutionary hypothesis for the group. We show that songbird diversification began in the Oligocene, but accelerated in the early Miocene, at approximately half the age of most previous estimates. This burst of diversification occurred coincident with extensive island formation in Wallacea, which provided the first dispersal corridor out of Australia, and resulted in independent waves of songbird expansion through Asia to the rest of the globe. Our results reconcile songbird evolution with Earth history and link a major radiation of terrestrial biodiversity to early diversification within an isolated Australian continent.

ARTICLE

Received 28 Jan 2016 | Accepted 26 Jul 2016 | Published 30 Aug 2016

DOI: 10.1038/ncomms12709 OPEN

Tectonic collision and uplift of Wallacea triggered the global songbird radiation

Robert G. Moyle1,*, Carl H. Oliveros1,*,w, Michael J. Andersen2, Peter A. Hosner3, Brett W. Benz4, Joseph D. Manthey1, Scott L. Travers1, Rafe M. Brown1 & Brant C. Faircloth5

Songbirds (oscine passerines) are the most species-rich and cosmopolitan bird group, comprising almost half of global avian diversity. Songbirds originated in Australia, but the evolutionary trajectory from a single species in an isolated continent to worldwide proliferation is poorly understood. Here, we combine the rst comprehensive genome-scale DNA sequence data set for songbirds, fossil-based time calibrations, and geologically informed biogeographic reconstructions to provide a well-supported evolutionary hypothesis for the group. We show that songbird diversication began in the Oligocene, but accelerated in the early Miocene, at approximately half the age of most previous estimates. This burst of diversication occurred coincident with extensive island formation in Wallacea, which provided the rst dispersal corridor out of Australia, and resulted in independent waves of songbird expansion through Asia to the rest of the globe. Our results reconcile songbird evolution with Earth history and link a major radiation of terrestrial biodiversity to early diversication within an isolated Australian continent.

1 Biodiversity Institute and Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence, Kansas 66045, USA. 2 Department of Biology and Museum of Southwestern Biology, University of New Mexico, Albuquerque, New Mexico 87131, USA. 3 Department of Biology, University of Florida, Gainesville, Florida 32611, USA. 4 Division of Vertebrate Zoology, American Museum of Natural History, New York, New York 10024, USA. 5 Department of Biological Sciences and Museum of Natural Science, Louisiana State University, Baton Rouge, Louisiana 70803, USA. * These authors contributed equally to this work. w Present address: Department of Biological Sciences, Louisiana State University, Baton Rouge, Louisiana 70803, USA. Correspondence and requests for materials should be addressed to R.G.M. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709

Avian diversity is concentrated conspicuously in a single clade, the oscine passerines (songbirds; B5,000 species), which constitute a major component of avifaunas

around the world1. Early molecular phylogenetic studies and the fossil record support the hypothesis that songbirds originated in Australia25, but most other aspects of early songbird evolution remain uncertain6. Prior research suggested songbird diversication scenarios that are largely uncoupled from Earth history, including extensive diversication of lineages in New Guinea before its emergence as a landmass7,8 and long-distance dispersal to Africa or Asia when no dispersal corridors existed2,7,911 (but see refs 1214). However, these inferences may be compromised by reliance on unresolved phylogenetic relationships and a controversial biogeographic time calibration. Molecular phylogenetic studies have identied major lineages of songbirds, but these studies also revealed rapid, early bursts of diversication that have proven intractable to all analyses2,9,15. Studies using the largest multi-gene data sets could not resolve many relationships within two major songbird radiations, the Corvides and Passerides2,8,16,17 (sensu Cracraft18). For example, the most extensive character sampling to date yielded songbird phylogenies with low support (that is, o70%

bootstrap support) for 450% of nodes8,15. Two major genome-scale phylogenetic studies of birds included only a few representatives from the oscine passerines19,20.

Attempts to understand the evolutionary history of songbirds have also been hampered by a sparse fossil record that provides few clues about the age and geographic origin of major clades5. Revelation of a likely Gondwanan origin for all passerine birds3,4,21 initiated a paradigm shift in biogeographic inquiry for the group and enabled an explicit timescale for passerine evolution based on geologic events during the breakup of the Gondwanan landmasses2,3. Using these biogeographic calibrations and partially resolved molecular phylogenies, studies have identied key events in crown-oscine evolution2,3,7,9,10, such as (1) a Palaeocene or Eocene origin in Australasia and early diversication of the Corvides among proto-islands of New Guinea; (2) an African origin of the largest clade of Passerides via dispersal across the Indian Ocean from Australia; and (3) multiple Palaeogene long-distance dispersal events to Asia. However, this set of palaeobiogeographic hypotheses produced an evolutionary chronology incongruent with other aspects of geologic history, such as the Miocene formation of New Guinea and Wallacea22,23. Recent studies using a variety of fossil calibration points offered an alternative time frame for bird diversication that is substantially younger, but they included few songbirds19,20 or did not resolve basal songbird relationships12,13.

To test previous hypotheses of songbird diversication, we collected genome-scale DNA sequence data to reconstruct songbird relationships, and we placed this phylogeny in a temporal context with fossil-derived calibration points. Phylogenetic analysis of these data produced a well-supported hypothesis of relationships among oscines that is highly concordant across multiple analytical approaches. Our biogeographic reconstructions support early songbird diversication in Australia in the Oligocene, followed by rapid diversication and geographic expansion via SE Asia early in the Miocene, coincident with extensive island formation in Wallacea.

ResultsWe prepared reduced representation genomic libraries for 104 songbird species (Supplementary Data 1), representing B70%

of songbird families, using a sequence capture protocol (http://www.ultraconserved.org

Web End =www.ul http://www.ultraconserved.org

Web End =traconserved.org ) that targeted ultraconserved elements (UCEs)24. We obtained an average of 4.7 106 reads for each individual after

discarding reads with low-quality scores (Supplementary Data 1). Trimming reads with adaptor contamination yielded an average read length of 98.9 bp across all individuals. From these reads, we assembled contigs corresponding to 3,0964,644 UCE loci (mean 4,286.8) for each individual having an average length of

695.9 bp and average coverage of 47.9 . As expected, the sample

taken from a toepad clipping of a museum specimen (Mohoua albicilla) yielded the least amount of data, with the lowest number of UCE loci (3,096) and the lowest average contig length (268.1 bp). Sequence capture yielded 4,155 UCE loci present in at least 75% of taxa and an alignment length of just under2.5 million bases (incomplete matrix). We also created a stringent data set of 515 loci recovered in all taxa with an aligned length of just over 375,000 bases (complete matrix).

Concatenation and coalescent approaches to species tree inference yielded well-resolved and highly concordant topologies (Fig. 1; Supplementary Figs 1 and 2). Bayesian and maximum likelihood methods produced nearly identical trees and congruent support indices (Supplementary Fig. 1). Of 103 internal nodes, 84 had support of 1.0 posterior probability and 100% bootstrap support. Only six nodes had o70% bootstrap support, but still had high posterior probability (40.98). SVDquartets recovered a topology highly congruent with Bayesian and maximum likelihood estimates with only 12 nodes receiving o70%

bootstrap support (Supplementary Fig. 2). Gene tree-based coalescent methods recovered trees with lower nodal support and were affected by bias caused by missing data (Supplementary Figs 3 and 4, Supplementary Discussion). Our results differed markedly from previous studies (Supplementary Fig. 5). As an example, at the base of the Passerides we found strong support for sequential branching of the New Guinea endemic Cnemophilidae and Melanocharitidae followed by the New Zealand endemic Callaeidae and then a sister pairing of the Picathartidae (Africa and Asia) and Petroicidae (Australasia). These families have been placed in various congurations near the Passerides and Corvides in previous studies2,7,9,13,15 but their unanticipated afnities require reinterpretation of previous biogeographic hypotheses.

A relaxed clock analysis of the complete matrix using the earliest fossil identiable as a crown-oscine25 and secondary calibrations derived from a fossil-calibrated phylogenomic analysis of birds19 produced a time frame (Fig. 1) more recent than hypothesized by most studies (but see refs 12,13,20). The rst songbird lineages diverged in the Oligocene (basal oscines; Fig. 1), but the vast majority of major clades diversied in the Miocene. A large concentration of short internodes in multiple clades of the Corvides and Passerides cluster at 1320 Myr ago. An alternative dating analysis based on secondary calibrations from a more recent study20 yielded slightly older date estimates (Supplementary Fig. 6).

Biogeographic analyses using geologically informed constraints overwhelmingly support early songbird diversication in Australia (Fig. 1; Supplementary Figs 710). Early diversication in the Corvides and Passerides also occurred within Australia in the Miocene, in contrast to previous ndings that the Corvides arose and diversied within proto-islands of New Guinea in the Eocene7 or Oligocene8. Our biogeographic results derive from the incorporation of recent geological models for the formation of New Guinea23,26 (see Supplementary Discussion) as a constraint on the biogeographic analysis (that is, New Guinea was not allowed as an ancestral range before its existence as a landmass). The exact timing and details of competing models differ, but all agree that New Guinea emerged as a substantial island no earlier than 12 Myr ago23, and possibly as late as 5 Myr ago26. Unconstrained analyses (Supplementary Fig. 7) imply New Guinea origins for nodes much deeper in the tree7,8 but these are inconsistent with the geological record.

2 NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709 ARTICLE

40 0

30

20

10

A B C D E F G H I

Menura

Aegithina

Cormobates

Basal Oscines Corvides Passerides

Core-Passerides

Climacteris

Ptilonorhynchus

Vireo

Malurus

Ailuroedus

Pardalotus

Acanthorhynchus

Meliphaga

Timeliopsis

Foulehaio

Orthonyx

Pomatostomus

Cinclosoma Ptilorrhoa

Eulacestoma

Toxorhamphus

Muscicapa

Coracina

Picathartes

Pericrocotus

Corvus

Mohoua

Daphoenositta

Psophodes

Melanocharis

Petroica

Oreoica

Falcunculus

Pachycephala

Oriolus

Oreocharis

Pteruthius

Sturnus

Vidua

Machaerirhynchus

Artamus

Rhagologus

Prionops

Rhipidura

Loboparadisea

Lamprolia

Monarcha

Ploceus

Paradisaea

Struthidea

Modulatrix

Lanius

Cnemophilus

Oedistoma

Philesturnus

Chaetops

Estrilda

Hyliota

Hirundo

Eopsaltria

Irena

Microeca

Bombycilla

Dulus

Locustella

Cinclus

Culicicapa

Regulus

Sitta

Troglodytes

Promerops

Leptocoma

Certhia

Chloropsis

Peucedramus

Orthotomus

Prunella

Passer

Fringilla

Progne

Motacilla

Chelidorhynx

Parus

Remiz

Abroscopus

Panurus

Eremophila

Nicator

Macrosphenus

Sylvia

Sylvietta

Cisticola

Oxylabes

Hylia

Donacobius

Aegithalos

Acrocephalus

Pnoepyga

Bleda

Pycnonotus

Timalia

Phylloscopus

Seicercus

Psaltriparus

Erythrocercus

Tesia

Ma

0

New Guinea

Orogeny

Figure 1 | Time-calibrated phylogeny of songbirds. Topology based on concatenated maximum likelihood analysis of 4,155 UCE loci and date estimates based on fossil-derived calibrations. Nodes with squares around ancestral range indicate 70% or greater maximum likelihood bootstrap support, circles indicate o70% bootstrap support. All nodes are supported by posterior probability greater than 0.98 in Bayesian analyses. Geological events are indicated by vertical grey bars and labelled below. Stars indicate nodes calibrated for dating analyses. Coloured squares at tips indicate distribution of clade for that exemplar and correspond to coloured world map (A, Australia; B, New Zealand; C, New Guinea; D, Wallacea; E, S and SE Asia; F, Africa; G, New World; H, Palaearctic; I, Madagascar). Ancestral areas at internal nodes inferred with DEC j model and New Guinea excluded as an ancestral area before

15 Myr ago. Palaeogeographic reconstructions adapted from Hall27 depict the distribution of land and sea across Wallacea and northern Australasia during early songbird diversication. Abbreviations: Plio. Pliocene; Pleis. Pleistocene. See (Supplementary Fig. 6) for dating with alternative calibrations.

Wallacea Uplift

New Guinea

Emergence

NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709

DiscussionOur biogeographic analyses indicate that all dispersal out of the Australian region, beginning at ca. 23 Myr ago, occurred via Asia followed by rapid colonization of other continents (Fig. 1). An early lineage of Passerides dispersed to Asia ca. 24 Myr ago, followed by several lineages of Corvides and Passerides starting ca. 16 Myr ago. This rapid colonization and diversication produced the majority of short internodes that plagued previous analyses. The time frame of dispersal out of Australia is consistent with the Sula spur of Australia colliding with proto-Sulawesi in the early Miocene, an event which initiated substantial island formation in Wallacea and provided the rst links from Australasia to Sundaland22,27. Subsidence of newly formed land in Wallacea by15 Myr ago reduced the dispersal corridor and may explain the lack of relictual endemics in the region22. Previous studies using biogeographic calibrations inferred this cluster of dispersal events to occur in the Eocene and Oligocene2,7 when Australia was isolated by thousands of kilometres of open ocean.

A recently published timescale for avian diversication12 obtained node ages within songbirds that were substantially more recent than most previous studies, yet older than our timing estimates. Although their study also concluded that the more recent time frame of songbird diversication obviates the need to invoke long-distance dispersal to explain the expansion of songbirds out of the Australian region, they did not provide an explicit biogeographic hypothesis of how oscine lineages were transferred to Southeast Asia during the late Eocene to Oligocene; perhaps because these continental regions were still separated by a substantial marine gap and deep water trench during this period22,28. Moreover, the microplate hypothesis which they cite as a possible mechanism for transporting Australian continental fragments (and presumably songbird lineages) west to Wallacea in the MiocenePliocene is now considered implausible based on the most recent palaeogeological modelling22. Their biogeographic results are similar to ours despite substantial disparity in phylogenetic relationships compared with our results. Although the backbone of the phylogeny they presented used published genomic data19, all oscine relationships were reconstructed with two protein-coding nuclear genes and node support was not given. Their results also indicated substantial diversication in Australia before dispersal to Asia, but New Guinea was not coded as a separate area, thus these results are not directly comparable to our biogeographic results.

A recent hypothesis for songbird diversication identied New Guinea as an ancestral range for many nodes deep in the songbird tree, especially within the Corvides7. These nodes were dated as Eocene to Oligocene in age, long before New Guinea existed as a substantial landmass, and so proto-Papuan islands were identied as the source of substantial early diversication of the Corvides. Our ancestral range analyses that were naive to the existence of the New Guinea landmass corroborate these biogeographic results, but our analyses that factor in the timing of New Guinea formation support an alternative hypothesis, namely that most early diversication occurred within Australia but many of these Australian lineages disappeared and/or dispersed to New Guinea more recently29,30. We believe that unconstrained biogeographic analyses are biased by extinctions caused by the extensive aridication of Australia (Supplementary Discussion). This aridication, which began in the Miocene but continued throughout the uplift of New Guinea, likely caused many wet-forest-adapted songbirds in Australia to become extinct or restricted to wet forest habitats in New Guinea, as has been documented in other taxa31. For example, southern beech (subgenus Brassospora) that once occurred in formerly widespread Australian wet forests is now absent from Australia,

but it remains a diverse component of the montane wet forests in New Guinea (see Supplementary Discussion for additional examples). Both hypotheses invoke substantial assumptions about diversication processes (for example, origin of a major avian clade and substantial extant diversity on ephemeral islands versus substantial extinction and range shifts); however, our data provide support for the hypothesis that New Guinea played no role in early songbird diversication, but instead was a region of major subsequent diversication of relictual, wet-forest-adapted Australian lineages29.

Several studies have advocated an African origin of the core-Passerides, resulting from direct colonization of Africa from Australasia rather than via Asia10,32. This hypothesis is based on a perceived preponderance of African taxa branching from the base of the passeridan clade, and an early time frame of diversication that allowed a dispersal scenario across now-submerged landmasses and archipelagoes in the Southern Indian Ocean. Our phylogeny, time estimates, and biogeographic reconstruction cast doubt on this dispersal route. After ca. 43 Myr ago, any dispersal across the southern Indian Ocean became increasingly unlikely, because accelerated spreading at the Southeast Indian Ridge sundered Broken Ridge, part of a putative dispersal corridor across the southern Indian Ocean29, and propelled Australia towards Sundaland33. Rather, our proposed diversication chronology supports the more traditional dispersal route through Asia for the Passerides2,4. Our estimates place all diversication and dispersal of the Passerides in the Miocene, contemporary with extensive island formation in Wallacea22 that provided a direct route of colonization from Australasia to Asia.

Specic phylogenetic relationships also cast doubt on direct dispersal to Africa. Previous studies reconstructed Picathartidae as sister to the core-Passerides9,10 or Corvides12. However, we found strong support that Picathartidae is sister to Petroicidae, an Australasian family; uncertainty regarding relationships within Picathartidae render its origin ambiguous15. Thus, Picathartidae provides no evidence for an African origin of core-Passerides, unless we also assume that the ancestor of the Petroicidae re-crossed the southern Indian Ocean to re-colonize Australasia. The African endemic Hyliota has also been cited as evidence of an African origin of the core-Passerides32, yet this inference relied on an unresolved position near the base of the Passerides. We do not recover Hyliota at the base of the core-Passeridesrather it is the sister lineage to the Stenostiridae, a family found in Africa and Asia. Thus, Hyliota also provides no direct evidence of an African origin for core-Passerides.

Our study provides a well-resolved phylogeny of songbirds, claries the relationships of all major lineages, and provides a time frame of evolution that implicates key geological drivers of diversication. Songbirds originated in the Eocene but diversied within an isolated Australian continent in the Oligocene. Tectonic collision and uplift of Wallacea as Australia drifted north allowed dispersal to Asia and proliferation of songbird species across the rest of the world, where most of their diversity now occurs. Although New Guinea presently contains high songbird species richness, our data provide an alternative to the proto-Papuan hypothesis for early songbird diversication7,8 and suggest that New Guinea has, instead, served as an evolutionary refuge for Australian lineages that have diversied more recently within the island29. Substantial songbird diversity in New Guinea likely developed recently in response to Papuan orogeny, consistent with diversication patterns for other groups in New Guinea34 and worldwide35,36.

Methods

Laboratory techniques. We extracted and puried DNA from fresh muscle or liver tissue, or toepad clips from museum specimens using the Qiagen DNeasy Blood and Tissue Kit following the manufacturers protocol. We quantied DNA

4 NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709 ARTICLE

extracts using a Qubit 2.0 Fluorometer, and we sheared 500 ng of DNA of each sample (except for one sample extracted from the toepad of a museum study skin, which was not sheared) to 400600 bp in 50 ml volume using a Covaris S220 sonicator at 175 W peak incident power, 2% duty factor, and 200 cycles per burst for 45 s. We performed 14 reactions of end repair, A-tailing, and adaptor ligation on

Solid Phase Reversible Immobilization beads using Kapa Biosystems Library Prep kits following the procedure of Faircloth et al.24 (described in detail at http://ultraconserved.org

Web End =http://ultraconserved.org ). We ligated universal iTru stubs37 instead of sample-specic adaptors to allow for dual indexing, and added a second AMPure XP bead clean up at 1.0 volume after stub ligation. We incorporated iTru

dual-indexes37 to library fragments using a 17-cycle PCR with NEB Phusion High-Fidelity PCR Master Mix.

We quantied libraries using a Qubit 2.0 Fluorometer and pooled sets of eight samples at equimolar ratios for enrichment. We performed sequence capture and post-enrichment amplication following standard protocols24 using the Mycroarray MYbaits kit for Tetrapods UCE 5K version 1, which targets 5,060 UCE loci. This procedure involved hybridizing biotinylated RNA probes with pooled libraries for 24 h, after which we enriched DNA targets using streptavidin-coated beads. We amplied fragments with a 17-cycle PCR amplication step using NEB Phusion High-Fidelity PCR Master Mix. After post-enrichment amplication, we quantied libraries using an Illumina Eco qPCR System and a commercial library quantication kit (Kapa Biosystems), combined 96 libraries into a 10 mM pool, and sequenced the pooled libraries in a high output, paired-end run of 100 cycles on an Illumina HiSeq 2500 System at the University of Kansas Genome Sequencing Core.

Data assembly. We de-multiplexed raw reads using CASAVA ver. 1.8.2 and trimmed low-quality bases and adaptor sequences from reads using illumiprocessor ver. 1 (https://github.com/faircloth-lab/illumiprocessor

Web End =https://github.com/faircloth-lab/illumiprocessor) which batch processed reads using Scythe (https://github.com/vsbuffalo/scythe

Web End =https://github.com/vsbuffalo/scythe) and Sickle (https://github.com/najoshi/sickle

Web End =https://github.com/ https://github.com/najoshi/sickle

Web End =najoshi/sickle ). We used the Python package PHYLUCE38 for subsequent data processing. We assembled cleaned reads into contigs using Trinity ver. r2013.08.14 (ref. 39) and extracted contigs for each taxon that matched UCE loci. We assembled an incomplete data set containing UCE loci that were present in at least 75% of all 106 taxa for maximum likelihood (ML), Bayesian (BI), and SVDquartets40 analyses. For gene tree-based species tree analyses, we assembled complete data sets consisting of loci common to all taxa being analysed. For each data set, we aligned each locus using MAFFT41, and we trimmed resulting alignments to allow missing nucleotides at the anks of each alignment only if at least 65% of taxa contained data, which is the default option in PHYLUCE. We further trimmed uncertain alignment regions using Gblocks42 with default parameters except for the minimum number of sequences for a ank position in Gblocks, which we set at 65% of taxa.

Protein-coding genes can provide convergent signal and base-composition bias in phylogenomic data sets19. We undertook two approaches to address this potential concern. First, using BEDTools43 we aligned the UCE probes to known chicken protein-coding genes and identied UCE loci whose probe(s) 300 bp of

anking region overlap with these genes. Of the 4,155 UCE loci recovered in the incomplete matrix, only 316 (7.6%) overlapped with chicken protein-coding genes. We assembled a data set without these 316 loci. Second, we recoded the incomplete matrix as purines and pyrimidines (RY-coding). We analysed both data sets with a maximum likelihood approach (Supplementary Fig. 11; see below).

Phylogenetic analyses. We performed maximum likelihood (ML) inference on the concatenated loci of the incomplete and complete data sets using RAxML ver. 8.1.3 (ref. 44) assuming a general time reversible model of rate substitution and gamma-distributed rates among sites. We evaluated node support using 500 rapid bootstrap replicates. We tested for convergence of bootstrap replicates a posteriori using the autoMRE option in RAxML. For BI, we used Exabayes ver. 1.4.2 (ref. 45) with loci partitioned according to evolutionary model (determined with Cloudforest http://github.com/ngcrawford/cloudforest

Web End =http://github.com/ngcrawford/cloudforest ). We unlinked state frequencies, rate heterogeneity among sites, and the substitution matrix among partitions; we linked branch lengths across partitions. Exabayes analyses tended to converge on single topologies with no subsequent topology swaps accepted when using default settings. To more thoroughly search parameter space, we adjusted the temperature of the heated chains (heatFactor), topology proposals (for example, increase parsimonySPR and likeSPR) and parameters for various moves (parsSPRRadius, parsimonyWarp, likesprmaxradius, and likesprwarp). Changes in these settings often represented a trade-off between more thorough tree searching at the expense of greatly increased computation time, especially increasing the likelihood SPR settings. We eventually converged on settings with more liberal parsimony criteria (for example, radius 10,000, increased parsimony proposals (B20%), etc.) and moderate likelihood criteria (for example, radius 10, modest likelihood proposal frequency (B5%), etc.) that prevented getting stuck in local optima while remaining computationally tractable. All Exabayes analyses included multiple independent runs (2 runs, each with one cold and two heated chains); we assessed convergence between runs with the average standard deviation of split frequencies o0.01, plots of likelihood and parameter estimates, the potential scale reduction factor close to one (o1.1), and effective sampling size (ESS) 4200.

To obtain a species tree estimate that did not rely on prior inference of individual gene trees, we used SVDquartets40,46, a method that analyzes quartets of species in a coalescent framework using singular value decomposition of the matrix

of site pattern frequencies and then assembles a species tree from the quartets using a supertree method. This method was developed for analysis of unlinked single nucleotide polymorphism data, but extended to analysis of DNA sequences, especially large, multi-locus data sets. We generated 100 bootstrap replicates of the data using RAxML ver. 8.1.3 (ref. 44). For each bootstrap replicate we generated all quartets using the implementation of SVDquartets in a test version of Paup,ver. 4.0a146 (ref. 47) and then assembled the species tree using the quartet max-cut method48. For gene tree-based coalescent methods for species tree analysis, we performed gene-tree inference and bootstrapping with RAxML ver. 8.1.3 (ref. 44) using the Python package PHYLUCE38. We modied the PHYLUCE scripts to implement multi-locus bootstrapping49 (that is, sampling with replacement of loci and sites), and we generated 500 multi-locus bootstrap replicate sets of gene trees for each data set. On each replicate set of gene trees, we ran four methods using default parameters: STAR and STEAC as implemented in the R package phybase ver. 1.3 (ref. 50), MP-EST ver. 1.4 (ref. 51), and ASTRAL ver. 4.7.7 (ref. 52). ASTRAL is not strictly a coalescent method, but it is statistically consistent with the multispecies coalescent model52, and as such, other authors have included it in the family of coalescent-based gene tree methods53. Command line, Python, and R scripts used to process the data and run species tree analyses are available from https://github.com/carloliveros/uce-scripts

Web End =https://github.com/carloliveros/uce-scripts .

Divergence time estimation. Most previous analyses of passerine diversication2,7,8 used a putative biogeographic event as a calibration for dating (that is, the separation of NZ and Australia at B80 Myr ago to date the node separating

Acanthisitta from all other passerines), but incorporated no geological information in their biogeographic reconstructions. For example, Wallacea and New Guinea are unconstrained and allowed to occur as ancestral areas (distributions) anywhere in the phylogeny, even though these regions formed recently compared with continental areas and the inferred time frame of bird evolution22,23,54. We took the opposite approach for biogeographic analysis, using biogeography independent calibrations derived from fossils for dating and applying geologically informed palaeogeographic models for biogeographic reconstruction.

The fossil record for passerine birds is sparse and many described fossil taxa lack information necessary for use as informative calibration points55. A recent genomic study of birds19 used fossil calibrations from a variety of bird orders to calibrate the avian phylogeny. The resulting Jarvis et al.19 tree sparsely sampled passerines (ve species) but included nodes shared with our sampling, such as the split between Corvides and Passerides, which we included in our dating analysis as a normal distribution with a mean of 21 Myr ago and 95% condence interval of 1726 Myr ago based on their results. Another recent genomic study of birds20 produced a timescale similar to, but slightly older than that of Jarvis et al.19 We also used an alternative calibration (normal distribution with mean 28 Myr ago and 95% condence interval of 1838 Myr ago) from Prum et al.20 for the same node (split between Corvides and Passerides) in separate analyses to test the inuence of these different dates. The dates from Jarvis et al.19 have been criticized as enforcing a lower bound on the phylogeny that was too recent, which might force unrealistically recent dates on the entire phylogeny56. However, further analyses with older lower bounds produced almost identical dates for the passerine nodes57. Because secondary calibrations are a step removed from fossil calibration data, we also included a calibration based on fossils of the oscine Orthonyx kaldowinyeri25 for two reasons. First, multiple fossils of this distinctive taxon have been described, alleviating concerns about potential misidentication. Second, these fossils are the earliest identied crown-oscines (OligoMiocene), and are from the well-characterized Riversleigh World Heritage Area of Australia, which provides broad comparative context for dating the fossils. For this calibration we used a truncated Cauchy distribution58 with minimum age bound of 24 Myr ago and a relatively at distribution (offset P 1.0, scale c 5.0). To constrain the base of the tree, we

applied a maximum divergence of 45 Myr ago when using the Jarvis et al.19 secondary calibration and 55 Myr ago when using the Prum et al.20 secondary calibration. These lower bounds are older than the lower end of the 95% condence interval of dates for the split between oscines and suboscines in the respective papers. We inferred dated phylogenies with the complete UCE data set, a constrained ML topology, and two different sets of calibration points (Orthonyx fossil with either the Jarvis et al.19 or Prum et al.20 estimates for the Corvides-Passerides split and bounds on the base of the tree) using MCMCTree in PAML ver. 4.8 (ref. 59). Divergence times were estimated using the independent rates model in MCMCTree and a birthdeath process with species sampling. We applied the HKY85 GAMMA model of nucleotide substitution with four rate categories,

the most complex model implemented in MCMCTree, but one that accommodates rate variation among nucleotide sites. Inclusion of variable site rates is likely important for age estimation with UCE data because UCEs display a unique pattern of rate variation, with few substitutions in the core UCE region and more substitutions in anking regions. Because of the size of the data matrix, we used approximate likelihood calculations with branch lengths estimated using the baseml programme of PAML.

Ancestral range estimation. Our study contained relatively sparse species sampling over a large clade. To simply code the current distribution of each species included in the analysis might bias reconstructions towards regions with greater sampling. For example, we have sampled extensively in New Guinea and Pacic

NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709

archipelagos, but sparsely in Wallacea. Our choice of clade exemplars would thus be biased towards our sampling regions. To minimize this issue, we considered each species sampled in the phylogeny to be a representative of an entire clade and used two methods for coding areas. First, we coded each terminal based on the total geographic area occupied by the clade it represents. Second, we coded each terminal based on information about the putative origin of its clade from published phylogenetic studies (Supplementary Data 2). We could not reduce most clades to a single area of origin. When a clades origin was ambiguous, we used the clades full distribution, and we excluded areas only occupied by species that are deeply embedded inside the clades of interest.

We included nine biogeographic areas in our analyses (New Guinea (A), New Zealand (B), Australia (C), Wallacea (D), S and SE Asia (including Philippines; E), sub-Saharan Africa (F), New World (G), Palaearctic (including N. Africa; H), and, Madagascar (I)). Ancestral range estimation can be informed by realistic palaeogeographic models based on Earth history60,61. The formation of New Guinea was a critical event in Australasian palaeogeography, and New Guinea has featured prominently in previous biogeographic analysis of oscines7,8. All recent models for the formation of New Guinea indicate that substantial land area only existed after 15 Myr ago23,54, or according to one model only after B5 Myr ago26,34. Thus, ancestral range estimates that recover New Guinea at nodes earlier than the mid-Miocene may not accurately reect palaeogeographic reality. To incorporate palaeogeographic history, we reconstructed ancestral ranges with the constraint that New Guinea was not allowed to occur until 15 Myr ago, as supported by the geologic record. To examine the inuence of this constraint, we also ran all analyses naively (that is, with no geographic constraints).

To estimate ancestral ranges and model the geography of songbird diversication, we tted likelihood versions of the DEC61,62 (dispersal, extinction and cladogenesis) and DEC j models63 to the time-calibrated songbird tree using

BioGeoBEARS ver. 0.2.1 (refs 64,65). Likelihood implementation of DEC (DEC-LIKE) allows two types of anagenetic events: dispersal, or range expansion; and extinction, or range contraction. DEC-LIKE allows two types of cladogenic events. In sympatric events, each daughter species inherits the distribution of the ancestor. This does not imply sympatric speciation per se, only that speciation has occurred within the user-dened area. In cladogenic events, each daughter species inherits separate and non-overlapping portions of the ancestors range62,63. BioGeoBEARS also implements a modied version of the DEC model that includes a founder-event speciation parameter (j, or jump), which allows one daughter species to inherit the entire range of the ancestor, and the other daughter to disperse to a novel geographic area outside of the ancestor range. This DEC j

model may be particularly useful in island systems, where founder-event speciation is expected63. We t DEC-LIKE and DEC-LIKE j under four different

conditions: using total clade distributions or inferred clade origin terminal codings, as well as with and without a constraint on New Guinea emergence at15 Myr ago. For each of these four model conditions, we used the Akaike Information Criterion (AIC) to select if the DEC-LIKE or DEC-LIKE j model

best t the data.We chose DEC/DEC j as the most appropriate biogeographic models a priori

based on their reasonable assumptions. However, to explore whether model choice inuenced biogeographic interpretations, we also considered two alternate model families, DIVA66,67 (dispersal-vicariance analysis) and BAYAREA68. Similar to the DEC models, DIVA and BAYAREA models allow geographic ranges to evolve through time, but the specic types of events permitted are less appropriate when dened areas are large. Specically, DIVA does not allow the sympatric subset area inheritance of the DEC model, and DIVA is more suited to systems where dened areas are too small for divergence within an area. Similarly, BAYAREA does not allow the cladogenic area inheritance of the DEC model, another unrealistic assumption. Nonetheless, we also tted DIVA-LIKE, DIVA-LIKE j,

BAYAREA-LIKE, BAYAREA-LIKE j in BioGEOBEARS. For these models, we

also ran iterations using both total clade distributions and inferred clade origin terminal codings, and we ran iterations with and without a constraint on New Guinea emergence at 15 Myr ago (Supplementary Table 1).

Data availability. Nucleotide alignments are deposited at Dryad (http://dx.doi.org/doi:10.5061/dryad.nf01p

Web End =doi:10.5061/ http://dx.doi.org/doi:10.5061/dryad.nf01p

Web End =dryad.nf01p ), raw sequencing reads at the NCBI SRA (SRA BioSample Nos. SAMN04301695 to SAMN04301800), and UCE nucleotide sequences at NCBI Genbank (Accession Nos. KAAD00000000 to KAEE00000000). Command line, Python, and R scripts are available from https://github.com/carloliveros/uce-scripts

Web End =https://github.com/carloliveros/uce-scripts . All other data are available from the corresponding author upon request.

References

1. Gill, F. B. & Donsker, D. IOC World Bird List (v. 5.2) http://www.worldbirdnames.org

Web End =http://www.worldbirdnames.org (2015).

2. Barker, F. K., Cibois, A., Schikler, P., Feinstein, J. & Cracraft, J. Phylogeny and diversication of the largest avian radiation. Proc. Natl Acad. Sci. USA 101, 1104011045 (2004).

3. Ericson, P. G. P. et al. A Gondwanan origin of passerine birds supported by DNA sequences of the endemic New Zealand wrens. Proc. R. Soc. B Biol. Sci. 269, 235241 (2002).

4. Barker, F. K., Barrowclough, G. F. & Groth, J. G. A phylogenetic hypothesis for passerine birds: taxonomic and biogeographic implications of an analysis of nuclear DNA sequence data. Proc. R. Soc. B Biol. Sci. 269, 295308 (2002).

5. Mayr, G. The age of the crown group of passerine birds and its evolutionary signicancemolecular calibrations versus the fossil record. Syst. Biodivers. 11, 713 (2013).

6. Barker, F. K. in Living Dinosaurs: The Evolutionary History of Modern Birds (eds Dyke, G. & Kaiser, G.) 235256 (John Wiley & Sons, 2011).

7. Jnsson, K. A., Fabre, P.-H., Ricklefs, R. E. & Fjelds, J. Major global radiation of corvoid birds originated in the proto-Papuan archipelago. Proc. Natl Acad. Sci. USA 108, 23282333 (2011).

8. Aggerbeck, M., Fjelds, J., Christidis, L., Fabre, P. H. & Jnsson, K. A. Resolving deep lineage divergences in core corvoid passerine birds supports a proto-Papuan island origin. Mol. Phylogenet. Evol. 70, 272285 (2014).

9. Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K. & Mooers, A. O. The global diversity of birds in space and time. Nature 491, 444448 (2012).

10. Jnsson, K. A. & Fjelds, J. Determining biogeographical patterns of dispersal and diversication in oscine passerine birds in Australia, Southeast Asia and Africa. J. Biogeogr. 33, 11551165 (2006).

11. Ericson, P. G. P., Klopfstein, S., Irestedt, M., Nguyen, J. M. T. & Nylander, J. A A. Dating the diversication of the major lineages of Passeriformes (Aves). BMC Evol. Biol. 14, 8 (2014).

12. Claramunt, S. & Cracraft, J. A new time tree reveals Earth historys imprint on the evolution of modern birds. Sci. Adv. 1, e1501005 (2015).

13. Selvatti, A. P., Gonzaga, L. P. & Russo, C. A. D. M. A Paleogene origin for crown passerines and the diversication of the Oscines in the New World. Mol. Phylogenet. Evol. 88, 115 (2015).

14. Gibb, G. C. et al. New Zealand passerines help clarify the diversication of major songbird lineages during the Oligocene. Genome Biol. Evol. 7, evv196 (2015).

15. Burleigh, J. G., Kimball, R. T. & Braun, E. L. Building the avian tree of life using a large-scale, sparse supermatrix. Mol. Phylogenet. Evol. 84, 5363 (2015).

16. Fregin, S., Haase, M., Olsson, U. & Alstrm, P. New insights into family relationships within the avian superfamily Sylvioidea (Passeriformes) based on seven molecular markers. BMC Evol. Biol. 12, 157 (2012).

17. Alstrom, P. et al. Discovery of a relict lineage and monotypic family of passerine birds. Biol. Lett. 10, 20131067 (2014).

18. Cracraft, J. in The Howard & Moore Complete Checklist of the Birds of the World 4th Edition, Vol. 2 (eds Dickinson, E. & Christidis, L.) xviixvi (Aves Press, Eastbourne, UK, 2014).

19. Jarvis, E. D. et al. Whole-genome analyses resolve early branches in the tree of life of modern birds. Science 346, 13201331 (2014).

20. Prum, R. O. et al. A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526, 569573 (2015).

21. Cracraft, J. Avian evolution, Gondwana biogeography and the Cretaceous-Tertiary mass extinction event. Proc. Biol. Sci. 268, 459469 (2001).22. Hall, R. The palaeogeography of Sundaland and Wallacea since the Late Jurassic. J. Limnol. 72, 117 (2013).

23. van Ufford, A. Q. & Cloos, M. Cenozoic tectonics of New Guinea. Am. Assoc. Pet. Geol. Bull 89, 119140 (2005).

24. Faircloth, B. C. et al. Ultraconserved elements anchor thousands of genetic markers spanning multiple evolutionary timescales. Syst. Biol. 61, 717726 (2012).

25. Nguyen, J. M. T., Boles, W. E., Worthy, T. H., Hand, S. J. & Archer, M. New specimens of the logrunner Orthonyx kaldowinyeri (Passeriformes: Orthonychidae) fromthe Oligo-Miocene of Australia. Alcheringa 38, 245255 (2014).

26. Hill, K. C. & Hall, R. Mesozoic-Cenozoic evolution of Australias New Guinea margin in a west Pacic context. Geol. Soc. Am. Spec. Pap. 372, 265290 (2003).

27. Hall, R. in Biogeography and Geological Evolution of SE Asia (eds Hall, R. & Holloway, J. D.) 99131 (Backhuys Publishers, 1998).

28. Hall, R. Late Jurassic-Cenozoic reconstructions of the Indonesian region and the Indian Ocean. Tectonophysics 570-571, 141 (2012).

29. Schodde, R. in Evolution and Biogeography of Australasian Vertebrates (eds Merrick, J. R., Archer, M., Hickey, G. M. & Lee, M. S. Y.) 413458 (Australian Scientic Publishing, 2006).

30. Schodde, R. & Christidis, L. Relicts from Tertiary Australasia: undescribed families and subfamilies of songbirds (Passeriformes) and their zoogeographic signal. Zootaxa 3786, 501522 (2014).

31. Byrne, M. et al. Decline of a biome: evolution, contraction, fragmentation, extinction and invasion of the Australian mesic zone biota. J. Biogeogr. 38, 16351656 (2011).

32. Fuchs, J., Fjelds, J., Bowie, R. C. K., Voelker, G. & Pasquet, E. The African warbler genus Hyliota as a lost lineage in the Oscine songbird tree: Molecular support for an African origin of the Passerida. Mol. Phylogenet. Evol. 39, 186197 (2006).

6 NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms12709 ARTICLE

33. Mller, R. D., Gaina, C. & Clark, S. in Billion-Year Earth History of Australia and Neighbours in Gondwanaland (eds Veevers, J. J. & Morgan, P.) 1828 (Gemoc Press, 2000).

34. Toussaint, E. F. A. et al. The towering orogeny of New Guinea as a trigger for arthropod megadiversity. Nat. Commun. 5, 4001 (2014).

35. Chaves, J. A., Weir, J. T. & Smith, T. B. Diversication in Adelomyia hummingbirds follows Andean uplift. Mol. Ecol. 20, 45644576 (2011).

36. Ribas, C. C., Moyle, R. G., Miyaki, C. Y. & Cracraft, J. The assembly of montane biotas: linking Andean tectonics and climatic oscillations to independent regimes of diversication in Pionus parrots. Proc. R. Soc. Lond. 274, 23992408 (2007).

37. Glenn, T. C. et al. Adapterama I: universal stubs and primers for thousands of dual-indexed Illumina libraries (iTru & iNext), Preprint at http://biorxiv.org/ content/early/2016/06/15/049114 (2016).

38. Faircloth, B. C. PHYLUCE is a software package for the analysis of conserved genomic loci. Bioinformatics 32, 786788 (2015).

39. Grabherr, M. G. et al. Full-length transcriptome assembly from RNA-Seq data without a reference genome. Nat. Biotechnol. 29, 644652 (2011).

40. Chifman, J. & Kubatko, L. Quartet inference from SNP data under the coalescent model. Bioinformatics 30, 33173324 (2014).

41. Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol. Biol. Evol. 30, 772780 (2013).

42. Castresana, J. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol. Biol. Evol. 17, 540552 (2000).

43. Quinlan, A. R. & Hall, I. M. BEDTools: a exible suite of utilities for comparing genomic features. Bioinformatics 26, 841842 (2010).

44. Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 13121313 (2014).

45. Aberer, A. J., Kobert, K. & Stamatakis, A. ExaBayes: massively parallel bayesian tree inference for the whole-genome era. Mol. Biol. Evol. 31, 25532556 (2014).

46. Chifman, J. & Kubatko, L. Identiability of the unrooted species tree topology under the coalescent model with time-reversible substitution processes, site-specic rate variation, and invariable sites. J. Theor. Biol. 374, 3547 (2015).

47. Swofford, D. L. PAUP*. Phylogenetic analysis using parsimony (*and Other Methods) (2003).

48. Snir, S. & Rao, S. Quartet MaxCut: a fast algorithm for amalgamating quartet trees. Mol. Phylogenet. Evol. 62, 18 (2012).

49. Seo, T. K. Calculating bootstrap probabilities of phylogeny using multilocus sequence data. Mol. Biol. Evol. 25, 960971 (2008).

50. Liu, L. & Yu, L. Phybase: an R package for species tree analysis. Bioinformatics 26, 962963 (2010).

51. Liu, L., Yu, L. & Edwards, S. V. A maximum pseudo-likelihood approach for estimating species trees under the coalescent model. BMC Evol. Biol. 10, 302 (2010).

52. Mirarab, S. et al. ASTRAL: genome-scale coalescent-based species tree estimation. Bioinformatics 30, i541i548 (2014).

53. Liu, L., Xi, Z., Wu, S., Davis, C. & Edwards, S. V. Estimating phylogenetic trees from genome-scale data. Ann. N. Y. Acad. Sci. 1360, 3653 (2015).54. Cloos, M., Weiland, R. J., Warren, P. Q. & McMahon, T. P. Collisional delamination in New Guinea: The geotectonics of subducting slab breakoff. Geol. Soc. Am. Spec. Pap. 400, 151 (2005).

55. Parham, J. F. et al. Best practices for justifying fossil calibrations. Syst. Biol. 61, 346359 (2012).

56. Mitchell, K. J., Cooper, A. & Phillips, M. J. Comment on Whole-genome analyses resolve early branches in the tree of life of modern birds. Science 349, 1460 (2015).

57. Cracraft, J. et al. Response to Comment on Whole-genome analyses resolve early branches in the tree of life of modern birds . Science 349, 1460 (2015).

58. Inoue, J., Donoghue, P. C. J. & Yang, Z. The impact of the representation of fossil calibrations on bayesian estimation of species divergence times. Syst. Biol. 59, 7489 (2010).

59. Yang, Z. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol. Evol. 24, 15861591 (2007).

60. Ree, R. H., Moore, B. R., Webb, C. O. & Donoghue, M. J. A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59, 22992311 (2005).

61. Ree, R. H. & Smith, S. A. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 414 (2008).

62. Clark, J. R. et al. A comparative study in ancestral range reconstruction methods: retracing the uncertain histories of insular lineages. Syst. Biol. 57, 693707 (2008).

63. Matzke, N. J. Model selection in historical biogeography reveals that founder-event speciation is a crucial process in island clades. Syst. Biol. 63, 951970 (2014).

64. Matzke, N. J. BioGeoBEARS: BioGeography with Bayesian (and Likelihood) Evolutionary Analysis in R Scripts (University of California, Berkeley, Berkeley, CA, 2013).

65. Matzke, N. J. Probabilistic Historical Biogeography: New Models for Founder-event Speciation, Imperfect Detection, and Fossils Allow Improved Accuracy and Model-Testing (University of California, 2013).

66. Ronquist, F. Dispersal-vicariance analysis: a new approach to the quantication of historical biogeography. Syst. Biol. 46, 195203 (1997).

67. Yu, Y., Harris, A. J. & He, X. S-DIVA (Statistical Dispersal-Vicariance Analysis): a tool for inferring biogeographic histories. Mol. Phylogenet. Evol. 56, 848850 (2010).

68. Landis, M. J., Matzke, N. J., Moore, B. R. & Huelsenbeck, J. P. Bayesian analysis of biogeography when the number of areas is large. Syst. Biol. 62, 789804 (2013).

Acknowledgements

We thank the curators, staff, and eld collectors at the following institutions for tissue samples included in this project: Burke Museum, University of Washington; Field Museum of Natural History; Peabody Museum, Yale University; American Museumof Natural History. Mike Alfaro, Jake Esselstyn, Robb Brumeld and Kris Krishtalka provided comments on the manuscript. This work was funded by a Strategic Initiatives Grant from the University of Kansas Research Investment Council to RGM and RMB, grants to RGM (NSF DEB-1241181 and DEB-1146345), and startup funds to BCF from Louisiana State University. The COBRE Genome Sequencing Core Laboratory, funded by National Institutes of Health (NIH) award number P20GM103638, provided laboratory facilities and services. We thank the KU Advanced Computing Facility (partially funded by NSF grant CNS 1337899 to A.T. Peterson) for the use of their facilities and portions of this research were conducted with high-performance computing resources provided by Louisiana State University (http://www.hpc.lsu.edu

Web End =http://www.hpc.lsu.edu).

Author contributions

R.G.M., M.J.A., C.H.O., P.A.H. and B.W.B. conceived of and designed the research. C.H.O., M.J.A., P.A.H. and J.D.M. collected the data. B.C.F. developed probe sets and the informatics pipeline. R.G.M. and C.H.O. performed phylogenetic analyses. P.A.H. performed biogeographic analyses and estimated mitochondrial substitution rates. R.G.M. performed dating analyses. B.W.B. contributed geological interpretations, palaeoecological summaries, and biogeographic synthesis. R.G.M. wrote the manuscript with contributions from all authors.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Moyle, R. G. et al. Tectonic collision and upliftof Wallacea triggered the global songbird radiation. Nat. Commun. 7:12709 doi: 10.1038/ncomms12709 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:12709 | DOI: 10.1038/ncomms12709 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7