Report

Phylogenomics and Morphology of Extinct Paleognaths Reveal the Origin and Evolution of the Ratites Highlights d

Nuclear genome fragments from extinct elephant bird species were recovered

d

A stable phylogenomic time tree for the Palaeognathae was inferred

d

A Laurasian origin of Palaeognathae is supported by molecular and morphological data

d

Ancestral paleognaths had small body size (3.5–5 kg) and probably were volant

Authors Takahiro Yonezawa, Takahiro Segawa, Hiroshi Mori, ..., Fumihito Akishinonomiya, Eske Willerslev, Masami Hasegawa

Correspondence [email protected] (E.W.), [email protected] (M.H.)

In Brief Yonezawa et al. recover nuclear genome fragments from extinct elephant birds and reconstruct a stable phylogenomic time tree for the Palaeognathae. Their tree based on morphological characters places the fossil paleognaths from the Northern Hemisphere as the basal lineages. This evidence suggests a Laurasian origin of Palaeognathae.

Accession Numbers AP014697 AP014698

Yonezawa et al., 2017, Current Biology 27, 68–77 January 9, 2017 ª 2017 Elsevier Ltd. http://dx.doi.org/10.1016/j.cub.2016.10.029

Current Biology

Report Phylogenomics and Morphology of Extinct Paleognaths Reveal the Origin and Evolution of the Ratites Takahiro Yonezawa,1,2,3,22 Takahiro Segawa,4,5,22 Hiroshi Mori,6,22 Paula F. Campos,7,8 Yuichi Hongoh,9 Hideki Endo,10 Ayumi Akiyoshi,5 Naoki Kohno,11,12 Shin Nishida,13 Jiaqi Wu,1,14 Haofei Jin,1 Jun Adachi,2,3 Hirohisa Kishino,14 Ken Kurokawa,6 Yoshifumi Nogi,5 Hideyuki Tanabe,3 Harutaka Mukoyama,15 Kunio Yoshida,10 Armand Rasoamiaramanana,16 Satoshi Yamagishi,17 Yoshihiro Hayashi,11,17 Akira Yoshida,18,19 Hiroko Koike,20 Fumihito Akishinonomiya,10,17,21 Eske Willerslev,7,* and Masami Hasegawa1,2,3,23,* 1School

of Life Sciences, Fudan University, SongHu Road 2005, Shanghai 200438, China Institute of Statistical Mathematics, Midori-cho 10-3, Tachikawa City, Tokyo 190-8562, Japan 3School of Advanced Sciences, SOKENDAI (The Graduate University for Advanced Studies), Hayama-cho, Kanagawa 240-0193, Japan 4Center for Life Science Research, University of Yamanashi, 1110 Shimokato, Chuo, Yamanashi 409-3898, Japan 5National Institute of Polar Research, Midori-cho 10-3, Tachikawa City, Tokyo 190-8562, Japan 6Department of Biological Information, Tokyo Institute of Technology, Ookayama 2-12-1, Meguro-ku, Tokyo 152-8550, Japan 7Centre for GeoGenetics, Natural History Museum of Denmark, University of Copenhagen, Øster Voldgade 5–7, 1350 Copenhagen K, Denmark 8CIMAR/CIIMAR, Centro Interdisciplinar de Investigac ¸ a˜o Marinha e Ambiental, Universidade do Porto, Rua dos Bragas, 289 Porto, 4050-123 Portugal 9Department of Biological Sciences, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Ookayama 2-12-1, Meguro-ku, Tokyo 152-8550, Japan 10The University Museum, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-0033, Japan 11National Museum of Nature and Science, Amakubo 4-1-1, Tsukuba City, Ibaraki 305-0005, Japan 12Graduate School of Life and Environmental Sciences, University of Tsukuba, Tennoudai 1-1-1, Tsukuba City, Ibaraki, 305-8572, Japan 13Biology, Science Education, Faculty of Education and Culture, University of Miyazaki, Gakuen-Kibanadai-Nishi 1-1, Miyazaki, Miyazaki 889-2192, Japan 14Graduate School of Agricultural and Life Sciences, University of Tokyo, Yayoi 1-1-1, Bunkyo-ku, Tokyo 113-8657, Japan 2The

(Affiliations continued on next page)

SUMMARY

The Palaeognathae comprise the flightless ratites and the volant tinamous, and together with the Neognathae constitute the extant members of class Aves. It is commonly believed that Palaeognathae originated in Gondwana since most of the living species are found in the Southern Hemisphere [1–3]. However, this hypothesis has been questioned because the fossil paleognaths are mostly from the Northern Hemisphere in their earliest time (Paleocene) and possessed many putative ancestral characters [4]. Uncertainties regarding the origin and evolution of Palaeognathae stem from the difficulty in estimating their divergence times [1, 2] and their remarkable morphological convergence. Here, we recovered nuclear genome fragments from extinct elephant birds, which enabled us to reconstruct a reliable phylogenomic time tree for the Palaeognathae. Based on the tree, we identified homoplasies in morphological traits of paleognaths and reconstructed their morphologybased phylogeny including fossil species without molecular data. In contrast to the prevailing theories, the fossil paleognaths from the Northern Hemisphere 68 Current Biology 27, 68–77, January 9, 2017 ª 2017 Elsevier Ltd.

were placed as the basal lineages. Combined with our stable divergence time estimates that enabled a valid argument regarding the correlation with geological events, we propose a new evolutionary scenario that contradicts the traditional view. The ancestral Palaeognathae were volant, as estimated from their molecular evolutionary rates, and originated during the Late Cretaceous in the Northern Hemisphere. They migrated to the Southern Hemisphere and speciated explosively around the Cretaceous-Paleogene boundary. They then extended their distribution to the Gondwana-derived landmasses, such as New Zealand and Madagascar, by overseas dispersal. Gigantism subsequently occurred independently on each landmass. RESULTS AND DISCUSSION Phylogenomic Time Tree Despite enthusiastic investigation and debate, some fundamental questions about the Palaeognathae (volant tinamous and flightless ratites including ostriches, cassowaries, and rheas) such as their geographical origin, the main driving force for the establishment of their current geographic distribution

15Faculty

of Vetrinary Medicine, Nippon Veterinary Science and Life Science University, Kyonancho 1-7-1, Musashino City, Tokyo 180-8602, Japan 16Department of Biological Anthropology and Paleontology, Faculty of Science, University of Antananarivo, BP 906, Antananarivo 101, Madagascar 17Yamashina Institute for Ornithology, Konoyama 115, Abiko City, Chiba 270-1145, Japan 18Tokyo University of Information Sciences, Onaridai 4-1, Wakaba-ku, Chiba City, Chiba, 265-8501, Japan 19The Research Institute of Evolutionary Biology, Kamiyoga 2-4-28, Setagaya-ku, Tokyo, 158-0098, Japan 20The Kyushu University Museum, Hakozaki 6-10-1, Higashi-ku, Fukuoka City, Fukuoka 812-8581, Japan 21Tokyo University of Agriculture, 1737 Funako, Atsugi-shi, Kanagawa 243-0034, Japan 22Co-first author 23Lead Contact *Correspondence: [email protected] (E.W.), [email protected] (M.H.) http://dx.doi.org/10.1016/j.cub.2016.10.029

(by the breakup of Gondwana or by ‘‘overseas sweepstakes dispersal’’), and the evolutionary process of their gigantism remain unanswered. Although the phylogenetic relationships among the paleognaths have been gradually resolved based on paleontological [2–5] and molecular evidence [1, 2, 6], the geographic origin of the clade and evolutionary processes remain unclear. Conflicting hypotheses proposed in previous studies [1–3] indicate that the branching pattern of the extant paleognath tree itself cannot fully resolve the uncertainty about these processes. In addition, sparse taxon sampling is inevitable because Palaeognathae is an old clade with only a small number of extant members. It is therefore expected that inclusion of extinct species such as moas and elephant birds in the analysis would contribute to a clearer picture of the evolution of the Palaeognathae [1, 2, 7, 8]. Previous divergence time estimates based on mitochondrial genomes or multiple nuclear genes, even including the extinct ratites such as moas or elephant birds, were strongly dependent on taxon sampling, in particular inclusion or exclusion of the tinamous, which have long branches [2]. Here we describe the extraction of DNA from sub-fossil bones of elephant birds (Aepyornis maximus and Mullerornis sp.; Table S1, Supplemental Experimental Procedures) found in Madagascar and the successful recovery of nuclear genome fragments with up to 73,716 nucleotide sites available for phylogenetic analysis, as well as nearly complete mitochondrial genomes of up to 15,381 base pairs. Based on concatenated sequences of nuclear and mitochondrial genes, we reconstructed a phylogenomic time tree (Figure 1A; Supplemental Experimental Procedures). The tree topology (Figure S1) was generally consistent with that of previous studies [1, 2, 6, 7], and all nodes of the paleognaths had high support values (>80%; nuc+mt data), except for the position of the rhea. Both the mitochondrial and nuclear data consistently supported the sister-group relationship between the elephant birds and the kiwis, in agreement with Mitchell et al. [2]. This result might suggest that an extremely large egg relative to body size is a shared derived or synapomorphic character of the elephant birds and the kiwis (Figure 2A). Kiwi eggs weigh 300–500 g, and it has been estimated that Aepyornis species produced eggs of more than 9 kg [9] (see also DataBaseFig1 at http://aepyornis.paleogenome.jp/). The morphometric similarity in the coxa between these two groups, as reported previously [9], is also regarded as a synapomorphic character related to the large egg size. The larger ratio of length to maximum width of the pre-acetabular area in relation to the

whole coxa suggests that the large egg of the elephant bird was placed not only in the posterior space of the pelvic cavity but also in the anterior space like the extant kiwi. Our phylogenomic tree based on the super-matrix with very long sequence data of 871,449 sites across all taxa (sequence lengths varies among taxa because of the missing data) provided stable divergence time estimates. The divergence between the Palaeognathae and the Neognathae was 110.0 million years ago (mya) (95% confidence interval [CI]: 104.7–115.5 mya), that between the ostrich and the other paleognaths was 79.6 mya (95% CI: 76.5–82.6), and the major lineages of the Palaeognathae successively diverged 70.6–62.0 mya. These estimates were little influenced by the taxon sampling (Figure 1B; Figure S2; Table S3). The emergence of the crown Aves (Neornithes) was estimated to be 104.7–115.5 mya, and it predated the times of the stem groups of the crown Aves among Ornithurae such as the ichthyornithids (86.5 mya) and hesperornithids (100.5 mya). For evaluating the stability of our estimates, we also used the ichthyornithids or hesperornithids as maximum constraints of the crown Aves for the divergence time estimations. Since the soft boundary method [10] was applied in this study, our estimate grossly predated these maximum constraints, and these constraints gave only a limited effect on the estimate (Figure 1B; Figure S2; Table S3). The genome-scale data yield stable time estimates irrespective of the assumption of the prior distributions including fossil calibrations, taxon sampling, and so on (Supplemental Experimental Procedures). Furthermore, the confidence intervals were considerably smaller than those in the previous studies [1, 2], but our estimated divergence times were not fully consistent with those previously reported [1, 2, 11, 12]. Our estimates are essentially close to those of Jarvis et al. [12]; the first splits within paleognaths, neognaths, and neoaves are especially close. However, there are some differences for the first splits within the extant Aves (
100 mya [12] and
110 mya in this study) and within the Galloanseres (
66 mya [12] and
75 mya in this study). Moreover, the estimates by Prum et al. [11] are grossly younger than those in this study. For example, the first split within crown Aves is
73 mya in [11] and
110 mya in this study, and that within the paleognaths is
51 mya in [11] and
80 mya in this study. It seems that these differences stem from the setting of the fossil calibrations. Since this study focuses on the divergence times of Palaeogathae, we gave strong constraints on the Neognathae, which are independent of the primary focus of this study. Conversely, Jarvis et al. [12] Current Biology 27, 68–77, January 9, 2017 69

A

Quaternary

Triassic Late

Cretaceous

Jurassic Early

M.

Late

Early

Late

Paleogene

Neogene

Pal. Eocene Oli.

Mio.

B Pleistocene Pliocene

First split within Crown Palaeognathae

First split within Crown Aves

Divergence Times (mya) Figure 1. Divergence Times of Aves as Inferred from Nuclear and Mitochondrial Genome Data of 873 kbp (A) Divergence times were estimated using the Bayesian relaxed clock method. Tree topology was fixed in advance based on the phylogenomic tree (Figure S1). Nodal numbers indicate the divergence times (in mya), and nodal bars indicate 95% highest posterior densities. The nodes with gray error bars indicate the calibrated nodes (Table S2). The node with the green error bar and the node with the orange error bar indicate the first splits within the crown Aves and within the crown Palaeognathae, respectively. Detailed information of the Aepyornis maximums and Mullerornis used in this study are summarized in Table S1. (B) Stability of divergence time estimates with respect to taxon sampling and fossil calibrations. Posterior means and 95% confidence intervals (indicated by error bars) of the first splits within the crown Aves (in green outline) and within the crown Palaeognathae (in orange outline) are shown, respectively (Figure S2; Table S3). ‘‘No calibrations’’ means that the root age of the crown Aves has no fossil constraint. ‘‘Ichthyornis (86.5 mya) as maximum bound’’ means the root age of the crown Aves was assumed to be younger than 86.5 mya. ‘‘Enaliornis (100.5 mya) as maximum bound’’ means the root age of the crown Aves was assumed to be younger than 100.5 mya. ‘‘Excl. Tinamou’’ indicates tinamous were excluded from the analyses. ‘‘Excl. Neognaths’’ indicates neognaths were excluded from the analyses. ‘‘Aves’’ indicates the non-avian outgroup (reptiles) was excluded from the analyses. ‘‘All’’ indicates all taxa (ratites, tinamous, neognaths, and reptiles) were involved in the analyses. See also Figure S1, Figure S2, Table S1, Table S2, and Table S3.

and Prum et al. [11] carefully scrutinized and evaluated the reliability of fossil records and used multiple fossil records within the Neognathae only for the minimum boundaries. They gave only one maximum boundary at the root within Aves (and one minimum boundary at the root as well). Since it is generally difficult to set the maximum boundary from the fossil record, especially when long ghost lineages exist [13], their calibrations look conservative in terms of the paleontological point of view. However, as recent studies [12, 14, 15] have suggested, if there is only one maximum calibration, the estimated times are highly dependent on it. This tendency was remarkable especially when the outgroup was excluded, as was done by Prum et al. [11], probably because the exact place of the root cannot be identified in the tree (Supplemental Experimental Procedures and DataBase70 Current Biology 27, 68–77, January 9, 2017

Fig3). The position of the Aves root seems especially difficult to identify, since the average branch length of Palaeognathae was grossly shorter than that of Neognathae. If the strictest constraint was assumed on such an unidentifiable root, the root position on the branch connecting Palaeognathae and Neognathae could not go backward (older age), but instead, this branch pushed the root of Palaeognathae to a younger age. Since the taxon sampling from Palaeognathae was sparse if no calibration was given within Palaeognathae, the impact of the rooting problem on the time estimation was remarkable. In such a situation, the root age of Aves as well as the divergence times within Palaeognathae will be grossly underestimated. For this reason, we applied the following approach for the fossil calibration. Since we focus on the divergence times within Palaeognathae in this study, the constraints within Palaeognathae

A

B

Figure 2. Ancestral States Reconstruction by Molecular Evolutionary Rates (A) Correlation between body mass (BM) and egg weight (EW) among Palaeognathae. The regression line (log10EW = 0.669 3 log10BM 0.2174: R2 = 0.983) was estimated from extant Palaeognathae, excluding kiwis. Gray dashed lines indicate 95% CI (Tables S4 and S5). Kiwis and Aepyornis are outliers (orange dots). Body mass and egg weight of the common ancestor of kiwis and Aepyornis (red dot) was estimated as described in the main text; it is suggested that the common ancestor was also an outlier. (B) Comparison of mitochondrial substitution rates among Aves (Table S5). The boxplots indicate the medians, the lower and upper quartiles, and the minimum and the maximum values. The dots on the left side of each boxplot indicate estimated mitochondrial substitution rates of respective branches. There are statistically significant differences between the substitution rates of the flightless birds (ratites and penguins) and those of the volant birds, and between those of the flightless birds and the ancestral Palaeognathae. However, there was no significant difference between the substitution rates of the volant birds and the ancestral Palaeognathae. **p < 0.01. (C) Correlation between avian body weight and mitochondrial substitution rate. The regression line as inferred from the extant crown Aves (see taxonomy at upper right for explanation of dot colors) is log10BM = 0.0208x + 4.2127 (R2 = 0.3636) (Table S5). The estimated mitochondrial substitution rates of the ancestral paleognaths (black dots) were plotted on this regression line. The width of the vertical green shaded bar corresponds to the range of the estimated mitochondrial substitution rates in 3rd codon positions of the ancestral paleognaths. The height of the horizontal blue shaded bar indicates the estimated body masses based on the regression line. The body masses of the ancestral paleognaths predicted by this procedure are 1,500–2,800 g. (A) and (C) are shown in detail in DataBaseFig1 and DataBaseFig2 at http://aepyornis.paleogenome.jp/, respectively. Illustrations by Takashi Oda. See also Tables S4 and S5.

C

Current Biology 27, 68–77, January 9, 2017 71

should be minimal. Therefore, we gave strong constraints mainly within Neognathae. Although the first appearance of the fossil in a particular lineage does not necessarily indicate the real divergence times [13], we used the oldest known fossils of crown groups as the approximations of the emergence times of the crown groups. Accordingly, our time estimation method assumes strong constraints to several particular internal nodes. However, as seen in DataBaseFig3C, our method actually gives only minor effect to other internal nodes, especially the first split within Aves, as well as the splits within Palaeognathae, compared with previous studies [11, 12]. Our time estimates are stable irrespective of the taxon sampling and fossil calibrations, indicating that our estimation does not depend very much on a particular assumption. Recently, Claramunt and Cracraft [18] developed a novel method to provide a prior distribution on an internal node by describing the likelihood of the age of the oldest fossil, the number of fossil occurrences, and the time span encompassing those occurrences. Although the impact of the non-uniform prior distribution of the internal nodes was demonstrated [10], there was no objective method to determine the shapes of the prior distribution. Therefore, Claramunt and Cracraft’s method to shape the prior distribution is epochal and has a potential deep impact in this field. However, even in such a case, the rooting issue discussed above seems inevitable. Our estimates and Claramunt and Cracraft’s estimates based on genome data are extremely close for the first splits within Neognathae and within Neoaves, but there are some differences for the first splits within the crown Aves as well as within Palaeognathae. These differences may have been caused by the rooting issue. Ancestral States Reconstruction It has been suggested that gigantism in the ratites evolved independently multiple times [6], and homoplasy (convergence or parallelism) makes it difficult to reconstruct the phylogenetic tree and ancestral states based only on morphological data [19]. Therefore, we took two independent approaches insusceptible to the morphological homoplasies for reconstructing the ancestral states of the paleognaths. The first approach is to take account of the correlation between the evolution of morphological traits and molecular substitution rates (Supplemental Experimental Procedures). The stable divergence time estimates enabled us to accurately calculate the molecular substitution rates. The nucleotide substitution rates of mitochondrial genes were significantly higher in the volant birds than in the flightless birds (Figure 2B). As hypothesized in a previous study [20], this difference is probably attributable to the larger amount of free radicals generated by the higher metabolic rate of the volant birds. Interestingly, the estimated mitochondrial substitution rates in the ancestral paleognaths were much higher than those of the extant flightless ratites and were similar to those of modern volant birds (Figure 2B). We also identified a negative correlation between the mitochondrial substitution rate and body size among Aves (Figure 2C), as previously reported [21]. Taking account of this correlation [22], the ancestral body weights of the Palaeognathae were estimated to be 3.8–5.5 kg based on our time tree (Table S5). This estimate is within the range of the volant birds (Figure 2C) (see also Maynard Smith [23], who suggested that flying birds could not exceed
15 kg 72 Current Biology 27, 68–77, January 9, 2017

in weight based on the calculation of the power of the muscle necessary for flight as a function of body weight). Although indirect, it suggests that the ancestral paleognaths were able to fly. The second approach is to use morphological characters with no homoplasy (Supplemental Experimental Procedures). The tree topology with high nodal support values obtained in this study provided a good opportunity to identify and exclude the characters with homoplasy, putatively caused by convergence, from the morphological traits of the extant paleognaths. Finally, 34 characters were selected from the matrix used in previous studies [2, 16] to reconstruct a morphology-based maximum-likelihood tree of the representatives of the Palaeognathae, including two fossil species, Lithornis sp. and Emuarius sp. (Figure 3A). The extinct lithornithids, such as Lithornis, Paracathartes, and Pseudocrypturus, inhabited the Northern Hemisphere from the Late Paleocene to the Middle Eocene and have been considered to be volant birds based on their morphology [24]. Although it is widely accepted that the lithornithids are the closest relatives of the extant volant tinamous [2, 3], our phylogenetic analyses strongly supported a basal position of Lithornis sp. among the Palaeognathae, together with the ostrich (bootstrap value 92%; Figure 3A). This result is congruent with the above estimate based on the molecular evolutionary rate that the ancestral paleognaths would have been volant birds. The phylogenetic positions of other fossil paleognaths (Palaeotis sp., Paracathartes sp., Pseudocrypturus sp., and Diogenornis sp.) were also inferred, but with only 10 non-homoplasy morphological characters selected from those used by Houde [5] (Figure 3B) and with 92 non-homoplasy morphological characters selected from Worthy et al. [17] (Figure 3C). As a result, all of the fossils from the Northern Hemisphere (Lithornis, Palaeotis, Paracathartes, and Pseudocrypturus) were placed in basal positions of the Palaeognathae. These phylogenetic placements are in agreement with the Lithornis-cohort hypothesis proposed by Houde [4, 5], who implied based on morphological arguments that the ancestors of the ratites were volant birds distributed in the Northern Hemisphere. Although they are paraphyletic, hereafter we refer to these species as the Northern Hemispheric paleognaths. While this manuscript was under revision, we learned that Nesbitt and Clarke [25] also reported the basal positions of the Lithornithidae and Palaeotis among the Palaeognathae based on morphological data. They used the molecular tree topologies of the extant paleognaths as the constraint. In this process, they implicitly assumed the monophyly of the extant paleognaths. Although our backbone method does not assume the monophyly of the extant paleognaths, both Nesbitt and Clarke [25] and our study obtained consistent results. In contrast, all of the fossils from the Southern Hemisphere (Diogenornis and Emuarius) were nested within the Notopalaeognathae (tinamous, rheas, emus, cassowaries, and kiwis [26]) (Figures 3A–3D). The appropriateness of the phylogenetic positions of fossil species can also be evaluated by our stable time estimates. If the Lithornithidae are phylogenetically close to the tinamou as suggested by previous studies [2, 16], the divergence times between the moa and the tinamou should be earlier than the oldest fossil record of the Lithornithidae (the oldest reliable fossil of Lithornithidae is 60.6–61.57 mya [12], and the oldest putative Lithornithidae based on the isolated incomplete bone that is

A

B

C

D

Figure 3. Phylogenetic Positions of Fossil Palaeognathae and Evolution of Their Geographic Distribution (A) Maximum-likelihood tree as inferred from the morphologic data matrix originally summarized by Mayr [16]. The thick black branches indicate the backbone of the constrained tree. The phylogenetic positions of the fossil species (indicated y) without molecular data were examined in these constraints. Clades indicated by red branches consist of fossil species only. Asterisks indicate species distributed in the Southern Hemisphere. The nodal numbers indicate ML bootstrap probability/Bayesian posterior probability. (B) Maximum-likelihood trees as inferred from the reconstructed morphologic data matrix from Houde [5]. Figure conventions are as in (A). (legend continued on next page)

Current Biology 27, 68–77, January 9, 2017 73

mya 110

E. Cretaceous

100

90

80

Late Cretaceous

70

60

50

Paleo.

40

Pseudocrypturus

Lithornis cohort

30

Eocene

20

Oligo.

10

0

Miocene

†

Pliocene

Pleistocene

Paracathartes † Lithornis † Palaeotis † Diogenornis †

Struthio camelus

Struthionidae Rheidae Tinamidae

Dinornithidae

Casuariidae

Pterocnemia pennata Rhea americana Tinamus major Crypturellus Eudromia elegans Pachyornis australis Emeus crassus Anomalopteryx didiformis Megalapteryx didinus Dinornis giganteus Casuarius bennetti Casuarius casuarius

Emuarius †

Dromaiidae Apterygiadae Aepyornithidae

Dromaius novaehollandiae Apteryx haastii Apteryx owenii Apteryx australis mantelli Mullerornis sp. Aepyornis maximus

Establishment of Antarctic circumpolar current

66mya

56mya

23mya

Figure 4. Palaeognathae Genomic Time Tree and Body Size Size of a circle on a given node is proportional to estimated body weight (Table S5). The phylogenetic positions of fossil species are indicated by dashed lines (their divergence times are arbitrary). The colors of the branches indicate geographic distribution. Hypothetical ancestral distribution and fossil records of the paleognaths are indicated on the paleomaps (triangles indicate volant birds; squares indicate ratites; detailed information on fossil records is shown in DataBaseTable1 and DataBaseFig4 (http://aepyornis.paleogenome.jp/). The thick red vertical line indicates the K–Pg boundary. The paleomaps were downloaded from http://www.odsn.de/odsn/services/paleomap/paleomap.html. Illustrations by Takashi Oda. See also Figure S3 and Table S5.

argued to be insufficiently diagnosable is
66 mya [24, 27]). The divergence time between the moa and the tinamou was estimated to be 50.2–56.1 mya, which is much younger than the oldest lithornithids. Even when this node was calibrated to be earlier than 66 mya, the highest posterior density of the divergence time was 50.8–55.7 mya, which is still younger than the age of reliable lithornithids (Table S3). This finding reinforces our argument that lithornithids represent one of the basal lineages among paleognaths and are not phylogenetically close to the tinamou. The ostrich, Struthio camelus, was basal among the crown paleognaths in our phylogenomic time tree, as in previous studies (Figure 1A; Figure S1) [1, 2]. Although both the extant ostriches and the oldest ostrich fossils (from the Early Miocene) are located in Africa [28], other fossils show that the ostrich was

widely distributed in Eurasia from the Late Miocene to the Holocene [29]. The timing of the Early Miocene African ostrich postdates a connection between Africa and Eurasia [5]. Based on the basal positions of the Northern Hemispheric paleognaths among the Palaeognathae, we suggest that the Palaeognathae originated in the Northern Hemisphere (Laurasia-derived continents), not in Gondwana-derived landmasses (Figure 3D; Figure 4; Figure S3; and Supplemental Experimental Procedures). Recently, Claramunt and Cracraft [18] carried out extensive reconstruction of the ancestral geographic distribution of birds. Although they indicated the South American origin of the Palaeognathae by using the maximum parsimony (MP) method (100%), the Northern Hemispheric origin of the Palaeognathae was moderately supported by the maximum likelihood (ML) method (see their Figure S3A). Since the ML and Bayesian

(C) Maximum-likelihood trees as inferred from the morphologic data matrix from Worthy et al. [17]. Figure conventions are as in (A). Nodal support values for 92 and 68 morphological characters without homoplasy (the characters coded as ‘‘missing data’’ convergently acquired in the flightless birds were included and excluded, respectively) are shown below and above the branches. (D) Reconstruction of the ancestral geographic distribution areas. The species indicated in orange are distributed in the Northern Hemisphere (Laurasia-derived landmasses). The species indicated in blue are distributed in the Southern Hemisphere (Gondwana-derived landmasses). Fossil species without molecular data are indicated y. Species indicated by circles were used for the reconstruction of the ancestral geographic distribution. The pie charts in these circles are proportional to the posterior probabilities of the distribution areas. Species indicated by squares are fossil species and were not involved in the analysis. Phylogenetic positions of fossil species were based on (A)–(C). The phylogenetic relationship among Lithornis, Palaeotis, and crown Palaeognathae was assumed as a trifurcation. Although Houde [5]’s data preferred a sister relationship between Palaeotis and ostrich [74/0.91], the phylogenetic position of Palaeotis is unclear because of many missing data. The possible alternative positions of Palaeotis are shown on the tree based on Mayr [16]’s data (blue arrows in A). The divergence times of the fossil species in this figure are arbitrary. The phylogenetic positions of fossil species without branches (e.g., Remiornis) are unknown. See also Figure S3.

74 Current Biology 27, 68–77, January 9, 2017

methods take account of branch lengths (the geological ages, in the case of time-calibrated trees), the fossils chronologically close to the internal nodes provide valuable information in reconstructing the ancestral states. Their ML ancestral reconstruction took account more of the fossil information compared to the MP method, which ignores the branch lengths. The geographic information of the fossil Asian ostrich is compatible with a Northern Hemispheric origin of the Palaeognathae. The basal position of the Lithornis cohort supported in this study strengthens the Northern Hemispheric origin hypothesis (Figure 3D; Figure S3). Origin and Evolution of Palaeognathae A novel evolutionary hypothesis arises from the results presented in this study. We suggest that the Palaeognathae originated in the Northern Hemisphere after branching from the Neognathae 104.7–115.5 mya (the late Early Cretaceous; Figure 3D; Figure 4). This time estimate is earlier than the emergence of the stem groups of the Ornithurae (e.g., ichthyornithids, 86.5 mya, or hesperornithids, 100.5 mya) but later than the emergence of the Jehol Biota (120–130.7 mya), which was the most diversified Mesozoic avian fauna and yielded the earliest ornithomorphs [30]. In our phylogenetic inference (Figures 3A–3C), the family Lithornithidae was recognized as the basal lineage of the Palaeognathae, and it is plausible that they form a paraphyletic group (Figure 3D). Our phylogenetic hypothesis supports the idea of the Lithornis cohort, from which the ratites independently evolved multiple times [4, 5]. Therefore, the extant Palaeognathae probably evolved from the Lithornithidae that were widely distributed from North America to Eurasia. The earliest divergence in the extant Palaeognathae (between the ostrich and others) was estimated to be 79.6 mya (95% CI: 76.5–82.6). After branching from the ostrich, the ancestral notopaleognaths migrated to the Southern Hemisphere between 79.6 mya (95% CI: 76.5–82.6) and 70.6 mya (95% CI: 67.9–73.0) (Figure 3D). While the ancestral paleognaths were most probably volant, there must have been a barrier to the presumed migration of tinamou-like ancestral birds from Northern to Southern Hemisphere. We expect that this migration would have been an extremely rare event that occurred by chance—a process referred to as overseas sweepstakes dispersal [31]. Theoretically, the migration to the Southern Hemisphere could have occurred via the African route rather than via the American route. However, considering the branching pattern, timings, and the geographic distribution pattern of the Notopalaeognathae, it is more plausible that the common ancestor migrated occasionally from North America to South America via the Panama strait that existed between the two continents at that time (79.6–70.6 mya) [32], for the following two reasons. The first reason is that South American continent was connected to Antarctica during this time, but the African continent was already separated from South America and Antarctica as early as 100 mya [33]. Taking into account the phylogenetic relationship among notopaleognaths, the South American route hypothesis requires four steps of overseas migrations, while the African route hypothesis requires one more step (five steps of overseas migrations). Therefore, the South American route hypothesis is more parsimonious than the African route hypothesis. The second reason is the fossil record. The fossil Palaeognathae in South America

(Diogenornis fragilis) suggests that they were distributed in South America as early as the Paleocene (56 mya [24]). It has been suggested that Diogenornis fragilis is phylogenetically close to rheas. Conversely, the fossil record of African paleognaths is poor in the Paleogene. Only a single possible paleognath, Eremopezus, is known from the Late Eocene (36 mya; see [24]). The phylogenetic position of this genus is unclear, and this genus is suggested to belong to Neognathae rather than to Palaeognathae [34]. For these reasons, although we cannot exclude the African route hypothesis, the South American route hypothesis is more plausible (see also Figures S3B–S3D for a more detailed reconstructed ancestral geographic distribution). The very short internal branches in the basal positions of the Notopalaeognathae and the unstable phylogenetic placement of the rhea suggest rapid diversification, and our time estimate suggests that this diversification was near the Cretaceous– Paleogene (K–Pg) boundary. Currently, there is no fossil record of the Palaeognathae from the Mesozoic era, but fossil records emerged abruptly from the Paleogene (near the K–Pg boundary). This is similar to the situation of the Neognathae. The oldest fossil neognath, Vegavis iaai (Anatoidea), is reported from just before the K–Pg boundary (66–68 mya) [35]. Genomic data for the Neognathae also suggest a radiation near the K–Pg boundary [11, 12]. Interestingly, similar tendencies, i.e., the radiation around the K–Pg boundary at the genetic level and the emergence of crown taxa in the fossil record after the K–Pg boundary, are also seen in the placental mammals [36, 37]. It is plausible that, although the ancestral Neornithes and placental mammals had already diverged as distinct lineages in the Cretaceous period, the ecological niches were occupied by the stem taxa (Mesozoic birds and eutherian mammals), and that their distribution and population sizes were very limited. The mass extinction of Mesozoic stem groups in the K–Pg boundary (see, e.g., Longrich et al. [38]) triggered the expansion of population sizes of Neornithes and placental mammals, as well as their morphologic diversification. Gigantism probably began after the migration of the ancestral volant paleognaths to the Southern Hemisphere. As mentioned before, the fossil ratite Diogenornis is suggested to have a close affinity with the rhea ([5, 24] and this study), and this fossil species is considered to date from the Paleocene [24]. It suggests that the first gigantism had already occurred at least by the end of the Paleocene. Although Gondwana had already broken into smaller landmasses, South America, Antarctica, and Australia were still connected. Because Antarctica was humid and temperate at that time [39], the ancestral notopaleognaths could expand their distribution into the whole of South America, Antarctica, and Australia. It is highly plausible that further migration events to other Gondwana-derived landmasses such as Madagascar and New Zealand also occurred by overseas sweepstakes dispersals from regions around Antarctica. Fossil ratites from the Middle Eocene to the Early Oligocene were found in Antarctica [40]. Since these fossils consist of nothing more than two toe knuckles, it may be hard to consider them to represent diagnostics of ratites, but if they are ratites, the existence of the paleognaths in Antarctica is consistent with our hypothesis. The divergence times among intercontinental lineages range from 70.6 mya (Late Cretaceous) to 53.3 mya (Early Eocene), and the Current Biology 27, 68–77, January 9, 2017 75

divergence events after the Late Eocene (
35 mya) are limited to intra-continental events (Figures 3 and 4). It suggests that these Antarctica-centered migration events might have lasted until the time when Antarctica became totally glaciated after the establishment of the Antarctic Circumpolar Current about 34 mya. The fossil paleognaths reported from the Southern Hemisphere, except for the fossil tinamous from South America, are currently limited to flightless ratites (DataBaseFig4). However, our estimations of the ancestral body masses suggest that the ancestral Notopalaeognathae were small (4.3–5.5 kg; Table S5) and most probably volant. Gigantism probably occurred independently in each landmass, as suggested in previous studies [1, 2, 6]. The ostrich might have evolved in Eurasia and then migrated to the African continent after the collision of Africa and Eurasia in the early Miocene. In conclusion, we suggest that the current geographic distribution of the Palaeognathae was shaped mainly by overseas sweepstakes dispersal and by continental configuration during the Late Cretaceous and the Paleogene. ACCESSION NUMBERS The accession numbers for the Aepyornis maximus and Mullerornis sp. mitochondrial genome sequences reported in this paper are DDBJ/ENA/GenBank: AP014697 and AP014698, respectively. The accession number for the MiSeq read data is DDBJ Sequence Read Archive: DRA002843.

SUPPLEMENTAL INFORMATION Supplemental Information includes three figures, five tables, and Supplemental Experimental Procedures and can be found with this article online at http://dx.doi.org/10.1016/j.cub.2016.10.029. AUTHOR CONTRIBUTIONS F.A., A.Y., Y. Hayashi, S.Y., T.Y., and M.H. conceived the project. A.R., H. Mukoyama, A.Y., S.N., and H. Koike collected fossil samples. K.Y. conducted the C14 dating. T.S., P.F.C., A.A., T.Y., and E.W. conducted ancient DNA experimental work. H. Mori and K.K. carried out the bioinformatics analyses. T.Y., N.K., J.W., H.J., J.A., H. Kishino, and M.H. performed the molecular and morphological evolutionary analyses. H.E., N.K., H.T., Y. Hongoh, and Y.N. provided biological and geological information. T.Y., T.S., H. Mori, Y. Hongoh, H.E., N.K., J.W., H. Kishino, F.A., E.W., and M.H. coordinated the preparation of the manuscript and the final editing. ACKNOWLEDGMENTS This work was started as a project of the Elephant Bird Research Group founded by F.A., and we wish to express our thanks to all members of the Research Group, especially to the late Professor Satoshi Horai (SOKENDAI) for his involvement in the early stages of this project. The manuscript was improved greatly through constructive comments from four anonymous reviewers, and we would like to express our sincere appreciation to them. Thanks are also due to Makoto Manabe (National Museum of Nature and Science), Takeshi Nakano (The Atelier VIGITA), Miyako Tsurumi (Yamashina Institute for Ornithology), Masaki Eda (Hokkaido University), Shozo Mihara (Tokyo Metropolitan University), and Tsimihole Tovondrafale (University of Toliara) for their support and to Takashi Oda for drawing the illustrations used in the figures. This work was supported by Grants-in-Aid for Scientific Research (25440219 and Innovative Area ‘‘Genome Science’’ 221S0002) from the Japan Society for the Promotion of Science (JSPS) to M.H., the Transdisciplinary Research Integration Center (TRIC) of the Research Organization of Information and Systems, the Center for the Promotion of Integrated Sciences

76 Current Biology 27, 68–77, January 9, 2017

(CPIS) of SOKENDAI (to J.A.), Sumitomo Life Insurance Company, the Society for Livestock Studies, and the Yamashina Institute for Ornithology. Received: April 7, 2016 Revised: July 26, 2016 Accepted: October 14, 2016 Published: December 15, 2016 REFERENCES 1. Haddrath, O., and Baker, A.J. (2012). Multiple nuclear genes and retroposons support vicariance and dispersal of the palaeognaths, and an Early Cretaceous origin of modern birds. Proc. Biol. Sci. 279, 4617–4625. 2. Mitchell, K.J., Llamas, B., Soubrier, J., Rawlence, N.J., Worthy, T.H., Wood, J., Lee, M.S., and Cooper, A. (2014). Ancient DNA reveals elephant birds and kiwi are sister taxa and clarifies ratite bird evolution. Science 344, 898–900. 3. Johnston, P. (2011). New morphological evidence supports congruent phylogenies and Gondwana vicariance for palaeognathous birds. Zool. J. Linn. Soc. 163, 959–982. 4. Houde, P. (1986). Ancestors of ostriches found in the Northern Hemisphere suggest a new hypothesis for origin of ratites. Nature 324, 563–565. 5. Houde, P. (1988). Paleognathous Birds from the Early Tertiary of the Northern Hemisphere (Nuttall Ornithological Club). 6. Harshman, J., Braun, E.L., Braun, M.J., Huddleston, C.J., Bowie, R.C., Chojnowski, J.L., Hackett, S.J., Han, K.L., Kimball, R.T., Marks, B.D., et al. (2008). Phylogenomic evidence for multiple losses of flight in ratite birds. Proc. Natl. Acad. Sci. USA 105, 13462–13467. 7. Baker, A.J., Haddrath, O., McPherson, J.D., and Cloutier, A. (2014). Genomic support for a moa-tinamou clade and adaptive morphological convergence in flightless ratites. Mol. Biol. Evol. 31, 1686–1696. 8. Cooper, A., Lalueza-Fox, C., Anderson, S., Rambaut, A., Austin, J., and Ward, R. (2001). Complete mitochondrial genome sequences of two extinct moas clarify ratite evolution. Nature 409, 704–707. 9. Endo, H., Akishinonomiya, F., Yonezawa, T., Hasegawa, M., Rakotondraparany, F., Sasaki, M., Taru, H., Yoshida, A., Yamasaki, T., Itou, T., et al. (2012). Coxa morphologically adapted to large egg in aepyornithid species compared with various palaeognaths. Anat. Histol. Embryol. 41, 31–40. 10. Inoue, J., Donoghue, P.C., and Yang, Z. (2010). The impact of the representation of fossil calibrations on Bayesian estimation of species divergence times. Syst. Biol. 59, 74–89. 11. Prum, R.O., Berv, J.S., Dornburg, A., Field, D.J., Townsend, J.P., Lemmon, E.M., and Lemmon, A.R. (2015). A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526, 569–573. 12. Jarvis, E.D., Mirarab, S., Aberer, A.J., Li, B., Houde, P., Li, C., Ho, S.Y., Faircloth, B.C., Nabholz, B., Howard, J.T., et al. (2014). Whole-genome analyses resolve early branches in the tree of life of modern birds. Science 346, 1320–1331. 13. Yonezawa, T., Nikaido, M., Kohno, N., Fukumoto, Y., Okada, N., and Hasegawa, M. (2007). Molecular phylogenetic study on the origin and evolution of Mustelidae. Gene 396, 1–12. 14. Cracraft, J., Houde, P., Ho, S.Y., Mindell, D.P., Fjeldsa˚, J., Lindow, B., Edwards, S.V., Rahbek, C., Mirarab, S., Warnow, T., et al. (2015). Response to Comment on ‘‘Whole-genome analyses resolve early branches in the tree of life of modern birds’’. Science 349, 1460. 15. Mitchell, K.J., Cooper, A., and Phillips, M.J. (2015). Comment on ‘‘Wholegenome analyses resolve early branches in the tree of life of modern birds’’. Science 349, 1460. 16. Mayr, G. (2014). The middle Eocene European ‘‘ratite’’ Palaeotis (Aves, €ontol Z. 89, 503–514. Palaeognathae) restudied once more. Pala

17. Worthy, T.H., Mitri, M., Handley, W.D., Lee, M.S.Y., Anderson, A., and Sand, C. (2016). Osteology supports a stem-galliform affinity for the giant extinct flightless bird Sylviornis neocaledoniae (Sylviornithidae, Galloanseres). PLoS ONE 11, e0150871. 18. Claramunt, S., and Cracraft, J. (2015). A new time tree reveals Earth history’s imprint on the evolution of modern birds. Sci. Adv. 1, e1501005. 19. Springer, M.S., Meredith, R.W., Eizirik, E., Teeling, E., and Murphy, W.J. (2008). Morphology and placental mammal phylogeny. Syst. Biol. 57, 499–503. 20. Gillooly, J.F., McCoy, M.W., and Allen, A.P. (2007). Effects of metabolic rate on protein evolution. Biol. Lett. 3, 655–659. 21. Lartillot, N., and Poujol, R. (2011). A phylogenetic model for investigating correlated evolution of substitution rates and continuous phenotypic characters. Mol. Biol. Evol. 28, 729–744. 22. Lartillot, N., and Delsuc, F. (2012). Joint reconstruction of divergence times and life-history evolution in placental mammals using a phylogenetic covariance model. Evolution 66, 1773–1787. 23. Maynard Smith, J. (1968). Mathematical Ideas in Biology (Cambridge University Press). 24. Mayr, G. (2009). Paleogene Fossil Birds (Springer). 25. Nesbitt, S.J., and Clarke, J.A. (2016). The anatomy and taxonomy of the exquisitely preserved Green River Formation (Early Eocene) lithornithids (Aves) and the relationships of Lithornithidae. Bull. Am. Mus. Nat. Hist. 406, 1–91. 26. Yuri, T., Kimball, R.T., Harshman, J., Bowie, R.C., Braun, M.J., Chojnowski, J.L., Han, K.L., Hackett, S.J., Huddleston, C.J., Moore, W.S., et al. (2013). Parsimony and model-based analyses of indels in avian nuclear genes reveal congruent and incongruent phylogenetic signals. Biology (Basel) 2, 419–444. 27. Parris, D.C., and Hope, S. (2002). New interpretations of birds from the Hornerstown and Navesink formations, New Jersey. In Proceedings of the 5th International Meeting of the Society of Avian Paleontology and Evolution, Z. Zhou, and F. Zhang, eds. (Science Press), pp. 113–124. , C., Senut, B., Pickford, M., and Mein, P. (1996). [The 28. Mourer-Chauvire oldest member of the genus Struthio (Aves, Struthionidae) Struthio coppensi n. sp., from the Miocene of Namibia]. C. R. Acad. Sci. 322, 325–332. 29. Hou, L., Zhou, Z., Zhang, F., and Wang, Z. (2005). A Miocene ostrich fossil from Gansu Province, northwest China. Chin. Sci. Bull. 50, 1808–1810.

30. Wang, M., Zheng, X., O’Connor, J.K., Lloyd, G.T., Wang, X., Wang, Y., Zhang, X., and Zhou, Z. (2015). The oldest record of ornithuromorpha from the early cretaceous of China. Nat. Commun. 6, 6987. 31. Simpson, G.G. (1940). Mammals and land bridges. J. Wash. Acad. Sci. 30, 137–163. 32. Pindell, J.L., and Kennan, L. (2009). Tectonic evolution of the Gulf of Mexico, Caribbean and northern South America in the mantle reference frame. In The Origin and Evolution of the Caribbean Plate (Special Publications 328), K.H. James, M.A. Lorente, and J.L. Pindell, eds. (Geological Society), pp. 1–55. 33. Eagles, G., and Ko¨nig, M. (2008). A model of plate kinematics in Gondwana breakup. Geophys. J. Int. 173, 703–717. , C., and Cheneval, J. (1983). Les Sagittariidae fossiles 34. Mourer-Chauvire (Aves, Accipitriformes) de l’Oligoce`ne des Phosphorites du Quercy et du rieur de Saint-Ge rand-le-Puy. Geobios 16, 443–459. Mioce`ne infe 35. Clarke, J.A., Tambussi, C.P., Noriega, J.I., Erickson, G.M., and Ketcham, R.A. (2005). Definitive fossil evidence for the extant avian radiation in the Cretaceous. Nature 433, 305–308. 36. dos Reis, M., Inoue, J., Hasegawa, M., Asher, R.J., Donoghue, P.C., and Yang, Z. (2012). Phylogenomic datasets provide both precision and accuracy in estimating the timescale of placental mammal phylogeny. Proc. Biol. Sci. 279, 3491–3500. 37. O’Leary, M.A., Bloch, J.I., Flynn, J.J., Gaudin, T.J., Giallombardo, A., Giannini, N.P., Goldberg, S.L., Kraatz, B.P., Luo, Z.X., Meng, J., et al. (2013). The placental mammal ancestor and the post-K-Pg radiation of placentals. Science 339, 662–667. 38. Longrich, N.R., Tokaryk, T., and Field, D.J. (2011). Mass extinction of birds at the Cretaceous-Paleogene (K-Pg) boundary. Proc. Natl. Acad. Sci. USA 108, 15253–15257. 39. Pross, J., Contreras, L., Bijl, P.K., Greenwood, D.R., Bohaty, S.M., Schouten, S., Bendle, J.A., Ro¨hl, U., Tauxe, L., Raine, J.I., et al.; Integrated Ocean Drilling Program Expedition 318 Scientists (2012). Persistent near-tropical warmth on the Antarctic continent during the early Eocene epoch. Nature 488, 73–77. 40. Tambussi, C.P., Noriega, J.I., Gazdzicki, A., Tatur, A., Reguero, M.A., and Vizcaino, S.F. (1994). Ratite bird from the Paleogene La Meseta Formation, Seymour Island, Antarctica. Pol. Polar Res. 15, 15–20.

Current Biology 27, 68–77, January 9, 2017 77