Ahypertrophied ossified sternum characterizes all living birds,

Transcription

1 On the absence of sternal elements in Anchiornis (Paraves) and Sapeornis (Aves) and the complex early evolution of the avian sternum Xiaoting Zheng a,b, Jingmai O Connor c,1, Xiaoli Wang a, Min Wang c, Xiaomei Zhang b, and Zhonghe Zhou c,1 a Institute of Geology and Paleontology, Linyi University, Linyi, Shandong , China; b Shandong Tianyu Museum of Nature, Pingyi, Shandong , China; and c Key Laboratory of Vertebrate Evolution and Human Origins of the Chinese Academy of Sciences, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing , China Contributed by Zhonghe Zhou, June 14, 2014 (sent for review April 13, 2014) Anchiornis (Deinonychosauria: Troodontidae), the earliest known feathered dinosaur, and Sapeornis (Aves: Pygostylia), one of the basalmost Cretaceous birds, are both known from hundreds of specimens, although remarkably not one specimen preserves any sternal ossifications. We use histological analysis to confirm the absence of this element in adult specimens. Furthermore, the excellent preservation of soft-tissue structures in some specimens suggests that no chondrified sternum was present. Archaeopteryx, the oldest and most basal known bird, is known from only 10 specimens and the presence of a sternum is controversial; a chondrified sternum is widely considered to have been present. However, data from Anchiornis and Sapeornis suggest that a sternum may also have been completely absent in this important taxon, suggesting that the absence of a sternum could represent the plesiomorphic avian condition. Our discovery reveals an unexpected level of complexity in the early evolution of the avian sternum; the large amount of observable homoplasy is probably a direct result of the high degree of inherent developmental plasticity of the sternum compared with observations in other skeletal elements. Maniraptora Mesozoic Jehol flight histology Ahypertrophied ossified sternum characterizes all living birds, and enlarged sterna are also present in other flying vertebrates (Ornithodira: Pterosauria, Mammalia: Chiropterygidae) (1). However, the presence of a sternum in Archaeopteryx, long considered the oldest and most primitive bird known, is controversial it is absent in every known specimen (2, 3). An ossified sternum is also absent in every reported specimen of Sapeornis chaoyangensis (n = 10), the largest known Early Cretaceous bird and one of the most primitive birds with a shortened tail ending in a pygostyle (Aves: Pygostylia: Sapeornithiformes) (4, 5). The sternum in living birds provides a large surface area for the attachment of the two most important flight muscles: the m. pectoralis and m. supracoracoideus (6). Thus, where an ossified sternum is absent in fossil birds, the element is still inferred to be present but cartilaginous (5, 7). The absence of a large ossified sternum and other skeletal differences (e.g., absence of a procoracoid process) suggest that flight capabilities would be severely limited in basal birds (8, 9). Sternal ossifications are absent in Troodontidae (Maniraptora: Paraves), a clade of dinosaurs considered closely related to birds (10, 11). In most phylogenetic analyses, Troodontidae and Dromaeosauridae form a clade (Deinonychosauria) that is the sister group of Aves (12). However, the fossil record of troodontids remained highly fragmentary until the recent discovery of a wealth of small, feathered taxa from the Late Jurassic and Early Cretaceous of China. Among these fossils was Anchiornis huxleyi, asmall, feathered troodontid from the Jurassic that has alternatively been resolved as an archaeopterygid; however, this conclusion removes this clade from Aves (13, 14). Published specimens of Anchiornis also do not preserve sternal elements of any kind (n = 3) (13 15). The absence of an ossified sternum in Troodontidae, Archaeopteryx, and Sapeornithiformes is counterintuitive to the inferred importance of the sternum as part of the avian flight apparatus and at odds with the phylogenetic distribution of ossified sterna among maniraptoran theropods. All other groups that are or have been considered closely related to birds (Scansoriopterygidae, Dromaeosauridae, and Oviraptorosauria) possess paired, ossified sternal plates that fuse into a singular element (sternum) late in ontogeny in at least some taxa (e.g., dromaeosaurid Microraptor, oviraptorosaur Ingenia) (16 19). Admittedly, the sternum is not one of the best-known skeletal elements in these clades; the presence of sternal plates is affected by ontogeny and even in adults, these thin, plate-like elements are often not preserved (19). This observation has lead to the assumption that the absence of sternal elements in Archaeopteryx, Sapeornis, and troodontids could potentially be due to preservational or ontogenetic bias in the fossil record (5, 7). To test this hypothesis for troodontids, we use Anchiornis, for which a large number of specimens are available at the Shandong Tianyu Museum of Nature (STM). This museum holds the largest collections of any single dinosaurian taxon in the world (more than 200 specimens each of Anchiornis and Microraptor), providing a unique environment in which we can investigate assumptions based on previously limited material. This study represents the largest published dataset of a single extinct dinosaurian taxon to date. Here, we examine a large number of previously unpublished specimens referable to Anchiornis and Sapeornis for sternal elements and discuss the implications of the absence of sternal ossifications at the base of the avian clade. Methods Two hundred twenty-six specimens of Anchiornis sp. (SI Appendix, Table S1) were studied at the STM, Shandong Province, China; 96 specimens of Significance We have observed more than 200 specimens of Anchiornis, the earliest known feathered dinosaur, and nearly 100 specimens of Sapeornis, one of the basalmost birds, and recognize no sternal ossifications. We propose that the sternum may have been completely lost in these two taxa (and Archaeopteryx as well) based on histological analysis and the excellent preservation of soft-tissue structures, thus suggesting the absence of a sternum could represent the plesiomorphic avian condition. Our discovery reveals an unexpected level of complexity and high degree of inherent developmental plasticity in the early evolution of the avian sternum. Author contributions: X. Zheng and Z.Z. designed research; X. Zheng, J.O., X.W., and X. Zhang performed research; J.O. and M.W. analyzed data; and J.O. and Z.Z. wrote the paper. The authors declare no conflict of interest. 1 To whom correspondence may be addressed. zhouzhonghe@ivpp.ac.cn or jingmai. oconnor@gmail.com. This article contains supporting information online at /pnas /-/DCSupplemental PNAS September 23, 2014 vol. 111 no. 38

2 Sapeornis chaoyangensis were studied (SI Appendix, Table S2). We selected a comparably sized taxon with an ossified sternum for comparison and to determine the extent of preservational bias; 88 specimens of Jeholornis (Aves: Jeholornithiformes), a long bony-tailed bird considered only more derived than Archaeopteryx within Aves, were studied (SI Appendix, Table S3). Published data were also included for all three taxa. The quality of preservation, femur length (measured by using stainless hardened digital calipers), and presence of sternal elements (sternal plates and ribs) and gastralia (these delicate elements act as a second proxy for quality of preservation) were noted for each specimen (SI Appendix, Tables S1 S3). Specimens near the lower and upper size limits were sampled histologically to confirm the presence of an ontogenetic series (SI Appendix, SI Methods). Clearly, no very young juveniles are present in the sample; no specimen preserves juvenile features such as small size, a large orbit, or incompletely ossified periosteal surface. EVOLUTION Results Anchiornis specimens (including previously published material; n = 229) range in femur length (as a proxy for size) from 40 to 93 mm (SI Appendix, Table S1). No specimen preserves sternal ribs or plates (Fig. 1); 78% preserve gastral elements, but in only 10% is the basket even moderately well preserved. Sapeornis (n = 106) specimens range in femoral length from 50 to 90 mm (SI Appendix, Table S2). Of the sample, no specimen preserves sternal elements (including sternal ribs), 50% preserve gastralia, but only 13% preserve complete or nearly complete baskets (SI Appendix, Table S2 and Fig. S1). Jeholornis (n = 95) specimens preserve femora ranging from 50 to 85 mm long (SI Appendix, Table S3). Approximately half the specimens (52%) preserve sternal ossifications and slightly less preserve gastralia (44%). The sternum is similar to that of the dromaeosaurid Microraptor gui (SI Appendix, Fig. S2F), except it is proportionately shorter (quadrangular), as in confuciusornithiforms (SI Appendix, Fig. S3C), with small craniolateral processes, and large, simple lateral processes (SI Appendix, Fig. S3A). In some specimens, clefts are located on the rostral and caudal midline, indicating that fusion of the two sternal plates was incomplete (e.g., STM2-16, 2-39). An additional pair of ossifications is preserved, associated with the sternum in multiple specimens Fig. 1. Three specimens of Anchiornis sp. preserving the complete or nearly complete gastral basket but no sternal ossifications including sternal ribs: STM0165 (A); STM0-120 (B); STM0-52 (C). (Scale bars: 1 cm.) cor, coracoid; fem, femur; fur, furcula; gas, gastralia; isc, ischium; pub, pubis; sca, scapula. Zheng et al. PNAS September 23, 2014 vol. 111 no

3 (SI Appendix, Fig. S3A) and is inferred to have articulated on the lateral surface, functioning as lateral trabeculae although in no specimen are the lateral trabeculae fused to the sternum (20), and their in vivo articulation with the sternal body is unknown (20, 21). Four to five pairs of sternal ribs articulated with the sternum; costal facets appear not to be developed (present in some nonavian theropods, the basal bird Confuciusornis, and ornithuromorphs) (22 24). Histological samples were taken from both the upper and lower size limit; STM15-32, a larger specimen of Sapeornis, and STM0-8, the largest sampled Anchiornis, both preserve an avascular outer circumferential layer (OCL) marked by multiple lines of arrested growth (LAGs) and a well-developed ICL (SI Appendix, SI Methods and Figs. S4 S9). The first LAG in Sapeornis STM15-32 is a double LAG, separated from the closely packed LAGs in the OCL, indicating that Sapeornis took several years to reach maturity and then continued to grow, albeit slowly (SI Appendix, Fig. S5). No previously sampled specimen of Sapeornis preserves LAGs (25, 26) to our knowledge. STM15-32 preserves a fully fused tibiotarsus and fused proximal carpometacarpus and tarsometatarsus (SI Appendix,Fig.S10), which together with histological data strongly suggests this specimen is a skeletally mature adult (27). Histology confirms that no young juvenile Sapeornis or Anchiornis are present in the sample; even the smallest specimens (Anchiornis STM0-5, Sapeornis STM15-6) preserve an ICL, indicating medullary expansion (bone remodeling) had already occurred at least once (SI Appendix, Figs. S8 and S9). Notably, despite its smaller size relative to STM15-70, the histology of STM15-32 appears to be more mature, with an OCL marked by several LAGs (SI Appendix,Figs.S5B and S7). We suggest that this observation may be indicative of sexual dimorphism in Sapeornis. Discussion Based on this study conducted on the largest published sample of specimens referable to a single dinosaurian taxon (Anchiornis, n = 229; Sapeornis, n = 106; Jeholornis, n = 95) (SI Appendix, Tables S1 S3), we consider the absence of sternal elements in Anchiornis and Sapeornis to be a true feature of these taxa and not an artifact of preservation or ontogeny. The known collections of Jehol specimens are heavily biased toward more complete specimens, which should increase the likelihood of sternal elements being preserved in the available material (71% of the Anchiornis specimens sampled are approximately >90% complete, compared with 62% of all Sapeornis specimens and 52% of all Jeholornis specimens; more fragmentary specimens are rarely less than 50% complete). Sternal plates and ribs are absent in all specimens of Anchiornis and Sapeornis regardless of preservational quality. Sternal morphology is affected by ontogeny, but sternal ossifications appear fairly early in basal birds, present in all known enantiornithine hatchlings and subadult jeholornithiform and ornithuromorph specimens (21, 28 30). In the one specimen of subadult confuciusornithiform, although there is no ossified sternum, a chondrified sternum is preserved (31). Sternal ribs ossify even earlier than the sternal plates in paravians, but this feature is also consistently absent in all specimens of Anchiornis and Sapeornis, further indicating a chondrified sternum was most likely not present. Specimens of both Anchiornis and Sapeornis occupy a considerable size range (SI Appendix, Tables S1 and S2); although there is a known bias toward juveniles in the fossil record of larger dinosaurs (32), this observation should not account for their absence in at least Sapeornis because sternal ossifications typically appear fairly early in ontogeny and subadults not juveniles appear to dominate the fossil record of Mesozoic birds. Histological analysis confirms our sample includes skeletally mature specimens, indicating that the absence of sternal plates is not an ontogenetic artifact. In light of the large number of specimens used for each taxon in this study, many of which are complete and fully articulated, and some of which boast exceptional preservation including soft-tissue impressions such as feathers and stomach contents, we feel there is sufficient evidence to conclude that a sternum, chondrified or ossified, was truly absent in both Anchiornis and Sapeornis. Ossification is driven by BMP signaling (33); to ossify a cartilaginous element requires only a simple change in transcription factors. The mechanical stress of volant activity would therefore be expected to readily induce the evolution of an ossified sternum in animals with a cartilaginous precursor, especially given that an ossified sternum is present in more basal maniraptorans, indicating the genetic potential for this feature. However, younger Jiufotang Formation sapeornithiforms, separated from their Yixian relatives by a period of 8 million years, also do not preserve any evidence for sternal ossifications, further suggesting no cartilaginous element was present at any stage during sapeornithiform evolution. However, given the vagaries of taphonomy and preservation and the delicate nature of the thin sternum even when ossified, we cannot conclude unequivocally that a chondrified sternum was absent in either Anchiornis or Sapeornis. In contrast, the adult long boney-tailed Jehol bird Jeholornis has fully fused sternal plates (SI Appendix, Fig. S3A). Of the sampled specimens, one-half (52%) preserve a sternum including the subadult holotype of Jixiangornis CDPC (29). Comparison with this comparably sized sympatric taxon supports interpretations that the absence of a sternum in Sapeornis is not preservational or the result of a sampling bias. Compensatory Morphologies. It is commonly accepted that Archaeopteryx and Sapeornis were volant (4, 34), thus they may have compensated for the absence of a sternum through other morphologies. In crocodilians, several muscles attach to the gastralia and the cranial pair of gastralia articulates with the sternum (35). This plesiomorphic condition is retained, and the gastralia are observed to articulate with the sternum in some theropods and basal birds (Jeholornis,Confuciusornis,Eopengornis,Parabohaiornis; SI Appendix, Fig. S11 A and B). Gastralia are notably highly modified within Theropoda compared with other groups of amniotes, and proximally fused gastralia have been proposed to function similarly to the sternum in tyrannosaurids (35). Gastralia are absent in living birds, considered functionally redundant in the presence of the large ossified sternum that characterizes Neornithes (35), thus bridging the possibility that the large gastral basket in Sapeornis (5) may have functioned as a compensatory feature in the absence of a sternum and that the extensive gastral basket of basal birds with sterna may also have participated in supporting or reinforcing the sternum and the flight muscles (SI Appendix, Fig. S1). However, the absence of a sternum in Sapeornis and Anchiornis would have left the gastral basket free, suggesting it would have lacked the rigidity to support large flight muscles. If the fused cranial row in tyrannosaurids functioned similar to the sternum of more derived theropods (35), potentially the same may be true regarding Archaeopteryx and Sapeornis. Although the first pair of gastralia in Sapeornis appears to be similarly fused, the gastral basket is otherwise void of modifications to suggest it was the attachment site of robust musculature. We do not fully understand how the gastralia may have supported the musculature necessary for volant activity, but in the absence of a sternum, these muscles clearly would have had to find attachment elsewhere. The morphology of the coracoid in Sapeornis also differs from that of other basal birds in which this bone definitively articulates with a sternum. Potentially the large coracoid of Sapeornis may also have compensated for the absence of the sternum in some way (7). Compared with other Cretaceous birds, the coracoid is proportionately wide and short (SI Appendix, Fig.S2E). The morphology in Sapeornis, which is similar to that of Anchiornis (SI Appendix, Fig.S2C) andarchaeopteryx, is plesiomorphic to a larger group of theropods, including nonavian paravians with an ossified sternum. Caudipteryx, with its simple oval sternal Zheng et al.

4 plates, also has a plesiomorphic proportionately short coracoid (SI Appendix, Fig. S2B). In contrast, the supposedly volant Microraptor has a proportionately more narrow and elongate coracoid relative to other derived maniraptorans, somewhat resembling that of basal birds such as Jeholornis (SI Appendix, Fig. S2 A and D and Table S4). It is interesting to note that in all supposedly volant taxa with sterna (Microraptor, Jeholornis, Confuciusornithiformes, and ornithothoracines), the coracoid is elongated relative to the plesiomorphic condition, suggesting these two features are correlated. In the plesiomorphic theropod coracoid morphology, present in Archaeopteryx and Sapeornis, the distal margin is convex, especially along the distolateral margin, whereas in Microraptor and more derived avian taxa, the sternal margins are straight for articulation with the sternum (e.g., Jeholornis, Confuciusornis, ornithothoracines). This convexity is particularly pronounced in troodontid taxa without sterna (e.g., Mei long, Anchiornis), which also have proportionately short coracoids (slightly wider than long; SI Appendix, Table S4). Proximally, the sternolateral margin of the coracoid in Sapeornis is expanded along the distal half (SI Appendix,Fig. S2E), which may also have served to provide a larger proximal attachment surface for the pectoral muscles. Unfortunately, this hypothesis is also not supported by any local rugosity, tubercle, or other indicator on the coracoid to suggest that this convexity was, in fact, a site of muscle attachment. The Absence of Sternum in Archaeopteryx. The iconic first bird Archaeopteryx is known from 10 published skeletal specimens (36). Of these specimens, six are nearly complete and articulated (the Berlin, Eichstätt, London, Munich, Solnhofen, and Thermopolis specimens), four of which also preserve feather impressions. No specimen preserves evidence of sternal plates or ribs, although a complete or nearly complete gastral basket is preserved in several specimens (well preserved in the Berlin, Eichstätt, Munich, Solnhofen, and Thermopolis specimens); preservation of the latter, which is formed of numerous small, delicate bones, suggests that the absence of sternal elements is not taphonomic. What was described as the sternum in the Munich specimen (2) has been reinterpreted as part of the coracoid (3). However, because this element has such a centripetal role in neornithine-powered flight and is present in other nonavian maniraptorans, it was considered that a cartilaginous sternum was present (37). Because of the small sample size (n = 10), the absence of an ossified sternum may potentially be due to preservational or ontogenetic bias. Archaeopteryx is in many ways morphologically similar to the nonavian maniraptorans Anchiornis and Xiaotingia (13). These taxa share more than the absence of a sternum in all known specimens: They have similar skull morphologies with small conical unserrated teeth; a short, quadrangular coracoid; a tetraradiate ischium; and a gastral basket composed of pairs of gastralia. These taxa have even been resolved together in an archaeopterygid clade outside of Dromaeosauridae + EVOLUTION Fig. 2. A simplified cladogram of derived maniraptoran theropods showing sternal morphology. The size of the circle reflects body size; blue indicates the absence of sternal ossifications, also indicated by short dashed branches (yellow, present). The number inside the circle indicates the number of gastralia (long dash indicates branch where gastralia are absent). The smaller circle inside the larger circle for Neornithes is meant to indicate the extreme size range occupied by fossil and extant members of this clade. Zheng et al. PNAS September 23, 2014 vol. 111 no

5 Troodontidae (10), highlighting the similarity between these taxa. We have provided convincing evidence that the absence of an ossified sternum in Anchiornis is not due to ontogenetic or preservational bias; we further propose that in light of the excellent preservation of soft tissue in numerous specimens, it is safe to conclude a chondrified sternum was also absent. Considering the morphological similarity between Anchiornis and Archaeopteryx and the absence of a sternum in Sapeornis, we argue a chondrified sternum may also have been absent in the earliest bird, Archaeopteryx. Sternal Morphology and Avian Origins. Sternal plates (unfused and cartilaginous) are plesiomorphic for the tetrapod crown group, only known to be lost in turtles and various limbless clades like snakes and caecilians (1). We have provided strong evidence that Anchiornis and Archaeopteryx, taxa inferred to cross the dinosaurianavian transition (Fig. 2), had no sternum. This observation has important repercussions for theories regarding volant activity in the earliest birds. Given the phylogenetic proximity of Troodontidae to the base of Aves and the absence of a sternum in Archaeopteryx,itis most parsimonious to interpret that this feature was the plesiomorphic avian condition. Its absence in the basal pygostylian Sapeornis may further suggest the avian lineage leading to Ornithothoraces also did not have a sternum; this observation suggests the sterna in Jeholornis and Confuciusornis could have evolved independently and may not be strictly homologous. Alternatively, current popular phylogenetic hypotheses regarding basal bird relationships are incorrect (7); some analyses have resolved Sapeornis as basal to Jeholornis (38, 39), which would suggest a single origin event for the sternum within Aves (but multiple origins for the pygostyle). Placing Archaeopteryx in Troodontidae would also require less steps with regards to sternal evolution. However, among derived maniraptorans, the sternum is highly homoplastic, as evidenced by the repeated parallel evolution of features such as a keel and lateral trabeculae; this observation highlights the apparent plasticity of this compound synstotic element (21), exemplified by the novel ossification pattern evolved by some enantiornithines (21). The absence of a sternum in Archaeopteryx and Sapeornis goes beyond its importance to the flight apparatus: This element is present in all tetrapods with forelimbs (with the unique exception of turtles in which the ribs are enlarged). Intuitively, we must infer a cartilaginous sternum was present not only because of its phylogenetic distribution but also because we can infer no reasonable benefit to volant activity gained through the loss of this feature, which provides the major point of attachment for the primary flight muscles (6). Most researchers tend to agree that Archaeopteryx had extremely limited flight capabilities. Although Sapeornis possesses features that are considerably more advanced such as its small pygostyle, long rectrices arranged into a fan, elongate forelimbs, and carpometacarpus with reduced manual digits (5, 26), derived features are notably absent from the shoulder girdle. This observation may suggest that Sapeornis retained a primitive form of flight and was not capable of a powerful up stroke, as inferred for Archaeopteryx (8). Conclusions Using the largest published collection of any dinosaurian taxon to date, we confirm the absence of sternal ossifications in Sapeornis and Anchiornis. Furthermore, despite the bias toward complete or exceptionally preserved material, there is no evidence of a chondrified sternum, widely assumed to be present in the absence of an ossified element in taxa such as Archaeopteryx. Based on this comparison, we suggest that Archaeopteryx, the oldest and most basal bird, also had no sternum, cartilaginous or otherwise. This observation suggests that this feature was the plesiomorphic condition in Aves, which, in turn, suggests that basal bird sterna may not be truly homologous. ACKNOWLEDGMENTS. We thank S. Zhang (Institute of Vertebrate Paleontology and Paleoanthropology, IVPP) for assistance with the histology and A-J. Shi (IVPP) for producing the illustrations. This research was supported by National Basic Research Program of China 973 Program Grant 2012CB821906; National Natural Science Foundation of China Grants , , and ; and the Chinese Academy of Sciences. 1. Parker WK (1867) A Monograph on the Structure and Development of the Shouldergirdle and Sternum in the Vertebrata (The Ray Society, London), pp Wellnhofer P (1993) Das siebte exemplar von Archaeopteryx aus den Solnhofer Schichten [The seventh specimen of Archaeopteryx from the Solnhofen Limestone]. Archaeopteryx 11:1 47. German. 3. Wellnhofer P, Tischlinger H (2004) Das Brustbein von Archaeopteryx bavarica Wellnhofer eine Revision [The sternum of Archaeopteryx bavarica Wellnhofer 1993: A revision]. Archaeopteryx 22:3 15. German. 4. Zhou Z, Zhang F (2002) Largest bird from the Early Cretaceous and its implications for the earliest avian ecological diversification. Naturwissenschaften 89(1): Zhou Z, Zhang F (2003) Anatomy of the primitive bird Sapeornis chaoyangensis from the Early Cretaceous of Liaoning, China. Can J Earth Sci 40(5): Gill FB (2007) Ornithology (Freeman, New York), 3rd Ed, pp O Connor JK, Chiappe LM, Bell A (2011) Pre-modern birds: Avian divergences in the Mesozoic. Living Dinosaurs: The Evolutionary History of Birds, eds Dyke GD, Kaiser G (Wiley, Hoboken, NJ, pp Jenkins FA, Jr (1993) The evolution of the avian shoulder joint. Am J Sci 293A: Gatesy SM, Dial KP (1996) From frond to fan: Archaeopteryx and the evolution of short-tailed birds. Evolution 50(5): Xu X, You H, Du K, Han F (2011) An Archaeopteryx-like theropod from China and the origin of Avialae. Nature 475(7357): Xu X, Norell MA (2004) A new troodontid dinosaur from China with avian-like sleeping posture. Nature 431(7010): Turner AH, Pol D, Clarke JA, Erickson GM, Norell MA (2007) A basal dromaeosaurid and size evolution preceding avian flight. Science 317(5843): Hu D, Hou L, Zhang L, Xu X (2009) A pre-archaeopteryx troodontid theropod from China with long feathers on the metatarsus. Nature 461(7264): Xu X, et al. (2009) A new feathered maniraptoran dinosaur fossil that fills a morphological gap in avian origin. Chin Sci Bull 54(3): Longrich NR, Vinther J, Meng Q, Li Q, Russell AP (2012) Primitive wing feather arrangement in Archaeopteryx lithographica and Anchiornis huxleyi. Curr Biol 22(23): Osmólska H, Currie PJ, Barsbold R (2004) Oviraptorosauria. The Dinosauria, eds Weishampel DB, Dodson P, Osmólska H (Univ of California Press, Berkeley), 2nd Ed, pp Zhang F, Zhou Z, Xu X, Wang X, Sullivan C (2008) A bizarre Jurassic maniraptoran from China with elongate ribbon-like feathers. Nature 455(7216): Xu X, et al. (2003) Four-winged dinosaurs from China. Nature 421(6921): Norell MA, Makovicky PJ (1997) Important features of the dromaeosaur skeleton: Information from a new specimen. Am Mus Novit 3215: Zhou Z, Zhang F (2003) Jeholornis compared to Archaeopteryx, with a new understanding of the earliest avian evolution. Naturwissenschaften 90(5): Zheng X, Wang X, O Connor J, Zhou Z (2012) Insight into the early evolution of the avian sternum from juvenile enantiornithines. Nat Commun 3: Chiappe LM, Ji S, Ji Q, Norell MA (1999) Anatomy and systematics of the Confuciusornithidae (Theropoda: Aves) from the Late Mesozoic of northeastern China. Bull Am Mus Nat Hist 242: You H-L, et al. (2010) A second ornithuromorph from the Changma Basin, Gansu Province, northwestern China. Acta Palaeontol Pol 55(4): Godfrey SJ, Currie PJ (2004) A theropod (Dromaeosauridae, Dinosauria) sternal plate from the Dinosaur Park Formation (Campanian, Upper Cretaceous) of Alberta, Canada. Feathered Dragons: Studies on the Transition from Dinosaurs to Birds, eds Currie PJ, Koppelhus EB, Shugar MA, Wright JL (Indiana Univ Press, Bloomington, IN), pp Erickson GM, et al. (2009) Was dinosaurian physiology inherited by birds? Reconciling slow growth in archaeopteryx. PLoS ONE 4(10):e Gao C-H, et al. (2012) A subadult specimen of the Early Cretaceous bird Sapeornis chaoyangensis and a taxonomic reassessment of sapeornithids. J Vertebr Paleontol 32(5): O Connor JK, Wang M, Zheng X-T, Wang X-L, Zhou Z-H (2014) The histology of two females Early Cretaceous birds. Vertebr Palasiat 52(1): Chiappe LM, Ji S, Ji Q (2007) Juvenile birds from the Early Cretaceous of China: Implications for enantiornithine ontogeny. Am Mus Novit 3594: Ji Q, et al. (2002) A new avialian bird - Jixiangornis orientalis gen. et sp. nov. - from the Lower Cretaceous of western Liaoning, NE China. Nanjing Shi Da Xue Bao 38(6): Zhou S, Zhou Z-H, O Connor JK (2013) Anatomy of the Early Cretaceous Archaeorhynchus spathula. J Vertebr Paleontol 33(1): Zhang F, Zhou Z, Benton MJ (2008) A primitive confuciusornithid bird from China and its implications for early avian flight. Sci China Ser D Earth Sci 51(5): Erickson GM, Currie PJ, Inouye BD, Winn AA (2006) Tyrannosaur life tables: An example of nonavian dinosaur population biology. Science 313(5784): Zheng et al.

6 33. Wang Y, Zheng Y, Chen D, Chen Y (2013) Enhanced BMP signaling prevents degeneration and leads to endochondral ossification of Meckel s cartilage in mice. Dev Biol 381(2): Burgers P, Chiappe LM (1999) The wing of Archaeopteryx as a primary thrust generator. Nature 399: Claessens LPAM (2004) Dinosaur gastralia; origin, morphology, and function. JVertebr Paleontol 24(1): Wellnhofer P (2008) Archaeopteryx. Der Urvogel von Solnhofen [Archaeopteryx: The first bird of Solnhofen] (Friedrich Pfeil, München). 37. Chiappe LM (2007) Glorified Dinosaurs: The Origin and Early Evolution of Birds (Wiley, Hoboken, NJ), pp Zhou Z-H, Zhang F-C, Li Z-H (2009) A new basal orithurine (Jianchangornis microdonta gen. et sp. nov.) from the Lower Cretaceous of China. Vertebr Palasiat 47(4): Zhou Z-H, Zhang F-C, Li Z-H (2010) A new lower cretaceous bird from China and tooth reduction in early avian evolution. Proc R Soc Lond B Biol Sci 277(1679): EVOLUTION Zheng et al. PNAS September 23, 2014 vol. 111 no

7 Zheng et al. On the absence of sternal elements in Anchiornis (Paraves: Troodontidae) and Sapeornis (Aves: Pygostylia) and the complex early evolution of the avian sternum Supplementary Information 1. Table S1-S4 2. Histology description 3. Figures S References Tables Table 1. List of Anchiornis specimens from the Shandong Tianyu Museum of Nature as well as published material (n = 229) sorted by size. The number of gastralia is only listed when the preserved basket is relatively complete. Brackets indicate incomplete elements. Parentheses indicate number of preserved pairs of gastralia. Collection No. Complete Articulated Soft-tissue Gastralia (pairs) Femur length (>90%) (mm) STM 0-32 x x x (>10) 40 STM 0-39 x x x x 40 STM x x x (13) 40 STM x x 40 STM x x x 40 STM 0-5 x x x 41 STM STM x x x x 41 STM 0-56 x x x x 42 STM x x x x (>8) 42 STM x x x x 43

8 STM x x x x 43 STM x x x x 43 STM IVPP V x - x 43.2 holotype STM 0-1 x x x 44 STM 0-57 x x x 44 STM 0-85 x x x (>7) 44 STM x x x (>11) 44 STM x x 44 STM x x x (>8) 44 STM 0-38 x x x 45 STM 0-60 x x 45 STM x x x (>5) 45 STM x x x x (>5) 45 STM 0-65 x x x x 46 STM x x x x 46 STM x x x x 46 STM x x x 46 STM 0-10 x x x 47 STM 0-59 x x x x 47 STM 0-82 x x x x 47 STM x x x 47 STM x x x x 47 STM x 48 STM x x x (14-15) 48 STM x 48

9 STM 0-58 x x x 49 STM 0-62 x x x 49 STM 0-77 x x x 49 STM 0-79 x x x x 49 STM 0-54 x x x 49.5 STM 0-12 x x x x 50 STM 0-16 x x x x 50 STM 0-24 x x 50 STM 0-30 x x x x (10) 50 STM 0-33 x x 50 STM 0-41 x x x x 50 STM 0-74 x x x (>7) 50 STM 0-75 x x 50 STM x x 50 STM x x x 50 STM x x x (>5) 50 STM x x x x (>8) 50 STM x x x x 50 STM x x x 50 STM x x x (>5) 50 STM x x x (11) 50 STM x x x x 50 STM STM 0-78 x x x 51 STM x x x 51 STM 0-20 x x x 52

10 STM 0-21 x x 52 STM 0-27 x x x x 52 STM 0-84 x 52 STM x x x x 53 STM x x x 54 STM x x x x 54 STM x x x 54 STM 0-76 x x x x 55 STM 0-53 x x x x 56 STM x x x (>8) 56 STM 0-52 x x x x (>11) 57 STM x x x x 57 STM 0-97 x x x 58 STM x x 58 STM x x x 58 STM x 59 STM 0-17 x x x x 60 STM 0-42 x x x 60 STM 0-44 x x x (>12) 60 STM 0-46 x 60 STM 0-55 x x x x 60 STM 0-64 x x x x 60 STM 0-66 x 60 STM 0-69 x x x x (>5) 60 STM 0-86 x x (>6) 60 STM 0-89 x x x x 60

11 STM 0-95 x x (>7) 60 STM x x x x 60 STM x x x x (>10) 60 STM x x (>8) 60 STM x x (>7) 60 STM x x (>4) 60 STM x x x x (>8) 60 STM x x x x 60 STM x 60 STM x x x (>11) 60 STM 0-25 x (6) 62 STM 0-73 x x x 62 STM 0-87 x x (>8) 62 STM x x (>5) 62 STM x x x x 62 STM x x x 62 STM x x x x (~12-14) 63 STM x x x (>7) 63 STM 0-7 x x 64 STM 0-71 x x x x (>5) 64 STM 0-92 x x x (>10) 64 STM x x x x 64 STM x x x x 64 STM 0-4 x x x 65 STM 0-9 x x x 65 STM 0-40 x x x 65

12 STM 0-50 x x x x 65 STM 0-70 x 65 STM x x x x 65 STM x x x x (>10) 65 STM x x x 65 STM x x 65 STM x x x (>8) 65 STM 0-61 x x 66 STM x x x 66 STM x 66 LPM-B00169 x x x 66.2 STM 0-14 x x x (13) 67 STM 0-37 x x x (>10) 67 STM 0-49 x 67 STM x x 67 STM 0-19 x x x x (7) 68 STM 0-47 x x x x (>10) 68 STM x x x x 68 STM x x x 68 STM x x x (>6) 68 STM x x x 68 STM x x x 68 STM 0-93 x x x 69 STM x x x 69 STM x x x x 69 STM 0-2 x x x 70

13 STM 0-31 x x x x 70 STM 0-34 x x x 70 STM 0-72 x x 70 STM 0-91 x 70 STM STM x x x x 70 STM x x x x 70 STM x x 70 STM x x x x 70 STM x x x x 70 STM x x x 70 STM x x x 70 STM x x x 70 STM x x x x 70 STM 0-36 x x 71 STM 0-48 x x x x (>10) 71 STM x x x x (>8) 71 STM STM x x x x 72 STM x x x x (>8) 72 STM x x x x (>10) 72 STM 0-29 x x x x 73 STM x x x x 73 STM x x x 73 STM 0-13 x x 73.5 STM 0-18 x x x x (7) 74

14 STM x x 74 STM 0-3 x x x x 75 STM 0-23 x x 75 STM 0-28 x x x 75 STM x x x x (>6) 75 STM x x x 75 STM x x x x (>5) 75 STM x x x x 75 STM x x x 75 STM 0-99 x x x (10-13) 76 STM 0-90 x x x 79 STM 0-35 x x x x (>9) 80 STM x x x x 80 STM 0-8 x x x x 86 STM 0-26 x x x (11) 93 STM 0-6 x x STM 0-11 x x STM 0-15 x x STM 0-45 STM 0-68 x x x (>11) STM 0-94 x STM 0-98 x x x STM STM x STM x x STM x x

15 STM x STM x x STM x STM x x (>6) STM STM x x x STM x STM x STM x x x STM x STM x x x STM x x (>6) STM x x STM x x STM x x x (>7) STM x x x x STM x x x x (>5) STM STM 0-22 x x x BNHMC Ph828 x x STM 0-81 x x [ 32 ] STM 0-80 x x x x [ 38 ] STM 0-83 x [ 39 ] STM 0-67 x x (>5) [ 43 ] STM x x [ 46 ] STM 0-51 x [ 50 ]

16 STM x x x x (>8) [ 50 ] STM x x x [ 50 ] STM x(>10) [ 50] STM x x x x [ 53 ] STM 0-63 x x (>8) [ 60 ] STM x x x x (>10) [ 63 ] STM x x (>12) [35] STM x x (>6) [46] STM 0-96 x x x [48] STM 0-88 x x (>9) [66] Table 2. List of Sapeornithiformes specimens from the Shandong Tianyu Museum of Nature as well as published sapeornithiform material (n = 106); all published specimens are considered assignable to Sapeornis chaoyangensis (O Connor et al., 2011; Pu et al., 2013; Zhang et al., 2013). The number of gastralia is only listed when the preserved basket is relatively complete. Brackets refer to incomplete elements. Collection No. Complete (>90%) Articulated Soft tissue Gastralia (pairs) Femur length (mm) STM x x x x (>9) 50 STM x x x x (>12) 54 STM x x x (>7) 54 STM x x x 55 STM x x x 55 STM x x 56 STM x x x (12) 56 DNHM D3078 x x x STM 15 6 x x x 57 STM

17 STM x 58 IVPP V13396 S. angustis x x - x (15-16) 58.3 STM x x x x (>10) 60 STM 16 1 x x 60 STM x x 62 STM 15 8 x x x 63 STM x x x 63 STM x x x x (>7) 63 STM x (>5) 64 STM x x x x (>10) 64 LPM B00018 Shenshiornis - x - x(3) 64 STM x x x 65 STM 16 9 x x 65 41HIII0405 x x x x (11) 65 STM x x 67 STM x x 68 STM x x x x 68 STM x x x 68 STM x x 68 STM x x 69 STM 15 5 x x x x 70 STM 15 7 x x x 70 STM x x x 70 STM x x x 70 STM x x 70 STM x x x x (>8) 70

18 STM x x x 70 STM x x x (>10) 70 STM x x x 70 STM 16 5 x x x ( >7) 70 STM x x 70 STM x x x 71 STM 16 7 x x (>6) 72 STM x x x 73 STM 16 6 x x 73 IVPP V13275 x x - x 74 STM 15 2 x 75 STM x x x 75 STM x x x 75 STM x x 75 STM x x x (>12) 75 STM x x x 75 STM x (>8) 75 STM x x x 75 STM x x x 75 STM x x 75 STM x x x 75 STM x x x 75 IVPP V x - x (15) 75 STM x x x 76 STM x x x x (11) 76 STM x x 76

19 STM x 77 STM x x 77 STM 15 9 x x x 78 STM x x x 78 STM x 78 STM 15 4 x x x x 80 STM x x x 80 STM x x 80 STM x 80 STM x 80 STM x x x (>10) 80 STM x x x (>8) 80 IVPP V12698 holotype x 80.4 STM x x x 81 STM x x x 82 STM x x 83 STM x x 85 STM x x 85 STM 15 3 x x 90 STM x 90 STM x x x 93 STM x x x STM STM x x STM x x STM x x

20 STM x STM STM x STM STM 16 3 x STM 16 8 x STM x x STM STM x STM STM 15 1 x x x [ 45 ] STM 16 4 x x [ 58 ] STM 16 2 x x (>10) [ 70] STM x [63] STM x x x x (>11) CDPC Didactylornis x x - x DNHM D1197 x x - x(11) DNHM D x Table S3. List of Jeholornithiformes specimens from the Shandong Tianyu Museum of Nature as well as published jeholornithiform material (n = 95). The number of gastralia is only listed when the preserved basket is relatively complete. Brackets refer to incomplete elements. Collection number Complete (>90%) Articulated Soft-tissue Gastralia (pairs) Sternum Femur length (mm) STM3-31 x x 50 STM2-45 x x x 53 STM2-10 x x x (5) x 54

21 STM3-19 x x 54 STM STM2-1 x x 55 STM2-18 x x x x (5) 55 STM3-5 x 55 LPM0193 Shenzhouraptor x x x x x 55.4 IVPP V x x - x 55.6 STM2-9 x x x x 56 STM2-55 x x 56 STM2-17 x x x 57 STM3-4 x x x 57 STM3-30 x 57 SDM J. palmapenis - X X x STM2-14 x x x 58 STM STM STM3-21 x 58 STM2-34 x x x x x 59 STM2-52 x x 59 STM2-32 x x x 60 STM2-41 x x x x 60 STM3-9 x 60 STM STM2-42 x x 63 IVPP V13533 x x x x?? 64 STM2-40 x 64

22 STM3-3 x x x 64 STM3-6 x x 65 STM2-33 x x x 66 STM2-36 x x x x x 66 STM3-1 x 66 STM2-7 x x x x (5) x 67 YFGP-yb2 x x - - x 67.5 STM2-8 x x x x (8) x 68 STM2-39 x x x 68 STM STM3-32 x x 68 STM2-5 x x x x x 70 STM2-12 x x x 70 STM2-13 x x x 70 STM2-25 x x x x x 70 STM2-26 x x x 70 STM2-29 x x x x x 70 STM2-51 x x x 70 STM3-16 x x x (6) 70 STM3-27 x x 70 STM2-37 x x x x 71 STM3-2 x x 71 CDPC Jixiangornis X x - x (>3) x 71.9 STM2-20 x x 73

23 STM2-54 x x x 73 STM STM3-23 x 73 IVPP V13274 holotype x x - x (6) x 75 STM2-6 x x x x x 75 STM2-30 x x 75 STM2-47 x x x (8) x 75 STM2-49 x x 75 STM3-17 x x 75 STM3-28 x x 75 STM3-33 x 75 STM STM3-20 x x x 76 STM2-19 x x (6) x? 77 STM2-23 x x x 77 STM3-7 x 77 STM2-21 x x x ( >5 ) 78 STM2-35 x x x 79 STM2-46 x x x 79 STM2-15 x x x 80 STM2-38 x x x x x 80

24 STM3-25 x x 80 STM2-4 x x x 85 STM2-24 x x x x x 100 STM2-22 STM2-44 x STM3-8 STM3-10 STM3-12 STM3-14 STM3-18 STM3-24 STM3-29 STM2-28 x x STM2-31 x x x x [35] STM2-27 x x x [40] STM2-53 x [40] STM2-50 x x [46] STM2-11 x x x [50] STM2-3 x [53] STM2-16 x x x (5) x [69] Table S4. Comparative data regarding size, size of gastral basket, proportions of the coracoid, and presence of a sternum in derived maniraptorans. Femoral length (mm) is

25 used as a proxy for size and was typically measured from the specimen preserving the most complete gastral basket for each taxon. We normalized the number of gastralia for size by dividing the minimum number of gastral pairs by the length of the femur in the associated specimen (when possible). Taxon Clade Sternum Femur L Gastralia Gastralia /Fem. L Coracoid W:L Allosaurus Certatosauria absent Sinocalliopteryx Compsognathidae unknown Caudipteryx Oviraptorosauria unfused Potentially fused 0.05 Velociraptor Dromaeosauridae late in ontogeny Sinornithoides Troodontidae absent Microraptor gui Dromaeosauridae fused Anchiornis Troodontidae absent Mei long Troodontidae absent Archaeopteryx Aves absent fused, additional Jeholornis prima Aves: Jeholornithiformes ossifications Confuciusornis Aves: sanctus Confuciusornithiformes fused Sapeornis Aves: Sapeornithiformes absent Eopengornis Aves: Enantiornithes fused Parabohaiornis Aves: Enantiornithes fused, keeled Archaeorhynchus Aves: Ornithuromorpha fused, keeled *Measurements for Allosaurus were taken from Madsen (1976); measurements for Velociraptor were taken from Norell and Mackovicky (1999). Histology Methods We selected specimens from the upper and lower size limit of our sample: STM0-5, STM0-8 and STM0-93 were sampled for Anchiornis; STM15-6, STM15-32 and STM15-70 were sampled for Sapeornis (Fig. S4). Two bone samples were taken from each of the specimens when possible, as close to midshaft as preservation allowed: the humerus and femur were sampled in all specimens except STM15-32, in which only the femur could be sampled, and STM0-8, in which the tibia was sampled and not the femur (Fig. S4). The samples were taken using a micro-saw and were embedded in EXAKT

26 Technovit 7200 one-component resin and allowed to dry for 24 hours. The samples were then cut and polished until the desired optical contrast was reached. The samples were viewed under normal and polarized light using a Leica DM-RX polarizing microscope. Measurements were taken with the computer software ImageJ 1.43r. Histological terminology is mainly sensu de Ricqlès (1976) and Chinsamy-Turan (2005). We follow Erickson et al. (2009) and consider channels as an indicator of the extent of vascularization. Results Anchiornis STM0-5 (small individual, femur length 41 mm): the humeral compacta is thin (3.2 mm thick) formed by a narrow (24 µm), even inner circumferential layer (ICL) of endosteally derived lamellated bone with well organized osteocyte lacunae and a thick outer layer of woven textured bone, densely packed with haphazardly arranged osteocyte lacunae. Vascularization is primarily longitudinal and some small primary osteons are present, located closer to the ICL than the periosteal surface. Areas of the compacta are a dark brown, indicating bacterial invasion. The femoral compacta varies in thickness ( mm). Like the humerus, there is a thin ICL and an outer layer with densely packed, haphazardly arranged osteocyte lacunae; the femur differs in that the upper layer is poorly vascularized with only a few longitudinal canals and small primary osteons and the haphazardly arranged osteocyte lacunae are relatively less densely packed. One area of the bone shows signs of heavy remodeling, with huge secondary osteons (diameter ranging from µm) (Fig. S8a). In another region of the compacta the upper layer is formed of strongly parallel fibered bone, clearly identifiable under polarized light, with fat osteocyte lacunae and moderate vascularization (Fig. S8b).

27 Anchiornis STM0-93 (medium sized individual, femur length 69 mm): The humeral compacta possesses a thin, even ICL of endosteally derived lamellated bone tissue followed by a thick upper layer of woven bone with longitudinal and reticular vascularization. The osteocyte lacunae are plump and lack organization but lack the chaotic distribution observed in STM0-5. There are numerous primary osteons and no LAGs are present. An isolated region with several secondary osteons is recognized. The secondary osteons are densely packed just outside the ICL. The femur also shows a thin ICL and a thick upper layer of woven textured bone with primarily longitudinal vascularization. Closer to the ICL the woven textured layer is nearly haversian, with numerous large secondary osteons; this haversian layer varies in thickness, in some areas approaching the periosteal surface. Secondary osteons are partially eroded by the ICL in some places indicating that this remodeling occurred before the medullary expansion in which the current ICL was deposited. Compared with the humeral cortex, vascularization is lower in the femur, and there are fewer anastomosing circular and radial canals. Anchiornis STM0-8 (large individual, femur length 86 mm): The humerus and tibia compacta are formed by three distinct layers: a think ICL of endosteally derived lamellated bone, a thick middle layer of strongly vascularized fibrolamellar bone, and a thin outer circumferential layer (OCL) of avascular parallel fibered bone marked by LAGs. The osteocyte lacunae are plump and haphazardly arranged except in the OCL and ICL where they are more organized and flattened. The humeral section (thickness varies from mm) shows both longitudinal, plexiform, and reticular vascularization

28 (Fig. S9). A few primary osteons are present, and all located in the inner half of the middle layer. The ICL preserves secondary osteons (Fig. S9a). The number of lines of arrested growth (LAGs) varies around the circumference of the compacta; there are anywhere from one to five LAGs present (Fig. 5a). The first LAG is a double LAG in most places; the outermost LAGs are closely spaced. The OCL, marked by the first LAG, is avascular, and the osteocyte lacunae become well increasingly organized approaching the periosteum. Sharp s fibers, oriented almost perpendicular to the periosteum, cross the majority of the OCL (Fig. S9b). Vascularization is primarily longitudinal in the tibia compacta (thickness varies from mm), which is nearly entirely formed of primary osteons. The number of LAGs is not uniform around the whole cortex, and the number varies from two to four; the first is followed by an annulus of lamellated bone. The OCL bears several simple longitudinal canals, a few of which interrupt a LAG. Sapeornis STM15-6 (smaller individual, femur length 57 mm): The compacta of the humerus (poorly preserved, thickness unknown) reveals a thin ICL, a middle layer of more woven textured bone, and an outer layer of more parallel fibered bone. The middle layer is highly vascularized with longitudinal and reticular canals, primary osteons, and densely packed, haphazardly arranged osteocyte lacunae. The upper layer, not a true OCL, is parallel fibered, with osteocyte lacunae that are flatter, more organized, and arranged parallel to the bone surface. There are only longitudinal canals, some of which are located on the periosteal surface, indicating active bone deposition. The thickness of the middle and outer layer vary around the circumference of the bone, their relative thickness being inversely proportionate to each other. The femur is largely the same

29 except with a thicker, more uneven ICL, and very little reticular vascularization. In one area the bone tissue just outside the ICL is avascular. Sapeornis STM15-32 (medium individual, femur length 75 mm): Only the femur was sampled; the compacta is thick ( mm) and displays a thick ICL of endosteally derived lamellated bone with plump but well organized osteocyte lacunae, a middle layer of more woven textured bone with primary and secondary osteons and longitudinal and reticular vascularization, and an OCL of poorly vascularized more parallel fibered bone with two to four LAGs near the outer surface. Only two LAGs are visible in some areas but the outer LAG splits into a double LAG in some places, and even appears to split a second time revealing four LAGs in others. The contact between the middle layer and OCL is indistinct; the bone rather shows a gradual reduction in vascularization and increase in organization of the collagen fibers towards the periosteal surface. Sapeornis STM15-70 (larger individual, femur length 81 mm): the compacta of the humerus shows a thick ICL (1/3 the width of the compacta) followed by an even thicker outer layer of woven-textured bone with longitudinal and reticular vascularization and numerous primary osteons. This layer is interrupted nearly half way (closer to medullary surface) by a single LAG. The bone tissue on either side of this LAG is largely the same, except that the circular anastomoses are more frequently observed in the outer cortex while reticular anastomoses are more common medial to the LAG. Secondary osteons are not present. The femoral section is poorly preserved and the histology can only be

30 observed in one area; the bone is woven textured with plump disorganized osteocyte lacunae, numerous primary osteons, and primarily longitudinal vascularization.

31 Supplemental Figures Figure S1. Close up of Sapeornis chaoyangensis preserving the complete or nearly complete gastral basket but no sternal ossifications. A, IVPP V13276; B, IVPP V13396 (subadult specimen, former holotype of S. angustis ). Scale bars equal ten millimeters. Anatomical abbreviations (not listed in Figure 1 caption): l, left; r, right.

32 Figure S2. Details of thoracic girdle in derived maniraptorans: A, coracoid and scapula in Jeholornis (Aves) STM2-49; B, coracoids in Caudipteryx STM4-3; C, scapulocoracoid in Anchiornis STM0-31; D, coracoid and sternal plate in Microraptor STM5-50 (young subadult specimen, sternal plates unfused); E, Sapeornis (Aves) STM15-45; F, sternum in Microraptor IVPP V Scale bars equal 10 mm. Anatomical abbreviations (not listed in Figure 1 caption): ct, coracoidal tubercle; f, coracoidal fenestra; pc, procoracoid process; sca, scapula; sn, supracoracoidal nerve foramen; stn, sternum.

33 Figure S3. Basal bird sterna. A 1, Jeholornis STM2-46 (Jeholornithiformes), A 2, IVPP V13274; B, Eopengornis STM24-1 (Enantiornithes); C, Jinzhouornis IVPP V12352 (Confuciusornithiformes); D, juvenile Longipteryx IVPP V12552 (Enantiornithes); E, Archaeorhynchus IVPP V17091 (Ornithuromorpha). Scale bars equal 10 mm. Anatomical abbreviations (not listed in Figure 1, S2 captions): ao, accessory ossification; cl, craniolateral process; hyp, hypocleidium; it, intermediate trabecula; lp, lateral process; lt, lateral trabecula; xp, xiphoid process.

34 Figure S4. Mature specimens with arrows indicating where histological samples were collected. A, Anchiornis STM0-93; B, Sapeornis STM15-32.

35 Figure S5. Mature long bone histology of Anchiornis and Sapeornis under normal light. A, transverse humeral section of a mature specimen of Anchiornis (STM 0-8) showing the well developed ICL and OCL with several LAGs (ranging from one to five throughout section); B, transverse femoral section of a mature specimen of Sapeornis (STM15-32), showing the middle layer of woven textured bone sandwiched between the poorly vascularized ICL and OCL; note the first LAG is a double LAGs.

36 Figure S6. Complete transverse sections of femur and humerus of the smallest sampled specimen of Sapeornis (STM 15-6) under normal light. A, femoral compacta mainly composed by woven textured bone tissue highly vascularized by longitudinal canals; B, humeral histology is largely the same as the femur, except with less reticular vascularization.

37 Figure S7. Humeral and femoral histology of a large specimen of Sapeornis (STM 15-70). A, complete transverse section of humerus under normal light; the medullary cavity is lined by a thick and poorly vascularized ICL, and the outer compacta is formed by woven bone tissue with longitudinal and reticular canals, interrupted by a LAG; B, transverse section of the femur under polarized light. The compacta is largely formed by woven bone tissue and the canals are mainly longitudinally oriented. No LAGs are visible in the poorly preserved femoral section.

38 Figure S8. Details of the transverse femoral section of the smallest sampled specimen of Anchiornis (STM 0-5) under polarized light. A, region of compacta with large secondary osteons, indicating heavy remodeling of the bone tissue; B, peripheral region of the compacta showing the collagen fibers becoming increasingly parallel as they approach the outmost cortex.

39 Figure S9. Details of the transverse humeral section of the largest sampled specimen of Anchiornis (STM 0-8). A, compacta near the ICL showing developed secondary osteons under normal light; B, compacta near the outermost cortex where Sharp s fibers run almost perpendicular to the bone surface and nearly cross the entire OCL under normal light; C, cross section of the compacta, which is well vascularized by longitudinal canals with three LAGs present in the OCL where a single longitudinal canal is visible. Figure S10. Fusion in Sapeornis STM A, left carpometacarpus, appears well fused with exception of suture between alular and major metacarpals, which is also present in other known specimens of Sapeornis and potentially represents the adult condition; B, left distal tibiotarsus and proximal tarsometatarsus, both fully fused with their respective tarsals. Anatomical abbreviations: al, alular metacarpal; c, proximal carpal; ma, major metacarpal; mi, minor metacarpal; tbt, tibiotarsus; tmt, tarsometatarsus.

40 Figure S11. Gastralia in basal birds: A, Jeholornis STM2-47; B, Confuciusornis STM13-52; C, pengornithid enantiornithine Eopengornis STM24-1; D, bohaiornithid enantiornithine Parabohaiornis IVPP V18960, note cranial pair of gastralia articulates with the xiphoid process of the sternum; E, basal ornithuromorph Archaeorhynchus IVPP V Scale bars equal 10 mm. Anatomical abbreviations (not listed in Figure 1, S2-3 captions): cav, caudal vertebrae; ili, ilium.