Digital restoration of the pectoral girdles of two Early Cretaceous birds and implications for early-flight evolution

  1. Shiying Wang
  2. Yubo Ma
  3. Qian Wu
  4. Min Wang  Is a corresponding author
  5. Dongyu Hu  Is a corresponding author
  6. Corwin Sullivan
  7. Xing Xu  Is a corresponding author
  1. Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, China
  2. CAS Center for Excellence in Life and Paleoenvironment, China
  3. University of Chinese Academy of Sciences, China
  4. University of Alberta, Canada
  5. Shenyang Normal University, China
  6. Philip J. Currie Dinosaur Museum, Canada
  7. Centre for Vertebrate Evolutionary Biology, Yunnan University, China

Abstract

The morphology of the pectoral girdle, the skeletal structure connecting the wing to the body, is a key determinant of flight capability, but in some respects is poorly known among stem birds. Here, the pectoral girdles of the Early Cretaceous birds Sapeornis and Piscivorenantiornis are reconstructed for the first time based on computed tomography and three-dimensional visualization, revealing key morphological details that are important for our understanding of early-flight evolution. Sapeornis exhibits a double articulation system (widely present in non-enantiornithine pennaraptoran theropods including crown birds), which involves, alongside the main scapula-coracoid joint, a small subsidiary joint, though variation exists with respect to the shape and size of the main and subsidiary articular contacts in non-enantiornithine pennaraptorans. This double articulation system contrasts with Piscivorenantiornis in which a spatially restricted scapula-coracoid joint is formed by a single set of opposing articular surfaces, a feature also present in other members of Enantiornithines, a major clade of stem birds known only from the Cretaceous. The unique single articulation system may reflect correspondingly unique flight behavior in enantiornithine birds, but this hypothesis requires further investigation from a functional perspective. Our renderings indicate that both Sapeornis and Piscivorenantiornis had a partially closed triosseal canal (a passage for muscle tendon that plays a key role in raising the wing), and our study suggests that this type of triosseal canal occurred in all known non-euornithine birds except Archaeopteryx, representing a transitional stage in flight apparatus evolution before the appearance of a fully closed bony triosseal canal as in modern birds. Our study reveals additional lineage-specific variations in pectoral girdle anatomy, as well as significant modification of the pectoral girdle along the line to crown birds. These modifications produced diverse pectoral girdle morphologies among Mesozoic birds, which allowed a commensurate range of capability levels and styles to emerge during the early evolution of flight.

Editor's evaluation

The authors provide new 3D fossil findings in Sapeornis, an avialan that lived during the Early Cretaceous period, a key node in our understanding of the evolution of avian flight. The functional reconstruction of two critical skeletal elements of the avian flight apparatus, the scapula and coracoid, enables the authors to hypothesize how the evolution of the scapula and coracoid enabled the modern avian wing stroke. The new 3D morphological reconstruction enables future integrative studies of Sapeornis flight performance based on biomechanical, muscle physiological, and aerodynamic principles and is thus a key building block to inform our general understanding of the evolution of avian flight.

https://doi.org/10.7554/eLife.76086.sa0

Introduction

The evolution of powered flight in birds was one of the great transformations in vertebrate history and involved a suite of dramatic anatomical changes that were required to produce a functional flight apparatus (Dudley and Yanoviak, 2011; Padian, 1985; Rayner, 1988; Videler, 2005; Burch, 2014; Jasinoski et al., 2006). The pectoral girdle, a skeletal structure that connects the forelimb to the trunk, is a key component of the flight apparatus, and its function and evolutionary history have been extensively studied (Baier et al., 2007; Bock, 2013; Novas et al., 2020; Senter, 2006; Burch, 2014; Jasinoski et al., 2006; Ostrom, 1976). The morphology of the pectoral girdle in Late Cretaceous enantiornithine birds is well known from several three-dimensionally preserved specimens (Atterholt et al., 2018; Chiappe and Walker, 2002; Chiappe et al., 2007). However, most Early Cretaceous bird fossils are essentially two-dimensionally preserved as slab specimens, and accordingly do not offer a full anatomical picture of the flight apparatus, a limitation that greatly hinders studies of early flight. Here, we reconstruct the pectoral girdles of the non-ornithothoracine bird Sapeornis chaoyangensis (PMoL-AB00015) and the enantiornithine bird Piscivorenantiornis inusitatus (IVPP V 22582) using computed tomography and three-dimensional visualization. Our renderings are the first three-dimensional ones of the pectoral girdle for these two Early Cretaceous birds and reveal some important anatomical details for understanding pectoral girdle evolution. One main objective of our study was to better understand the evolution of the scapula-coracoid articulation and triosseal canal on the line to crown group birds because the form of the scapula-coracoid articulation has traditionally been used to distinguish between enantiornithines and euornithines (ornithuromorphs), whereas the nature of the triosseal canal has implications for the course of the tendon of M. supracoracoideus and thus for the mechanics of the upstroke during flight.

The pectoral girdle underwent dramatic changes in the shape, position, and orientation of each component element along the line to crown birds (Xu, 2002). In early-diverging theropods (Figure 1A), the scapula and coracoid lie obliquely on the ribcage with the anatomically cranial (anterior) edge of the coracoid significantly lower than the trunk vertebral column; the glenoid fossa is caudoventrally (posteroventrally) directed; the scapula has a large craniodorsally (anterodorsally) oriented acromion process; and the coracoid is a laterally facing semicircular plate with a small coracoid tubercle (called the biceps tubercle in some studies, and a precursor to the acrocoracoid process of birds). In early-diverging maniraptoriform theropods, the coracoid is somewhat biplanar, being divided by a line of deflection into two subtriangular areas that we term the sternal wing (called the distal ramus in Xu, 2002) and the scapular wing (called the proximal ramus in Xu, 2002) of the coracoid. In some maniraptoriform species, the line of deflection is marked by a ridge originating from the coracoid tubercle, on the coracoid’s lateral surface. In other species, a distinct ridge is absent, and only the deflection itself defines the boundary between the scapular and sternal wings. The deflection causes the originally laterally facing lateral surface of the coracoidal sternal wing to be directed somewhat cranially (anteriorly) and ventrally. The scapular wing comprises the glenoid fossa, the scapular articulation, and the thin sheet normally housing the supracoracoid foramen, which accommodates the supracoracoid nerve. In early-diverging pennaraptorans (Figure 1B, E, and F), the glenoid of the scapulocoracoid is close to the trunk vertebral column and faces laterally; the scapula has a small acromion process; the scapular blade is nearly parallel to the trunk vertebral column, with its lateral surface tilted dorsally; and the coracoid has a large coracoid tubercle, a large sternal wing, and a small scapular wing. The originally lateral surface of the sternal wing faces cranially (anteriorly) and that of the scapular wing is directed craniodorsally, so that the coracoid more closely resembles an inverted ‘L’ than a semicircular plate in lateral view. In most avialans (Figure 1C and G), the glenoid fossa is dorsolaterally oriented; the scapular acromion process protrudes farther cranially; the scapular blade is twisted so that the anatomically lateral surface of the proximal portion faces dorsally, whereas that of the distal portion faces dorsolaterally; and the coracoid is a strut-like structure with several derived features (e.g., the coracoid tubercle is enlarged to form the acrocoracoid process; the sternal wing is elongated in a craniodorsal-caudoventral direction; the scapular wing is highly reduced, bringing the scapular articulation and glenoid fossa extremely close to the sternal wing; and in many species, the scapular wing gives rise to a procoracoid process). In species with an extremely small scapular wing, the supracoracoid foramen is either absent or located in the sternal wing. Figure 1 illustrates the key morphological features of the pectoral girdle in different theropod groups and the terms used in this article, though we admit that a complete evolutionary picture of the theropod pectoral girdle has yet to be presented and there are different opinions on what terms should be used for certain structures (Xu, 2002; Novas et al., 2021).

The position of the pectoral girdle and the form of the coracoid in different theropod groups.

(A–C) Skeletal silhouettes showing the anatomical position of the pectoral girdle in (A) the early-diverging theropod Coelophysis, (B) the early-diverging pennaraptoran Microraptor, and (C) the modern bird Columba. The M. supracoracoideus is illustrated in (C) but typically covered by the M. pectoralis, which is not illustrated. (D–G) Illustrations of the left coracoids of (D) Coelophysis (modified from Tykoski, 1998), (E) the early-diverging pennaraptoran Sinornithosaurus (modified from Xu et al., 1999), (F) the early-diverging avialan Archaeopteryx (modified from Wellnhofer et al., 2009), and (G) the early-diverging avialan Jeholornis (based on STM 2-49 and IVPP V 13886). Coracoid of Coelophysis in lateral view, coracoids of other taxa in ventral view.

Results

The nearly complete pectoral girdle elements of S. chaoyangensis PMoL-AB00015 and P. inusitatus IVPP V 22582 have been three-dimensionally rendered in detail based on computed tomography scan data (Figure 2, Figure 3, and Figure5). However, the bones have been compressed during fossilization, so the renderings do not precisely capture the original morphology. Originally, the furcula of Sapeornis PMoL-AB00015 was probably slightly curved caudally (despite being straight in our rendering), and the angle between the scapular and sternal wings of the coracoid that contacted the scapula and the sternum was probably smaller than in our rendering. Because of these distortions, the pectoral girdle of Sapeornis based on digitally articulating our uncorrected renderings is characterized by a larger distance between the two coracoids, and a more ventrally oriented glenoid fossa, than would have been present in the skeleton of the living animal (Gao et al., 2012). However, these errors do not affect our major conclusions.

Pectoral girdle bones of Sapeornis chaoyangensis PMoL-AB00015.

(A–D) Left scapula in lateral, dorsal, medial (costal), and ventral views. (E–H) Left coracoid in ventral, dorsal, lateral, and cranial views. (I–L) Furcula in cranial, caudal, lateral, and ventral views. The black arrows in (J) and (L) indicate the concave surface for the tendon of M. supracoracoideus.

Pectoral girdle bones of Piscivorenantiornis inusitatus IVPP V 22582.

(A–D) Left scapula in lateral, dorsal, medial (costal), and ventral views. (E–H) Right coracoid in ventral, dorsal, lateral, and cranial views. (I–L) Furcula in cranial, caudal, left, and ventral views.

Osteology of the pectoral girdle of Sapeornis PMoL-AB00015

The cranial part of the left scapula is curved medially and ventrally. The scapular blade is slightly twisted about its longitudinal axis, so that the anatomically lateral surface of the proximal portion faces dorsally, whereas that of the distal portion faces dorsolaterally as in extant birds (Figure 2). The acromion is short. As in the dromaeosaurid Rahonavis (Forster et al., 2020), a broad flange protrudes medially from the acromion (Figure 2), adding to a previously known set of derived similarities shared by Rahonavis and some early-diverging avialans (Novas et al., 2018). As in Jeholornis (Zhou and Zhang, 2003a), the scapular glenoid fossa faces mainly ventrally but also slightly laterally, showing more lateral deflection than in deinonychosaurs (e.g., Sinovenator and Rahonavis) (Forster et al., 2020) and Archaeopteryx (Zhou and Zhang, 2003a). The articular surface for the coracoid consists of two parts: a deeply concave main surface situated craniomedial to the glenoid facet on the ventral surface of the scapula and a slightly concave subsidiary surface positioned more craniomedially (Figure 4). A weak tubercle lies on the ventrolateral margin of the scapular blade, possibly representing the muscle insertion site for M. subscapulare (Figure 2; Gianechini et al., 2018). A short and shallow groove (Figure 2) on the medial surface of the scapula, close to the glenoid fossa and subparallel to the ventral margin, probably represents another muscle insertion site.

Comparison of scapula and coracoid morphology across various paravian taxa.

Each panel shows articulated left scapula and coracoid in ventral view (on left) and opposing articular surfaces of left scapula and coracoid (on right, with cranial direction toward top of figure for both scapula and coracoid). (A) Sinovenator changii (mirrored), (B) Sapeornis chaoyangensis, (C) Piscivorenantiornis inusitatus, (D) Tyto alba, (E) Egretta garzetta, and (F) Pavo muticus.

The coracoid is in general similar to that of non-avialan pennaraptorans in having a large scapular wing, with the glenoid fossa and scapular articulation well separated from the sternal wing. The caudal margin of the sternal wing is slightly convex and lacks a visible articular facet for the sternum (Figure 2), rather than being straight to concave and bearing a sternal facet as in Jeholornis (Wang et al., 2020a) and most ornithothoracines (Atterholt et al., 2018; Wang and Zhou, 2017a). The lack of a sternal facet lends further support to previous inferences that an ossified sternum is genuinely absent in this early pygostylian lineage (Zheng et al., 2014). In living birds, the ossified sternum provides the major attachments for M. supracoracoideus and M. pectoralis; while in Sapeornis, the large coracoid may have served to provide a proximal attachment surface for these muscles (Zheng et al., 2014). On the ventral surface of the coracoid, a ridge extends from the acrocoracoid process to the distomedial corner of the bone, clearly demarcating the scapular and sternal wings of the coracoid (Figure 2) as in some maniraptoriform theropods (e.g., Sinornithosaurus) (Xu, 2002). The sternal wing is short, having a ratio of cranial-caudal length to medial-lateral width at the caudal margin of only 1.17. This is close to the value for Archaeopteryx (~1.15), but differs from those for the more elongated sternal wings of Jeholornis, Confuciusornis, and most ornithothoracines (generally >1.5). The scapular wing is large, as in most non-avialan pennaraptorans, but in contrast to the reduced scapular wing seen in such dromaeosaurids as Microraptor and in the avialans Archaeopteryx and Jeholornis (Wang et al., 2020aWang et al., 2020a; Wellnhofer et al., 2009; Xu et al., 2003). In most euornithines, the scapular wing is likewise small, and curves ventrally and gives rise to a narrow procoracoid process at the craniomedial corner (e.g., Egretta garzetta; Figure 4E). In some euornithine species (e.g., Tyto alba; Figure 4D), however, the scapular wing is thin and relatively large, somewhat similar to the condition in non-enantiornithine pennaraptorans. As in Jeholornis (Wang et al., 2020a), the supracoracoid foramen (Figure 2) is large and positioned within the scapular wing.

The acrocoracoid process is short and blunt, and extends slightly above the midpoint of the coracoidal glenoid fossa as in Jeholornis and Confuciusornis (Turner et al., 2012Wang and Zhou, 2018bWang and Zhou, 2018b; Wang et al., 2020a; Zhou and Zhang, 2003b; Zhou and Zhang, 2003b). In enantiornithines (Panteleev, 2018; Chiappe and Walker, 2002), the acrocoracoid process extends slightly above the dorsal margin of the coracoidal glenoid fossa. In most euornithines (e.g., Figure 4D–F), this process is proportionally longer and extends much further beyond the glenoid than in non-euornithine birds. In non-avialan theropods and Archaeopteryx, by contrast, the acrocoracoid process (frequently described as coracoid tubercle or biceps tubercle) is located cranioventral to the glenoid fossa (Mayr et al., 2005; Novas et al., 2021). The acrocoracoid process of Sapeornis forms a shelf-like structure projecting dorsally, cranially, and laterally from the lateralmost part of the coracoid, a condition not known in other birds. A small shallow fossa with an irregular surface, located at the medial end of the cranioventral surface of the acrocoracoid process (Figure 2), may have provided an attachment point for a coracoclavicular ligament connecting the coracoid and furcula (Figure 5). In some volant extant birds, by contrast, the coracoclavicular ligament attaches to the cranial edge of the medial surface of the acrocoracoid process (Ghetie, 1976).

Simplified phylogeny with hypothetical steps in pectoral girdle evolution.

The pectoral girdles of Sapeornis chaoyangensis, Piscivorenantiornis inusitatus, and Pavo muticus (from top to bottom) are shown in cranial, dorsal, and left lateral views. The pink lines in the dorsal and lateral views represent the tendon of M. supracoracoideus, and the gray line in the dorsal view of the Sapeornis rendering represents the coracoclavicular ligament that connects the coracoid and furcula. Phylogenetic framework following Wang et al., 2018a.

The glenoid fossa is on the craniolateral corner of the dorsal face of the coracoid and wraps onto the cranial margin. The scapular articular surface is situated entirely on the coracoid’s cranial margin and is not separated from the coracoidal glenoid fossa by any distinct border. The coracoidal glenoid fossa is weakly concave and faces dorsocaudally and only slightly laterally, whereas the lateral tilt of the fossa is greater in late-diverging birds (Figure 4).

Like the opposing surface on the scapula, the scapular articular surface on the coracoid is divided into main and subsidiary parts. The former is weakly convex, to match the concave main articular surface for the coracoid on the scapula, whereas the latter extends along the cranial margin of the scapular wing of the coracoid and contacts the subsidiary coracoidal articular surface on the scapula.

The robust, craniocaudally compressed furcula presumably would not have been as flexible as those of volant extant birds, which have mediolaterally compressed rami and in at least some cases act as a spring during flapping flight (Boggs et al., 1997; Nesbitt et al., 2009). The omal third of the ramus is slightly narrower than the remainder and terminates in a blunt epicleidium. In dorsal view, the epicleidium is twisted laterally by about 80° relative to the ramus. Similar torsion can also be observed in some other early birds (e.g., Confuciusornis) and in some deinonychosaurs (e.g., Halszkaraptor, Buitreraptor) (Cau et al., 2021; Gianechini et al., 2018). The lateral surface of the epicleidium is concave, making the epicleidium C-shaped in dorsal view. This concave surface possibly accommodated the tendon of the supracoracoideus muscle, which would have passed through the partially closed triosseal canal and over the small acrocoracoid process to reach its insertion on the humerus (Figure 5). As in Confuciusornis (Wu et al., 2020) and Xiaotingia (Xu et al., 2011), a small caudal projection is located on the medial side of the epicleidium and probably articulated with the acromion of the scapula. The short, slender hypocleidium is broken away, but the probable outline of the hypocleidium is indicated in Figure 2I–K based on another specimen of Sapeornis (IVPP V 19058).

Osteology of the pectoral girdle of the P. inusitatus IVPP V 22582

The pectoral girdle is closely comparable in morphology to those of other enantiornithines (Hu et al., 2015a; Zhang et al., 2014; Chiappe and Walker, 2002). The long and robust acromion of the scapula is separated by a neck from the coracoidal articular surface. The cranial part of the medial surface of the acromion forms a flat articular surface for the furcula, as in many other enantiornithines (e.g., Rapaxavis and Halimornis) (Chiappe et al., 2002; O’Connor et al., 2011). The subtriangular scapular glenoid fossa faces more laterally than in Sapeornis and non-avialan theropods (Brusatte et al., 2013; Funston et al., 2020), although the orientation of the fossa is nevertheless also somewhat ventral. The slightly concave coracoidal articular surface lies cranial and medial to the scapular glenoid fossa, and is nearly perpendicular to the latter. This surface corresponds to the main coracoidal articular surface on the scapula of non-enantiornithine pennaraptorans, and that the subsidiary coracoidal articular surface is absent. The scapular blade is curved ventrally and tapered caudally.

The coracoid is subtriangular in dorsal or ventral view. The acrocoracoid process is rounded and minimally developed, as is typical in enantiornithines (Atterholt et al., 2018; Zhang et al., 2014; Panteleev, 2018). The coracoidal glenoid fossa is craniocaudally elongated and faces caudally and somewhat dorsolaterally. The slightly convex scapular articular surface is smaller than the coracoidal glenoid fossa, and is situated dorsomedial to the latter as in other ornithothoracines. In Sapeornis and non-avialan theropods, the scapular articular surface is proportionally larger and situated directly medial to the coracoidal glenoid fossa. This surface corresponds to the main scapular articular surface on the coracoid of non-enantiornithine pennaraptorans, and that the subsidiary scapular articular surface is absent. On the medial side of the glenoid fossa, between the acrocoracoid process and scapular articular surface, lies a small incisure that is present in most enantiornithines (Hu et al., 2015bHu et al., 2015b; Panteleev, 2018; Wang et al., 2016d). This structure is identified as the sulcus M. supracoracoideus, through which the M. supracoracoideus tendon passed (Hu et al., 2012; Panteleev, 2018). A large impression (Figure 3) associated with attachment of Lig. acrocoracohumerale is located on the cranioventral surface of the acrocoracoid process, above the level of the ventral margin of the glenoid fossa, and faces cranially and slightly laterally. In extant birds, by contrast, the equivalent feature is located well craniodorsal to the glenoid fossa and faces more laterally. The sheet-like component of the scapular wing of the coracoid, termed the ‘medial crest’ by Panteleev, 2018, is extremely narrow and terminates at the base of the scapular articular surface. The loss of the subsidiary scapular articular surface is connected to the reduction of the coracoid’s scapular wing. The scapular wing is perforated by a small supracoracoid foramen, as in most enantiornithines (Atterholt et al., 2018; Chiappe et al., 2007; Panteleev, 2018; Wang et al., 2016a). The neck of the sternal wing is proportionally shorter than the equivalent structure in most extant birds (Panteleev, 2018).

The furcula is robust and generally Y-shaped, with an interclavicular angle of only about 50° and a long hypocleidium, as in most enantiornithines (Hu et al., 2015a; Zhang et al., 2014). The epicleidium is expanded both craniocaudally and mediolaterally to form a dorsally facing articular facet for the coracoid, as in other enantiornithines (Atterholt et al., 2018). The midshaft of each ramus is ‘L’ shaped in cross section owing to the presence of a deep caudolateral groove, another characteristic of Enantiornithes (Atterholt et al., 2018; Chiappe et al., 2007; Close et al., 2010). The omal part of this groove faces somewhat laterally due to torsion of the ramus and tapering of the cranial surface of the ramus from the lateral side. The groove extends ventrally to the end of the hypocleidium, producing a high keel on the caudal surface of the hypocleidium between the right and left grooves. The bilaterally compressed form of the hypocleidium is shared with many enantiornithines (Wang et al., 2014; Wang et al., 2020b), but differs from the craniocaudal compression seen in some oviraptorosaurs, some troodontids, and Sapeornis (Nesbitt et al., 2009; Xu and Norell, 2004).

Discussion

S. chaoyangensis PMoL-AB00015 and P. inusitatus IVPP V 22582 provide significant new information on the pectoral girdle morphology of these two Early Cretaceous birds. This information bears, in particular, on the following issues pertaining to the early evolution of the avialan pectoral girdle.

Morphology of the scapula-coracoid articulation, the scapular wing of the coracoid, and the procoracoid process

The scapula-coracoid articulation is variable in morphology among pennaraptoran theropods. In non-avialan pennaraptorans, the structure of the articulation between the scapula and coracoid is not well known. This is mainly because the two elements tend to fuse, albeit normally at a relatively late ontogenetic stage, to form a scapulocoracoid. For example, the scapula and coracoid are tightly sutured together, but not fused, in a juvenile specimen of Velociraptor mongoliensis (MPC-D100/54), and a fused scapulocoracoid is seen in an adult specimen (IGM 100/986) (Hone et al., 2012; Norell and Makovicky, 1999). Complete fusion of the scapulocoracoid, leaving no visible suture, is the usual condition in adult individuals (e.g., Citipati osmolskae IGM 100/1004, Anzu wyliei CM 78 001) (Lamanna et al., 2014; Norell et al., 2018). Among non-ornithothoracine avialans, all known jinguofortisids, and all known confuciusornithiforms except a single subadult Eoconfuciusornis zhengi specimen (IVPP V 11977), possess a fused scapulocoracoid (Wang et al., 2018a; Wu et al., 2021). The occurrence of scapula-coracoid fusion in Archaeopteryx is controversial, but the scapula and coracoid are separately preserved in some specimens (Kundrát et al., 2018; Mayr et al., 2005; Wellnhofer et al., 2009; Wu et al., 2021). In sapeornithiforms and jeholornithiforms, the scapula and coracoid are not fused (Zhou and Zhang, 2003b; Zhou and Zhang, 2003b). Among ornithothoracines, a fused scapulocoracoid is known only in flightless paleognaths (Wu et al., 2021).

In several deinonychosaurs (e.g., Sinovenator and Rahonavis), the scapula and coracoid not only fail to fuse but remain loosely rather than tightly connected, contacting one another via smooth articular surfaces rather than via a firm interdigitating suture as in other non-avialan pennaraptorans (Forster et al., 2020). In Sinovenator (Figure 4A), the glenoid fossa of the coracoid is smaller than the scapular articular surface as in other non-avialan theropods. The reverse is true in crown group birds, and also in many stem birds in which this anatomical region is well known, such as Piscivorenantiornis (Figure 3), Elsornis (Chiappe et al., 2007), Mirarce (Atterholt et al., 2018), and Gansus (Wang et al., 2016c). The scapular articular surface is located on the scapular wing of the coracoid and consists of two parts: a broadly convex main articular surface and a flat to shallowly concave subsidiary articular surface. The corresponding coracoidal articular surfaces on the scapula are both concave. The main and subsidiary articulations together form what is termed here a double articulation between the scapula and the coracoid. Bambiraptor also has a double articulation, in which the smaller, shallowly convex subsidiary articular surface on the coracoid is located mediodorsally to the flat main articular surface and is separated from the latter by a low ridge. The corresponding coracoidal articular surfaces on the scapula are shallowly concave, as in Sinovenator. This implies a gap between the main articular surfaces on the scapula and coracoid, which in life was presumably filled with cartilage. The dromaeosaurid Rahonavis possesses a double articulation, with a slightly concave main articular surface for the coracoid as in Sinovenator (Forster et al., 2020). As described above, the scapula-coracoid articulation of Sapeornis resembles that of Sinovenator in that the coracoid has a convex main scapular articular surface (matching the scapula’s concave main coracoidal articular surface) and a subsidiary articular surface that would have contacted the cranioventral margin of the acromion process of the scapula. Sapeornis thus has a double articulation, consistent with the plesiomorphic overall shape of the scapulocoracoid, as in non-avialan pennaraptorans. Avialans other than enantiornithines have a broadly uniform type of scapula-coracoid articulation, although some morphological variation is present. In Jeholornis and Fukuipteryx, the coracoid has a concave main articular surface for the scapula (Imai et al., 2019; Turner et al., 2012), a feature that has been proposed as an apomorphy of the Euornithes (Ornithuromorpha) in previous studies (Wang and Zhou, 2017a). In most euornithines, the coracoid indeed has a deep cotyla that receives a corresponding convexity on the scapula (Figure 4E). In some crown birds, however, the main scapular articular surface on the coracoid is flat to slightly convex (Figure 4D and F), as in many enantiornithine birds and in some non-avialan theropods, such as Sinovenator. In most euornithines and Jeholornis, the subsidiary articular surface for the scapula is situated partly on the procoracoid process, which is a projection of the dorsomedial margin of the small scapular wing of the coracoid.

Enantiornithines have a strikingly different scapula-coracoid articulation from other pennaraptorans. The presence of a single, convex articular surface, fitting into a cotyla on the cranial end of the scapula, has been widely accepted as a unique feature of the coracoid of enantiornithine birds (Chiappe and Walker, 2002; Panteleev, 2018; Wang and Zhou, 2019). Conversely, modern birds have been considered to display the opposite condition, with the main scapular articular surface on the coracoid being concave and that on the scapula convex. This purported discrepancy is the source of the clade name Enantiornithes, meaning ‘opposite birds.’ As mentioned above, however, a convex scapular articular surface is also seen on the coracoids of some non-avialan theropods, such as Sinovenator, and in some crown birds that are secondarily evolved (Mayr, 2021), though the convexity is less prominent in these taxa than in late-diverging enantiornithine birds. Furthermore, the scapular articular surface on the coracoid is shallowly concave in some enantiornithines, such as pengornithids (e.g., IVPP V 18687 and V 18632). Consequently, the ‘opposite’-type scapula-coracoid articulation is neither present in all enantiornithine birds nor unique to the Enantiornithes. However, our study indicates that the enantiornithine scapula-coracoid articulation is indeed unique, but for a different reason: extreme reduction of the scapular wing of the coracoid and consequent loss of the subsidiary articular surface for the scapula in all enantiornithines (including pengornithids). This results in a single articulation in enantiornithines, in which the coracoid bears only one spatially restricted articular surface for the coracoid. The single articulation of enantiornithines is thus smaller in area, as well as morphologically simpler, than the double articulation of other pennaraptorans.

Combining anatomical details revealed by this study with information from the literature, three important modifications to the scapula-coracoid articulation may be inferred to have occurred among avialans: (1) loss of fusion between the scapula and coracoid in the majority of adult avialans, (2) displacement of the main articular surface for the scapula to a position extremely close to the base of the acrocoracoid process in a clade comprising Jeholornis and pygostylians (reversed in Sapeornis to the primitive condition of having the main articular surface relatively distant from the acrocoracoid process), and (3) establishment of the unique single articulation by extreme reduction of the scapular wing of the coracoid in enantiornithine birds. All non-enantiornithine pennaraptorans, including crown birds, have a double articulation connecting the scapula and coracoid, and in the majority of non-enantiornithine birds, a procoracoid process is present to buttress the double articulation and contribute to the triosseal canal.

Architecture of the triosseal canal

The triosseal canal facilitates powered flapping flight in modern birds by forming a passage to admit the tendon of M. supracoracoideus, a muscle that contributes to humeral elevation and longitudinal rotation (Baumel and Witmer, 1993; Poore et al., 1997). However, the name ‘triosseal canal’ is misleading given the variable composition of this structure in living birds (Livezey and Zusi, 2006). The annotation provided for the triosseal canal by Baumel and Witmer, 1993 explicitly described variation across taxa in the canal’s architecture. In most extant birds, the furcula, coracoid, and scapula all participate (hence the name of the canal) in forming a fully enclosed bony passage. Typically, the epicleidium of the furcula forms the craniomedial wall of the triosseal canal, the acrocoracoid process of the coracoid forms the lateral wall, and the procoracoid process of the coracoid and the cranial margin of the acromion of the scapula form the caudomedial wall. However, in many extant birds (e.g., Phalacrocorax capillatus and E. garzetta), the bony triosseal canal is only partially closed, in that the furcula lacks a bony contact with the scapula but is bound to the latter by Lig. scapuloclaviculare dorsale (Baumel and Witmer, 1993). In Rhea, the Lig. acrocoracoacromiale bridges the tendon of M. supracoracoideus, contributing to a what is functionally a ‘triosseal canal’ (Novas et al., 2021). In certain other birds, a closed bony canal is formed by the coracoid and scapula only, with no contribution from the furcula, or even formed by the coracoid alone via an ossified bridge connecting the acrocoracoid and procoracoid processes (e.g., Upupa epops and Columba livia) (Baumel and Witmer, 1993). Therefore, the triosseal canal is not necessarily formed by all three pectoral elements and is not necessarily a fully enclosed bony passage. Also, the procoracoid process is absent in certain volant crown birds that possess a triosseal canal, including Pavo muticus (Figure 4) and Colius striatus (Mayr, 2021), as well as in the Late Cretaceous galliform-like genus Palintropus (Longrich, 2009). Accordingly, the procoracoid process cannot be considered an essential constituent of the triosseal canal.

Among stem birds, a triosseal canal is widely accepted as present in early-diverging euornithines (Mayr, 2017; Wang et al., 2016b; Zhou and Wang, 2017). In most euornithine specimens (e.g., Yixianornis and Gansus) (Clarke et al., 2006; Wang et al., 2016c), the acrocoracoid process is medially hooked and a prominent procoracoid process is present, features that suggest the existence of a typical, fully enclosed bony triosseal canal formed by the scapula, coracoid, and furcula. Many previous studies have denied the presence of a triosseal canal in non-euornithine birds because of the lack of a long medially hooked acrocoracoid process and a procoracoid process (Novas et al., 2021; Wang and Zhou, 2017a). Although a small and pointed procoracoid process has been reported in Protopteryx (Zhang and Zhou, 2000; Chiappe et al., 2020), this cannot be confirmed in the provided figures and preservation of the omal region is poor. Other studies have argued that a triosseal canal is present in enantiornithines, albeit based on limited evidence (Kurochkin et al., 2013; Zhang and Zhou, 2000). Mayr, 2017 argued that the tendon of M. supracoracoideus ran along the medial side of the acromion in enantiornithine birds, rather than along the lateral side as in crown birds, which would imply that the supracoracoideus pulley system was differently configured in enantiornithines than in euornithines.

Our renderings of the pectoral girdles of Sapeornis and Piscivorenantiornis indicate the presence of a partially enclosed triosseal canal in these early stem birds. In Sapeornis, the lateral wall of the triosseal canal is formed by the acrocoracoid process, the medial wall by the furcula, and the caudomedial wall by the scapular wing of the coracoid and the cranial margin of the acromion of the scapula. In these respects, the triosseal canal of Sapeornis has essentially the same structure as in most extant flying birds. In Piscivorenantiornis, the scapular wing of the coracoid is extremely reduced and lacks a procoracoid process. The caudomedial wall of the triosseal canal is formed by the cranioventral margin of the long acromion process of the scapula and the floor of the sulcus M. supracoracoideus of the coracoid, as in some extant birds that lack a procoracoid process, for example, Corvus corax and P. muticus (Figure 4). Both Sapeornis and enantiornithines have features of the pectoral girdle (e.g., dorsolaterally projecting acrocoracoid process in Sapeornis, and elongate acromion process and widely spaced acrocoracoid processes in enantiornithines) that imply the lack of a coracoid-furcula contact (Mayr, 2017; Novas et al., 2021). Absence of such a contact is a plesiomorphic feature, widely observed among non-avialan theropods (Currie and Zhiming, 2001; Klingler, 2020; Lü, 2003). Thus, Sapeornis and enantiornithines have only a partially enclosed bony triosseal canal, though a ligament could have completed the enclosure of the canal as in some modern birds (Ghetie, 1976). Absence of the cranial bony wall of the triosseal canal presumably would not affect the upstroke in these birds as the acrocoracoid process redirects the tendon of M. supracoracoideus (Baumel and Witmer, 1993).

Pectoral girdle evolution and early flight

Mapping major aspects of pectoral girdle morphology onto an avialan phylogenetic tree suggests that three important evolutionary steps can be defined along the line to the modern flight apparatus and reveals some distinctive features characterizing certain avialan clades (Figure 5). Step I occurs at the base of the clade comprising Jeholornis and pygostylians, and involves torsion and elongation of scapular blade (ratio of scapular length to femoral length about 0.9, compared to 0.68–0.81 in Archaeopteryx [Rauhut et al., 2018], 0.68 in Anchiornis [Hu et al., 2009], and 0.58–0.73 in several dromaeosaurids [Burnham et al., 2000; Hwang et al., 2002; Makovicky et al., 2005]); elongation of sternal wing of coracoid (ratio of cranial-caudal length of sternal wing to medial-lateral width of caudal margin of sternal wing greater than 1.5; reversed to primitive condition in Sapeornis) (Zhou and Zhang, 2003b) shifting of scapular articular surface and glenoid fossa of coracoid to position extremely close to base of acrocoracoid process; reduction in area occupied by scapula-coracoid articulation compared to condition in non-avialan theropods; elongation of acrocoracoid process, which is situated at dorsoventral level of coracoidal glenoid fossa; reduction in distance between left and right coracoids, indicating a relatively narrow and deep chest; reduction in angle between scapula and coracoid (Novas et al., 2021) and establishment of partially closed triosseal canal. Step II occurs at the base of Ornithothoraces and involves cranial extension and thickening of the acromion process (Novas et al., 2021) reduction of the interclavicular angle (generally to less than 65°) (Hu et al., 2015b) and reduction of the supracoracoid foramen. Step III occurs in early-diverging Euornithes and involves increased downward curving of caudal end of scapular blade (O’Connor et al., 2016) shifting of glenoid fossa of scapula onto the external surface of bone, causing the fossa to face dorsolaterally (Wellnhofer et al., 2009) appearance of the procoracoid process on the scapular wing of coracoid (Clarke et al., 2006) medial curving and further elongation of the acrocoracoid process (Novas et al., 2021) further reduction in angle between the scapula and coracoid (Wellnhofer et al., 2009) and the complete bony enclosure of triosseal canal. Regarding distinctive pectoral girdle features in particular taxa, Jeholornis has an unusual combination of a prominent procoracoid process and a large supracoracoid foramen (Lefèvre et al., 2014; Turner et al., 2012; Wang et al., 2020a), Sapeornis has a dorsolaterally oriented acrocoracoid process, and Enantiornithes is characterized by an extremely small scapular wing of the coracoid, a single scapula-coracoid articulation, elongation of the hypocleidium, presence of caudal grooves on the furcular rami and a keel on the caudal surface of the hypocleidium, and further solidification of the furcula-scapula articulation.

The variation in pectoral girdle morphology seen among early birds is suggestive of a similarly wide diversity of flight capabilities and modes, an inference supported by previous studies (Close and Rayfield, 2012; Heers and Dial, 2012; Novas et al., 2021). The position of the acrocoracoid process, or coracoid tubercle, and the orientation of the glenoid fossa are functionally important because the former is a key determinant of the course of the M. supracoracoideus tendon and the latter has a major effect on the range of motion of the wing (Novas et al., 2020; Novas et al., 2021). In volant crown birds, the tendon ascends through the triosseal canal, passes laterally over the acrocoracoid process, and ultimately inserts on the dorsal tubercle near the proximal end of the humerus. The medial surface of the acrocoracoid process forms the lateral wall of the triosseal canal and acts as a pulley to redirect the M. supracoracoideus tendon. Because the pulley is situated above the insertion point when the humerus is depressed, the force generated by the ventrally positioned belly of the M. supracoracoideus elevates the humerus, rather than protracting the humerus as in early-diverging theropods (Burch, 2014). In contrast to non-avialan theropods, volant crown birds are characterized by a well-developed acrocoracoid process located above the level of the glenoid fossa, which has a sub-horizontal major axis and faces laterodorsally (Novas et al., 2020). The M. supracoracoideus is the main elevator of the wing, and the wing moves approximately dorsoventrally at the shoulder (Novas et al., 2020; Novas et al., 2021). In Archaeopteryx and deinonychosaurs (e.g., Buitreraptor and Sinovenator), the hypertrophied coracoid tubercle would likewise have acted as a pulley for the M. supracoracoideus tendon (Novas et al., 2021), but the pulley would have been located below the level of the glenoid fossa and approximately at the dorsoventral level of the insertion point when the humerus was depressed. Thus, the M. supracoracoideus would have protracted the humerus, as in flightless extant paleognaths (Novas et al., 2020; Novas et al., 2021; Jasinoski et al., 2006). The glenoid fossa of Archaeopteryx, non-avialan pennaraptorans, and flightless paleognaths faces laterally and has a sub-vertical major axis, indicating that the movements of the forelimb at the shoulder joint are, or in the case of extinct taxa would have been, predominantly cranial-caudal (Novas et al., 2020; Ostrom, 1974).

In Sapeornis and most other non-ornithothoracine avialans (e.g., Jeholornis and Confuciusornis), the acrocoracoid process is slightly above the midpoint of the coracoidal glenoid fossa (Zhou and Zhang, 2003b), and consequently would have been above the insertion point when the humerus was depressed, as in living birds. The more dorsal location of the acrocoracoid process would have caused the tendon of M. supracoracoideus to be slightly dorsally displaced relative to its position in Archaeopteryx and non-avialan pennaraptorans (Mayr et al., 2005; Turner et al., 2012; Wang and Zhou, 2018b). The triosseal canal is located mediocranial to the glenoid fossa, whereas in extant birds the triosseal canal is located more directly medial to the glenoid fossa. In Sapeornis, the vector of the tension exerted by the M. supracoracoideus on the humerus would therefore have been directed cranially and somewhat dorsomedially, causing the muscle to promote protraction, elevation, and pronation of the wing during the upstroke. The angle with the vertical formed by the major axis of the glenoid fossa is larger than in flightless paleognaths, but smaller than in volant extant birds (Novas et al., 2020). This suggests that the wing may have moved in a craniodorsal-caudoventral downstroke, unlike either the dorsal-ventral downstroke of extant volant birds or the largely cranial-caudal humeral movements of flightless paleognaths and presumably also of Archaeopteryx and non-avialan pennaraptorans. Several studies Mayr, 2017; Olson and Feduccia, 1979 have suggested that the well-developed M. deltoideus, which would have inserted broadly on the deltopectoral crest and humeral shaft, played the main role in the wing elevation in Sapeornis and other non-ornithothoracine birds. This seems consistent with the finding in this study that the M. supracoracoideus of Sapeornis would have pulled the humerus cranially (Figure 5) rather than acting primarily as a wing elevator, and with the finding that Lig. acrocoracohumerale in Sapeornis had a relatively horizontal orientation, so that the dorsal shoulder musculature would have been largely responsible for preventing ventral dislocation of the humeral head when M. pectoralis was strongly activated. Nevertheless, the presence of the triosseal canal indicates that most non-ornithothoracine birds possessed some incipient capacity for powered, flapping flight. In Piscivorenantiornis, the acrocoracoid process is only slightly higher than the coracoidal glenoid fossa as in most non-ornithothoracine avialans, but nevertheless is considerably higher than the coracoid’s scapular articular surface as in euornithines, due to the craniodorsally caudoventrally elongated shape of the coracoidal glenoid fossa. The orientation of the major axis of the glenoid fossa falls within the range seen in volant extant birds, and the triosseal canal is located medial to the glenoid fossa. Therefore, the wing movements of Piscivorenantiornis would have been more like those of volant extant birds than those of Sapeornis, indicating stronger flight capabilities in enantiornithines than in non-ornithothoracine birds.

In general, our study reveals additional lineage-specific variations in pectoral girdle anatomy as well as an overarching pattern of significant modification of the pectoral girdle along the line to crown birds. The morphological diversity seen across the pectoral girdles of Mesozoic birds presumably resulted in a commensurate range of flight capabilities and modes in early-flight evolution. The wing movements of Sapeornis would have differed from those of extant volant birds, highlighting the need to consider the possible effect of wing kinematics when reconstructing the flight ability of early birds.

Materials and methods

Institutional abbreviations

Request a detailed protocol

PMoL, Paleontological Museum of Liaoning, Shenyang, China; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences (CAS), Beijing, China; STM, Shandong Tianyu Museum of Natural History, Linyi, China.

S. chaoyangensis PMoL-AB00015 is a nearly complete semi-articulated skeleton collected from the Lower Cretaceous Jiufotang Formation at Yuanjiawa Village, Dapingfang Town, Chaoyang County, Liaoning Province, China. It is probably an adult individual based on skeletal fusion features (e.g., closed neurocentral sutures in all vertebrae, sacral vertebrae fused to form a synsacrum, distal carpals fused with metacarpals to form a carpometacarpus, proximal tarsals fused with tibia to form a tibiotarsus, and distal tarsals fused with metatarsals to form a tarsometatarsus). P. inusitatus IVPP V 22582 is a disarticulated skeleton collected from the Jiufotang Formation near Dapingfang Town (Wang and Zhou, 2017Wang and Zhou, 2017; Wang et al., 2016a). The skeleton shows the same fusion features as PMoL-AB00015 and probably also represents an adult individual. These two specimens are the main data sources for this study. For comparison, we also examined the skeletons of several extant birds and fossil theropods, including T. alba IVPP OV 954, E. garzetta IVPP OV 1631, P. muticus IVPP OV 1668, and the troodontid Sinovenator changii IVPP V 12615. The specimens were scanned using a GE v|tome|x m300&180 micro-computed tomography scanner (GE Measurement & Control, Wuntsdorf, Germany) and a 225 kV micro-computerized tomography scanner (developed by the Institute of High Energy Physics, CAS), both housed at the Key Laboratory of Vertebrate Evolution and Human Origins of the Chinese Academy of Sciences. Three-dimensional segmentation of the computed tomography data was performed using the software package Mimics (19.0).

Data availability

All data are available in the article.

References

  1. Book
    1. Baumel JJ
    2. Witmer LM
    (1993)
    Osteologia
    In: Baumel JJ, King AS, Breazile JE, Evans HE, Vanden Berge JC, editors. Handbook of Avian Anatomy: Nomina Anatomica Avium 2nd Ed: Publications of the Nuttall Ornithological Club. Acedemic Press. pp. 45–124.
    1. Burnham DA
    2. Derstler KL
    3. Currie PJ
    4. Bakker RT
    5. Ostrom JH
    (2000)
    Remarkable new birdlike dinosaur (theropoda: maniraptora) from the upper cretaceous of montana
    University of Kansas Paleontological Contributions 13:1–14.
  2. Book
    1. Chiappe LM
    2. Walker CA
    (2002)
    Skeletal morphology and systematics of the cretaceous euenantiornithes (ornithothoraces: enantiornithes
    In: Chiappe LM, Witmer LM, editors. Mesozoic Birds: Above the Heads of Dinosaurs. Berkeley, California: University of California Press. pp. 559–588.
  3. Book
    1. Ghetie V
    (1976)
    Anatomical Atlas of Domestic Birds
    Editura Academiei Republich Socialiste Romania.
    1. Norell MA
    2. Makovicky PJ
    (1999)
    Important features of the dromaeosaurid skeleton II: information from newly collected specimens of Velociraptor mongoliensis
    American Museum Novitates 3282:1–45.
  4. Book
    1. Novas FE
    2. Agnolin FL
    3. Egli FB
    4. Lo Coco GE
    (2020) Pectoral girdle morphology in early-diverging paravians and living ratites: implications for the origin of flight
    In: Pittman M, Xu X, editors. Pennaraptoran Theropod Dinosaurs Past Progress and New Frontiers: Bulletin of the American Museum of Natural History. BioOne Complete. pp. 345–353.
    https://doi.org/10.1206/0003-0090.440.1.1
    1. Ostrom JH
    (1976)
    Some hypothetical anatomical stages in the evolution of avian flight
    Smithsonian Contributions to Paleobiology 27:1–21.
    1. Padian K
    (1985)
    The origins and aerodynamics of flight in extinct vertebrates
    Palaeontology 28:413–433.
    1. Senter P
    (2006)
    Scapular orientation in theropods and basal birds and the origin of flapping flight
    Acta Palaeontologica Polonica 51:305–313.
  5. Thesis
    1. Tykoski RS
    (1998)
    The osteology of syntarsus kayentakatae and its implications for ceratosaurid phylogeny
    University of Texas.
    1. Wang M
    2. Zhou ZH
    3. O’Connor JK
    4. Zelenkov NV
    (2014)
    A new diverse enantiornithine family (Bohaiornithidae fam. Nov.) from the lower cretaceous of China with information from two new species
    Vertebrata PalAsiatica 52:31–76.
  6. Book
    1. Wang M
    2. Zhou Z
    (2017) The evolution of birds with implications from new fossil evidences
    In: Maina JN, editors. The Biology of the Avian Respiratory System. Cham: United Kingdom: Springer Nature. pp. 1–26.
    https://doi.org/10.1007/978-3-319-44153-5_1
  7. Book
    1. Wellnhofer P
    2. Haase F
    3. Chiappe LM
    (2009)
    Archaeopteryx: The Icon of Evolution
    Munich: Verlag Dr. Friedrich Pfeil.
  8. Book
    1. Xu X
    (2002)
    Deinonychosaurian Fossils from the Jehol Group of Western Liaoning and the Coelurosaurian Evolution Ph.D. Dissertation
    Beijing: Chinese Academy of Sciences.

Decision letter

  1. David Lentink
    Reviewing Editor; University of Groningen, Netherlands
  2. George H Perry
    Senior Editor; Pennsylvania State University, United States
  3. Jingmai O'Connor
    Reviewer
  4. Fernando E Novas
    Reviewer

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Digital restoration of the pectoral girdles of two Early Cretaceous birds, and implications for early flight evolution" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and George Perry as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Jingmai O'Connor (Reviewer #1); Fernando E. Novas (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. Please comply with the comments and suggestions to your best ability, mark all changes using a blue font to indicate changes in the revised manuscript, and respond in a point-by-point fashion. This is essential to enable the reviewing editor to fully evaluate the merit of your revision.

Essential revisions:

Overall, both the reviewers and the editors are excited by the quality of the research, the following weaknesses and comments remain to be addressed.

General Comments:

1. Overall, the presentation of the 3D morphological data in the figures is not sufficiently easy to interpret for the general eLife reader and as such this manuscript is focust on specialist only in its current format. To make the manuscript acceptable for eLife the following changes are needed. (i) There is a strong need for an introductory figure introducing all key morphological concepts in the context of the entire inferred body plan of the animals studied. This introductory figure should also show the assumed behavioral – locomotion – context that the work used to interpret the functional anatomy of the fossils. (ii) All the current figures have specialist labeling, the labeling and figures require avatars showing the location of the bones in the body plan with common wording and labels any eLife reader can understand. Only once the labels have been introduced with common anatomical nomenclature, can shorthand names be used when absolutely essential for the rigor of the science. Whenever common nomenclature is sufficient, that is preferred so it is easier for the general eLife reader to interpret and understand the findings presented.

2. It is unclear if the PMOL specimen is a juvenile or subadult? >> Please avoid using acronyms in the manuscript, eLife provides as much space as the authors need to not use acronyms and make the manuscript more accessible to eLife's entire readership. << Discuss the matter of fusion between the scapulocoracoid in Sapeornis and how that may (or may not) affect the musculature. This is mentioned for other taxa in the discussion but apparently not for Sapeornis.

3. Please expand the description to include the detail that a general eLife reader needs to comprehend the morphology. The current descriptions are too vague. For example, the manuscript states the scapula is twisted, but information on how it is twisted is missing in the text and the figures – clear avatars would help the general reader see this in the figures. Another example "A broad flange projects costally…" It is unclear what the shape of the flange is, how far it projects, how long it is, etc. This lack of quantitative information and description makes the paper too subjective for the eLife readership to follow and it makes it hard for specialists to fully appreciate the reported findings and integrate it in their future research.

4. The division between the scapular and sternal portions of the coracoid are currently confusing and give the impression it is almost arbitrary, especially because the sternal portion in birds still articulates with the scapula in some taxa in figure 3, e.g. enantiornithines. If this way of describing the two parts of the plesiomorphic coracoid is commonly used in the specialist literature, please keep in mind this manuscript has been submitted to eLife, which serves an extraordinary broad audience. So in addition to clarifying these matters in the figures and text, please add citations so the eLife readership can also find the pertinent specialist context in the literature that may motivate some of the authors choices. However, in case it is new, we don't recommend moving forward with this way of describing, or rather dividing, the coracoid and instead serve the broad readership. One of our technicalconcerns is that if this section houses the supracoracoidal nerve foramen, then it doesn't quite work when the SCNF is in the neck of the coracoid in enantiornithines (see the Buffetaut 1998 specimen and Lecho specimens) and some modern birds. It would be more helpful to either mark the glenoid and acrocoracoid etc. in a distinct color instead of coloring the two supposed parts of the coracoid. This would better illustrate the revised text.

5. It is necessary to be more consistent in using either anterior/cranial and posterior/caudal to serve eLife's readership, please also introduce any differences between your field's preferred choices and common choices in other biological fields so the manuscript better serves all readers. Please also visually introduce the chosen definitions in figure one in the context of the body plan so any reader can follow and integrate your contribution crossdisciplinary. Similarly, in Sapeornis the two major surfaces of the coracoid are described as dorsal and ventral but it changes to cranial and caudal in Piscivorenantiornis despite the fact the orientation of the coracoid doesn't change according to Figure. 4 (and also Baumel uses dorsal and ventral for these two surfaces). Consistency in directional terms will improve the readability of this manuscript although for non-avian theropods different terms are used, we suggest using all avian terms and putting the theropod directional language in parentheses when talking about taxa like Sinovenator and illustrating these different perspectives in the new introductory figure 1 which will probably have to be a multipaneled figure to serve eLife's readership). Please also add a short section in the beginning in which the directional terminology used throughout is explained explicitly in the text with clear references to the visual illustration in the new figure 1.

6. To further help eLife's readership and specialist alike, the new figure 1 should also include the following. Comparative images of scapular and coracoid shape in more taxa would, even if they are not based on digital data, which is why we recommend adding these in the new introductory figure 1. E.g. Archaeopteryx, Jeholornis, Confuciusornis are all mentioned but there are no images to help orient the reader and interpret the comparisons that are made due to missing critical visual information that is assumed known – which is not the case for eLife's readership since the journal serves the entire biological (neuroscience, medical as well as several other disciplines including biophysics, biomathematics, bioinspired engineering etc.). This should also be paired with indications of the range of motion that are mentioned in the discussion for greater clarity. The discussion should also mention the limitations in how these ranges of motion have previously been established and how this limits the current analysis in a general biomechanical functional framework. Additionally, visually indicating the attachment of major muscles discussed in the new introductory figure 1, in the context of the body plan, would also be very helpful to orient the reader and follow the discussion. To help the reader further, the introduction, results and Discussion sections should reference to this introductory figure whenever a new concept is discussed in the text (please don't assume the readers find these connections obvious). eLife does not have length restrictions for papers, so the authors can invest words and figures to serve the readers.

7. There seems to be no acknowledged that the morphology of the enantiornithine pectoral girdle in 3D already known from Late Cretaceous specimens, but that the new morphology presentation is clearer for Early Cretaceous specimens, and it can now be recognized that many Late Cretaceous morphologies are also present in Early Cretaceous specimens, which previously could not be recognized without 3D CT data. Please provide this context in the introduction and discussion.

8. It would be helpful to discuss the homoplasy affecting the evolution of these two bones using non specialist wording, so the general implications are clearer to all readers. Additionally, it would be great to discuss how the absence of an ossified sternum would affect the flight musculature in Sapeornis and what further research could provide more decisive evidence.

9. In the discussion, when all the morphological transitions are discussed, citations should be provided for where these have been discussed previously. And where possible, please illustrate them in the new introductory figure 1 so all readers can comprehend the significance of the current research in the literature context.

Detailed Comments:

1. In figure 2 the Fura abbreviation is missing from caption; also indicate where the furcula contacts the scapula in B and D. Ideally as few as possible abbreviations are used in all figures, it would better serve the reader to use the figure space that eLife provides to use the full names and make the figures highly readable to all. In particular regarding the key morphological aspects researched and discussed. It is understandable that the authors may choose to abbreviate aspects that are only indicated for the specialist reader and are not essential to interpret and understand the work, however one could question if indicating non-essential information is even needed. So, abbreviations should be used sparingly with a bias towards making all labels readable for all – without abbreviation.

2. Please add the coracoclavicular ligament in Figure 4 add to the other models (currently only Sapeornis) and show the triosseal canal in lateral view as well (or rather the path of the supracoracoideus ligament). Because it is very hard to see the location of the triosseal canal in the figure as is, because the illustrations are proportionately small. Again eLife provides all figure space needed to make the figures readable to both specialist (who could not read it) and the rest of eLife's readership (who need even clearer figures).

3. Line 29 – The statement that some taxa have "one area" and others a "double articulation" is confusing. Please specify which is "the double articulation" mentioned as widely present among pennaraptorans? Please provide all context (visual, text, references) needed to follow the point.

4. Line 32 – Please resolve the readers confusion regarding the sharp cut in the descriptive process. E.g. the treatment of the presence of triosseal canal among birds, without mentioning the (eventual) relation with the "only one vs double articulation" condition analyzed before.

5. Line 33 – Is this "transitional stage" first recognized here? What are the nodes representing the transition? Please provide all context (visual, text, references) needed to follow the point.

6. Line 34-37 – Please explain the importance of "only one vs double scap-cor articulation" and the "transitional stage of triosseal canal" in the context of the anatomical modifications that occurred in the line to birds. How do the new observations on these two aspects modify and expand previous hypotheses on the origin and evolution of bird flight? Please provide all context (visual, text, references) needed to follow the point.

7. Line 44 – Consider citing Ostrom 1976 for his role in this line of research.

8. Line 50-51 – Please, consider the paper by Imai et al., 2019 on Fukuipteryx. May these authors already have presented a 3D reconstruction of the scapular girdle of the kind present here for other taxa? Being first is not considered important for an eLife publication, as a true first rarely exist in science, we are all building off the research by others and this work should be clearly cited and discussed so that science robustly advances.

9. Line 74-75 – There may be some context missing regarding the diversity in ideas of how dromaeosaurid relates to Rahonavis. Novas et al., (2018. "Postcranial osteology of a new specimen of Buitreraptor". Cretaceous Research) discuss features that Rahonavis shares with birds more derived than dromaeosaurids and Archaeopteryx (particularly the size and shape of the acromial process). This work suggests that Rahonavis is a long-tailed avialan. Whereas the new work seems to support support the idea that the taxon is not a dromaeosaurid. Discussing how the new work may enrich or revise previous perspectives with citations will help the general reader better understand the contribution made in the context of the literature. Please revise the text in a fashion that you consider most rigorous and aligned with the new fossil insights.

10. Line 80 – Please clarify if this may be related with the ventral surface of the acromial process.

11. Line 81 – It may help the reader to also indicate the position with respect to the glenoid. E.g. aside from being "more medially positioned", is it anteriorly/posteriorly/at level with the glenoid? Please provide all context (visual, text, references) needed to follow the point.

12. Line 111 – Consider revising this as follows: "…the acrocoracoid process (frequently described as biceps or coracoid tubercle)". Considering Archaeopteryx is an avialan sharing a common ancestor with Sapeornis and the rest of the birds, for which the term "acrocoracoid" is used. Afterall is a matter of nomenclature, and there is agreement that such tubercle is homologous. Please provide context in the text using references to other readers outside the field can follow.

13. Line 113 – Was this morphology confirmed in other specimens of Sapeornis? Is it natural or is it the result of postmortem compression? Please provide all context (visual, text, references) needed to follow the point, considering not all eLife readers are familiar with how postmortem compression may typically be visible in fossils.

14. Line 114 – Please, check the notes regarding Figure 1G,H for correctness and clarity.

15. Line 115-118 – Such a fossa on the acrocoracoid process is also seen in Buitreraptor and Rhea (colloquially it is a little like a volcano with a crater). Novas et al., (2021) interpreted this fossa as the site of attachment of the acrocoraco-acromial ligament. The acrocoracoacromial ligament forms a bridge under which the m. supracoracoideus slides. This ligament is also present in modern flying birds, but it is cranially covered by the omal end of the furcula. Please better integrate the literature context in alignment with the perspective that the new data provides and clearly delineate quantitative evidence versus qualitative interpretation and any further work that may be needed to rigorously conclude.

16. Line 131 – Please provide an extra figure, a clear line drawing that relates back to the new introductory figure 1, depicting the correspondence of the articular areas? This is important for other authors studyingh these features in other paravians. Please provide all context (visual, text, references) so other workers can build off your research.

17. Line 140-141- Based on Figure 4 (Sapeornis), the triosseal canal width, could the tendon of the m. supracoracoideus have been substantially thicker than in other avialans, considering the reconstructed diameter of the canal? Please provide all context (visual, text, references) so other workers can build off your research.

18. Line 154-155 – Please condiseder and integrate in your discussion that Rahonavis may be a long-tailed avialan, instead of a deinonychosaur. To support your perspective based on the presented data, please provide all context (visual, text, references) so the readers can follow your interpretations and comprehend any limitations and open discussion points so science advances robustly.

19. Line 206- Please provide the species name so the general reader can follow.

20. Line 237- Considering Rahonavis exhibits derived features shared with birds more derived than Archaeopteryx, which seem absent in deinonychosaurs and unenlagiines, the manuscript could evaluate the scapular features of Rahonavis, to provide more concludive support to either deinonychosaur or avialan affinities of this taxon. Please provide this in a context that aligns best with the new 3D fossil information and functional morphological analyses.

21. Line 273-275- The fossils presented support this condition. We wonder, however, if enantiornithes may have had (in life) a procoracoidal section (cartilaginous) for contacting with the well-developed acromial region of the proximal scapula. A "proto-procoracoid" or "supracoracoid canal" seems to have been present and well developed in Buitreraptor, and presumably served as a "channeling surface" for the m. supracoracoid towards the proximal end of humerus. Crown birds seem to have a procoracoid, and a similar channeling surface is thought to be present among basal paravians. We thus wonder if Enantiornithes may have had such structure, but in an unossified form. Please consider these options and clarify the manuscript to better substantiate your interpretation of the new 3D fossil evidence.

22. Line 290-300 – There is no need to repeat this since it was clearly explained in previous lines.

23. Line 313-315 – is "roofed" the correct wording? May the procoracoid be forming the caudomedial wall of the canal, instead of a roof? Please help the general reader correctly interpret the fossil evidence and clarify the text and figure labeling and orientation of the reader to fully resolve this.

24. Line 322 – Please consider that in Rhea the acrocoraco-acromial ligament bridges the tendon of m. supracoideus, enabling it to functionally act as a "triosseal canal" as discussed by Novas et al., 2021.

25. Line 338-340 – The following is confusing "in basal paravians as well as in living birds the m. supracoracoid runs along the LATERAL side of the acromion" for the reader. The medial surface of the later one is contiguous with the medial surface of scapular blade, thus the medial surface of the acromion contacts with the underlying thorax. Hence it seems more logical to consider the supracoracoid doesn´t run between the scapula and ribs. Please reconsider the wording and clarify the text (as needed with a figure and references) so the general reader can follow the authors reasoning or correct the error.

26. Line 348-349 – The presence of a triosseal canal is defined by the contact between the acrocoracoid and epicleidium. If such a contact does not exist, then such a "triosseal" canal is not defined. Hence this is also a semantic problem, please resolve so a general reader can comprehend.

27. Regarding the function of the scapular girdle in paravians: The deflection of the tendon may have been present in basal paravians, and the "inner half" of the triosseal canal seems to have already been present in basal paravians. The shape of the coracoid of Sapeornis (the development and position of acrocoracoid) seems to resemble that of Archaeopteryx and Buitreraptor, hence it may be reasonable to assume the "pulley system" may have functioned in a similar way in all these forms. Hence one could consider the term "triosseal canal" to be inadequate for Sapeornis. Further, a similar pulley-shaped morphology for coracoid was also reported for basal paravians (Buitreraptor, Archaopteryx; Novas et al., 2021). Please discuss how the new 3D fossil evidence may or may not support this line of thought, in case you end up agreeing, please revise the text accordingly and otherwise please clarify the text and cite supporting literature so the reader can follow the authors reasoning.

28. Line 375-376 – Figure 4 seems to suggest that Sapeornis does exhibit a derived acrocoracoid, because it is positioned at level with the glenoid, and not in front of it. Hence it is not fully clear why the "extension of acrocoracoid process dorsal to level of coracoidal glenoid fossa" (a condition that seems absent in Sapeornis) is interpreted as characteristic of Step I. In case you end up agreeing, please revise the text accordingly and otherwise please clarify the figure and text and cite supporting literature so the reader can follow the authors reasoning.

29. Line 403-407 – Novas et al., (2020, 2021) suggest that the pulley was already operative in basal paravians such as Buitreraptor and Archaeopteryx, but the m. supracoracoideus acted as a humeral protractor (as in living ratites). This condition is intermediate between the humeral depressor action (in basal theropods) and humeral elevator role (in extanct birds). Novas et al., (2020, 2021) did not state that the "m. supracoracoideus would have acted straightforwardly as a wing depressor". Please resolve.

30. Line 419-422 – The beautifully illustration of the coracoid of Sapeornis does not show a condition of the acrocoracoid that is clearly different from Archaeopteryx. The acrocoracoid process does not seem to be in front of the glenoid, but almost at the same level. So, could it be more reasonable to consider the m. supracoracoideus in Sapeornis to have functioned as a humeral protractor, not as an elevator? Please resolve.

31. Line 720 (including the labels in Figure 1H) – Is the border closer to the sternum? If so, it seems "distal" may be more appropriate in the text (considering the glenoid represents the "proximal end" of the coracoid). Please clarify and resolve so the general reader can follow the authors reasoning in the context of the literature.

Reviewer #1 Recommendations for the authors:

Great manuscript, it is really exciting to see more 3D data becoming available for important taxa like Sapeornis and for the morphology of the musculature in these early birds to begin to be discussed (and also reconstructed, see comment below – it is a bit difficult to visualize this information for someone who is not a myologist like myself).

Is the PMOL specimen a juvenile or subadult? Discuss the matter of fusion between the scapulocoracoid in Sapeornis and how that would affect the musculature (if at all). This is mentioned for other taxa in the discussion but not for Sapeornis itself

I would suggest the description include greater detail to clarify the morphology. As written the description is a little vague (examples, the scapula is twisted, twisted how? A broad flange projects costally…. what is the shape of the flange? how far does it project, how long is it, etc.).

I find the division between the scapular and sternal portions of the coracoid confusing and almost arbitrary, especially because the sternal portion in birds still articulates with the scapula in some taxa in figure 3 (including enantiornithines). If this way of describing the two parts of the plesiomorphic coracoid is commonly used in the literature and I am simply unaware of this, then please add citations. However, if this is new, I would not recommend moving forward with this way of describing or rather dividing the coracoid. I understand that it seems superficially that this is the portion of the coracoid that becomes reduced but thats not quite accurate. If this section houses the supracoracoidal nerve foramen then it doesn't quite work when the SCNF is in the neck of the coracoid in enantiornithines (see the Buffetaut 1998 specimen and Lecho specimens) and some modern birds. It would be more helpful to either mark the glenoid and acrocoracoid etc in a distinct color (rather than color the two supposed parts of the coracoid). This would better supplement the text.

It is necessary to be more consistent in using either anterior/cranial and posterior/caudal (choose one). Similarly in Sapeornis the two major surfaces of the coracoid are described as dorsal and ventral but it changes to cranial and caudal in Piscivorenantiornis despite the fact the orientation of the coracoid doesn't change according to Figure. 4 (and also Baumel uses dorsal and ventral for these two surfaces). Consistency in directional terms will improve the readability of this manuscript (although for non-avian theropods different terms are used, I would suggest using all avian terms and putting the theropod directional language in parentheses when talking about taxa like Sinovenator). I think it is worth adding a short section in the beginning in which the directional terminology used throughout is explained explicitly.

Comparative images of scapular and coracoid shape in more taxa would be helpful, even if they are not based on digital data. Archaeopteryx, Jeholornis, Confuciusornis are all mentioned but there are no images to help the reader interpret the comparisons that are made. This could also be paired with indications of the range of motion that are mentioned in the discussion for greater clarity. Additionally, indicating the attachment of major muscles discussed would also be very helpful.

It should be acknowledged that the morphology of the enantiornithine pectoral girdle in 3D was well known from Late Cretaceous specimens but that now the morphology is clearer for Early Cretaceous specimens and it can now be recognized that many Late Cretaceous morphologies are also present in Early Cretaceous specimens but previously could not be recognized without 3D CT data.

It would be helpful to discuss the homoplasy affecting the evolution of these two bones. Additionally, it would be great to discuss how the absence of an ossified sternum would affect the flight musculature in Sapeornis.

In the discussion, when all the morphological transitions are discussed, citations should be provided for where these have been discussed previously.

Figure 2 Fura abbreviation is missing from caption; also indicate where the furcula contacts the scapula in B and D.

In Figure 4 add the coracoclavicular ligament to the other models (currently only Sapeornis) and show the triosseal canal in lateral view as well (or rather the path of the supracoracoideus ligament); it is very hard to see the location of the triosseal canal in the figure as is because the illustrations are proportionately small.

Reviewer #2 Recommendations for the authors:

Line 29 – I am confused with the statement that some taxa have "one area" and others a "double articulation". Could you specify which is "the double articulation" mentioned as widely present among pennaraptorans?

Line 32 – Here produces a sharp cut in the descriptive process, with the treatment of the presence of triosseal canal among birds, without mentioning the (eventual) relation with the "only one vs double articulation" condition analyzed before.

Line 33 – Is this "transitional stage" firstly recognized here? Could you precise which are the nodes representing such transition?

Line 34-37 – I agree with the authors in the conclusions they express in this paragraph. However, I strongly suggest to explain the importance of the "only one vs double scap-cor articulation" and the "transitional stage of triosseal canal" in the anatomical modifications occurred in the line to birds. How their observations on these two aspects modify and expand previous hypotheses on the origin and evolution of bird flight?

Line 44 – Let me suggest to cite here Ostrom 1976, which I believe is the founder of this line of studies

Line 50-51 – Please, check the paper by Imai et al., 2019 on Fukuipteryx. Do these authors already present a 3D reconstruction of the scapular girdle of the kind you present here for other taxa? Anyway, to be the first or not is secondary, and present manuscript represents a formidable progress in our knowledge on the flight apparatus of early birds

Line 67 – These warnings are necessary and welcome. Many papers on scapulocoracoid in paravian theropods overlook describing the state of preservation of the available materials.

Line 74-75 – I am respectful of the interpretation that current authors are following about the dromaeosaurid affiliation of Rahonavis. However, let me say that Novas et al., (2018. "Postcranial osteology of a new specimen of Buitreraptor". Cretaceous Research) emphasized about the presence of many features that Rahonavis shares with birds more derived than dromaeosaurids and Archaeopteryx (particularly the size and shape of the acromial process!), suggesting that Rahonavis is, in fact, a long-tailed avialan. Interestingly, you are here noting on a quite derived morphology of the acromion, being present in Jeholornis and Rahonavis, thus lending support to the idea that the later taxon is not a dromaeosaurid.

Line 80 – Is it related with the ventral surface of the acromial process?

Line 81 – Let me suggest to indicate also the position with respect to the glenoid. Aside from being "more medially possitioned", is it anteriorly/posteriorly/at level with the glenoid?

Line 111 – Let me suggest to change this part of the phrase, saying something like the following: "…the acrocoracoid process (frequently described as biceps or coracoid tubercle)". The reason is that Archaeopteryx is an avialan sharing a common ancestor with Sapeornis and the rest of the birds, for which the term "acrocoracoid" is used. Afterall is a matter of nomenclature, and all of us agree that such tubercle is homologous.

Line 113 – Did you check such morphology in other specimens of Sapeornis? Is it natural or is it the result of postmortem compression?

Line 114- Please, check some notes on Figure 1G,H

Line 115-118 – Such a fossa on the acrocoracoid process is also seen in Buitreraptor and Rhea, for example (it is something like a volcano with a crater on its top). Novas et al., (2021) interpreted this fossa as the site of attachment of the acrocoraco-acromial ligament. The acrocoracoacromial ligament forms a bridge under which the m. supracoracoideus slides. This ligament is also present in modern flying birds, but it is cranially covered by the omal end of the furcula.

Thus, keep in mind this in your considerations that the fossa on the acrocoracoid process served for attachment point for a coracoclavicular ligament connecting the coracoid and furcula (an interpretation that I am not dismissing).

Line 131 – Could you provide an extra figure (a simple line drawing) depicting the correspondence of articular areas? This could be important for other authors in searching for these features in other paravians.

Line 140-141- This is an interesting interpretation. By observing Figure 4 (Sapeornis), the triosseal canal results wide. I wonder if the tendon of the m. supracoracoideus was very thick, or thicker than in other avialans, based on the reconstructed diameter of the canal.

Line 154-155 – keep in mind that Rahonavis may be a long-tailed avialan, instead of a deinonychosaur.

Line 206- Please, insert species name.

Line 237- Let me express again that Rahonavis exhibits many derived features shared with birds more derived than Archaeopteryx, which are absent in deinonychosaurs and unenlagiines. Probably the present manuscript may represents a good place to evaluate the scapular features of Rahonavis, lending support to either deinonychosaur or avialan affinities of this taxon.

Line 273-275- I agree, of course, with this conclusion. Fossil bones are clear in indicating this condition. I wonder, however, if enantiornithes had (in life) a procoracoidal portion (cartilaginous) for contacting with the well developed acromial region of proximal scapula. A "proto-procoracoid" or more properly a "supracoracoid canal" was present and well developed in Buitreraptor, and presumably served as a "channeling surface" for the m. supracoracoid in its course towards the proximal end of humerus. Crown birds have procoracoid, and a similar channeling surface was present among basal paravians. Enantiornithes may have had such structure, but in an unossified condition.

Line 290-300- I believe it is unnecessary to repeat. Authors have already clearly explained this in previous lines.

Line 313-315 – s it right to say "roofed"? Is the procoracoid forming the caudoMEDIAL wall of the canal, instead of a roof? Apologies if I am wrong.

Line 322- In Rhea the acrocoraco-acromial ligament bridges the tendon of m. supracoideus, functionally acting as a "triosseal canal". See Novas et al., 2021 for discussions on these asepcts.

Line 338-340- I am surprised for this statement: in basal paravians as well as in living birds the m. supracoracoid runs along the LATERAL side of the acromion. The medial surface of the later one is contiguous with the medial surface of scapular blade, thus the medial surface of the acromion contacts with the underlying thorax. The supracoracoid doesn´t run between the scapula and ribs!

Line 348-349- The presence of a triosseal canal is defined by the contact between the acrocoracoid and epicleidium. If such a contact does not exist, then such a "triosseal" canal is not defined. of course, this is a semantic problem.

A completely different matter is the function that this region of the scapular girdle played in paravians: a well defined though is present on the craniomedial surface of coracoid in Buitreraptor which may have act as a "proto-procoracoid" for chanalizing the m. scapulocoracoid. Thus, the deflection of its tendon was probably operative in basal paravians, and the "inner half" of the triosseal canal was already present even in basal paravians. The shape of the coracoid of Sapeornis (especially the development and position of acrocoracoid) resembles that of Archaeopteryx and Buitreraptor, thus it is expectable that the "pulley system" was postioned and functioned in a similar way in all these forms.

In sum, I agree with authors regarding the function of this region of scapular girdle, but the term applied ("triosseal canal") for Sapeornis seems inadequate. Besides, but no less important, a similar pulley-shaped morphology for coracoid was also identified for basal paravians (Buitreraptor, Archaopteryx; Novas et al., 2021).

Line 375-376- Notably, as you shown in Figure 4, Sapeornis does not exhibit a derived condition of acrocoracoid, because it is at level with the glenoid, not in front of it. Thus, I don´t understand why the "extension of acrocoracoid process dorsal to level of coracoidal glenoid fossa" (a condition that seems absent in Sapeornis), is interpreted as characteristic of Step I.

Line 403-407- Novas et al., (2020, 2021) said that the pulley was already operative in basal paravians such as Buitreraptor and Archaeopteryx, but the m. supracoracoideus acted as a humeral PROTRACTOR (as in living ratites). This condition is intermediate between the humeral depressor action (in basal theropods) and humeral elevator role (in extanct birds).

Novas et al., (2020, 2021) did not say that the "m. supracoracoideus would have acted straightforwardly as a wing depressor".

Line 419-422- Based on what is beautifully illustrated in present manuscript, the coracoid of Sapeornis does not show a condition of the acrocoracoid sharply different from that in Archaeopteryx. In other terms, the acrocoracoid process is not in front of the glenoid, but almost at the same level. So, I will say the m. supracoracoideus in Sapeornis functioned as a humeral protractor, not as an elevator.

Line 720 (and labels inserted in Figure 1H)- Is this the border closer to the sternum? Then, it has to be "distal" using the terminology applied on the text (the glenoid represents the "proximal end" of the coracoid).

https://doi.org/10.7554/eLife.76086.sa1

Author response

General Comments:

1. Overall, the presentation of the 3D morphological data in the figures is not sufficiently easy to interpret for the general eLife reader and as such this manuscript is focust on specialist only in its current format. To make the manuscript acceptable for eLife the following changes are needed. (i) There is a strong need for an introductory figure introducing all key morphological concepts in the context of the entire inferred body plan of the animals studied. This introductory figure should also show the assumed behavioral – locomotion – context that the work used to interpret the functional anatomy of the fossils. (ii) All the current figures have specialist labeling, the labeling and figures require avatars showing the location of the bones in the body plan with common wording and labels any eLife reader can understand. Only once the labels have been introduced with common anatomical nomenclature, can shorthand names be used when absolutely essential for the rigor of the science. Whenever common nomenclature is sufficient, that is preferred so it is easier for the general eLife reader to interpret and understand the findings presented.

Following the reviewers’ suggestion, we provided an introductory figure to illustrate the key morphological concepts, and we even added a paragraph to introduce the general context; we also added cartoon illustration to show the anatomical position of the structures that we described in the manuscript; we deleted nearly all shorthand names and replace them with the common nomenclature. In doing this, we are able to help the general eLife readers to understand the manuscript more easily as the editor suggested.

2. It is unclear if the PMOL specimen is a juvenile or subadult? >> Please avoid using acronyms in the manuscript, eLife provides as much space as the authors need to not use acronyms and make the manuscript more accessible to eLife's entire readership. << Discuss the matter of fusion between the scapulocoracoid in Sapeornis and how that may (or may not) affect the musculature. This is mentioned for other taxa in the discussion but apparently not for Sapeornis.

Following the reviewers’ suggestion, we added information on the ontogenetic stage of both PMoL-AB00015 and IVPP V 22582. Both specimens are probably adult individuals based on skeletal fusion features.

Following the reviewers’ suggestion, we have taken out all acronyms from the manuscript.

Following the reviewers’ suggestion, we added data on whether Sapeornis has fused scapulocoracoid. When the M. supracoracoideus contracts, the triosseal canal supports the tendon in almost the same way whether coracoid fused with the scapula or not. Thus, we do not think the fusion state would strongly affect the musculature.

3. Please expand the description to include the detail that a general eLife reader needs to comprehend the morphology. The current descriptions are too vague. For example, the manuscript states the scapula is twisted, but information on how it is twisted is missing in the text and the figures – clear avatars would help the general reader see this in the figures. Another example "A broad flange projects costally…" It is unclear what the shape of the flange is, how far it projects, how long it is, etc. This lack of quantitative information and description makes the paper too subjective for the eLife readership to follow and it makes it hard for specialists to fully appreciate the reported findings and integrate it in their future research.

Following the reviewers’ suggestion, we added new figures and new text to help the readers understand more easily the morphology. However, in some cases, we are unable to provide details (e.g., the flange is badly preserved, and we are not able to tell its exact shape and size)

4. The division between the scapular and sternal portions of the coracoid are currently confusing and give the impression it is almost arbitrary, especially because the sternal portion in birds still articulates with the scapula in some taxa in figure 3, e.g. enantiornithines. If this way of describing the two parts of the plesiomorphic coracoid is commonly used in the specialist literature, please keep in mind this manuscript has been submitted to eLife, which serves an extraordinary broad audience. So in addition to clarifying these matters in the figures and text, please add citations so the eLife readership can also find the pertinent specialist context in the literature that may motivate some of the authors choices. However, in case it is new, we don't recommend moving forward with this way of describing, or rather dividing, the coracoid and instead serve the broad readership. One of our technicalconcerns is that if this section houses the supracoracoidal nerve foramen, then it doesn't quite work when the SCNF is in the neck of the coracoid in enantiornithines (see the Buffetaut 1998 specimen and Lecho specimens) and some modern birds. It would be more helpful to either mark the glenoid and acrocoracoid etc. in a distinct color instead of coloring the two supposed parts of the coracoid. This would better illustrate the revised text.

Following the reviewers’ suggestion, we explicitly defined the scapular and sternal wings of the coracoid in the text and we also illustrated the two wings in the newly added figure. Such a division was originally proposed by Xu (2002) in which the author suggests a major modification in coracoid evolution is the appearance of biplanar morphology. More specifically, in ornithomomiosaurs and other maniraptoriforms, the distal portion of the coracoid (called distal ramus in Xu 2002, but sternal wing in this manuscript because ramus normally refers to an elongated structure) is deflected from the proximal ramus (the scapular wing in this manuscript), and this distal ramus (the sternal wing) becomes larger in size and with a nearly straight edge for articulating the sternum in late-diverging maniraptoriforms such as the dromaeosaurid Sinornithosaurus (see the following figure). In some species, there is a ridge emanating from the coracoid tubercle to delimit the boundary between the scapular and sternal wings (see the following figure), and in other species, a distinct ridge is absent, but a deflecting zone is present. It is the sternal wing that turns into the main body of the strut-like coracoid in most avialans; the scapular wing is highly reduced, with its sheet-like medial portion turning into the procoracoid process in many birds and the lateral portion (the main portion for articulating the scapula) shortened and moving to the base of the acrocoracoid process. The supracoracoid foramen is present within the scapular wing in most species, though in species that has a highly reduced scapular wing, the supracoracoid foramen is present in the sternal wing (in some species this foramen is absent). Nevertheless, following the reviewers’ suggestion, we also colored some other important parts of the coracoid such as the glenoid and acrocoracoid.

5. It is necessary to be more consistent in using either anterior/cranial and posterior/caudal to serve eLife's readership, please also introduce any differences between your field's preferred choices and common choices in other biological fields so the manuscript better serves all readers. Please also visually introduce the chosen definitions in figure one in the context of the body plan so any reader can follow and integrate your contribution crossdisciplinary. Similarly, in Sapeornis the two major surfaces of the coracoid are described as dorsal and ventral but it changes to cranial and caudal in Piscivorenantiornis despite the fact the orientation of the coracoid doesn't change according to Figure. 4 (and also Baumel uses dorsal and ventral for these two surfaces). Consistency in directional terms will improve the readability of this manuscript (although for non-avian theropods different terms are used, we suggest using all avian terms and putting the theropod directional language in parentheses when talking about taxa like Sinovenator and illustrating these different perspectives in the new introductory figure 1 which will probably have to be a multipaneled figure to serve eLife's readership). Please also add a short section in the beginning in which the directional terminology used throughout is explained explicitly in the text with clear references to the visual illustration in the new figure 1.

Following the reviewers’ suggestion, we added a paragraph and a figure to introduce the main structures and their definitions in this manuscript and also the anatomical and directional terminology. Particularly, following the reviewers’ suggestion, we used avian terms and added nonavian theropod terms in the parentheses when talking about nonavian theropods. It should be noted that even among birds, the dorsal and ventral surface of coracoid display significantly variable orientation among different avialan groups: in most ornithothoracine birds, the dorsal surface faces dorsocaudally, but in other birds including Sapeornis, it faces nearly caudally.

6. To further help eLife's readership and specialist alike, the new figure 1 should also include the following. Comparative images of scapular and coracoid shape in more taxa would, even if they are not based on digital data, which is why we recommend adding these in the new introductory figure 1. E.g. Archaeopteryx, Jeholornis, Confuciusornis are all mentioned but there are no images to help orient the reader and interpret the comparisons that are made due to missing critical visual information that is assumed known – which is not the case for eLife's readership since the journal serves the entire biological (neuroscience, medical as well as several other disciplines including biophysics, biomathematics, bioinspired engineering etc.). This should also be paired with indications of the range of motion that are mentioned in the discussion for greater clarity. The discussion should also mention the limitations in how these ranges of motion have previously been established and how this limits the current analysis in a general biomechanical functional framework. Additionally, visually indicating the attachment of major muscles discussed in the new introductory figure 1, in the context of the body plan, would also be very helpful to orient the reader and follow the discussion. To help the reader further, the introduction, results and Discussion sections should reference to this introductory figure whenever a new concept is discussed in the text (please don't assume the readers find these connections obvious). eLife does not have length restrictions for papers, so the authors can invest words and figures to serve the readers.

Following the reviewers’ suggestion, we added a few more images of several other theropods and illustrated the ranges of motion and the muscles for the readers to have a better understanding of the whole issue, and we also referred to the introductory figure when we talk about the features that are not easily understood by the readers. However, we want to note that we did not illustrate the attachment of all major muscles except the ones we discussed in this manuscript for two reasons: 1. We are afraid to illustrating the muscles that are not discussed in this manuscript might mislead the readers; 2. We have an ongoing project specifically discussing the forelimb motion and the major flight muscles.

7. There seems to be no acknowledged that the morphology of the enantiornithine pectoral girdle in 3D already known from Late Cretaceous specimens, but that the new morphology presentation is clearer for Early Cretaceous specimens, and it can now be recognized that many Late Cretaceous morphologies are also present in Early Cretaceous specimens, which previously could not be recognized without 3D CT data. Please provide this context in the introduction and discussion.

Following the reviewers’ suggestion, we have done appropriate amendments on this.

8. It would be helpful to discuss the homoplasy affecting the evolution of these two bones using non specialist wording, so the general implications are clearer to all readers. Additionally, it would be great to discuss how the absence of an ossified sternum would affect the flight musculature in Sapeornis and what further research could provide more decisive evidence.

Following the reviewers’ suggestion, we added a brief discussion on the major homoplasies present in early-diverging avialans. We add a brief discussion on how the absence of ossified sternum would affect the flight musculature in Sapeornis. However, this manuscript focuses on the pectoral girdle morphology in early-diverging avialans and its implication for early flight, and as we noted above, we have a separate project specifically discussing the forelimb motion and the major flight muscles among early-diverging avialans including Sapeornis. Hopefully, these two projects together would contribute to a better understanding of early flight.

9. In the discussion, when all the morphological transitions are discussed, citations should be provided for where these have been discussed previously. And where possible, please illustrate them in the new introductory figure 1 so all readers can comprehend the significance of the current research in the literature context.

We have updated all the necessary references.

Reviewer #2 Recommendations for the authors:

Line 29 – I am confused with the statement that some taxa have "one area" and others a "double articulation". Could you specify which is "the double articulation" mentioned as widely present among pennaraptorans?

We provided an introductory figure and added a brief description on the general morphology of pectoral girdles among different theropod groups. Simply speaking, the scapula-coracoid articulation has only one connection in all theropods, but this connection is highly localized and display only one surface in enantiornithines and a few crown birds, but it has one main surface and one subsidiary surface in other pennaraptoran theropods including most non-enantiornithine birds (though in these birds, the main surface is more localized, and the subsidiary facet is even smaller compared to non-avialan theropods).

Line 32 – Here produces a sharp cut in the descriptive process, with the treatment of the presence of triosseal canal among birds, without mentioning the (eventual) relation with the "only one vs double articulation" condition analyzed before.

We revised the abstract, and hopefully, the revised version reads better.

Line 33 – Is this "transitional stage" firstly recognized here? Could you precise which are the nodes representing such transition?

Yes, it is. We added a brief statement to improve clarity.

Line 34-37 – I agree with the authors in the conclusions they express in this paragraph. However, I strongly suggest to explain the importance of the "only one vs double scap-cor articulation" and the "transitional stage of triosseal canal" in the anatomical modifications occurred in the line to birds. How their observations on these two aspects modify and expand previous hypotheses on the origin and evolution of bird flight?

Following the reviewer’s suggestion, we revised the abstract.

Line 44 – Let me suggest to cite here Ostrom 1976, which I believe is the founder of this line of studies

Thank you for the references, which are now included in the current draft.

Line 50-51 – Please, check the paper by Imai et al., 2019 on Fukuipteryx. Do these authors already present a 3D reconstruction of the scapular girdle of the kind you present here for other taxa? Anyway, to be the first or not is secondary, and present manuscript represents a formidable progress in our knowledge on the flight apparatus of early birds

This sentence was rephrased

Line 67 – These warnings are necessary and welcome. Many papers on scapulocoracoid in paravian theropods overlook describing the state of preservation of the available materials.

We thank the reviewer for the encouraging comments.

Line 74-75 – I am respectful of the interpretation that current authors are following about the dromaeosaurid affiliation of Rahonavis. However, let me say that Novas et al., (2018. "Postcranial osteology of a new specimen of Buitreraptor". Cretaceous Research) emphasized about the presence of many features that Rahonavis shares with birds more derived than dromaeosaurids and Archaeopteryx (particularly the size and shape of the acromial process!), suggesting that Rahonavis is, in fact, a long-tailed avialan. Interestingly, you are here noting on a quite derived morphology of the acromion, being present in Jeholornis and Rahonavis, thus lending support to the idea that the later taxon is not a dromaeosaurid.

Following the referee’s suggestion, we put a special note on the newly discovered similarity between Rahonavis and Jeholornis. However, we will not discuss in this manuscript the systematic position of Rahonavis.

Line 80 – Is it related with the ventral surface of the acromial process?

No, the main articular surface is not related with the ventral surface of the acromial process.

Line 81 – Let me suggest to indicate also the position with respect to the glenoid. Aside from being "more medially possitioned", is it anteriorly/posteriorly/at level with the glenoid?

Made necessary amendments on this.

Line 111 – Let me suggest to change this part of the phrase, saying something like the following: "…the acrocoracoid process (frequently described as biceps or coracoid tubercle)". The reason is that Archaeopteryx is an avialan sharing a common ancestor with Sapeornis and the rest of the birds, for which the term "acrocoracoid" is used. Afterall is a matter of nomenclature, and all of us agree that such tubercle is homologous.

This sentence was rephrased according to the comment.

Line 113 – Did you check such morphology in other specimens of Sapeornis? Is it natural or is it the result of postmortem compression?

Yes, we did. This feature is also confirmed in 41HIII0405 and IVPP V 19058. We consider that it is a real feature, not the result of compression.

Line 115-118 – Such a fossa on the acrocoracoid process is also seen in Buitreraptor and Rhea, for example (it is something like a volcano with a crater on its top). Novas et al., (2021) interpreted this fossa as the site of attachment of the acrocoraco-acromial ligament. The acrocoracoacromial ligament forms a bridge under which the m. supracoracoideus slides. This ligament is also present in modern flying birds, but it is cranially covered by the omal end of the furcula.

Thus, keep in mind this in your considerations that the fossa on the acrocoracoid process served for attachment point for a coracoclavicular ligament connecting the coracoid

and furcula (an interpretation that I am not dismissing).

We were aware that both acrocoraco–acromiali and coracoclavicular ligaments are potentially attached to this fossa. However, we did not find any local rugosity, tubercle, or other indicator for acrocoraco–acromiali ligament on the proximal ends of the scapula, which is present in modern birds. Therefore, we reckoned that the acrocoraco–acromiali ligament is absent or weakly developed. On the other hand, as in most flight living birds, the well–developed furcula of Sapeornis will maintain the distance between the right and left shoulders of the pectoral girdle. This requires strong ligaments bounded the proximal ends of the furcula and coracoid. Thus, we considered that this fossa is the site of the attachment of the coracoclavicular ligament.

Line 131 – Could you provide an extra figure (a simple line drawing) depicting the correspondence of articular areas? This could be important for other authors in searching for these features in other paravians.

We colored the articular surfaces and the glenoid in Figure 4.

Line 140-141- This is an interesting interpretation. By observing Figure 4 (Sapeornis), the triosseal canal results wide. I wonder if the tendon of the m. supracoracoideus was very thick, or thicker than in other avialans, based on the reconstructed diameter of the canal.

In living birds, the triosseal canal has a large diameter. Some tendons can be surprisingly thick, but some are very thin. Therefore, it is difficult to confirm whether the tendon of Sapeornis is thicker than other taxa. If so, it may be related to elevating of the enlarged wing. However, I suspect the M. supracoracoideus tendon is too thin to fill much of the canal. It may slide around within the canal quite extensively.

Line 154-155 – keep in mind that Rahonavis may be a long-tailed avialan, instead of a deinonychosaur.

we choose to not use this reference (Forster et al., 2020) and add two new references (Brusatte et al., 2013; Funston et al., 2020) to the statement.

Line 206- Please, insert species name.

We have added “mongoliensis” to the text.

Line 237- Let me express again that Rahonavis exhibits many derived features shared with birds more derived than Archaeopteryx, which are absent in deinonychosaurs and unenlagiines. Probably the present manuscript may represents a good place to evaluate the scapular features of Rahonavis, lending support to either deinonychosaur or avialan affinities of this taxon.

Following the referee’s suggestion, we put a special note on the newly discovered similarity between Rahonavis and Jeholornis. However, we will not discuss in this manuscript the systematic position of Rahonavis.

Line 273-275- I agree, of course, with this conclusion. Fossil bones are clear in indicating this condition. I wonder, however, if enantiornithes had (in life) a procoracoidal portion (cartilaginous) for contacting with the well developed acromial region of proximal scapula. A "proto-procoracoid" or more properly a "supracoracoid canal" was present and well developed in Buitreraptor, and presumably served as a "channeling surface" for the m. supracoracoid in its course towards the proximal end of humerus. Crown birds have procoracoid, and a similar channeling surface was present among basal paravians. Enantiornithes may have had such structure, but in an unossified condition.

The coracoid is known as an endochondral bone developing from cartilaginous precursors. As a part of the coracoid, the cartilage procoracoid process will be replaced by bone at the later development stage. So far, no cartilage procoracoid process has been found in adult living birds. Therefore, it is unlikely that Enantiornithes has a cartilage procoracoid process.

Line 290-300- I believe it is unnecessary to repeat. Authors have already clearly explained this in previous lines.

As suggested, we have deleted this paragraph.

Line 313-315 – s it right to say "roofed"? Is the procoracoid forming the caudoMEDIAL wall of the canal, instead of a roof? Apologies if I am wrong.

We changed “roof” to “caudomedial wall”.

Line 322- In Rhea the acrocoraco-acromial ligament bridges the tendon of m. supracoideus, functionally acting as a "triosseal canal". See Novas et al., 2021 for discussions on these asepcts.

Line 338-340- I am surprised for this statement: in basal paravians as well as in living birds the m. supracoracoid runs along the LATERAL side of the acromion. The medial surface of the later one is contiguous with the medial surface of scapular blade, thus the medial surface of the acromion contacts with the underlying thorax. The supracoracoid doesn´t run between the scapula and ribs!

Yes, the reviewer is correct. We rejected the assumption of Mayr (2017). The supracoracoideus pulley system in enantiornithine birds was generally the same as that of living birds and other early branches of paravians.

Line 348-349- The presence of a triosseal canal is defined by the contact between the acrocoracoid and epicleidium. If such a contact does not exist, then such a "triosseal" canal is not defined. of course, this is a semantic problem.

A completely different matter is the function that this region of the scapular girdle played in paravians: a well defined though is present on the craniomedial surface of coracoid in Buitreraptor (which may have act as a "proto-procoracoid" for chanalizing the m. scapulocoracoid. Thus, the deflection of its tendon was probably operative in basal paravians, and the "inner half" of the triosseal canal was already present even in basal paravians. The shape of the coracoid of Sapeornis (especially the development and position of acrocoracoid) resembles that of Archaeopteryx and Buitreraptor, thus it is expectable that the "pulley system" was postioned and functioned in a similar way in all these forms.

In sum, I agree with authors regarding the function of this region of scapular girdle, but the term applied ("triosseal canal") for Sapeornis seems inadequate. Besides, but no less important, a similar pulley-shaped morphology for coracoid was also identified for basal paravians (Buitreraptor, Archaopteryx; Novas et al., 2021).

We disagree that if such a contact does not exist, then such a "triosseal" canal is not defined. As we stated in the previous vision, the triosseal canal is not necessarily formed by all three pectoral elements, and not necessarily a fully enclosed bony passage. The annotation provided for the triosseal canal by Baumel and Witmer (1993) explicitly described variation across taxa in the canal’s architecture.

We believe a partially enclosed structure in non–euornithine birds should be also considered as a triosseal canal as long as it functions as changing the direction of the tendon of the m. supracoracoideus and elevating the wing. In Sapeornis, the acrocoracoid process is slightly above the midpoint of the coracoidal glenoid fossa, and consequently above the insertion point during the humeral depression as in the living bird. The vector of the tension is therefore exerted by m. supracoracoideus on the humerus is directed cranially and somewhat dorsomedially. Thus, the term “triosseal canal” is applied for Sapeornis.

We agree that acrocoracoid process (coracoid tubercle) are functioned in Archaeopteryx, Buitreraptor, and other non–avialan pennaraptorans. The term “sulcus m. supracoracoideus” is more appropriate for the groove (or cavity) on the coracoid for the tendon of M. supracoracoideus in these groups.

Line 375-376- Notably, as you shown in Figure 4, Sapeornis does not exhibit a derived condition of acrocoracoid, because it is at level with the glenoid, not in front of it. Thus, I don´t understand why the "extension of acrocoracoid process dorsal to level of coracoidal glenoid fossa" (a condition that seems absent in Sapeornis), is interpreted as characteristic of Step I.

In Sapeornis and most other non–ornithothoracine avialans (e.g., Jeholornis and Confuciusornis), the acrocoracoid process is slightly above the midpoint of the coracoidal glenoid fossa. We have done appropriate amendments in lines 530-571.

Line 403-407- Novas et al., (2020, 2021) said that the pulley was already operative in basal paravians such as Buitreraptor and Archaeopteryx, but the m. supracoracoideus acted as a humeral PROTRACTOR (as in living ratites). This condition is intermediate between the humeral depressor action (in basal theropods) and humeral elevator role (in extanct birds).

Novas et al., (2020, 2021) did not say that the "m. supracoracoideus would have acted straightforwardly as a wing depressor".

We are very sorry for the mistake. We have done appropriate amendments in lines 530-571.

Line 419-422- Based on what is beautifully illustrated in present manuscript, the coracoid of Sapeornis does not show a condition of the acrocoracoid sharply different from that in Archaeopteryx. In other terms, the acrocoracoid process is not in front of the glenoid, but almost at the same level. So, I will say the m. supracoracoideus in Sapeornis functioned as a humeral protractor, not as an elevator.

The acrocoracoid process of Archaeopteryx is well below the glenoid. We agree that the m. supracoracoideus of Sapeornis would have acted primarily as a humeral protractor, but we think it would have pulled the humerus dorsomedially more or less. We have done appropriate amendments on this. please check also throughout the text at lines 530-571.

Line 720 (and labels inserted in Figure 1H)- Is this the border closer to the sternum? Then, it has to be "distal" using the terminology applied on the text (the glenoid represents the "proximal end" of the coracoid).

No, this is the omal end. As we mentioned above, we modified figures to help the readers understand them more easily the morphology.

https://doi.org/10.7554/eLife.76086.sa2

Article and author information

Author details

  1. Shiying Wang

    1. Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China
    2. CAS Center for Excellence in Life and Paleoenvironment, Beijing, China
    3. University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Conceptualization, Visualization, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6067-0303
  2. Yubo Ma

    University of Alberta, Edmonton, Canada
    Contribution
    Writing – review and editing
    Competing interests
    No competing interests declared
  3. Qian Wu

    1. Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China
    2. CAS Center for Excellence in Life and Paleoenvironment, Beijing, China
    3. University of Chinese Academy of Sciences, Beijing, China
    Contribution
    Writing – review and editing
    Competing interests
    No competing interests declared
  4. Min Wang

    1. Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China
    2. CAS Center for Excellence in Life and Paleoenvironment, Beijing, China
    Contribution
    Writing – review and editing
    For correspondence
    wangmin@ivpp.ac.cn
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8506-1213
  5. Dongyu Hu

    Shenyang Normal University, Shenyang, China
    Contribution
    Conceptualization, Writing – review and editing
    For correspondence
    hudongyu@synu.edu.cn
    Competing interests
    No competing interests declared
  6. Corwin Sullivan

    1. University of Alberta, Edmonton, Canada
    2. Philip J. Currie Dinosaur Museum, Wembley, Canada
    Contribution
    Writing – review and editing
    Competing interests
    No competing interests declared
  7. Xing Xu

    1. Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China
    2. CAS Center for Excellence in Life and Paleoenvironment, Beijing, China
    3. Shenyang Normal University, Shenyang, China
    4. Centre for Vertebrate Evolutionary Biology, Yunnan University, Kunming, China
    Contribution
    Conceptualization, Writing – original draft, Writing – review and editing
    For correspondence
    xingxu@vip.sina.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4786-9948

Funding

National Natural Science Foundation of China (41688103)

  • Xing Xu
  • Dongyu Hu

National Natural Science Foundation of China (42288201)

  • Xing Xu
  • Dongyu Hu

National Natural Science Foundation of China (42072030)

  • Xing Xu
  • Dongyu Hu

International Partnership Program of Chinese Academy of Sciences (132311KYSB20180016)

  • Xing Xu

Natural Sciences and Engineering Research Council of Canada (Discovery Grant RGPIN-2017-06246)

  • Corwin Sullivan

University of Alberta

  • Corwin Sullivan

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Xiaoqing Ding for specimen preparation; Minghui Ren for illustration; Yun Feng and Yemao Hou for help with computed tomography scanning; and Paul Rummy for discussion. We thank the reviewers Jingmai K O’Connor and Fernando E Novas for their helpful and constructive comments that greatly contributed to improving this manuscript. This work was supported by the National Natural Science Foundation of China (41688103, 42288201, 42072030), the International Partnership Program of Chinese Academy of Sciences (132311KYSB20180016), Natural Sciences and Engineering Research Council of Canada funding (Discovery Grant RGPIN-2017-06246), and start-up funding awarded by the University of Alberta to CS.

Senior Editor

  1. George H Perry, Pennsylvania State University, United States

Reviewing Editor

  1. David Lentink, University of Groningen, Netherlands

Reviewers

  1. Jingmai O'Connor
  2. Fernando E Novas

Publication history

  1. Received: December 3, 2021
  2. Preprint posted: December 10, 2021 (view preprint)
  3. Accepted: March 30, 2022
  4. Accepted Manuscript published: March 31, 2022 (version 1)
  5. Version of Record published: April 21, 2022 (version 2)

Copyright

© 2022, Wang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 542
    Page views
  • 149
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Shiying Wang
  2. Yubo Ma
  3. Qian Wu
  4. Min Wang
  5. Dongyu Hu
  6. Corwin Sullivan
  7. Xing Xu
(2022)
Digital restoration of the pectoral girdles of two Early Cretaceous birds and implications for early-flight evolution
eLife 11:e76086.
https://doi.org/10.7554/eLife.76086

Further reading

    1. Ecology
    2. Evolutionary Biology
    Longhui Zhao et al.
    Research Article Updated

    Many animals rely on complex signals that target multiple senses to attract mates and repel rivals. These multimodal displays can however also attract unintended receivers, which can be an important driver of signal complexity. Despite being taxonomically widespread, we often lack insight into how multimodal signals evolve from unimodal signals and in particular what roles unintended eavesdroppers play. Here, we assess whether the physical movements of parasite defense behavior increase the complexity and attractiveness of an acoustic sexual signal in the little torrent frog (Amolops torrentis). Calling males of this species often display limb movements in order to defend against blood-sucking parasites such as frog-biting midges that eavesdrop on their acoustic signal. Through mate choice tests we show that some of these midge-evoked movements influence female preference for acoustic signals. Our data suggest that midge-induced movements may be incorporated into a sexual display, targeting both hearing and vision in the intended receiver. Females may play an important role in incorporating these multiple components because they prefer signals which combine multiple modalities. Our results thus help to understand the relationship between natural and sexual selection pressure operating on signalers and how in turn this may influence multimodal signal evolution.

    1. Computational and Systems Biology
    2. Evolutionary Biology
    Louisa Gonzalez Somermeyer et al.
    Research Article Updated

    Studies of protein fitness landscapes reveal biophysical constraints guiding protein evolution and empower prediction of functional proteins. However, generalisation of these findings is limited due to scarceness of systematic data on fitness landscapes of proteins with a defined evolutionary relationship. We characterized the fitness peaks of four orthologous fluorescent proteins with a broad range of sequence divergence. While two of the four studied fitness peaks were sharp, the other two were considerably flatter, being almost entirely free of epistatic interactions. Mutationally robust proteins, characterized by a flat fitness peak, were not optimal templates for machine-learning-driven protein design – instead, predictions were more accurate for fragile proteins with epistatic landscapes. Our work paves insights for practical application of fitness landscape heterogeneity in protein engineering.