Forelimb musculature and function in the therizinosaur Nothronychus (Maniraptora, Theropoda)

2.2.1. M. serratus superficialis (SS) level I inference

M. serratus superficialis is a broad sheet in crocodylians and birds, but the exact architecture of this muscle is variable in archosaurs (Burch, 2014; Jasinoski et al., 2006). Jasinoski et al. (2006) described M. serratus superficialis as a single muscle that originates on the posterior cervical and anterior dorsal ribs and inserting posterior margin of the scapula in crocodylians. Burch (2014) treated this muscle as a single unit, but Jasinoski et al. (2006) described discrete pars cranialis and pars caudalis divisions that may be connected by an aponeurosis in neognath birds. In Cygnus and Meleagris, the cranial head inserts on a small tubercle on the posterior margin of the scapula. M. serratus superficialis pars caudalis originates on dorsal ribs behind pars cranialis. It inserts on the posterior margin of the scapula.

Striations in the distal scapula presumably reflect the attachment of M. serratus superficialis in Maiasaura (Dilkes, 2000). Burch (2014, 2017) phylogenetically reconstructed M. serratus superficialis in Tawa and Majungasaurus. Jasinoski et al. (2006) referred two proposed discrete heads to M. serratus superficialis pars cranialis and M. serratus superficialis pars caudalis. The origination in Tawa is modeled based on extant archosaurs, arising from the posterior cervical and anterior dorsal ribs (Burch, 2014). She described a scar in the posteroventral margin of the scapula distal groove in the ventral scapulae of many coelurosaurs that can be referred to the insertion of this muscle. Its development varies from a small tubercle to an elongate groove. Such a scar is not present in Tawa or other non‐tetanurines, so insertion of M. serratus superficialis has no osteological correlate in these taxa. Burch proposed an elongate attachment to the posteroventral margin in Tawa, similar to lizards (Russell & Bauer, 2008) and crocodilians (Meers, 2003). A shallow groove distal to the midpoint of the scapula is inferred as the insertion in Majungasaurus and other ceratosaurs (Burch, 2017). Jasinoski et al. (2006) used the observed origin on the posterior cervical and anterior dorsal ribs in birds to infer a similar attachment in dromaeosaurs. They used the presence of a small tubercle in the scapulae of the oviraptorid Ingenia similar to Cygnus and Meleagris as evidence for a discrete pars cranialis, and, therefore, a division of this muscle in maniraptorans. They noted the lack of any scarring in dromaeosaurs that can be associated with this muscle, but M. serratus superficialis pars cranialis is reconstructed as inserting on a point above the glenoid fossa, as in Cygnus and Meleagris. M. serratus superficialis pars caudalis is inferred as inserting on the posterior margin of the scapular margin above pars cranialis, but there is no osteological correlate (Jasinoski et al., 2006). This muscle would have retracted the scapula.

The available ribs for N. graffami are disarticulated and there are no features on the ribs, such as uncinate processes, that can attributed to the origin for M. serratus superficialis. Therefore, the muscle must have originated on the lateral surfaces of the ribs. The described tubercle that Burch (2014) hypothesized as the attachment for M. serratus superficialis pars cranialis is absent in Nothronychus. Hedrick et al. (2015) described the existence of a groove on the ventral margin on the distal scapula of N. mckinleyi. It is absent in N. graffami. Hedrick et al. regarded it as a distal “re‐establishment” of the more pronounced proximal groove, but is quite separate from it. Given its location on the scapula, the groove is interpreted as marking a broad insertion of M. serratus superficialis (Figure 3) as in Tawa, Majungasaurus, and dromaeosaurs (Burch, 2014, 2017; Jasinoski et al., 2006).

2.2.2. M. serratus profundus (SP) level I inference

M. serratus profundus has been described for crocodylians and birds (Burch, 2014; Jasinoski et al., 2006). It is deep to M. serratus superficialis in crocodylians (Russell & Bauer, 2008). Jasinoski et al. describe the muscle insertion on anterior margin of the scapula at a rugose region and Burch confirmed this observation. The muscle is described as originating on an aponeurosis attaching to the posterior dorsal, anterior dorsal vertebrae, and the lateral surface of the cervical and/or dorsal ribs. It inserts on the medial surface of the scapula in neognath birds.

M. serratus profundus is reconstructed for Tawa as in extant archosaurs such that it would have originated on the dorsal ribs adjacent to the associated vertebrae (Burch, 2014) and dromaeosaurs (Jasinoski et al., 2006). Jasinoski et al. expanded the origin to include the dorsal vertebrae in dromaeosaurs. Burch reconstructed the insertion on the distal scapula, but there are no scars delimiting its attachment. In dromaeosaurs, Jasinoski et al. used a crocodylian/lepidosaur model to reconstruct an insertion on the medial surface of the scapula. There is no direct evidence of M. serratus profundus in Maiasaura (Dilkes, 2000). The medial region of the scapula is marked by a broad, shallow distal excavation in Nothronychus (Figure 3), the distal end of which is reconstructed as the M. serratus profundus insertion as proposed for other theropods. The muscle would have protracted the scapula (Burch, 2014; Jasinoski et al., 2006).

2.2.3. M. sternocoracoideus level I′ inference

M. sternocoracoideus is present in birds, but may be homologous with M. costocoracoideus of crocodylians (Jasinoski et al., 2006). Jasinoski et al. describe two components, pars superficialis and pars profundus, that are only separated near the insertion. The former inserts along the posterior margin of the scapula, while the latter inserts on the medial scapulosternal ligament overlying the coracoid and sternum (Jasinoski et al., 2006; Nicholls & Russell, 1985). In birds, M. sternocoracoideus originates at the sternum, and sometimes on the sternal ribs (George & Berger, 1966; Jasinoski et al., 2006), but this architecture is equivocal in other theropods. They described it as inserting on the medial side of the coracoid including the coracoid process within an extensive depression, the impressio M. sternocoracoidei (Baumel & Witmer, 1993). George and Berger (1966), Russell and Bauer (2008), and Bradley et al. (2020) report an insertion on the medial to dorsal region of the coracoid near the contact with the sternum in lizards and birds.

Jasinoski et al. (2006) reconstructed M. sternocoracoideus for dromaeosaurs as originating on the lateral side of the sternal ribs and inserting on a triangular depression below the coracoid foramen on the medial side of the coracoid as in neognathine birds, reminiscent of pars superficial in crocodylians. M. sternocoracoideus is not restored for Tawa or Majungasaurus (Burch, 2014, 2017) as the sternum is rarely preserved (Bradley et al., 2020). Bradley et al., however, described a sternum referred to as Tawa. They reconstructed the M. sternocoracoideus externus head origin on the lateral side of the sternocoracoidal process, sternum and first two sternal ribs, as in birds (George & Berger, 1966). The origin of this muscle is generally and tentatively reconstructed as in dromaeosaurs (Jasinoski et al., 2006). The sternal plate is not preserved in Nothronychus, hence the M. sternocoracoideus origin cannot be specifically reconstructed. Bradley et al. reconstruct the insertion on the medial surface of the coracoid. The medial coracoid of Nothronychus is broadly excavated (Hedrick et al., 2015), and is interpreted as the insertion for M. sternocoracoideus (Figure 3). M. sternocoracoideus retracts the coracoid within the coracoid groove in the sternum (Bradley et al., 2020; Jasinoski et al., 2006).

2.2.4. M. rhomboideus (RH) level I′ inference

The configuration of M. rhomboideus is variable in theropods (Burch, 2014; Jasinoski et al., 2006). The topology of M. rhomboideus is heavily influenced by the orientation of the pectoral girdle. Crocodylians possess an almost vertical orientation, while birds are characterized by a horizontal orientation (Burch, 2014; Chiasson, 1962, 1984). The profundus division is only present in animals with a horizontal scapula (Burch, 2014; Jasinoski et al., 2006).

Lacking any osteological data, Burch (2014) did not distinguish between the divisions in non‐avian theropods, so a single muscle is reconstructed for Tawa. She reconstructed the scapular orientation intermediate between the vertical condition in crocodylians and horizontal in birds. Therefore, Burch suggested that the origin of M. rhomboideus in Tawa was intermediate between birds and crocodiles, attaching to the cervico‐dorsal fascia and adjacent neural spines. It would have inserted on the anterior half of the distal region of the medial scapular blade (Burch, 2014) in contrast to an origination on the anterior margin of the scapula reconstructed for dromaeosaurs (Jasinoski et al., 2006).

Jasinoski et al. (2006) tentatively reconstructed M. rhomboideus superficialis and M. rhomboideus profundus in dromaeosaurs as similar to birds. M. rhomboideus superficialis would have originated from the anterior dorsal vertebrae, as in birds, such as pigeons (Chiasson, 1984). It is modeled as inserting on the anterior edge of the scapula, probably extending onto the epicleidium of the furcula. It originates from neural spines of the posterior cervical and anterior dorsal vertebrae, but Jasinoski et al. (2006) limited it to the anterior dorsal vertebrae in dromaeosaurs as a result of the horizontal orientation of the reconstructed scapula. They tentatively reconstructed it as inserting on the dorsal edge of the blade of the distal scapula.

Brochu (2003) noted the insertion of the rhomboideus musculature on the anterior border of the scapula in birds (Chiasson, 1984; George & Berger, 1966), but was uncertain about the significance of the rugose anterior scapular margin in tyrannosaurs. The acromion is incomplete in Nothronychus, so a bird‐like configuration (Chiasson, 1984) along the dorsal margin is tentative, similar to dromaeosaurs, following Jasinoski et al. (2006). This topology is in contrast to the more distal insertion described for crocodilians (Chiasson, 1962; Meers, 2003).

Nothronychus almost certainly possessed an intermediately directed scapula as reconstructed in non‐maniraptoran theropods (Burch, 2014; Jasinoski et al., 2006; Senter & Robins, 2015), therefore, the topology of M. rhomboideus was presumably also similar to the condition seen in Tawa and a single head is reconstructed, rather than two as in dromaeosaurs. M. rhomboideus would have originated from the spines of the posterior cervical and dorsal vertebrae and possibly the M. longissimus dorsi fasciae. Since Nothronychus possesses an intermediately oriented scapula, this muscle is reconstructed similar to Tawa in inserting on the medial blade of the distal scapula (Figure 3), following Burch (2014). There are, however, no specific scars on the scapula supporting this interpretation. The muscle would have protracted the scapula (Burch, 2014; Jasinoski et al., 2006).

2.2.5. M. levator scapulae (LS) level II inference

M. levator scapulae is absent in birds, (Burch, 2014; Jasinoski et al., 2006), but present in crocodylians, so Jasinoski et al. make it equivocal in theropods. In crocodylians, the muscle originates at the anterior cervical ribs and fascia of the cervical muscles (Jasinoski et al., 2006). It inserts along the anterior margin of the scapula (Burch, 2014; Chiasson, 1962; Meers, 2003). Jasinoski et al. and Burch describe it as inserting on the anterior margin of the scapula posterior to the acromial process in crocodylians, sometimes in an elongate sulcus. They observe this groove in Ceratosaurus, Sinraptor, Megalosaurus, Tarbosaurus, and tetanurines, but not basal theropods, such as Tawa. Jasinoski et al. did not reconstruct M. levator scapulae in dromaeosaurs, but considered it possibly present. There is scarring in the anterior margin of Maiasaura that could be attributed either to this muscle or to M. trapezius (Dilkes, 2000). Burch (2014) considered M. levator scapulae present in Tawa, despite the lack of definitive scarring, given the existence of such a groove in crocodylians and more derived theropods.

Scarring attributed to M. levator scapulae is observed in crocodylians and theropods more primitive than Nothronychus, but not more derived ones. There is a slight expansion along a rough dorsal margin with a very shallow parallel groove at its base in Nothronychus. The expanded margin is tentatively reconstructed as a combined insertion for M. levator scapulae and M. trapezius in Nothronychus, possibly expanding onto the epicleidium, where a shallow groove is present. (Figure 3). The anterior dorsal blade is thickened distally, but it is not distinctly rugose. Unfortunately, the scapular acromion is incomplete in the available material of Nothronychus. M. levator scapulae would have rotated the scapula and served as a lateroflexor for the neck (Burch, 2014).

2.2.6. M. latissimus dorsi (LD) level I inference

This broad, sheet‐like muscle is present in birds (Chiasson, 1984) and crocodylians (Chiasson, 1962; Meers, 2003). Jasinoski et al. (2006) report that there are no definite muscle scars related to the origin in extant archosaurs. In most birds, it inserts posterior to the deltopectoral crest at a pit, sulcus, tubercle, or crest (Burch, 2014; Jasinoski et al., 2006).

Jasinoski et al. (2006) indicated that the origin and number of heads of M. latissimus dorsi is equivocal in dromaeosaurs. They prefer an origin at the dorsal vertebrae and a single head. Dromaeosaurs, as in extant crocodylians, birds, and many non‐avian theropods, possess a linear sulcus near the deltopectoral crest that would serve as an insertion for the muscle. Such a scar is also present in Tawa (Burch, 2014). Burch (2017) reconstructs an origin at dorsal vertebrae 1‐5 and insertion at a depression between tubercles in the posterolateral humerus in Majungasaurus.

M. teres major is described as a division of M. latissimus dorsi in at least some diapsids, and its presence as a discrete muscle in dinosaurs is considered equivocal (Dilkes et al., 2012). Dilkes (2000) reconstructed the origin of M. teres major on the posterolateral or posterior margin of the scapula in Maiasaura, but noted that there are no osteological correlates for this muscle. Lipkin and Carpenter (2008) reconstruct it as sharing an insertion with M. teres major, following the condition in crocodilians (Meers, 2003), however, neither Burch nor Jasinoski et al. reconstructed a separate M. teres major for other theropods.

Following the reconstruction in other theropods, the origin of M. latissimus dorsi is placed on the cervical and dorsal vertebrae in Nothronychus, but this reconstruction is tentative. It could have originated on the scapula as Dilkes (2000) proposed for Maiasaura. As in many theropods, including Tawa and the dromaeosaurs (Burch, 2014; Jasinoski et al., 2006), the posterior humerus of Nothronychus possesses a long sulcus that is interpreted as the insertion for M. latissimus dorsi (Figure 4). The sulcus is about midway down the shaft, somewhat more distal than the location illustrated for Tawa (Burch, 2014), presumably increasing the moment arm associated with retracting the humerus. The reconstructed insertion in Nothronychus is in the same general area, but more extensive than that sometimes modeled for Tyrannosaurus (Lipkin & Carpenter, 2008). M. latissimus dorsi would have retracted the humerus (Burch, 2014).

2.2.7. M. trapezius (T) level II′ inference

The topology of M. trapezius is similar to that of M. levator scapulae (Burch, 2014). Burch noted that it lacks any osteological scarring. This muscle is reconstructed as present in many non‐avian theropods (Burch, 2014), related to the vertical scapular orientation in crocodilians. Jasinoski et al. (2006) and Burch (2014) described it as lost in birds, associated with a nearly horizontal scapular orientation. Senter and Robins (2015), however, reconstruct most non‐avian theropods with nearly horizontal scapulae.

Burch (2014) reconstructed M. trapezius for Tawa using a crocodylian model, but Jasinoski et al. (2006) did not model it in dromaeosaurs, regarding it as highly equivocal in theropods. A reconstruction in Majungasaurus is similar (Burch, 2017). M. trapezius would have originated in thoracic muscular fascia and inserted along the dorsal margin of the scapula with M. levator scapulae in Tawa. Given the intermediate orientation of the scapula and the lack of a dorsal groove as seen in Majungasaurus (Burch, 2017), the reconstruction of M. trapezius with M. levator scapulae in Nothronychus follows the reconstruction in Tawa (Figure 3), but must be regarded as tentative. Both muscles may be lost as in other derived theropods. Burch (2014) indicates this muscle rotated the scapula and possibly contributed to protracting the forelimb.

2.2.8. M. pectoralis (P) level I′ inference

M. pectoralis is a large muscle in crocodylians and birds, but is complex (Jasinoski et al., 2006). The origin of M. pectoralis is variable among diapsids (Burch, 2014) but consistently on the sternum. Jasinoski et al. (2006) describe it as a broad sheet in crocodylians. In crocodilians, the muscle is complex (Meers, 2003). Typically, it originates on the sternum and thoracic costal cartilages (Chiasson, 1962) or sternal ribs (Jasinoski et al., 2006). Jasinoski et al. describe M. pectoralis as inserting on the ventral deltopectoral crest. Jasinoski et al. (2006) describe an M. pectoralis origin on the keel, lateral sternum, and furcula, sometimes including the sternocoracoclavicular membrane in birds. In pigeons, it originates on the sternum and furcula (Chiasson, 1984), while in ostriches, it originates at the coracoid (Burch, 2014). The muscle inserts as an aponeurosis into the anteroventral shaft of the humerus.

M. pectoralis is usually reconstructed as originating from cartilaginous elements of the ribs and the ventral sternum in theropods (Dilkes et al., 2012), possibly extending to the coracoid. Makovicky and Currie (1998) used the avian model of Dial et al. (1987) to support a partial origin of M. pectoralis on the furcula in Tyrannosaurus. They further suggested that the reduced furcula and, presumably M. pectoralis, was related to the reduction in forelimbs. Olson and Feduccia (1979) reconstructed a well‐developed M. pectoralis origin on the expanded furcula of Archaeopteryx that may be related to an enlarged M. supracoracoideus, both well developed in Nothronychus. In no case, however, is there any direct evidence for such an attachment. Conversely, Carpenter and Smith (2001) and Carpenter (2002) argued that M. pectoralis is a ventral muscle mass and the furcula was anterodorsal to the shoulder in non‐avian theropods. Therefore, it could not originate on the furcula in Tyrannosaurus, and they argued for an origin of M. deltoideus here. M. pectoralis would have originated on the scapula. However, no known extant animals have such an attachment (D.W. Dilkes, pers. comm.), so such an architecture seems unlikely. Padian (2004) argued that the sternal plate was probably cartilaginous in non‐avian theropods. Jasinoski et al. (2006) tentatively proposed an origin somewhere on the sternum, exclusive of the coracoid in dromaeosaurs. Burch (2014) did not definitively identify an origin for M. pectoralis in Tawa beyond the sternum, but tentatively suggested an origin of M. pectoralis on the coracoid, sternal ribs, or unossified sternal plates as possibilities in Tawa. Burch (2017) identified a rugosity on the ventral coracoid of Aucasaurus that she tentatively identified as an origin for M. pectoralis. Bradley et al. (2020) described the M. pectoralis origination on the ventral face of the sternum, mesosternum, and sternal ribs in lizards and crocodylians. They follow George and Berger (1966) in ascribing a restricted origin for M. pectoralis on the ventral margin on the sternum and enlarged M. supracoracoideus attachment most of the ventral sternum in birds. Bradley et al. reconstruct an enlarged origin for M. pectoralis and reduced M. supracoracoideus origin in Tawa, as in crocodylians. The origin was likely limited laterally by a low ridge in the proximal half of the sternum.

M. pectoralis is commonly reconstructed as inserting on the humerus (e.g., Burch, 2014; Jasinoski et al., 2006). Unlike the reconstruction in dromaeosaurs (Jasinoski et al., 2006), where discrete osteological correlates are absent, Burch (2014) described a depression in the medial deltopectoral crest of Tawa that she ascribed to a restricted insertion. The insertion in Majungasaurus is generally reconstructed on the medial deltopectoral crest (Burch, 2017). Dilkes (2000) reconstructed the insertion of M. pectoralis on the apex of the deltopectoral crest, continuous with a more proximal attachment for M. supracoracoideus and no M. supracoracoideus accessorius in Maiasaura. A similar point of attachment was proposed for dromaeosaurs (Jasinoski et al., 2006).

M. pectoralis is reconstructed in Nothronychus as originating on the anterior face of the furcula extending to the coracoid (Figure 3) and possibly the sternal plates if present, but there are no definite scars and this architecture is highly speculative. The insertion for M. pectoralis is on the deltopectoral crest in all archosaurs (Burch, 2014; Chiasson, 1984; Jasinoski et al., 2006). Nothronychus is characterized by a very shallow excavation, so the attachment is reconstructed similarly (Figure 4). Barsbold (1976) reconstructs the insertion on a medial rugose surface on the extremity of the deltopectoral crest in Therizinosaurus, but this surface probably represents separate insertions for M. supracoracoideus and M. supracoracoideus anterior. The insertion for M. pectoralis was reduced to a tendon on the anterior surface in Nothronychus. Nicholls and Russell (1985) suggested that the insertion of M. pectoralis and M. supracoracoideus might be separate in Struthiomimus, describing a thickened extremity of the deltopectoral crest. Separation of these two muscles is apparent in Nothronychus, as well. M. pectoralis would have protracted and adducted the humerus (Burch, 2014).

2.2.9. M. subscapularis (SBS) level I inference

M. subscapularis is very closely associated with M. scapulohumeralis posterior in crocodylians (Jasinoski et al., 2006). In extant crocodylians (Chiasson, 1962), ostriches, and iguanas (Jasinoski et al., 2006), a single head of M. subscapularis originates on the medial side of the scapula. It inserts on the posterior tuberosity of the humerus (Chiasson, 1962; Jasinoski et al., 2006). The muscle divides into two heads in pigeons (Chiasson, 1984) and this architecture is observed in most other birds (Jasinoski et al., 2006). One head originates on the medial side and the other on the posterolateral margin of the ventral third of the scapula. They fuse and insert on posterior tubercle of the humerus (Chiasson, 1984; Jasinoski et al., 2006), proximal to the heretofore undescribed fossa pneumotricipitalis in non‐avian theropods (Baumel & Witmer, 1993; Jasinoski et al., 2006).

Separate M. subscapularis and M. subcoracoideus, as in most birds (Dilkes, 2000), were retained in reconstructions of this muscle in Tawa (Burch, 2014) and Majungasaurus (Burch, 2017). In Maiasaura, Dilkes merged it with M. subcoracoideus into a combined M. subcoracoscapularis following Romer (1944), as observed in turtles and crocodylians, referred to M. subscapularis (Chiasson, 1962). In Tawa, Majungasaurus, and dromaeosaurs, the muscle would have originated on the medial side of the scapula, as is typical for theropods (Burch, 2014; Jasinoski et al., 2006). Burch (2014) described a medial ridge in the proximal half of the scapula of Tawa. A similar ridge defining a depression was observed in the posteroventral margin of the scapula in Meleagris (Jasinoski et al., 2006). They suggested that it served as the M. subscapularis origin, with the base of the M. subscapularis caput mediale originating from a broad depression in the medial of the scapula in oviraptorids, dromaeosaurids, and troodontids in the same area as the medial longitudinal ridge as in Meleagris. Burch (2014), however, noted a similar ridge in crocodylians (Meers, 2003). It is part of the origin of M. scapulohumeralis posterior. She suggests that the M. subscapularis origin was displaced ventrally in maniraptoran theropods, presumably characterized by a horizontal scapula, but retains a more dorsal attachment in Tawa. Jasinoski et al. (2006) did not reconstruct the lateral head in dromaeosaurs as no scarring is apparent. Barsbold (1976) describes a broadly concave medial surface in the acromial process bounded posteriorly by an oblique ridge in scapula of Therizinosaurus. He interpreted this region as the origin for M. subscapularis.

Burch (2014) reconstructed the insertion on the internal tuberosity of the humerus in Tawa. A similar insertion was inferred for dromaeosaurs (Jasinoski et al., 2006). Typically, M. subscapularis and M. subcoracoideus merge into a common inserting tendon in archosaurs (Burch, 2017) as in pigeons (Chiasson, 1984). Jasinoski et al. (2006) used that topology in dromaeosaurs, but Burch suggested a separate insertion in abelisaurs, shown by a medial ridge in the internal tuberosity may be present in Majungasaurus.

Nothronychus is reconstructed with an intermediately oriented scapula, similar to Tawa, so the origin of a single head of M. subscapularis is similarly reconstructed within a broad excavation in the medial scapula. This reconstruction differs from that of Barsbold (1976) for Therizinosaurus in that M. subscapularis broadly originates on the medial surface of the scapula rather than limited to the medial surface of the acromial process. The parallel crests on the ventral margin of the scapula described previously in Nothronychus (Hedrick et al., 2015) is regarded here constitute the origin for M. scapulohumeralis posterior (Figure 3), following Burch (2014). This hypothesis agrees with the attachment proposed for Therizinosaurus (Barsbold, 1976). The insertion of M. subscapularis is reconstructed as on the proximal medial tuberosity adjacent to the head (Figure 4). This reconstruction follows that of Jasinoski et al. (2006) and Burch (2014) for dromaeosaurs and Tawa. Nothronychus lacks the dividing ridge of Majungasaurus (Burch, 2017), so presumably, M. subscapularis and M. subcoracoideus shared a common insertion tendon, as in pigeons (Chiasson, 1984) and most other archosaurs (Burch, 2017). In contrast to Maiasaura (Dilkes, 2000), the insertions of M. subscapularis and M. scapulohumeralis posterior are easily distinguished in Nothronychus. M. subscapularis would have adducted and rotated the humerus (Burch, 2014).

2.2.10. M. subcoracoideus (SBC) (level I inference)

This muscle is present in crocodylians and birds. Jasinoski et al. (2006) and Burch (2014) described M. subcoracoideus as synonymous with M. part of M. subcoracoscapularis in crocodylians, originating on the scapula following Romer (1944). Jasinoski et al. describe the origin of M. subcoracoideus caput dorsale on the coracoid over the coracoid foramen in most birds. M. subcoracoideus caput ventrale originates on the anterior margin of the coracoid. In many birds, caput dorsale and caput ventrale merge with M. subscapularis to insert on the posterior tubercle of the humerus.

In Tawa, Burch (2014) followed the reconstruction of Jasinoski et al. (2006) for dromaeosaurids in that M. subcoracoideus covers the coracoid foramen on the medial side of the scapula (Figure 3). It should be noted that the coracoid foramen in this specimen of N. graffami is incomplete on the posterodorsal side and the margins are damaged. Therefore, the apparent dimensions of the foramen are unclear. Jasinoski et al. (2006) regarded the second head as probably absent in dromaeosaurs, and Burch (2014) did not reconstruct it in Tawa. In any case, M. subcoracoideus would have fused with the tendon for M. subscapularis and inserted on the internal tuberosity of the humerus. Burch (2017) described a striated area in Majungasaurus anterior to the coracoid foramen that she ascribed to the M. subcoracoideus origin. Nothronychus is reconstructed similarly to other theropods (Figure 3). Barsbold (1976) similarly reconstructed the origin of M. subcoracoideus on the medial acromial process of the coracoid in Therizinosaurus. As in the reconstruction of Barsbold, M. subscapularis and M. subcoracoideus merged into a single tendon inserting on the internal tuberosity of the humerus in Nothronychus. This muscle would have rotated and adducted the humerus (Burch, 2014).

2.2.11. M. supracoracoideus (SC) (level I inference)

M. supracoracoideus variably arises from the coracoid, but may extend to the scapula and sternum in archosaurs (Burch, 2014). In some crocodylians, including alligators, one head originates on the medial surface of the procoracoid (Chiasson, 1962), whereas in others, two or three heads are present (Cong et al., 1998; Meers, 2003). In any case, it would have originated on the coracoid (Jasinoski et al., 2006). It would insert on the deltopectoral crest of the humerus. Chiasson (1984) described the origin in pigeons on the furcula and part of the sternum above M. pectoralis. The muscle then inserts on the deltopectoral crest of the humerus in crocodylians and birds (Chiasson, 1962, 1984). As reconstructed by Jasinoski et al. (2006) and Burch (2014), M. supracoracoideus appears equivalent to M. supracoracoideus intermedius as described in crocodylians (Meers, 2003).

The origin of M. supracoracoideus is described as sometimes including both the lateral surfaces of the acromial process of the scapula and coracoid in archosaurs. Dilkes (2000) described these as distinct heads in crocodylians. Jasinoski et al. (2006) limited the origin of M. supracoracoideus to the coracoid in dromaeosaurs, but it probably extends into the lateral surface of the scapula in Deinonychus in a broad excavation (Ostrom, 1969). Nicholls and Russell (1985) proposed a bipartite M. supracoracoideus in Struthiomimus similar to Chamaeleo and Alligator. One head would originate anterior to the glenoid fossa in a depression, the other along the scapular blade. Dilkes (2000) and Burch (2014) proposed that the origin of this muscle may have extended onto the scapula at a subacromial depression in many dinosaurs, including Maiasaura and Tawa. Jasinoski et al. (2006) regarded the acromial process as the origin for M. deltoideus clavicularis, but Burch (2014) identified separate origins for M. supracoracoideus accessorius and M. supracoracoideus in this area, limiting the origin of M. deltoideus clavicularis to the dorsal margin of the acromion.

M. supracoracoideus extends through a triosseal canal to attach to the humerus in birds (Chiasson, 1984). Such an architecture would serve to elevate and rotate the humerus during flight (Poore et al., 1997). Poore et al. also described it as notably absent in Archaeopteryx and some other Mesozoic birds described. The canal has not historically been observed in other theropods, so this muscle would protract the humerus in them (Burch, 2014). In Nothronychus, most of the acromial process is not preserved. The origin for M. supracoracoideus for Nothronychus may be similar to that inferred for Deinonychus (Figure 3), following the reconstruction for Tawa (Burch, 2014).

The insertion of M. supracoracoideus is variable in archosaurs (Burch, 2014; Jasinoski et al., 2006). Jasinoski et al. and Burch described the insertion as the proximal deltopectoral crest in crocodilians, whereas in many birds, it inserts on the anterior tubercle between the deltopectoral crest and the head. Jasinoski et al. observed a shallow, rugose excavation in this region in Velociraptor that they ascribe to the insertion for this muscle. Burch (2014) regarded this attachment as unlikely. She noted that it passes through the triosseal canal in birds (Chiasson, 1984), a structure not heretofore considered present in other theropods, which she suggested would make it nearly non‐functional. Therefore, she argued for an insertion at the tip of the deltopectoral crest, including a small depression on the lateral side in Tawa. Burch (2017) reconstructed the insertion of M. supracoracoideus on rugosities on the margin of the deltopectoral crest in Majungasaurus, similar to Tawa. No such concavity is present in the deltopectoral crest of Nothronychus, but I tentatively follow the latter reconstruction (Figure 4). The reconstructed insertion in Nothronychus onto the deltopectoral crest also agrees with that proposed for Tyrannosaurus (Lipkin & Carpenter, 2008), but there is no identifiable scar associated with the attachment of M. supracoracoideus longus, so it is considered continuous with M. supracoracoideus. Assuming no triosseal canal, M. supracoracoideus would have protracted and contributed to abducting the humerus (Burch, 2014). However, if it ran through a canal, it would have elevated and rotated the humerus (Poore et al., 1997), convergent with ornithurans, as such a canal is absent in dromaeosaurs and Archaeopteryx.

2.2.12. M. supracoracoideus accessorius (SCA) (level II inference)

M. supracoracoideus accessorius was proposed as a new term for M. deltoideus minor in birds, as it arises embryologically from the supracoracoideus group rather than the deltoid (Burch, 2014). M. deltoideus minor is regarded as a neomorph in birds with either no homolog in crocodylians (Burch, 2014; Dilkes, 2000; Jasinoski et al., 2006) or synonymized with M. deltoideus clavicularis (Diogo & Abdala, 2010). M. deltoideus scapularis, presumed related in crocodylians, arises from the anterior procoracoid and inserts on the deltopectoral crest (Chiasson, 1962).

M. supracoracoideus accessorius is reconstructed as originating at the subacromial depression of the scapula (Burch, 2014) and inserting on the deltopectoral crest in theropods. In Majungasaurus, it inserts on the proximal margin of the deltopectoral crest (Burch, 2017). In Nothronychus, this muscle may have originated at the posterior acromial depression (Figure 3), but this reconstruction is tentative, as this region is not preserved. It probably would have inserted on the proximal end of the deltopectoral crest of the humerus (Figure 4). This muscle would have protracted and abducted the humerus with M. supracoracoideus if the latter muscle did not run through a triosseal canal (Burch, 2014).

2.2.13. M. coracobrachialis (CB) (level I inference)

This muscle is well‐developed in archosaurs (Chiasson, 1962, 1984; Meers, 2003). Jasinoski et al. (2006) describes two M. coracobrachialis heads, brevis and longus. M. coracobrachialis of many authors appears equivalent to M. coracobrachialis brevis ventralis of Meers (2003). M. coracobrachialis brevis originates at the lateral surface of the coracoid (Burch, 2014; Jasinoski et al., 2006) or on the posterolateral margin of the procoracoid (Chiasson, 1962) in crocodylians. M. coracobrachialis longus was usually absent in crocodylians (Jasinoski et al., 2006). It was, however, sometimes present in adult Alligator and Crocodylus (Nicholls & Russell, 1985), where it originates on the coracoid process and inserts on the dorsal epicondyle of the humerus.

Dilkes (2000) noted that two discrete heads of M. coracobrachialis may not have developed in dinosaurs and only the anterior head was present. Both heads would have originated on the lateral coracoid in Maiasaura at an extensive rugosity within the subglenoid fossa. Generally, Jasinoski et al. (2006) reconstructed the origin of M. coracobrachialis at the subglenoid fossa adjacent to the glenoid fossa in dromaeosaurids. This fossa is present in most theropods, including ornithomimids (Nicholls & Russell, 1985; Osmólska et al., 1972). Burch (2014) followed this interpretation in reconstructing the origin of M. coracobrachialis in Tawa. Brochu (2003) described two rugosities on the ventral margin of the coracoid of Tyrannosaurus referred to muscle attachment points. In that taxon, one is 3 cm from the rim, while another is immediately adjacent to the glenoid fossa within the subglenoid fossa. Although the first one is more prominent, Brochu did not ascribe it to a specific muscle. Jasinoski et al. (2006) described this area as the origin of M. coracobrachialis longus in dromaeosaurs. It is close to the origin postulated for this muscle in ornithomimids by Nicholls and Russell (1985). The second is considered either the origin for a head of M. triceps brachii or M. sternocoracoideus (Brochu, 2003). Jasinoski et al. (2006) proposed this area as the origin for M. coracobrachialis brevis occupying the subglenoid fossa posterior to a subglenoid ridge extending obliquely from the biceps tubercle in dromaeosaurs. Burch (2014) reconstructed the subglenoid fossa as constituting part of the origin for M. coracobrachialis in Tawa, such as observed in pigeons (Chiasson, 1984). Discrete M. coracobrachialis longus and brevis heads (anterior and posterior heads in pigeons, Chiasson, 1984) were not inferred for Tawa (Burch, 2014), as there are no osteological correlates and phylogenetic support is lacking in non‐avian theropods. M. coracobrachialis is reconstructed as inserting on the medial surface of the deltopectoral crest in Tawa (Burch, 2014). The inferred attachment points of M. coracobrachialis in Majungasaurus are similar (Burch, 2017).

Baier et al. (2006) consider the biceps tubercle (coracoid tubercle) of non‐avian theropods homologous with the enlarged acrocoracoid process in the lateral coracoid of birds. It would have enlarged to contribute to the triosseal canal, deflecting the supracoracoid tendon. This ligament elevates the humerus in birds during flight. They consider M. biceps brachii and M. coracobrachialis both originating, along with the acrocoracohumeral ligament, at the biceps tubercle in non‐avian theropods in their reconstruction.

The coracoid of Nothronychus possesses a broad shallow concavity anterior to the glenoid fossa that is identified as the subglenoid fossa. Both rugosities described for Tyrannosaurus are present in Nothronychus, but the subglenoid ridge of dromaeosaurs is faint. Alternatively, Dilkes (2000) regarded this point as the origin for M. sternocoracoideus or M. costosternocoracoideus, with M. coracobrachialis originating dorsal to the biceps tubercle in Maiasaura. This reconstruction appears to follow the crocodylian topology, where M. costosternocoracoideus may be related to M. costocoracoideus of Meers (2003). Meers (2003) described the insertion of M. costocoracoideus as ventral to the origin for M. coracobrachialis in crocodylians, within the subglenoid fossa.

The subglenoid fossa of Nothronychus is reconstructed as the origin for M. coracobrachialis brevis (Figure 3), as is conventional for theropods, although the crocodylian topology followed by Dilkes (2000) cannot be definitively excluded. The secondary tubercle described by Brochu (2003) is proposed as the origin for M. coracobrachialis longus, approximately following Jasinoski et al. (2006), but this inference is highly tentative. Romer (1944) considered the absence of a separate M. coracobrachialis longus a synapomorphy for archosaurs and this same conclusion was reached by Maidment and Barrett (2011) in their discussion of ornithischian dinosaurs. The insertion in both Tawa and dromaeosaurs is regarded as medial to the deltopectoral crest (Burch, 2014; Jasinoski et al., 2006) at a site corresponding to the impressio coracobrachialis in the humerus of birds (Baumel & Witmer, 1993). The insertion is similarly interpreted as the medial surface of the humerus adjacent to the deltopectoral crest in Nothronychus (Figure 4). The muscle would have protracted the humerus (Burch, 2014).

2.2.14. M. scapulohumeralis anterior (SHA) (level I inference)

M. scapulohumeralis anterior is lost in crocodylians and ratites (Romer, 1922), but is present in some other extant birds and lizards (Burch, 2014; Dilkes, 2000; Jasinoski et al., 2006). It originates dorsal to the glenoid fossa and inserts distal to the fossa pneumotricipitalis in the humerus. M. scapulohumeralis inserts on the posterior humerus at the fossa pneumotricipitalis, close to M. subscapularis in birds.

Both Burch (2014) and Jasinoski et al. (2006) suggested that part of M. scapulohumeralis anterior originated at an oval knob in the posterior margin of the ventral scapula in dromaeosaurs and Tawa, as in Cygnus, chameleons, and iguanas. Insertion points for M. pectoralis and M. supracoracoideus were probably distinct in Nothronychus (Figure 4). In these cases, Jasinoski et al. (2006) and Burch (2014) will be followed, as they corroborate with extant archosaurs (Chiasson, 1962; Chiasson, 1984; Meers, 2003). Burch (2014) described a tendinous insertion of M. scapulohumeralis anterior in the posterior medial tuberosity, but stated that there is no specific correlate in the humerus of non‐avian theropods. She modeled it as inserting adjacent to a ridge in the posterior face of the proximal humerus in Tawa distal and lateral to the M. scapulohumeralis posterior insertion, but refrained from reconstructing it in Majungasaurus (Burch, 2017), due to the lack of scarring. Dilkes (2000) modeled this muscle as originating lateral side of the scapula at the acromion process and inserting in the posterior humerus with M. deltoideus clavicularis in Maiasaura.

The origin of M. scapulohumeralis anterior in Nothronychus may be associated with a tubercle medial to the origin for M. triceps brachii scapulae (Figure 3). There is a very shallow excavation in the posterior humerus in Nothronychus referred to the fossa pneumotricipitalis (Baumel & Witmer, 1993; Jasinoski et al., 2006) that is regarded as the insertion (Figure 4). Such a fossa, however, is unknown in other non‐avian theropods (Burch, 2014) and would have been convergent with ornithurans. There is no osteological correlate for a long‐fibered subdivision of this muscle in Nothronychus, and it is also absent in other theropods. This scar is absent in Tawa (Burch, 2014), although she described a light depression in this area of Herrerasaurus and Sanjuansaurus. This muscle would have retracted the humerus (Burch, 2014).

2.2.15. M. scapulohumeralis posterior (SHP) (level I inference)

M. scapulohumeralis posterior is well‐developed in both crocodylians and most extant birds (Burch, 2014; Jasinoski et al., 2006). Jasinoski et al. reports that it is associated with M. subscapularis. Notably, Brochu (2003) described all archosaurian scapulae as rugose at the margins, presumably serving as major muscle attachment points. Chiasson (1962) referred to M. scapulohumeralis posterior as arising from the posterior margin of the distal third of the scapula. In pigeons, M. scapulohumeralis posterior originates at the lateral surface of the scapula (Chiasson, 1984; Vanden Berge & Zweers, 1993), but this attachment may increase in other birds (Jasinoski et al., 2006). This muscle inserts distal to the posterior tuberosity of the humerus in Caiman (Jasinoski et al., 2006; Meers, 2003). In most extant birds, it inserts in the cruz ventrale fossa below the fossa pneumotricipitales (Vanden Berge & Zweers, 1993).

In Maiasaura, M. scapulohumeralis posterior was interpreted as originating at a flattened area with abundant striations, close to the posteroventral corner of the scapula, distal to the glenoid fossa (Dilkes, 2000). Jasinoski et al. (2006) reconstructed it with a restricted origin at the ventral posterior blade in dromaeosaurs due to the attachment of M. deltoideus scapularis. They noted that its posterior extent may have been constrained by M. deltoideus scapularis and that the insertion in dromaeosaurs was probably distal to the posterior tuberosity, as in birds. Burch (2014) described two parallel ridges medially and laterally forming a long ventral fossa or groove in the proximal scapula of Tawa, as in many theropods. She reconstructed M. scapulohumeralis posterior in Tawa, with a small origin near the medial ridge on the ventral scapula and the insertion distal to the internal tuberosity. The muscle inserted on the proximal posterior face of the humerus. Burch (2017) reconstructed the origin on a crest on the ventral margin of the scapula adjacent to the glenoid. It would have inserted on a rugosity in the internal tuberosity of the humerus.

Nothronychus probably possessed an intermediate‐oriented scapula marked by a ventral groove with lateral and medial ridges on either side (Hedrick et al., 2015). This groove extends anteriorly to a shallow excavation posterior to the glenoid fossa. For Nothronychus, the lateral ridge is interpreted as a broad origin for M. scapulohumeralis posterior (Figure 3), more similar to the expansive origin reconstructed for dromaeosaurs (Jasinoski et al., 2006) and Therizinosaurus (Barsbold, 1976) than the restricted attachment of Tawa (Burch, 2014). The origin probably extended into the ventral groove. This ventral groove narrows and shallows distally. The distal extension is interpreted as terminating with the ventral groove. However, Jasinoski et al. (2006) argued that the posterior extent of the origin of M. scapulohumeralis posterior would have been limited by M. deltoscapularis in dromaeosaurs. This configuration would make it more similar to most birds than crocodylians (Burch, 2014). The resulting reconstruction is somewhat more restricted than that of Brochu (2003), who suggested the tentative origin of M. scapulohumeralis posterior as occupying much of the lateral scapular blade, as in modern birds (Chiasson, 1984), but ultimately decided the attachment as ambiguous.

Both Jasinoski et al. (2006) and Burch (2014) reconstructed the insertion of M. scapulohumeralis posterior on the posterior face of the internal tuberosity in Tawa and dromaeosaurs, very similar to the reconstruction for Deinonychus (Ostrom, 1969), although he proposed no separation between M. scapulohumeralis posterior and M. scapulohumeralis anterior in the latter genus. Ostrom described the medial tuberosity (internal tuberosity) as more expanded in Deinonychus than other theropods and this generalization holds for Nothronychus. This topology is reduced from the broad insertion described for crocodilians (Meers, 2003). In Nothronychus, this region is shallowly excavated and rugose. Lipkin and Carpenter (2008) reconstructed the insertion of M. scapulohumeralis posterior in Tyrannosaurus in the same general area, but far more extensive than that proposed for Nothronychus. Following Jasinoski et al. (2006) and Burch (2014), the distal portion at the base of a very short posterior tuberosity is interpreted as the origin of M. scapulohumeralis posterior (Figure 4). Jasinoski et al. (2006) described a similar region in Velociraptor and Troodon, presumably serving as a similar attachment point. M. scapulohumeralis posterior would have retracted the humerus (Burch, 2014).

2.2.16. M. deltoideus clavicularis (DC) (level I inference)

Jasinoski et al. (2006) noted that M. deltoideus clavicularis of crocodylians is equivalent to M. propatagialis in birds. In crocodylians and birds, the origin extends from anterior margin of the acromial process along the scapula (Burch, 2014; Jasinoski et al., 2006; Meers, 2003). Jasinoski et al. (2006) and Burch (2014) both described an additional origin as extending onto the epicleidium of the furcula in most extant birds. It is described as inserting on the lateral deltopectoral crest in crocodylians (Burch, 2014; Jasinoski et al., 2006). The M. deltoideus clavicularis (as M. propatagialis) insertion is highly modified in many birds to attach to the carpals (Burch, 2014; Chiasson, 1962) or the forelimb musculature (Jasinoski et al., 2006).

Jasinoski et al. (2006) reconstructed M. deltoideus clavicularis as originating from an excavation on the lateral side of the acromial process of the scapula and epicleidium in dromaeosaurs, whereas Burch (2014) limited the origin to the anterior margin of the acromial process, extending to the furcula, in Tawa. In Majungasaurus, the origin is reconstructed as posterolateral to the subacromial depression (Burch, 2017). The acromial process is reduced in Maiasaura, but M. deltoideus clavicularis is reconstructed as originating in the same general area (Dilkes, 2000). These reconstructions on the lateral side of the scapular portion of the acromial process (equals scapular prominence) agree with that inferred for Struthiomimus (Nicholls & Russell, 1985). For Tawa and the dromaeosaurs (Burch, 2014; Jasinoski et al., 2006), M. deltoideus clavicularis is reconstructed as inserting on the lateral surface of the deltopectoral crest.

The acromial process in Nothronychus is very thin and only partially preserved. I tentatively follow Meers (2003) and Burch (2014) in limiting the origin of M. deltoideus clavicularis to the dorsal margin of the scapula, including the posterior margin of the acromial process, extending onto the dorsal furcula, where a faint groove is present (Figure 3). This muscle would insert on the lateral deltopectoral crest of the humerus in Nothronychus (Figure 4), where Hedrick et al. (2015) described an enlarged rugose protrusion in this area. This scar is not present in other therizinosaurs, so they suggested a pathological partial insertion of M. deltoideus. Barsbold (1976) hypothesized an insertion for M. deltoideus on a rugose surface on the lateral extremity of the deltopectoral crest in Therizinosaurus. This point is interpreted here as a partial insertion for M. supracoracoideus, with the M. deltoideus clavicularis attaching more posteriorly. M. deltoideus clavicularis would have abducted and contributed to protraction of the humerus (Burch, 2014).

2.2.17. M. deltoideus scapularis (DS) (level I inference)

The origin of M. deltoideus scapularis is modified from crocodylians to birds (Burch, 2014). In crocodylians, this muscle is described as broadly originating on the anterior surface of the procoracoid (Chiasson, 1962). Jasinoski et al. (2006) observed an origin in a shallow depression in the scapular blade. It inserts distal to the anterior tuberosity of the humerus. Burch (2014) described the M. deltoideus scapularis origin as shifting from the lateral surface of the distal scapula in crocodylians to the acromion process in birds. She observes an insertion at a point distal to greater tubercle in crocodylians. In birds, it inserts on most of the lateral deltopectoral crest.

Both Jasinoski et al. (2006) and Burch (2014, 2017) preferred a crocodylian model for M deltoideus scapularis in dromaeosaurs, Tawa, and Majungasaurus. Burch (2014, 2017) reconstructed the origin of M. deltoideus scapularis on the lateral distal scapula in Tawa and Majungasaurus. In Tawa, the muscle would originate in the lateral side of the distal scapula. She used the insertion in crocodylians to infer a limited attachment in a striated, oval depression within the posterior proximal humerus. In Majungasaurus, the blade does not narrow proximally, and Burch used this character to model an anterior extension of the muscle. Nicholls and Russell (1985) describe the anterolateral scapula of Struthiomimus as lacking any scarring that could be attributed to M. deltoideus scapularis attachment, but proposed an origin at this point based on Chamaeleo and Alligator. Jasinoski et al. (2006) used a crocodylian model to reconstruct the M. deltoideus scapularis, so that it originated on the anterodorsal scapular blade and inserted in an elongate depression in the dorsal surface of the deltopectoral crest extending down the shaft shared by dromaeosaurs, troodontids, and oviraptorids. Dilkes (2000) reconstructed M. deltoideus scapularis as originating at the acromion or on the lateral scapular blade and inserting on the proximal humerus or on the shaft in Maiasaura.

In Nothronychus, M. deltoideus scapularis is reconstructed as originating from the distal lateral scapula (Figure 3). Like Majungasaurus, the scapula does not narrow proximally, suggesting that the muscle extends anteriorly (Burch, 2017). There is a shallow excavation in the head of the humerus in Nothronychus, similar to Tawa, that probably marks the proximal extent of this muscle (Figure 4). Notably, Jasinoski et al. (2006) used the inferred attachment points of M. deltoideus scapularis in dromaeosaurs to argue that it could not lift the humerus above the scapula in these animals, as Gishlick (2001) determined for Deinonychus. Nothronychus is regarded as similarly constrained. M. deltoideus scapularis would have abducted and retracted the humerus (Burch, 2014).

2.2.18. M. triceps brachii (TB) (level I inference)

The morphology of M. triceps brachii is variable in tetrapods (Burch, 2014; Chiasson, 1962, 1984; Dilkes, 2000; Dilkes et al., 2012; Meers, 2003). Meers (2003) described five heads in crocodilians. Chiasson (1984) observed three reasonably well‐developed heads in pigeons, whereas Burch (2014) noted two major heads in birds. In birds, M. triceps brachii can be minimally subdivided into scapular and medial heads (Burch, 2014; Chiasson, 1984). A discrete coracoid head is described for pigeons (Chiasson, 1984), but both Jasinoski et al. (2006) and Burch (2014) regarded it as vestigial in birds. Some heads are more strongly inferred than others in theropods. All heads merge to insert at the olecranon process of the ulna in extant crocodylians (Burch, 2014; Dilkes, 2000; Jasinoski et al., 2006; Meers, 2003) and birds (Burch, 2014; Dilkes, 2000; Jasinoski et al., 2006), including pigeons (Chiasson, 1984).

Burch (2014) described M. triceps brachii caput scapulare as present in all archosaurs. Jasinoski et al. (2006) found that M. triceps brachii caput scapulare is the largest head of the M. triceps brachii complex in crocodylians. Jasinoski et al. describe a tendinous origin at an oval rugosity that sometimes expands into a tubercle for the muscle at the posterior margin of the scapula posterodorsal to the glenoid fossa (Burch, 2014). The origin for M. triceps brachii caput scapulare in neognaths is similar to crocodylians, although there may be an additional attachment at a rugose surface on the dorsal side of the humerus in some birds. Nicholls and Russell (1985) reconstructed the origin of this head at a pronounced scar on the supraglenoid buttress, resulting in a morphology similar to Chamaeleo.

Dilkes (2000) supported an M. triceps brachii caput scapularis origin at the supraglenoid buttress in Maiasaura. This muscle is reconstructed similarly by Burch (2014) for Tawa, for Majungasaurus (Burch, 2017), and by Jasinoski et al., for dromaeosaurs. Burch (2014) reconstructed the origin in Tawa at a small striated area posterior to the glenoid fossa, rather than an enlarged infraglenoid tubercle. She disagrees with previous models (Burch and Carrano (2012) that a rugose tubercle close to the glenoid in Carnotaurus and Aucasaurus represents the attachment for this head as the attachment is reduced in archosaurs. Burch (2017) inferred this structure as the origin for M. scapulohumeralis posterior. She placed the origin of M. triceps brachii caput scapulare between this tubercle and the glenoid in Ceratosaurus and Majungasaurus at a small rugose surface. Dromaeosaurs possess a pronounced oval scar dorsal to the scapula (Jasinoski et al., 2006), presumably serving as the origin for M. triceps brachii caput scapularis. This component appears equivalent to M. triceps longus lateralis as described for crocodilians (Meers, 2003) and reconstructed for Tyrannosaurus (Brochu, 2003 (as M. triceps scapularis).

M. triceps brachii caput coracoideum is large in crocodylians (Jasinoski et al., 2006), but is reduced or lost in neognaths (Burch, 2014; Jasinoski et al., 2006). Nicholls and Russell (1985) regarded this head as probably absent in Struthiomimus, suggesting a resulting increased freedom of movement in the humerus. Jasinoski et al. describe a tendinous origin from the posterolateral margin of the scapula above the glenoid. This head may be a mechanoreceptor rather than possessing a locomotive function in birds (Burch, 2014; Jasinoski et al., 2006; Rosser & George, 1985; Vanden Berge & Zweers, 1993).

The medial head of M. triceps brachii (Burch, 2014) is equivalent to M. humerotriceps (Chiasson, 1984; Dilkes, 2000). In both crocodylians and birds, it originates on the posterior shaft of the humerus (Jasinoski et al., 2006). The muscle divides proximally into two heads on both sides of M. scapulohumeralis in crocodylians and most birds except ostriches (Burch, 2014; Jasinoski et al., 2006). Jasinoski et al. describe the origin of one head at the fossa pneumotricipitales in the proximal humerus of neognaths.

There are no muscle scars associated with the origin of M. triceps brachii caput mediale in any known theropod (Burch, 2014), but it was probably present as it is observed in crocodylians and birds (Jasinoski et al., 2006). Its extent could only be estimated for Tawa (Burch, 2014). She reconstructed the origin as covering almost all of the diaphysis. Burch (2017) reconstructed the origin in most of the posteromedial humerus in Majungasaurus. A similar attachment was proposed for dromaeosaurs (Jasinoski et al., 2006). Jasinoski et al. inferred a bifurcated proximal end at the origin for these theropods.

M. triceps brachii caput laterale originates along the posterior border of the attachments of M. deltoideus clavicularis and M. humeroradialis at the lateral triceps ridge on the anterodorsal face of the humerus in crocodylians (Jasinoski et al., 2006; Meers, 2003). This head is absent in birds (Jasinoski et al., 2006).

Jasinoski et al. (2006) note that the triceps ridge is not present in dromaeosaurs. Therefore, they regarded the muscle as absent, similar to birds. It is reconstructed, however, as originating from an obvious triceps ridge on the humerus in Tawa (Burch, 2014). She noted that this ridge is widespread in theropods, but Jasinoski et al. (2006) did not reconstruct the muscle in dromaeosaurs, suggesting that the muscle was absent as the ridge is not clear. The M. triceps brachii caput laterale origin was inferred at a triceps ridge in Majungasaurus (Burch, 2017).

M. triceps brachii caput scapulare probably had a tendinous origin from a rugose, triangular excavation at the infraglenoid tubercle previously described (Hedrick et al., 2015) in Nothronychus (Figure 3). This attachment point is reduced in N. graffami relative to N. mckinleyi, suggesting considerable variability at this point. Nothronychus lacks a well‐developed anterior flange or supraglenoid buttress observed in ornithomimids (Nicholls & Russell, 1985) that would have limited dorsal movement of the humerus, presumably increasing the mobility of that upper arm in the therizinosaur. The excavation appears homologous with the smaller one observed for the scapula of the ornithomimids Struthiomimus (Nicholls & Russell, 1985) and Gallimimus (Osmólska et al., 1972) that was interpreted as the origin of the scapular head of the triceps. This architecture is seen in extant crocodylians (Burch, 2014) and birds, including pigeons (Chiasson, 1984), and follows reconstructions of Maiasaura (Dilkes, 2000), dromaeosaurs (Jasinoski et al., 2006), and Struthiomimus (Nicholls & Russell, 1985). The origin of M. triceps brachii caput mediale lacks any associated scarring, so its extent can only be estimated for Nothronychus. There is a low lateral triceps ridge in Nothronychus, so I follow the interpretation of Burch (2014) in proposing that M. triceps brachii caput laterale originated here (Figure 4).

In Tawa (Burch, 2014), the insertion is marked by a faintly striated region on the olecranon process, while in Velociraptor, the dorsal olecranon process is reported as quite rugose (Jasinoski et al., 2006; Norell & Makovicky, 1999). Nothronychus possesses a well‐developed olecranon process, as in other therizinosaurs (Hedrick et al., 2015), on the proximal ulna with a pronounced rugose tubercle on the posterior side serving as the insertion for the combined insertion for M. triceps brachii (Figure 5). Brochu (2003) described a distinct muscle, M. humeroulnophalangei, originating at this point in crocodilians and Tyrannosaurus, but this architecture would apparently interfere with the more commonly described M. triceps brachii insertion. M. triceps brachii would extend the antebrachium and contribute to extending the humerus (Burch, 2014).

2.2.19. M. biceps brachii (BB) (level I inference)

M. biceps brachii is present in crocodilians and birds (Chiasson, 1962, 1984), but its configuration and attachments have been described as variable and sometimes debated (Burch, 2014; Jasinoski et al., 2006). Meers (2003) and Jasinoski et al. (2006) described the origin of M. biceps brachii at the anterior margin of the coracoid in crocodilians, close to the attachment for M. coracobrachialis. Jasinoski et al. described M. biceps brachii arising from a facet or tubercle on the coracoid in all archosaurs. The tubercle is ventral to the glenoid in crocodylians and ratites, whereas it is dorsal to the glenoid in neognaths. They note a secondary origin in most neognaths marked by rugosities in the posteroventral side of the proximal humerus. Non‐avian reptiles possess two heads attaching to the coracoid (Burch, 2014). In most birds, except ostriches, it arises from the acrocoracoid process at a small depression (Jasinoski et al., 2006). Jasinoski et al. describe it meeting a second head in most birds originating as an aponeurosis from the proximal ventral side of the humerus from the facies bicipitales. Burch described the presence of a primary head in all studied archosaurs, but noted the presence of a secondary head attaching to the humerus in lepidosaurs and neognathous birds. In non‐avian reptiles and birds, the coracoid attachment is on a tubercle anterior to the glenoid (Burch, 2014). Meers (2003) described the insertion on the proximal radius following Chiasson (1984) and Baumel and Witmer (1993) at the radial tuberosity (equals humeroradialis tubercle in crocodylians and tuberculum bicipitale radii in birds). Burch (2014) notes a single radial insertion is typical in crocodylians. In some crocodylians, an additional insertion is present in the proximal ulna (Burch, 2014; Jasinoski et al., 2006; Reese, 1915). Baumel and Witmer (1993) noted that the radial tuberosity only receives M. biceps brachii in birds. Jasinoski et al. (2006) and Burch (2014) described a double insertion in the proximal ulna and radius in most birds.

The primary origin of M. biceps brachii is classically interpreted as the biceps (=coracoid) tubercle of the coracoid in archosaurs, but this interpretation has been challenged. In his description of Sphenosuchus, Walker (1977, 1990) argued that the biceps tubercle as traditionally identified in theropods, such as Deinonychus (Ostrom, 1974), was too posteriorly placed to be the origin for M. biceps brachii and M. coracobrachialis actually arises from this point. The resulting reconstruction of the origin of M. coracobrachialis was also suggested for Therizinosaurus (Barsbold, 1976). This interpretation was followed by Makovicky and Sues (1998) in their description of the coracoid in Microvenator and Norell and Makovicky (1999) for Velociraptor. In both descriptions, the classical biceps tubercle was renamed the coracoid tubercle as used by Osmólska et al. (1972) for Gallimimus. Burch (2014, 2017) used the classical interpretation of M. biceps brachii primarily originating at the biceps tubercle in Tawa and Majungasaurus. She noted that this tubercle is indistinct in some basal theropods (e.g., Coelophysis). Jasinoski et al. (2006) used a similar model for dromaeosaurs. A discrete humeral secondary attachment adjacent to the internal tuberosity of the humerus for M. biceps brachii is reconstructed for non‐avian theropods, including Tawa (Burch, 2014) and dromaeosaurs (Jasinoski et al., 2006), at a shallow excavation adjacent to the internal tuberosity. Burch (2017) reconstructed the insertion of M. biceps brachii at a distinct excavation distal to the internal tuberosity in Majungasaurus.

Dilkes (2000) reconstructed M. biceps brachii as originating at the biceps tubercle and inserting on the radius and ulna in Maiasaura. Norell and Makovicky (1999) described a tubercle on the ulna of at least one specimen of Velociraptor that Jasinoski et al. (2006) considered a probable insertion for M. biceps brachii. Burch (2014) described radial and ulnar proximal insertions in lepidosaurs, but no scar is typically reported for either point. She reported the presence of a bulge on the ulna of Tawa that she suggested may correspond to an ulnar insertion in addition to the radial insertion. In Majungasaurus, the ulnar tubercle is described as larger than the radial one, so Burch (2017) suggested a primary insertion on the ulna. Nicholls and Russell (1985) followed the classical origination of M biceps brachii on the biceps tubercle in Struthiomimus, but rejected a secondary attachment on the humerus, citing a lack of evidence.

Following Dilkes (2000), Jasinoski et al. (2006), and Burch (2014), the classical identification and interpretation of the biceps tubercle on the coracoid as a tendinous primary origin for M. biceps brachii is followed here for Nothronychus (Figure 3). The biceps (coracoid) tubercle is broad and somewhat low. It is close to a large, eroded foramen that is in the right location for the coracoid foramen. A pronounced excavation distal to the medial tuberosity is interpreted as a secondary origin for M. biceps brachii. Ulnar and radial insertions are reconstructed for Nothronychus (Figure 5), as proposed for Tawa (Burch, 2014). No proximal scar is present in the ulna of Nothronychus, but a proximal tubercle in the radius of N. graffami is proposed as representing a partial, combined, primary insertion of M. biceps brachii and M. brachialis (Hedrick et al., 2015). M. biceps brachii would have flexed the antebrachium (Burch, 2014).

2.2.20. M. humeroradialis (HR) (level II inference)

M. humeroradialis is present in many diapsids (Romer, 1944). Any homolog, however, is uncertain in birds (Burch, 2014; Jasinoski et al., 2006). Sometimes, it has been homologized with M. propatagialis (Meers, 2003), but Jasinoski et al. and Burch question this synonymy. Burch considers M. propatagialis more likely a homolog of M. deltoideus clavicularis using embryological data. Chiasson (1962), Jasinoski et al. (2006), and Meers (2003) described it as originating posterodistal to the deltopectoral crest in crocodylians, with Meers identifying a pronounced rugose scar. M. humeroradialis is described as inserting on a humeroradialis tuberosity in the proximal anterior radius (Chiasson, 1962; Jasinoski et al., 2006; Meers, 2003). Notably, none of these authors firmly identified an avian homolog.

Jasinoski et al. (2006) considered M. humeroradialis equivocal in dromaeosaurs. They described a low rugose tuberosity in the ulna of some maniraptorans, including Ingenia, Saurornitholestes, and Velociraptor, distal to the deltopectoral crest and the M. deltoideus scapularis insertion that they considered a probable origin. It is in the same general area as a potential scar in some crocodylians. Burch (2014) noted that this structure is uncommon in basal theropods. Ostrom (1969) suggested that a crest in this region might serve as an attachment for M. humeroradialis, M. brachialis, or another muscle in Deinonychus. Brochu (2003) was doubtful that it could be ascribed to M. brachialis. Jasinoski et al. (2006) did not identify a discrete tubercle in dromaeosaurs that could serve as the insertion, but Burch (2014) described a small humeroradial tubercle distal to the deltopectoral crest in some theropods, including Herrerasaurus, so reconstructed it in Tawa. She followed descriptions of this muscle in crocodylians by Chiasson (1962) and Meers (2003) in suggesting that M. humeroradialis passed through a tendinous loop in theropods. Ostrom (1969) tentatively suggested an insertion on a small ridge in the proximal ulna for Deinonychus.

The corresponding region distal to the deltopectoral crest of Nothronychus is marked by a low tubercle and this site is reconstructed as the origin for M. humeroradialis (Figure 4). The reconstruction differs from the hypothesis of Barsbold (1976) for Therizinosaurus, in that the scar parallel to the base of the deltopectoral crest is here interpreted as the insertion for M. latissimus dorsi in Nothronychus. There is a well‐developed humeroradial tubercle distal to that interpreted as the insertion for M. biceps brachii and M. brachialis that probably also served as the insertion for M. humeroradialis (Figure 5). M. humeroradialis would have flexed the antebrachium (Burch, 2014).

2.2.21. M. brachialis (BR) (level I inference)

M. brachialis varies in crocodilians and birds. Meers (2003) considered it originating on the distal margin of the deltopectoral crest and inserting in the radius with M. biceps brachii and M. humeroradialis in crocodylians. Jasinoski et al. (2006) described the origin distal to the deltopectoral crest apex. The insertion of M. brachialis is close to that of M. biceps brachii on the proximal end of the ulna and radius of crocodilians (Burch, 2014). Jasinoski et al. (2006), however, limit the insertion to the proximal radius. Burch (2014) described the origin of M. brachialis at an excavation proximal to the distal condyles in birds, the fossa musculus brachialis presumably increasing speed of contraction, but reducing the moment arm. It is restricted to the proximal ulna in most birds (Burch, 2014; Chiasson, 1962) at a distinct excavation that Baumel and Witmer (1993) refer to the impressio brachialis.

As there is no evidence of the distal anterior intercondylar depression in Tawa, or any other theropod Burch (2014) reconstructed M. brachialis using the crocodylian model. The lack of a well‐developed fossa brachialis suggests a similar topology in Coelophysis (Colbert, 1989). Ostrom (1969) supported an origin along the distal deltopectoral crest for Deinonychus. It would have originated midway along the diaphysis in their model. There is a well‐developed anterior intercondylar depression in the distal humerus common in more derived non‐avian theropods (Burch, 2017), including Majungasaurus, Tyrannosaurus Brochu (2003) and Deinonychus (Ostrom, 1969). This depression extends proximally from the coronoid fossa that permits acute flexion of the radius in most derived maniraptorans, except alvarezsaurs (Senter, 2005).

An avian model with M. brachialis originating at the fossa m. brachialis in the distal humerus is accepted for Nothronychus (Figure 4) following the interpretation of Burch (2017) for derived theropods, including birds. Although such an excavation is present in the distal humerus of Nothronychus, based on the development of the distal condyles and enlarged coronoid process, extreme flexion at the elbow was not possible. There is a shallow excavation distal to the articular surface of the ulna that is interpreted as the impressio brachialis, receiving M. brachialis, as in birds (Baumel & Witmer, 1993). If an avian model is used, an additional insertion sharing the radial tuberosity with M. biceps brachii on the radius of N. graffami was probably absent (Figure 5). Hedrick et al. (2015) probably used a crocodylian model in proposing a secondary insertion for M. brachialis onto the radius. M. brachialis would have flexed the antebrachium (Burch, 2014).

2.2.22. M. anconeus (AN) (level I′ inference)

A suite of muscles with very small originations on the lateral epicondyle (=ectepicondyle) of the humerus were reconstructed for Tawa (Burch, 2014, 2017), but lack discrete identifiable scars in both Tawa and Majungasaurus. Jasinoski et al. (2006) did not reconstruct this region. Nothronychus is reconstructed similarly. One such muscle is M. anconeus. It is closely related to M. extensor carpi ulnaris in most diapsids, sometimes sharing a tendon with the latter muscle and completely fused in some taxa (Burch, 2014). It is sometimes referred to M. flexor ulnaris in crocodylians (Meers, 2003) and M. ectepicondylo‐ulnaris in birds (Chiasson, 1984). Meers described it as originating on the lateral epicondyle in crocodylians, but Chiasson (1962) described three heads originating from the lateral epicondyle, the scapula, and from the procoracoid. In any case, it inserts on the lateral or posterior ulna in crocodylians and birds (Burch, 2014; Chiasson, 1962, 1984; Meers, 2003).

Burch (2014) tentatively reconstructed M. anconeus as unfused with M. extensor carpi ulnaris in Tawa. It would have originated on the distal lateral epicondyle as in birds and inserted on a well‐developed crest along the length of the lateral ulna. In Majungasaurus, it lacks a specific osteological correlate (Burch, 2017). M. anconeus/M. extensor carpi ulnaris is reconstructed as inserting along this crest, so is similar to that reconstructed for Tawa. In Nothronychus, M. anconeus/M. extensor carpi ulnaris is reconstructed as originating at about the same point on the lateral epicondyle (Figure 4), but there is no specific osteological correlate. The ridge described for Tawa is also expressed in Nothronychus as a low, rugose surface that extends most of the length of the shaft of the ulna. M. anconeus would have flexed the antebrachium (Burch, 2014).

2.2.23. M. extensor carpi ulnaris (ECU) (level III inference)

Homologies of M. extensor carpi ulnaris are unclear in archosaurs as reviewed by Burch (2014). It is spatially close to M. anconeus and the two sometimes share an origin in some diapsids (Burch, 2014). She considered this muscle absent in crocodylians, but tentatively suggested that it may be homologous with M. extensor digitorum longus (communis) in birds, following some authors (e.g., Haines, 1939) and Chiasson (1962). This muscle would insert on the proximal metacarpals in most tetrapods (Burch, 2014). She described it as originating on the lateral epicondyle of the humerus and inserting on the base of metacarpal II or III in extant birds. There may be a secondary origin from the proximal ulna in some birds (Burch, 2014; George & Berger, 1966).

Burch (2014) described M. extensor carpi ulnaris as inserting variably on the carpus of tetrapods. Carpal evolution in tetrapods complicates the problem. She noted that the muscle may insert on the pisiform and lateral metacarpal in most lizards, but in Varanus, it also inserts on the ulnare, following Haines (1939). Additionally, pisiform evolution is complex in theropods (Botelho et al., 2014). The element so identified is located distal to the end of the ulna in ornithomimids (Nicholls & Russell, 1985). It is often synonymized with the ulnare in the paleontological literature, as discussed by Chure (2001). He described this carpal as absent in the wrist of Allosaurus. Botelho et al. (2014) argued for a reduction in the pisiform in theropods followed by re‐acquisition using both paleontological and developmental data. They noted that this carpal, however, retains its muscular attachment. Burch (2014) stated that the pisiform is not present in birds and followed Kundrát (2008) in that the ulnare of birds is not homologous with other tetrapods.

In Tawa, M. extensor carpi ulnaris is reconstructed as originating on the distal lateral epicondyle of the humerus and extending to insert on the carpal classically identified as the pisiform and the base of metacarpal III (Burch, 2014). In Majungasaurus, however, this attachment is limited to metacarpal III with the loss of the pisiform, as in more derived theropods (Burch, 2017). However, this carpal could be the same as the ulnare in these animals (Kundrát (2008). Therefore, M. extensor carpi ulnaris is reconstructed as having originated on the distal lateral epicondyle of the humerus in Nothronychus (Figure 4), following Burch (2014) for Tawa. Assuming the loss of the pisiform, M. extensor carpi ulnaris is reconstructed as inserting on the ulnare and/or the base of the lateral metacarpal III (Burch, 2014, 2017) (Figure 6). M. extensor carpi ulnaris would have extended and abducted the wrist (Burch, 2014).

2.2.24. M. supinator (SU) (level II inference)

In crocodylians, M. supinator originates at the lateral epicondyle of the humerus and inserts along the anterior or anterolateral shaft of the radius (Burch, 2014; Meers, 2003). The muscle originates on the lateral epicondyle of pigeons and other birds (Burch, 2014; Chiasson, 1984). It inserts along much of the anterolateral length of the proximal shaft of the radius in pigeons and most other birds (Burch, 2014; Chiasson, 1984).

Burch (2014) reconstructed the origin of M. supinator as proximal on the epicondyle in basal theropods, including Tawa, as in birds. It would have inserted along much of the length of the radius in Tawa at a flattened anterior surface set off by low ridges. The muscle is reconstructed as inserting on a large anterior rugosity in Majungasaurus extending the length of the radius (Burch, 2017). I follow the proposed origin of M. supinator on the lateral epicondyle for Nothronychus (Figure 4). As the rugosity seen in Majungasaurus is absent, the muscle is reconstructed in Nothronychus similar to Tawa within a shallow anterior excavation with two parallel ridges on the shaft of the radius, the medial one of which serves as the insertion for M. supinator (Figure 5). M. supinator would have supinated and flexed the antebrachium (Burch, 2014). Senter and Robins (2005) and Senter (2006b) considered such rotation absent in non‐avian theropods as proximal morphologies of the radius and ulna would have prevented such motion, but the inferred presence of this muscle suggests some such movement was present in Nothronychus as Burch argued.

2.2.25. M. extensor carpi radialis (ECR) (level II′ inference)

M. extensor carpi radialis is present in all diapsids (Burch, 2014). It appears equivalent to M. extensor carpi radialis longus distinct from two heads of M. extensor carpi radialis brevis of Meers (2003). In crocodylians, it originates at the lateral epicondyle, between the origins of M. supinator and M. extensor digitorum longus (Burch, 2014). It inserts on the dorsal radiale (Chiasson, 1962; Meers, 2003). In birds, the wrist is highly modified and the origin of M. extensor carpi radialis attaches to the carpometacarpus near the base of metacarpal I (Burch, 2014).

An additional muscle described for crocodylians, M. extensor carpi radialis brevis, pars ulnaris (Meers, 2003) appears equivalent to M. ulnocarpiradialis reconstructed in Tyrannosaurus (Brochu, 2003). It originates along the length of the medial shaft of the ulna, but the reconstruction in Tyrannosaurus limits the attachment to a proximal tuberosity. There is no such tuberosity in the ulna of Nothronychus and it is not reconstructed in Tawa (Burch, 2014).

In Maiasaura, Dilkes (2000) suggested that M. extensor carpi radialis may have inserted on metacarpal III. Burch (2014) used a crocodylian model to reconstruct it in Tawa, originating on the lateral epicondyle and inserting on the radiale. Burch (2017) noted that M. extensor carpi radialis typically inserts on the radiale in basal theropods, including Tawa. In Majungasaurus, the muscle is modeled as attaching at a combined insertion with M. abductor radialis (Burch, 2017), as the radiale is absent in ceratosaurs. The radiale is unknown in Nothronychus, but it is described as present in the somewhat basal therizinosaurian Alxasaurus (Russell & Dong, 1993), so it is hypothesized as present in the former genus. Therefore, attachments of M. extensor carpi radialis in Nothronychus (Figure 6) are reconstructed as in Tawa. M. extensor carpi radialis would have adducted and extend the wrist. It was also involved in flexing the antebrachium (Burch, 2014).

2.2.26. M. abductor radialis (AR) (level II inference)

M. abductor radialis of Meers (2003) originates close to M. extensor carpi radialis, so has undergone many name changes (Burch, 2014). It may be equivalent to M. extensor carpi radialis profundus (Burch, 2014; Chiasson, 1962), but this muscle has a different origin on the distal radial shaft. M. abductor radialis is a bipartite muscle in lepidosaurs, but only a single one in crocodilians (Burch, 2014). She hypothesized a general reduction of the muscle in archosaurs. It originates on the lateral epicondyle in crocodilians (Meers, 2003). It fuses distally with M. extensor carpi radialis in birds and this may have been the case in more basal theropods, as well (Burch, 2014).

Burch (2014) reconstructed M. abductor radialis as reduced, if not fused, with M. extensor carpi radialis in Tawa and other theropods, as in crocodylians. M. abductor radialis is reconstructed as originating on the lateral epicondyle and inserting on the proximal half of the shaft of the radius in Tawa (Burch, 2014). It is reconstructed as either fused with M. extensor carpi radialis or sharing a common insertion on the distal radius in Majungasaurus (Burch, 2017). The muscle in Nothronychus is modeled similar to Tawa, as the radiale is considered present, following the condition in Alxasaurus (Russell & Dong, 1993). Therefore, it would have originated on the lateral epicondyle (Figure 4) and inserted on the radiale. M. abductor radialis would have abducted and contributed to flexing the antebrachium (Burch, 2014).

2.2.27. M. abductor pollicis longus (APL) (level I inference)

M. abductor pollicis longus is present in many diapsids, but given many different names (Burch, 2014; Meers, 2003). Burch designated it with a common term. She described it originating on the facing radial and ulnar shafts in crocodylians and most birds except Passeriformes. The muscle originates solely from the ulnar shaft in lepidosaurs (Haines, 1939), but in extant archosaurs, except Passeriformes, it also arises from the radial shaft (Meers, 2003). In the latter clade, M. abductor pollicis longus arises from the radius. In crocodylians, it inserts on the radiale, but this condition is derived for them (Burch, 2014). In birds, she described it inserting on the base of metacarpal I, as in lizards.

Burch (2014) reconstructed M. abductor pollicis longus with the double origin and inserting on a flange on metacarpal I in Tawa. It is reconstructed as originating adjacent to the crest for the interosseous membrane with a groove for passage of the tendon to metacarpal I in Majungasaurus (Burch, 2017). Following Burch (2014, 2017), M. abductor pollicis longus is reconstructed with a double head originating along low crests in the shafts of the radius and ulna of Nothronychus (Figure 5). The groove described for Majungasaurus, however, is absent. The muscle is modeled as inserting on the first metacarpal in Nothronychus (Figure 6). M. abductor pollicis longus would have extended and abducted the wrist and abducted digit I (Burch, 2014).

2.2.28. M. extensor digitorum longus (EDL) (level I′ inference)

M. extensor digitorum longus is very conservative in tetrapods (Burch, 2014). It appears to be equivalent to M. humerodorsalis (Chiasson, 1962). Burch (2014) described this muscle as arising from approximately the middle of the lateral epicondyle in birds (Chiasson, 1984) and crocodylians (Chiasson, 1962; Meers, 2003). It inserts on the base of all of the metacarpals except the last in crocodylians (Burch, 2014; Chiasson, 1962). In pigeons and other birds, it inserts in the carpometacarpus (Chiasson, 1984). Burch refined this description to an insertion on phalanx 1 of digits I and II for birds.

Burch (2014) used a functional argument to model M. extensor digitorum longus originating on the lateral epicondyle and inserting on the bases of each metacarpal of the three functional digits in Tawa. Digit IV was highly reduced in Tawa, so she proposed that the corresponding M. extensor digitorum longus attachment was already lost. Burch (2017), however, reconstructed a metacarpal IV attachment in Majungasaurus, given the exaggeration of that digit in abelisaurs and apparent extreme hyperextension. Barsbold (1976) proposed a similar insertion for M. extensor digitorum communis in the manus of Therizinosaurus, which is probably synonymous with M. extensor digitorum longus.

M. extensor digitorum longus is reconstructed in Nothronychus originating on the lateral epicondyle (Figure 4). As in Tawa, Nothronychus possessed three functional digits. It appears that musculature in theropods was lost prior to complete digit loss. Therefore, following Burch (2014), I reconstruct M. extensor digitorum longus of Nothronychus as inserting on the bases of all three metacarpals as in crocodylians (Figure 6). M. extensor digitorum longus would have extended the wrist (Burch, 2014).

2.2.29. M. pronator teres (PT) (level I inference)

M. pronator teres is widespread in tetrapods (Burch, 2014). It originates on the medial epicondyle, sometimes with fibers arising from the proximal ulna (Meers, 2003). In any case, it inserts along the medial shaft of the radius, and to some extent, onto the interosseous membrane in crocodylians (Chiasson, 1962; Meers, 2003). Burch noted that it inserts along more than half the proximal of the radius in archosaurs, generally. Plesiomorphically in lizards and turtles, it inserts on less than half the shaft distally, while in more derived diapsids, including crocodilians and paleognathous birds, it has a longer insertion (Burch, 2014). M. pronator teres has a similar origin in neognathous birds (Burch, 2014). She described the origin as the most proximal of those arising from the medial epicondyle in birds. This attachment is close to that for M. flexor carpi ulnaris in crocodylians (Meers, 2003).

In Tawa, Burch (2014) reconstructed the origin for M. pronator teres on the ridge above the medial epicondyle. She modeled a long insertion greater than half the length of the radius for M. pronator teres at a surface marked by parallel ridges. In Majungasaurus, it is reconstructed as inserting at a proximal rugosity on the anterior surface (Burch, 2017). M. pronator teres is reconstructed in Nothronychus following that for Tawa (Burch, 2014). The ridge on the medial epicondyle is also present in the distal humerus of Nothronychus, so the origin of this muscle is similarly reconstructed (Figure 4). Nothronychus lacks the inserting rugosity of Majungasaurus (Burch, 2017), but the presence of a pronounced longitudinal anteromedial ridge in the radius of Nothronychus indicates that this muscle had a long insertion, similar to Tawa (Burch, 2014) (Figure 5). M. pronator teres would have flexed and pronated the antebrachium (Burch, 2014).

2.2.30. M. pronator accessorius (PA) (level II inference)

M. pronator accessorius is present in lizards and most birds, but absent in crocodylians and most paleognaths (Burch, 2014). She described the origin as distally on the medial epicondyle adjacent to the origin for M. flexor digitorum longus superficialis. Its insertion is somewhat variable in diapsids, but she described the insertion as less than half the length of the radius plesiomorphically in lizards and some neognaths, while extending to most of the length in most birds.

Burch (2014) reconstructed the origin of M. pronator quadratus on the medial epicondyle of the humerus. Scarring is not observed in the radius of Tawa, and Burch (2014) tentatively reconstructed the insertion as one‐half the length of the distal radius. The origin for M. pronator accessorius in Nothronychus in the distal medial epicondyle (Figure 5) follows that for Tawa (Burch, 2014). In contrast to Tawa, there is a long, low longitudinal crest in the correct area to serve as the insertion for this muscle in Nothronychus, as Burch described for birds (Figure 5). M. pronator accessorius would have flexed and pronated the antebrachium (Burch, 2014).

2.2.31. M. pronator quadratus (PQ) (level I inference)

Burch (2014) supported the homology of M. pronator quadratus of many diapsids with M. ulnometacarpalis ventralis in birds, as described by Sullivan (1962) for chickens and Chiasson (1984) for pigeons. Burch described it as originating along the ventromedial shaft of the ulna in most turtles, and all lizards and crocodiles on more than half the length of diaphysis of the ulna, following Meers (2003). In diapsids except birds, it inserts on the ulnar side of the radial shaft, sometimes extending to the ventral carpals in turtles and some lizards (Burch, 2014). In birds, the origin is restricted to less than half the distal length of the shaft (Burch, 2014; George & Berger, 1966). Chiasson (1984) described it (as M. ulnometacarpalis ventralis) as originating on the middle third of the ulnar shaft in the pigeon and attaching to the base of the carpometacarpus as in birds in general (Burch, 2014).

Burch (2014) reconstructed M. pronator quadratus with an elongate origin in basal theropods, including Tawa, as in crocodylians. She models Tawa with a crocodylian radial shaft insertion. The ulna of Nothronychus possesses a long low crest that may correspond to the origin of M. pronator quadratus (Figure 5). There is a ridge associated with the interosseous ligament that probably also served as the M. pronator quadratus insertion onto the radius (Figure 5). This topology would suggest that M. abductor pollicis longus was separated from M. pronator quadratus by the interosseous ligament. M. pronator quadratus would have pronated the antebrachium and manus (Burch, 2014).

2.2.32. M. epitrochleoanconeus (EA) (level II inference)

M. epitrochleoanconeus is present in many diapsids, including turtles, lizards, and some birds (Burch, 2014). Her outgroup analysis implied that it was secondarily lost in crocodylians. It originates on the distal medial epicondyle in all of those taxa where it is present, close to M. pronator accessorius and M. flexor carpi ulnaris. The muscle inserts on the ventromedial side of the ulna, extending about one‐third to one‐half way down the shaft in most lizards and birds where it is present.

Burch (2014) tentatively reconstructed the origin of M. epitrochleoanconeus between those of M. pronator accessorius and M. flexor carpi ulnaris in Tawa similar to lizards. She indicated an restricted origin at the proximal half of the ulna in basal theropods. In Majungasaurus, Burch (2017) reconstructed the insertion extending on a ridge along the ulnar shaft. The reconstruction of the attachments of M. epitrochleoanconeus in Nothronychus on a low medial crest follow that for Tawa (Figure 5). In Majungasaurus, the insertion is described as extending distally in genera with reduced limbs (Burch, 2017; Livezey, 1992), but Nothronychus was not so characterized and the crest is short, so the reconstructed insertion is restricted to the proximal end. M. epitrochleoanconeus would have flexed the antebrachium (Burch, 2014).

2.2.33. M. flexor carpi ulnaris (FCU) (level II inference)

M. flexor carpi ulnaris is widespread in diapsids (Burch, 2014). Chiasson (1962) synonymized it with M. humero‐radialis lateralis in crocodylians. The muscle has a tendinous origination at the medial epicondyle in crocodylians (Meers, 2003). Meers described it as inserting on the pisiform, but Chiasson (1962) regarded it as inserting on the ulnare. Burch (2014), however, described it as inserting on the pisiform, sometimes with an accessory attachment on the ulnare in non‐avian theropods. The wrist in birds is modified and M. flexor ulnaris divides into two heads in pigeons, referred to pars cranialis and pars caudalis (Chiasson, 1984). These originate at the medial epicondyle and the proximal ulna, respectively. The first inserts generally on the carpometacarpus and a second on the ulnare that attaches to the base of flight feathers (Burch, 2014; George & Berger, 1966). Burch (2014) described it as inserting on the ulnare in birds, but noted that this bone is a neomorph and not homologous with the ulnare in other diapsids (Kundrát, 2008), functionally replacing it.

Burch (2014) reconstructed M. flexor carpi ulnaris in Tawa as originating through a single tendon on the medial epicondyle. Both Tawa and Nothronychus lack papillae on the ulna that can be associated with secondary flight feathers with remiges (Baumel & Witmer, 1993), so both are interpreted as lacking them. Therefore, they are reconstructed as lacking the accessory attachment of the muscle. As Tawa possessed a plesiomorphic carpus and manus, Burch (2014) reconstructed M. flexor carpi ulnaris as inserting on the pisiform and possibly additionally on the original ulnare. Burch (2017) reconstructed M. flexor carpi ulnaris inserting on a pit in the distal ulna of Majungasaurus. Dilkes (2000) proposed an insertion on the flexor side of metacarpal IV in Maiasaura.

I follow the plesiomorphic architecture for Nothronychus. Therefore, M. flexor carpi ulnaris is reconstructed originating on the medial epicondyle (Figure 4) and may well have inserted on an ulnare in Nothronychus. The carpals, are not preserved in Nothronychus, but they are known for Alxasaurus (Russell & Dong, 1993), so were likely similar in Nothronychus. The insertion of M. flexor carpi ulnaris on the pisiform is interpreted as lost, along with that discrete ossification. There is an anterior pit in the distal ulna of Nothronychus, however, as in Majungasaurus, and it is quite possible that M. flexor carpi ulnaris inserted there, as well. M. flexor carpi ulnaris would have flexed and adducted the wrist (Burch, 2014).

2.2.34. M. flexor digitorum longus (FDL) (level I inference)

M. flexor digitorum longus is present in most diapsids (Burch, 2014). It is probably homologous with the crocodylian M. flexor digitorum communis, humeral and ulnar heads of Chiasson (1962) and M. flexor digitorum (longus) superficialis and M. flexor digitorum (longus) profundus in pigeons (Chiasson, 1984). These two components merge into a single set of tendons (Burch, 2014). Avian terminology will be used here, following Burch. In crocodylians, M. flexor digitorum longus superficial head has a tendinous origin on the medial epicondyle. Meers (2003) subdivided M. flexor digitorum longus into a complex of muscles inserting onto the first three digits. M. flexor digitorum longus profundus head originates along the distal ulnar shaft (Burch, 2014; Chiasson, 1962).

M. flexor digitorum longus superficialis was identified in all birds except ratites (Burch, 2014). It originates from the medial epicondyle similar to crocodylians (Burch, 2014; Chiasson, 1984). M. flexor digitorum profundus originates along the proximal third of the ulnar shaft in pigeons (Chiasson, 1984), whereas Burch (2014) noted that the origin of M. flexor digitorum (longus) profundus is reduced to the proximal end of the ventromedial side of shaft of the ulna in neognaths from the more elongate origin in more basal archosaurs. She suggests that this change may be related to the loss of discrete digits and associated musculature. Burch (2014) described fusion of the two sets of tendons in some, but not all, birds, further suggesting that it is a basal character based on character distribution.

Both M. flexor digitorum longus superficialis and profundus fuse at the palmar aponeurosis in pigeons (Chiasson, 1984) and all extant diapsids except some birds (Burch, 2014). From there, tendons extend to all the digits to insert on the ungual phalanges of lizards and turtles. She describes crocodylians as reducing digits IV and V, resulting in the loss of the respective tendons (Burch, 2014; Meers, 2003). Burch follows Bever et al. (2011) in noting that the inserting tendon only inserts onto the ungual of digit II in most birds. A retained insertion on digit I is present in some birds and may be functionally linked.

M. flexor digitorum longus is probably equivalent to M. palmar communis reconstructed by Dilkes (2000) for Maiasaura. Both insert through the palmar aponeurosis onto the manus in crocodilians and most birds. Burch (2014) reconstructed a tendinous origin for M. flexor digitorum longus superficialis in Tawa on the medial epicondyle adjacent to the origin for M. flexor carpi ulnaris. Burch (2017) reconstructed the origin of M. flexor digitorum longus profundus along the anterior shaft of the ulna with a distal notch permitting the tendon to pass to the manus in Majungasaurus. She further indicated that as Tawa possessed a four fingered manus, as in crocodylians, the attachment of this muscle is more similar to them than to birds. Therefore, Burch reconstructed the muscle in Tawa as inserting on all four digits, except, possibly for the reduced digit IV.

Nothronychus is reconstructed with M. flexor digitorum longus superficialis originating from the medial epicondyle (Figure 4), as in Tawa (Burch, 2014). M. flexor digitorum longus profundus is modeled as originating from an anterior groove in the medial shaft of the ulna as in crocodilians and Majungasaurus (Burch, 2014, 2017). The groove in Majungasaurus is reduced, but appears split into shallow lateral and medial sulci in Nothronychus. Therefore, two tendons arising from M. flexor digitorum longus profundus are proposed in Nothronychus. In spite of the close proximity of digits II and III, separate from digit I, Nothronychus possessed three well‐developed, functional digits so all three inserting tendons were probably retained (Figure 6). M. flexor digitorum longus would have flexed the digits and wrist (Burch, 2014).

2.2.35. M. extensores digitores breves (EDB) (level I inference)

M. extensores digitores brevis is widespread in diapsids (Burch, 2014). Burch distinguishes two layers, a superficial layer (M. Extensores Digitores Brevis Superficialis [EDBS]) and a deep layer (M. Extensores Digitores Brevis Profundus [EDBP]). She described the superficial layer as originating on the proximal carpals and the deep layer arising from the metacarpals in lizards and crocodylians. They fuse into a single tendon and insert on the bases of the ungual phalanges of the respective digit. Meers (2003) treated the superficial muscles servicing each digit separately. Here, M. extensor digiti I superficialis originates in part on the radiale and then on the proximal metacarpal I. The muscle then inserts on the first ungual. M. extensor digit II superficialis originates on the radiale, as well. It also inserts on the base of the ungual at the base of the extensor process. M. extensor digiti III superficialis also originates on the radiale and inserts on the ungual. M. extensor digit IV superficialis usually separates into two heads. One originates on the distal ulna and ulnare, while the other arises from the radiale. They fuse into a common tendon to insert on the ungual. M. extensor digiti V superficialis originates on the distal ulna and ulnare. It inserts through fascia onto the respective ungual.

Meers (2003) described M. extensores digitores breves profundus separately in crocodilians, similar to the superficial extensors. M. extensor digiti I profundus originates from the base of metacarpal I and adjacent carpal ligaments. It inserts on the base of the ungual, M. extensor digiti II profundus originates on metacarpals I and II. It inserts along the entire second digit from the first phalanx to the base of the ungual. M. extensor digiti III profundus has two and sometimes three heads that variably originate from the radiale, ulnare, radius, and metacarpal II. All heads then insert onto the extensor processes of digit III.

The manus and associated M. extensores digitores breves in birds is extremely modified (Burch, 2014). She described those muscles associated with the first two digits inserting on digits I and II as M. extensor brevis alulae and M. extensor longus digiti majorus. Burch proposes that M. ulnometacarpalis dorsalis is an extensor, as well. All of these muscles are described in pigeons (Chiasson, 1984). M. extensor brevis alulae divides into two heads and originates from the dorsal extensor process, carpometacarpus, and the M. ulnometacarpus ventralis tendon. M. extensor longus digiti majorus originates on the radius. M. ulnometacarpalis dorsalis arises from the distal ulna and inserts on the proximal carpometacarpus at metacarpal III (Burch, 2014; George & Berger, 1966).

Burch (2014) noted that the extensor musculature in the first two to three digits is quite conservative in archosaurs. Therefore, she reconstructed the first two superficial extensors as originating from the dorsal surface of the radiale and the third and the fourth from the dorsal surface of the ulnare in Tawa. She reconstructed the origin of the deep portion at the bases of the respective metacarpal, extending to the metacarpal medial to it, as in extant crocodylians and lizards, and inserting on the bases of the unguals of the given digit. Burch (2017) indicated that M. extensores digitores breves may not have been present in Majungasaurus, as they are typically lost in lizards with reduced limbs. If they were present, she reconstructed them originating on the dorsal surfaces of metacarpals I‐III and inserting on the respective distal phalanges.

Nothronychus lacks the fourth digit, so all three remaining superficial extensors are proposed as originating from the radiale and the deep extensors arising from proximal metacarpals I and II as in crocodylians. Articulation of metacarpal II with III excludes a discrete origin on metacarpal III. Therefore, metacarpal III did not extend independently of II. Metacarpal I and II tendons would both insert on the respective unguals. As Nothronychus lacks digits IV and V, the associated musculature for these digits is not considered. The base of each metacarpal possesses a broad, shallow excavation interpreted as the respective origin, not described for Tawa, but observed in Nothronychus. Burch (2014) described a striated facet in the unguals of Tawa that she interprets as the insertion for M. extensores digitores breves. While the facet is not observed in Nothronychus, the muscle terminal insertion is reconstructed similarly. Intermediate insertions are present on the bases of the phalanges in crocodilians (Meers, 2003) and additional insertions of M. extensor digitorum profundus here in Nothronychus are quite possible (Figure 6). The resulting reconstruction is similar to that proposed for Therizinosaurus (Barsbold, 1976). Mm. extensors digitores breves would have extended the digits (Burch, 2014).

2.2.36. M. flexores digitores breves (FDB) (level III inference)

Burch (2014) separated M. flexor digitores breves into superficial (FDBS) and deep (FDBP) layers comparing them to the extensors. In lizards, she described the muscle as originating at an annular ligament. In crocodylians, they originate from the distal carpals and insert on some of the phalanges (Meers, 2003). In both lizards and crocodylians, all digital bellies, except digit I (Burch, 2014), split distally to accommodate M. flexores digitores breves profundus tendons. A possible avian homolog for M. flexores digitores breves would be M. flexor alulae (Burch, 2014). This muscle originates along the carpometacarpus and the M. flexor digitorum longus tendon, very similar to M. flexor digitores longus superficialis in crocodylians (Burch, 2014). M. flexores digitores breves profundus originates at the distal carpals and metacarpals in crocodylians and lizards and inserts on the first phalanges (Burch, 2014). She stated that the muscle subdivides, extending separately to each digit in birds.

Burch (2014) argued that the wrist in Tawa was more similar to lepidosaurs than crocodylians or birds and possessed an annular ligament, possibly constituting the origin for M. flexores digitores breves, but she reconstructed the origin on the distal carpals and metacarpals. In Majungasaurus, the muscle is reconstructed originating at a proximal lip in each metacarpal (Burch, 2017). She reconstructed a single insertion on a ventral tubercle on digit I proximal phalanx. Split insertions, as in Tawa, on digits II and III would have been present on the proximal corners of the proximal digits. Distal tendons may subdivide, accommodating M. flexores digitores breves profundus (Meers, 2003) to insert either separately on the sides of the flexor tubercles in some phalanges in crocodilians and turtles or singly onto the phalanges as in some lizards. In Burch's reconstruction of Tawa, she proposed separate insertions lateral and medial to the flexor tubercles, except M. flexor digitorum superficialis digiti I, which she models as inserting simply on the ventromedial side of the flexor tubercle without subdividing, as in lizards and birds. M. flexores digitores breves would have originated at the distal carpals and metacarpals. It is reconstructed as inserting at the flexor tubercles.

M. flexores digitores breves superficialis and profundus are reconstructed similarly in Nothronychus, so M. flexores digitores breves superficialis would originate at the distal carpals and insert after bifurcating on the phalanges (Figure 6). An annular ligament is present at the wrist, as modeled for Tawa. M. flexores digitores breves profundus presumably originated at the distal carpals and metacarpals I‐III and inserted at the flexor tubercles. Therefore, metacarpal II and III flexed together. M. flexores digitores brevis would have flexed the digits (Burch, 2014).

2.2.37. M. abductor pollicis brevis (APB) (level I inference)

M. abductor pollicis brevis may be homologous with M. abductor alulae in birds (Burch, 2014). In diapsids, she described the origin of this muscle as variable, including the radiale in crocodilians, the radius in some turtles, the base of the carpometacarpus in some birds, or the tendon of M. extensor carpi radialis in most birds. Burch reconstructed the origin of this muscle as at the radiale in Tawa, as the wrist is more similar to the plesiomorphic condition. The insertion is located near the base of phalanx I of digit I in most diapsids, except in crocodylians, where it inserts more proximally on the first metacarpal. Burch preferred the more distal insertion for Tawa.

The attachment points for Nothronychus are reconstructed as in Tawa, with an origin at the radiale and an insertion on a lateral tubercle on the base of phalanx I (Figure 6). The proximal insertion in Nothronychus is closer to crocodylians than the more distal attachment described for lizards and birds (Burch, 2014; Meers, 2003). M. abductor pollicis brevis would have abducted digit I (Burch, 2014).

2.2.38. M. abductor digiti minimi (ADM) (level I inference)

Burch (2014) described M. abductor digiti minimi as originating on the pisiform in turtles, lizards, and crocodilians and inserting on the lateral metacarpal in crocodilians or on the first phalanx in turtles and lizards. The pisiform, however, is lost in most derived theropods other than birds, where it is re‐established. In birds, the origin is probably shifted to the pseudoulnare, but they retain an insertion on the lateral metacarpal. In Tawa, this metacarpal would have been IV, but in Nothronychus, the lateral‐most metacarpal would have been III. Burch (2014) reconstructed this muscle as originating on the pisiform in Tawa and inserting on the lateral metacarpal IV.

Nothronychus is reconstructed such that M. abductor digiti minimi originated on the ulnare or pseudoulnare, and inserted on a lateral tubercle in the base of the lateral metacarpal III (Figure 6). Metacarpal III is closely associated with II along its length, however, so this muscle must only have permitted abduction of the combined wrist in contrast to abduction of digit IV and manus in Tawa (Burch, 2014).