Nothronychus
publication ID |
https://doi.org/10.1093/zoolinnean/zlad148 |
DOI |
https://doi.org/10.5281/zenodo.14851292 |
persistent identifier |
https://treatment.plazi.org/id/AF3F816A-CD1D-D60C-8F98-FA9AB261FBA7 |
treatment provided by |
Plazi |
scientific name |
Nothronychus |
status |
|
OSTEOLOGICAL DESCRIPTION
Taxa included in this contribution are Nothronychus mckinleyi ( AzMNH P2117 ) Kirkland and Wolfe (2001) and N. graffami ( UMNH 16420 ) Zanno et al. (2009). Much of the skeleton is convergent with Ornithischia, titanosaurs ( Wilson and Upchurch 1993), and many modern birds. Character polarity inferences are based on Zanno et al. (2009).
Nothronychus graffami is represented by a nearly complete skeleton, including girdles and limbs on one side or the other ( Figs 1 View Figure 1 , 3–4 View Figure 3 View Figure 4 ). In some cases, elements were mirrored so they are all the same side. Some elements are fused, but most are not. Fusion has been used as an index for maturity in some taxa, such as ceratosaurs, but this condition is only observed in ratites in extant birds ( Rowe and Gauthier 1990, Brochu 2003). For example, the left scapula and coracoid are fused in N. graffami ( Zanno et al. 2009) but the right is not in N. mckinleyi ( Hedrick et al. 2015) , suggesting some specific, individual, or side-to-side variation. On this basis, the specimen of N. graffami might be considered more mature than N. mckinleyi , but this idea is questionable. On the other hand, this trait is sometimes considered derived. In their description of Beipiaosaurus, Liao et al. (2021) note that fusion of the scapulocoracoid does not occur in more basal therizinosaurs, as well as the more derived Nothronychus and Erliansaurus . It is, however, inconsistent with phylogeny and might have been ontogenetic.
VERTEBRAE
The vertebral column of N. graffami ( Figs 5–8 View Figure 5 View Figure 6 View Figure 7 View Figure 8 ; Tables 1 View Table 1 , 2 View Table 2 ) was disarticulated, but the bones were found in association. Presacral vertebrae were all compressed and heavily fractured owing to extensive pneumaticity of these bones. In contrast, the sacral vertebrae, recovered ankylosed with the synsacrum, were intact and relatively undistorted. The caudal vertebrae were found near the hindlimbs and synsacrum, and were not severely compressed.
Cervical vertebrae
Two large N. mckinleyi cervical vertebrae are preserved ( Fig. 5 View Figure 5 ). Both are highly pneumatic and very similar to the Bissekty, Uzbekistan therizinosaur vertebrae (Sues and Averianov 2016). These vertebrae are transversely flattened with respect to the Bissekty specimens, but Sues and Averianov noted that anterior cervicals possess centra and articular surfaces that are laterally narrower than the posterior ones, but this may partially reflect taphonomic distortion in Nothronychus . On this basis, the preserved vertebrae of N. mckinleyi are considered anterior cervicals. Both centra are longer than high. In contrast to the Bissekty vertebrae, the neurocentral sutures are fused, supporting the conclusion that the specimen of N. mckinleyi was mature.
The cervical centra are shallowly amphicoelous. The anterior articular surfaces are nearly perpendicular to the long axis of the centra, but the posterior surface is slightly angled anteriorly, so the centra are almost rectangular in lateral view in contrast to the Bissekty specimens, where the anterior articular surface is more oblique (Sues and Averianov 2016). They are shallowly constricted laterally, but ventrally excavated. In one specimen, the lateral side of the centrum is visible. There are two pneumatic fossae, very similar in position and development to the Bissekty specimens. The sides are penetrated by multiple pneumatic foramina within the pneumatic fossae, as in Fukuivenator (Hattori and et al. 2021). A single pneumatic foramen invades the centrum of Erliansaurus ( Xu et al. 2002) . The midcervical centra bear two pneumatic foramina, whereas the postcervicals exhibit one ( Pu et al. 2013). These foramina are variable in number in Nothronychus between vertebrae and from right to left sides based, in part, on computed tomography (CT) data ( Smith et al. 2020). Sues and Averianov (2016) suggested that some foramina may be vascular in the Bissekty vertebrae. Well-developed anterior and posterior centrodiapopyseal laminae define boundaries for the pneumatic fossae. The anteroventral corner of the centrum is marked by a well-developed ventrolaterally directed parapophysis. As in the Bissekty vertebrae, longitudinal ridges extend posteriorly from the parapophyses, enclosing a deep longitudinally directed ventral groove that is deeper than the flatter ventral surface. In contrast to Erliansaurus ( Xu et al. 2002) there is no evidence of a ventral keel in Nothronychus .
The cervical vertebrae exhibit low neural spines as in Jianchangosaurus ( Pu et al. 2013) . The neural canal is circular anteriorly, but transversely narrow posteriorly. The prezygapophyses gradually narrow away from their bases. They are nearly triangular in cross-section, with shallow excavations laterally and dorsally. The articular facets are oriented dorsomedially. The postzygapophyses are broken near the base. They are trapezoidal in cross-section. As in the Bissekty vertebrae (Sues and Averianov 2016), the articular facets would have been lateral to the centrum. The epipophyses extend anteriorly as low ridges from the postzygapophyses, a derived character. The neural spine is long and low, very similar to that of the Bissekty material and Neimongosaurus ( Zhang et al. 2001) . Nothronychus possesses anteroventrally short neural spines that result in an X-shape dorsally, a derived character. The anterior and posterior faces possess interspinous recesses. Nothronychus possesses well-developed, wide diapophyses that project ventrolaterally over multiple lateral pneumatic fossa in the centra, a derived character.
Cervical rib
The preserved cervical rib is fused to a cervical vertebra in Nothronychus ( Fig. 5 View Figure 5 ), as in the Bissekty specimens (Sues and Averianov 2016). The capitulum is narrower than that seen in the Asian material. The tuberculum is fused with the parapophysis along its length, forming a bony web continuous with the capitulum. Therefore, there is no division between the two processes.
Dorsal vertebrae
There are three well-preserved dorsal vertebrae known for N. graffami ( Fig. 6 View Figure 6 ) exhibiting slight distortion. They are highly modified from the plesiomorphic theropod condition. Their exact positions in the vertebral column are unknown, but some relative position can be assigned based on quarry position, size, and relative height and orientation of the neural spines. The neural spines are much higher than in the cervical vertebrae as in Jianchangosaurus ( Pu et al. 2013) . Centra sizes of the presacral vertebrae decrease anteriorly from the synsacrum. They are roughly equidimensional, with tall neural spines in the middorsal position. The anterior dorsal vertebrae are smaller, the neural spines are shorter, and the centra become longer relative to the cross-sectional dimensions. As in Neimongosaurus ( Zhang et al. 2001) , the anterior surfaces are lower than the posterior ones. Hedrick et al. (2015) arranged them such that anteriorly there is a gradual increase in the height of the neural spine followed by an overall reduction passing posteriorly. One vertebra is anteroposteriorly taphonomically deformed, whereas the other two are laterally deformed. Laminae are identified following Wilson (1999). In contrast to Jianchangosaurus (Pu et al. 2012) and Beipiaosaurus ( Liao et al. 2021) , the dorsal vertebrae of Nothronychus are pneumatic, similar to Suzhousaurus ( Li et al. 2008) . The transverse processes form a spine table, a derived character.
Dorsal vertebra 1
This vertebra ( Fig. 6A, B View Figure 6 ) is smaller than the other succeeding vertebrae. The centrum is not as long as in the anterior dorsal vertebrae, but it is taller and wider. It is spool shaped, laterally excavated, and shallowly amphicoelous. The anterior articular surface is somewhat smaller than the posterior surface. The anterior facet is taller than wide, but this may be a result of taphonomic compression as the posterior facet is roughly equidimensional. A short ventral ridge is present on the ventral surface of the centrum, but this may be exaggerated by taphonomic distortion. There is a small neural canal that is nearly triangular. On either side, the pedicels are well developed, they merge above the canal but do not meet in the floor, resulting in a shallow groove. The prezygapophyses extend dorsolaterally and bear oval dorsomedially oriented articular facets, suggesting limited dorsoflexion and increased lateroflexion at this joint ( Smith 2015). The processes are connected by a wide, dorsally excavated web. The parapophyses are reduced in contrast to the condition in Fukuivenator ( Hattori et al. 2021) .
The neural spine is taller in relative proportions above the zygapophyses than is seen in the succeeding dorsals. The anterior margin of the neural spine is bifurcated, resulting in a shallow, anteriorly oriented groove. The anterior and posterior surfaces are rugose, indicating the presence of a supraspinous ligament as in Tyrannosaurus ( Brochu 2003) . A circular scar dorsal to the neural canal in some of the dorsal vertebrae ( Fig. 6 View Figure 6 ) indicates the presence of a second, interarcuate ligament as in the extant rhea and convergent with some sauropods ( Tsuihiji 2004), a derived character. Stevens and Parrish (1999) reported that such a compound nuchal ligament is present in extant birds and crocodylians, holding the neck in a horizontal position. Tsuihiji proposed that this ligament would exhibit maximal tension when the associated vertebrae are nearly horizontal to sloping ventrally, functioning as a brace. In the dorsal vertebrae of Nothronychus , such a compound supraspinous ligamental mechanism would have reduced energy requirements involved in dropping the abdomen such that more anterior vertebrae are below posterior ones as during feeding.
The tubercular facet projects only slightly laterally relative to the neural arch, whereas the capitular facet is low on the centrum and is indistinct. Well-developed anterior centroparapophyseal laminae extend from the centrum to the extremities of the prezygapophyses. The anterior margin of the transverse process is marked by a prezygodiapophyseal lamina. The neural spine bears lateral spinodiapophyseal laminae with transverse processes extending laterally dorsal to each neural canal. They are supported by paradiapophyseal laminae. The process is excavated by infraprezygopophyseal fossa, infradiapophyseal fossa, and infrapostzygapophyseal pneumatic fossae, a derived character.
Dorsal vertebra 2
This vertebra ( Fig. 6C, D View Figure 6 ) is considered anterior in the series relative to dorsal vertebra ‘3’. The preserved transverse process was taphonomically deformed and pressed dorsally against the neural spine ( Hedrick et al. 2015). Therefore, the dorsal surface of the process is concealed. As in the previous vertebra, the centrum is spool shaped and laterally excavated. A small lateral pleurocoel is present in the centrum. The pedicels are well developed and extend above the neural canal. The neural spine is taller above the level of the zygapophyses than in the posterior dorsal. Both vertebrae are marked by rugose anterior and posterior margins marking the presence of a supraspinous ligament. A narrow hypantrum is present dorsal to the articular facets.
A number of laminae are identifiable ( Hedrick et al. 2015). The ventral surface of the transverse process is marked by the prezygodiapophyseal and paradiapophyseal laminae. The laminae are divided by the infraprezyapophyseal fossa. Behind the second lamina, the infradiapopyseal and infrapostzygapopyseal laminae are continuous with each other. The infrapostzygapophyseal lamina extends from the posterior articular facet to the postzygapophyses. The postzygapophyses are broken near the base, but the base visibly merges with the hyposphene, the presence of the latter a derived character.
Dorsal vertebra 3
This vertebra ( Fig. 6E, F, G View Figure 6 ) is a posterior dorsal. It resembles the vertebra of the type specimen of N. mckinleyi ( Kirkland and Wolfe 2001) . Both vertebrae have tall pedicels and zygapophyses, substantial zygapophyseal laminae, and short neural spines. The centrum of the N. graffami specimen is taller than wide and the lateral surface lacks a well-defined pleurocoel. There is a distinct ventral keel present. These traits differ from the vertebra in N. mckinleyi , where the centrum is roughly as wide as tall, there is no keel, and there is a pronounced pleurocoel. The capitular facet is more dorsally located and less pronounced in N. graffami . A weak hyposphene/hypantrum is present in N. mckinleyi , but is absent in N. graffami . This N. graffami vertebra may be one or two positions posterior relative to that in N. mckinleyi vertebra, in keeping with the observation that pneumaticity in the vertebrae increases from posteriorly to anteriorly, a derived character. Such a pattern reflects the pneumatic hiatus in the series ( Smith et al. 2020). The transverse process is broken near the base. Most of the laminae are unclear, except for a partial centropostzygapophyseal lamina.
Sacral vertebrae
There are five primordial sacral vertebrae and a sixth sacrocaudal in Nothronychus ( Fig. 7 View Figure 7 ). Five are described for Jianchangosaurus ( Pu et al. 2013) and Suzhousaurus ( Li et al. 2008) , but six for Neimongosaurus ( Zhang et al. 2001) and Segnosaurus ( Barsbold and Perle 1980) . Alxasaurus ( Russell and Dong 1993) and Jianchangosaurus ( Pu et al. 2013) possess five sacral vertebrae, Enigmosaurus ( Barsbold 1983) has not more than six. Hattori et al. (2022) described four fused sacral and one isolated vertebra in Fukuivenator , but Azuma et al. (2016) originally argued for six. The fifth vertebra may correspond with the sacrocaudal vertebra in Nothronychus . Therefore, this character count must be considered variable within therizinosaurs ( Hedrick et al. 2015). The five sacral vertebrae are completely fused to each other at the centra and neural spines, forming a low median sacral crest, as in Enigmosaurus ( Zanno 2010) and similar to that described for Velociraptor ( Norell and Makovicky 1997) , but in contrast with Jianchangosaurus (Pu et al. 2012) . This trait, along with fusion of the neural spines, is considered a derived character. The S6 centrum is not fused to S5. Sacral ribs are fused to all but the last vertebra. Accordingly, the last vertebra between the ilia is best regarded as a sacrocaudal rather than a 6 th sacral as identified by Hedrick et al. (2015). The sacral vertebrae of Jianchangosaurus and, presumably, Beipiaosaurus ( Liao et al. 2021) , are not fused where only neural spines three and four are fused. At no point are they separated in Nothronychus . Lines of contact between the successive neural spines are faint. Estimated lengths are roughly equivalent for all neural spines, unlike Tyrannosaurus , where they tend to increase from S1 to S4 ( Brochu 2003). The neural spines are shallowly laterally excavated, narrowing posteriorly. The neural spine of the sacrocaudal is displaced anteriorly and incorporated into the ridge, similar to that of Shuvuuia ( Chiappe et al. 2002) .
The sacral vertebral centra are wider than tall. The neural canal is well preserved, with no evidence of distortion from crushing. The centra are dorsoventrally constricted and laterally excavated, but not ventrally, unlike the condition in Suzhousaurus ( Li et al. 2008) . The centra lack any pneumatic openings and CT data indicate no pneumatic invasion, as in the first four sacral vertebrae of Fukuivenator (Hattori et al. 2022) . S1 is shallowly procoelous. The sacrocaudal central articular surface is flat. The ventral surface is longer than the dorsal surface of the centrum, resulting in an obliquely inclined posterior articular contact with the first caudal vertebra. This condition is not a result of taphonomic distortion. As a result, the tail would be held posteriorly at a slightly obtuse angle with respect to the synsacrum. The prezygapophyses are laterally enlarged, bearing dorsally facing articular facets. They are divided by a narrow hypantrum. The diameter of the neural canal is small and taller than wide.
The centra are united to the ilia by a series of six sacral ribs that are supported by the transverse processes as in birds ( Brochu 2003). The ventral surfaces lack any excavation or grooves in contrast to the Bissekty therizinosaur centra ( Sues and Averianov 2015). The ventral sulcus noted as characteristic in sacral vertebrae of maniraptoran theropods ( Norell and Makovicky 1997) is absent in Nothronychus . In contrast to Alxasaurus ( Russell and Dong 1993) , all of the sacral vertebrae in Nothronychus are roughly the same length and lack ventral grooves. The S1 ventral surface of Jianchangosaurus bears a groove, whereas S3–5 of Falcarius bear grooves ( Zanno 2010). In contrast to the alvarezsaurs, there is no ventral keel in Nothronychus ( Chiappe et al. 2002) . The centra of both Neimongosaurus ( Zhang et al. 2001) and the Bissekty therizinosaur possess pneumatic openings that are not present in Nothronychus . There is a faint hyposphene under the postzygapophyses of the sacrocaudal. The neural canal is reduced to a vertically oriented slit. The pedicels are very short.
The sacral ribs are fused to the ilia and bridge adjacent vertebral centra as in Allosaurus ( Madsen 1976) and in contrast to Tyrannosaurus , where each rib contacts both adjacent centrum ( Brochu 2003). Sacral ribs 1–5 bear dorsal and ventral laminae on the anterior surface. These can be referred to the prezygodiapophyseal and centrodiapophyseal laminae, respectively. The last rib is much thinner than the previous ones and lacks accessory laminae. In sacral rib 1, the transverse process overlies the rib in a shallow depression. The first sacral rib extends anterolaterally from the S1 centrum and is the narrowest of the ribs. The second rib is mediolaterally wider and anteroposteriorly narrower. Sacral ribs 2–5 extend laterally and widen posteriorly, so the fourth sacral foramen is oval shaped, whereas the preceding ones are more circular. Sacral ribs 3 and 4 are medial to the acetabulum. The last sacral rib (#6) bridges sacral vertebra 5 and the caudosacral. It extends posterolaterally before curving anterolaterally to partially contact sacral rib 5. Sacral foramen 5 is more circular than 4.
Caudal vertebrae
The tail of N. graffami consists of 21 caudal vertebrae, not counting the sacrocaudal, with a transition point and loss of transverse processes at vertebra #13 ( Figs 8 View Figure 8 , 9 View Figure 9 ). Anything less than 40 is considered a derived character. Liao et al. (2021) counted at least 30 in Beipiaosaurus , apparently ending in a pygostyle ( Xu et al. 2003). Zhang et al. (2001) suggested between 25 and 30 caudal vertebrae in Neimongosaurus , but counted 22. Hattori et al. (2022) counted 28 caudals in Fukuivenator , but Azuma et al. (2016) referred to at least thirty. Traditionally, the transition point has been taken to mark the posterior extent of the m. caudofemoralis longus ( Gatesy 1995), but Persons and Currie (2010) used the presence of scarring on the haemal arches and descent of the transverse processes to the level of the midcentrum to identify its maximum posterior extent.This boundary is difficult to confidently identify in Nothronychus , but may occur around vertebra #7 or 8. Hedrick et al. (2015) suggested that the tail may have extended past the last preserved caudal vertebra. There is no direct evidence for that hypothesis, but it cannot be refuted. A low caudal vertebral count in therizinosaurs indicates a reduced tail relative to the plesiomorphic theropod condition.
A well-preserved caudal vertebra 1 ( Fig. 8 View Figure 8 ) is available for N. mckinleyi ( AzMNH 2117) . Unlike the condition described for Tyrannosaurus ( Brochu 2003) , no neurocentral sutures are visible in Nothronychus . The proximal-most caudal centra are shallowly amphicoelous ( Hedrick et al. 2015). They are laterally and ventrally excavated with no ventral ridges. The centra are box like, a derived character. A low ventral ridge is seen in caudal #5. Single large pleurocoels penetrate the lateral sides of the centra of the first four caudal vertebrae ( Zanno et al. 2009, Hedrick et al. 2015), indicating extension of air sacs into the base of the tail in life ( Smith et al. 2020) and in contrast to the apneumatic proximal caudal vertebrae of Fukuivenator (Hattori et al. 2022) and Jianchangosaurus (Pu et al. 2012) . These pneumatic foramina are a derived character. The first few caudal vertebrae of Neimongosaurus bear small pleurocoels ( Zhang et al. 2001). The anterior articular surface is taller than the posterior in all the proximal caudal vertebrae.
Stout transverse processes extend laterally from the first 13 caudal vertebrae at the level of the neural canal. The dorsal surfaces face posterodorsally. Hedrick et al. (2015) describe an embayment in this surface in the first caudal vertebra. The spines are supported ventrally by robust centrodiapophyseal laminae. In contrast to Tyrannosaurus ( Brochu 2003) , they do not expand distally. The anterior margin forms a ridge supporting the prezygapophyses. The prezygapophyses are long and extend anterodorsally similar to Neimongosaurus ( Zhang et al. 2001) , and as noted by Hedrick et al. (2015), extending almost to the height of the neural spine. A horizontal ridge demarcates the hypantrum. The processes bear medially oriented articular facets. The postzygapophyses are much shorter, with corresponding laterally oriented articular facets. They merge ventrally, but a vertical groove marks the hyposphene. The spines in the adjacent vertebrae are separated by large intervertebral foramina. The distal caudal vertebrae are deeply amphicoelous. The centra are short, a derived character; shallowly excavated, and lack a ventral ridge. The chevrons are long and blade like, a derived character.
Ribs
A number of dorsal ribs are known for N. graffami ( Fig. 10 View Figure 10 ). They cannot be assigned to any vertebrae with any confidence. All are apneumatic and lack foramina. The capitulum is borne on a long neck that extends almost perpendicularly from the shaft, suggesting an enlarged abdomen. The neck is narrower approaching the capitulum, but expands as it meets the shaft below the tubercle. The tubercle is concavo-convex and very thin. The two structures are posteriorly separated by a wide notch. A long ridge extends down the posterolateral corner of the shaft from the tubercle, excavating the posterior side into an elongate sulcus. The surface of the ridge becomes increasing rugose distally. It is very similar to a corresponding structure in Tyrannosaurus ( Brochu 2003) and may be the origin of the serratus musculature. The shaft is thin and widely curved. A pronounced ridge is present on the anterior surface of the shaft extending from the level of the capitulum down the shaft, corresponding to what Brochu (2003) described for Tyrannosaurus , but exaggerated over the latter taxon. They are associated with a costal groove. As in Tyrannosaurus , it terminates at about midshaft.
Chevrons
A number of variably sized chevrons are preserved ( Fig. 10 View Figure 10 ; Table 3 View Table 3 ). As in Tyrannosaurus ( Brochu 2003) , there are some anteroposterior processes at the heamal canal. The proximal ends meet medially above the canal. There is no ventral expansion.
Furcula
The furcula ( Fig. 11 View Figure 11 ) exhibits a considerable amount of intraspecific variation in theropods ( Chure and Madsen 1996). Hedrick et al. (2015) did not include the furcula (UMNH-VP 16420) of N. graffami in their description of the known material, but it is well developed. The Nothronychus specimen is more complete on the left side than the right, such that the distal right ramus is incomplete. This element is unpaired, with no indication of a midline suture. It is widely ‘U’-shaped, a derived character, with thicker rami than the ‘V’-shaped furcula of Velociraptor ( Norell and Makovicky 1997) . This morphology would presumably result in a wider chest in Nothronychus than Velociraptor , following the proposal for Daspletosaurus vs. Gorgosaurus ( Makovicky and Currie 1998) . It compares quite closely with that of Neimongosaurus ( Zhang et al. 2001) and Falcarius ( Zanno 2010) , but possesses a reduced hypocleidium, a derived character, and the enlarged apical ridge of some other theropods, such as Albertosaurus ( Makovicky and Currie 1998) , Allosaurus ( Chure and Madsen 1996) , and Oviraptor ( Nesbitt et al. 2009) . Both rami gradually narrow distally from the midline, as is typical for non-tyrannosaurid theropods. The body and rami are antero-posteriorly flattened in contrast to Coelophysis ( Tykoski et al. 2002) . In both Neimongosaurus and Nothronychus , the depth increases at the midline. The furcula of Nothronychus is non-pneumatic, in contrast to the dromaeosaurids Buitreraptor ( Makovicky et al. 2005) , and Bambiraptor ( Burnham 2004) , and the megaraptoran Aerosteon ( Sereno et al. 2008) . Both rami form an intraclavicular angle of about 130°, as in Neimongosaurus ( Zhang et al. 2001) , Coelophysis ( Tykoski et al. 2002) , and some specimens of Allosaurus ( Chure and Madsen 1996) , less than Beipiaosaurus ( Liao et al. 2021) , and more than those described for tyrannosaurids ( Makovicky and Currie 1998, Lipkin and Carpenter 2008) and Beipiaosaurus ( Nesbitt et al. 2009) . The furcula of Nothronychus is anteroposteriorly flat in contrast to the oval cross-section of Falcarius ( Nesbitt et al. 2009) . The left ramus and presumably the right are robust, becoming thicker towards the hypocleidium, in contrast to the trend in Iberomesornis , where the ramus thins approaching the hypocleidium ( Sanz et al. 2002). The left ramus is very straight in anterior view. Chure and Madsen (1996) described a longitudinal groove in the rami of at least some specimens of Allosaurus that they ascribe to the origin of M. supracoracoideus as is seen in roadrunners ( Berger 1952). Although this muscle may arise from this region in Nothronychus , no such groove is apparent, but this area is slightly expanded. Such a groove also remains undescribed for Tyrannosaurus , but a scar in this area that may correspond to the groove described in a specimen of Gorgosaurus ( Makovicky and Currie 1998) . Makovicky and Currie (1998) agree with the interpretation of Dial et al. (1987) that this margin represents the origin of the sternobrachialis head of M. pectoralis. Lipkin and Carpenter (2008) agree with the interpretation of this region in Allosaurus as a muscle attachment point.
The epicleidium does not appear to be taphonomically deformed. It is very similar to that in modern birds, including Aquila ( Baumel and Witmer 1993) . It is expanded posteriorly but not anteriorly. In anterior view, the epicleidium is much shallower than the ramus. The two form an angle of about 130°. The epicleidium extends laterally and curves over a deep lateral groove, similar to the enantiornithine Sinornis ( Sereno et al. 2002) . The spatial relations of the furcula, scapula, and coracoid would limit the scapula to a nearly horizontal orientation high on the ribcage as in Velociraptor ( Norell and Makovicky 1999) and typical for maniraptoran theropods ( Jasinoski et al. 2006). In Nothronychus , the distal end of the furcula proximal to the epicleidium adjacent to the articular facet for the coracoid is laterally deflected to form an apparent horizontal groove or notch in contrast to the straight rami in Neimongosaurus ( Zhang et al. 2001) . This region does not appear taphonomically deformed. Therefore, a tendon could have extended through this canal from the scapulocoracoid towards the humerus. If this tendon is related to M. supracoracoideus, the associated canal could be the triosseal canal. There is a ventrally inclined elongate acrocoracoideal facet on the posterior surface of the epicleidium extending to the acromial process, distally contacting the acromion of the scapulocoracoid. In anterior view, the canal extends distally to join a ventrally inclined elongate humeral articulating facet. The surface is somewhat rugose, but lacks any ridges. This canal is interpreted, in part, as a supracoracoideus sulcus accommodating the supracoracoideus tendon, hence a portion of a triosseal canal. Such an architecture would be unique in non-avian theropods and, if correctly interpreted, convergent with Neornithes. Clearly Nothronychus did not fly, so the function of such a development is uncertain. M. supracoracoideus powers the upstroke during flight ( Poore et al. 1997), so it may have been involved in abducting the humerus in Nothronychus .
The furcula of Nothronychus generally corroborates the proposal of Nesbitt et al. (2009) that therizinosaur furculae would be robust with a high intraclavicular angle. A robust furcula may have functioned as a spacer between the adjacent scapulocoracoids ( Ostrom 1976) or resisted forces generated during prey acquisition ( Makovicky and Currie 1998). Like Tyrannosaurus ( Lipkin and Carpenter 2008) , the furcula of Nothronychus could probably absorb some compressive stress within the rami. For tyrannosaurs, Makovicky and Currie (1998) indicated that the sternobrachialis head of M. pectoralis inserted on the furcula as in birds such as pigeons ( Chiasson 1984, Dial and Biewener 1993). Chure and Madsen (1996) suggested that the furcula served as the origin for M. supracoracoideus in Allosaurus .
Coracoid
The coracoid (UMNH-VP 16420) ( Fig. 11 View Figure 11 ; Table 4 View Table 4 ) is incomplete, but clearly semicircular, similar to Jianchangosaurus ( Pu et al. 2013) , rather than subrectangular as in Beipiaosaurus ( Liao et al. 2021) , Deinonychus ( Ostrom 1974) , and alvarezsaurs ( Chiappe et al. 2002). It is convex externally, concave internally, and very thin along its length. There is a low, wide biceps tubercle that is separated from the margin of the glenoid fossa by a shallow excavation, in contrast to Beipiaosaurus ( Liao et al. 2021) . Ostrom proposed a relationship between the development of the biceps tubercle and the length of the forelimb, but that correlation does not hold in Nothronychus . Traditionally, it has been associated with the origin of M. biceps brachii, but that interpretation is occasionally challenged, for example by Norell and Makovicky (1997), who supported an association with M. coracobrachialis. Perle (1979) supported an M. coracobrachialis origin in his description of Segnosaurus . The traditional interpretation ( Burch 2014) is followed here, but the alternative reconstruction remains possible. The ridge extending from the top of the glenoid fossa to the tubercle described in Deinonychus ( Ostrom, 1974) is reduced to a faint ridge in Nothronychus . The margins of the coracoid fenestra are eroded ( Hedrick et al. 2015). The glenoid fossa is supported by an anterior ridge similar to Velociraptor , along with a laterally directed osseous floor ( Norell and Makovicky 1997), but this structure is absent in Fukuivenator (Hattori et al. 2022) . The subglenoid fossa is reduced and linear. The glenoid fossa in both specimens of Nothronychus , but especially in N. graffami , opens posterolaterally. The surface is rugose, as in some large crocodylians ( Brochu 2003), suggesting a well-developed cartilage pad and synovial capsule, and also possibly related to maturity. There is a small tubercle anterior to the glenoid fossa that is interpreted as a muscle attachment point. The sternal process, sometimes referred to as the caudal process ( Zanno 2006), in the coracoid described in Velociraptor ( Norell and Makovicky 1997) , is reduced to a small point in Nothronychus , whereas it is longer in Neimongosaurus ( Zhang et al. 2001) , Jianchangosaurus (Pu et al. 2012) , and Terizinosaurus ( Barsbold 1976). In many theropods, this structure extends posterior to the glenoid (e.g. Tyrannosaurus ) ( Brochu 2003). Sereno (1999) proposed the posterior extension as a synapomorphy for coelurosaurs. It has also been noted in Allosaurus ( Madsen 1976) . Gauthier (1986) regarded it as a tetanurine synapomorphy. Hedrick et al. (2015) note that the extension is present in the left coracoid, but absent from the right in Nothronychus . The coracoid foramen is anterodorsally positioned from the glenoid, close to the contact with scapula. The acromial process is not preserved.
Scapula
The scapula of Nothronychus ( UMNH 16420) is long and strap like ( Fig. 11 View Figure 11 ; Table 4 View Table 4 ), widening in lateral view distal to the glenoid, as in Velociraptor ( Norell and Makovicky 1999) . The overall morphology is very similar to Erliansaurus ( Xu et al. 2002) . As in Erliansaurus , the scapula of Nothronychus is arched in lateral view, but straight in dorsal view, but lacks any dorsal or ventral flanges. It is broken distally and the acromion is incomplete. There is no apparent constriction above the glenoid as noted for Erliansaurus .
Hedrick et al. (2015) describe the scapula constituting about half of the glenoid fossa. This proportion would include the osseous floor. There is a reinforced, rugose lip on the dorsal margin of the fossa, that Hedrick et al. termed a caudal buttress. It corresponds with the triceps tubercle in Tyrannosaurus and birds ( Brochu 2003). This structure is also described in Falcarius ( Zanno 2006) and Jianchangosaurus ( Pu et al. 2013) , but in contrast to the latter taxon, both glenoid buttresses are roughly the same size in Nothronychus . It meets the corresponding coracoid lip on the anterior margin of the fossa, but is reduced at this point. The suture between the two bones is fused, but slightly elevated, corresponding to the ridge in Gallimimus ( Osmólska et al. 1972) . As in Erliansaurus ( Xu et al. 2002) , the scapular blade is gently arched along its length. In both specimens there is a longitudinal ventral groove bounded by lateral and medial ridges. The lateral ridge is more pronounced and is delineated by a lateral groove. Hedrick et al. (2015) described a small rugosity along the groove. These ridges merge about midway down the scapular blade to form a sharp posterior margin. The dorsal margin is rugose along its length. Distally, the lateral surface of the blade is markedly convex, corresponding with the swellings that Brochu (2003) described for Tyrannosaurus . The scapular cross-section of Nothronychus is generally oval, becoming more triangular proximally in cross-section along its length, but somewhat excavated medially, similar to the description of Beipiaosaurus ( Liao et al. 2021) . The lateral face becomes somewhat excavated above the glenoid, but this concavity shallows distally along the scapular blade.
The morphology of the scapulocoracoid and glenoid fossa is quite similar to Velociraptor ( Norell and Makovicky 1999, Hedrick et al. 2015). The scapulae were widely spaced with widely separated ventrally oriented glenoid fossae ( Fig. 11 View Figure 11 ), as was typical for maniraptoran theropods (Senter 2006). There is a laterally-directed osseous floor with anterior and posterior labia in the glenoid, as in Velociraptor . The humerus would have been capable of sub-horizontal elevation, but not above the dorsum, as in other theropods except birds. This limitation is supported by the lack of well-developed scars for the M. levator scapulae and M. trapezius in the dorsal margin of the scapula. These muscles are absent in birds, but there is evidence for them in ceratosaurs and tetanurans with obliquely oriented scapulae ( Jasinoski et al. 2006, Burch 2017). Brochu (2003) described the scapular margin of Tyrannosaurus as rugose and attributed this development to attachment of the rhomboid musculature, but it is probably associated with the M. levator scapulae and M. trapezius (Burch 2016). The dorsal margin of the scapula in Nothronychus , however, is semicircular and smooth, becoming increasingly acute distally.
Humerus
The forelimb of Nothronychus has sometimes been described as elongate as in other therizinosaurs ( Barsbold 1976), but a humerus/femur ratio of 0.54 falls within the lower range for maniraptoran theropods ( Senter and Parrish 2006). Liao et al. (2021) report the forelimb of Beipiaosaurus as longer than the hindlimb, but this relation is reversed in Nothronychus . The humerus ( Fig. 12A–J View Figure 12 ; Table 5 View Table 5 ) of Nothronychus (UMNH-VP 16420) is anteroposteriorly crushed. It widens proximally and distally. The shaft is stouter and straighter than Deinonychus ( Ostrom 1969) , a derived character. The head is cylindrical, like most theropods ( Brochu 2003) and overhangs the anterior and posterior faces. It cannot be described as bulbous like Beipiaosaurus , which is suggested as common in therizinosaurs ( Liao et al. 2021). It points medially away from the deltopectoral crest. As in Falcarius , a shallow posterior depression separates the head from the internal tuberosity ( Zanno 2006). The internal tuberosity is a rugose crest that angles medially from the head, a derived character. It is much shorter than in Deinonychus ( Ostrom 1969) . Liao et al. (2021) describe a proximal excavation extending to the posterior face separating it from the head in the humeri of Beipiaosaurus , Erlikosaurus , Neimongosaurus , and Segnosaurus . This character is present, but reduced in Nothronychus . Distal to the tuberosity is a notch that may be taphonomically exaggerated. There is a shallow sulcus dividing the head from the lateral process. The posterior groove described for many theropods is reduced ( Brochu 2003), but about the same thickness as the head, as in extant birds ( Brochu 2003). There is no development of an anterior tuberosity or posterior trochanter as described in Erliansaurus ( Xu et al. 2002) .
The proximal end has a transversely short deltopectoral crest that extends along about 30% of the proximal shaft. It is much smaller than in Terizinosaurus or Segnosaurus ( Barsbold 1976) and similar to Jianchangosaurus ( Pu et al. 2013) , Erliansaurus ( Xu et al. 2002) , Neimongosaurus ( Zhang et al. 2001) , and Beipiaosaurus ( Liao et al. 2021) . The margin is slightly rugose. In contrast to Tyrannosaurus , it merges smoothly with the lateral process as in Gallimimus ( Osmólska et al. 1972) . There is a wide excavation distal to the head on the anterior side.
There is a shallow longitudinal groove in the anterior face of the shaft merging with the proximal end of the M. brachialis fossa. The distal end is clearly crushed, but is deflected slightly anteriorly. The articular surface extends onto the anterior surface, as in Deinonychus ( Ostrom 1969) and Beipiaosaurus ( Liao et al. 2021) . The medial and lateral condyles are short and divided by a shallow trochlear notch. The notch is deeper anteriorly and distally than posteriorly. Both condyles are equally developed, in contrast to Gallimimus ( Osmólska et al. 1972) or Deinonychus ( Ostrom 1969) . The medial condyle angles slightly posteriorly. As in Deinonychus , the lateral epicondyle is reduced to a low lateral ridge, but the medial epicondyle is present as an enlarged knob. The olecranon fossa is very shallow and indistinct.
Radius
Like the ulna, the radius of Nothronychus is straight with an oval cross-section ( Fig. 12K–P View Figure 12 ; Table 6 View Table 6 ), similar to Beipiaosaurus ( Liao et al. 2021) and Jianchangosaurus ( Pu et al. 2013) . In contrast to Fukuivenator (Hattori et al. 2022) , it is much shorter than the humerus. The head is slightly expanded beyond the shaft. The articular surface is expressed as a concave cotyle with distinct lips anteriorly and laterally. A pronounced rugose tubercle represents the biceps tubercle. It is similar to, but larger than, that described for Neimongosaurus ( Zhang et al. 2001) . It is set off from the head by a concave surface in the anterior surface. This is in contrast to that in Tyrannosaurus , where there are two discrete tubercles with one distally displaced down the shaft ( Brochu 2003). The shaft is marked by a long groove on the anterior surface, but this is probably a result of postmortem deformation. Distally, the articulation is expanded posteriorly, forming a single condyle angling towards the ulna. There is a shallow ulnar notch medially. The posterior margin is excavated, but this may be a result of damage. A lateral pit corresponds to the sulcus tendinosa ( Baumel and Witmer 1993).
Ulna
The ulna (UMNH-VP 16420) ( Fig. 12Q–V View Figure 12 ; Table 6 View Table 6 ) is about 75% the length of the humerus, similar to Falcarius ( Zanno 2006) . Like other therizinosaurs ( Hedrick et al. 2015), the shaft is straight in contrast to other maniraptorans, where it bows laterally ( Gauthier 1986). The ulna is expanded proximally to form a trochlea and distinct olecranon process, a derived character. It is nearly triangular in proximal view, as in Velociraptor ( Norell and Makovicky 1999) . The olecranon process is displaced medially. A prominent rugose coronoid process extends distally from the olecranon process on the anterior face. As in Tyrannosaurus ( Brochu 2003) , a dorsal cotylar process is present on the lateral surface. This process constitutes part of the radial notch. The sigmoid notch is shallow and triangular, as in most theropods ( Brochu 2003). The proximal end is rugose. Ostrom (1969) suggested that this texture probably reflects the presence of a thin cartilaginous pad in Deinonychus . The distal end is expanded beyond the shaft. It is nearly triangular in distal view, but is distorted. The ulna is parallel with the radius forming a syndesmosis, a derived character.
Manus
Most of the manual elements ( Figs 13–17 View Figure 13 View Figure 14 View Figure 15 View Figure 16 View Figure 17 ; Table 7 View Table 7 ) are preserved, except for the carpals. Some are taphonomically distorted.
Metacarpal I
Metacarpal I (UMNH-VP 16420) was crushed ( Figs 13 View Figure 13 , 14 View Figure 14 ). It is the shortest of the three metacarpals, as found in most theropods ( Ostrom 1969). Unlike Deinonychus , where the first metacarpal shaft is wider than II or III, in Nothronychus , this element is nearly as wide as II, but wider than III. In contrast to Beipiaosaurus ( Liao et al. 2021) , it is considerably wider than tall, but this is partially a taphonomic artifact. In proximal view, the ventral surface is concave. The articular surface thickens medially. In dorsal view, there are lateral and medial spurs projecting proximally, partially enclosing distal carpal I. Therefore, movement of the first metacarpal independent of the distal carpals was probably limited, as Ostrom (1969) proposed for Deinonychus . However, unlike Deinonychus , the medial margin does not contribute to an extensive sweeping arc with distal carpal I and there is no change in surface texture that Ostrom used to propose a cartilaginous pad. Unfortunately, the carpals are not preserved in Nothronychus . There is a dorsally projecting excavation about 1/4 of the way down the shaft that makes up part of the boundary of a flattened facet presumably contacting metacarpal II, a derived character. If that is the case, metacarpals I and II were closely appressed, more so than reconstructed in Deinonychus . The lateral condyle is wedge like and larger than the medial one. Both condyles extend more ventrally than dorsally, permitting more flexion than hyperextension. The intercondylar sulcus is inclined, resulting in a medial rotation of the first digit with flexion, but little lateral movement of the first phalanx as proposed for Deinonychus ( Ostrom 1969) . The sulcus terminates in a deep flexor pit. The associated collateral pits are shallow.
Metacarpal II
Metacarpal II (UMNH-VP 16420) is the longest of the three metacarpals ( Figs 13 View Figure 13 , 14 View Figure 14 ). The shaft is dorsally crushed. It has an oval cross-section. The minimal diameter is about mid-shaft. The proximal and distal ends widen from the shaft. The distal lateral joint surface is wedge shaped, whereas the medial surface is more bulbous. These two components are separated ventrally by a short groove. In dorsal view, the proximal end is convex with a wide vertical ridge that forms the medial boundary of a faceted contact with metacarpal III. Distally, the condyles are both well-developed. The medial condyle is bulbous, but the lateral one is more wedge shaped. The medial condyle is vertical, but the lateral is inclined, resulting in a deep, wide intercondylar sulcus ventrally dividing the two. The ventral surface of the shaft is marked by an extensive flexor pit. The sulcus extends between the two condyles to join an extensor pit in the dorsal surface of the shaft. Like the situation in Deinonychus ( Ostrom 1969) , this architecture would have permitted little lateral movement. Within the condyles, the medial collateral pit is more distinct than the lateral one.
Metacarpal III
The third metacarpal is also dorsoventrally crushed ( Figs 13 View Figure 13 , 14 View Figure 14 ). It is more robust and straighter than the corresponding element in Deinonychus ( Ostrom 1969) or Velociraptor ( Norell and Makovicky 1999) . The proximal articular end is convex in dorsal view, but the medial region is damaged. There is a wide shallow dorsal joint surface marking the contact with metacarpal II. The ventral surface of the shaft bears a longitudinal groove, but this is probably a taphonomic artifact. Distally, the medial condyle is separated from the lateral condyle by a wide intercondylar sulcus forming a ginglymus, a derived character. The sulcus extends to shallow extensor and flexor pits. The medial condyle is circular with a shallow collateral pit.
Manual phalanges
All phalanges are preserved for N. graffami (UMNH-VP 16420), except III-3, which is preserved in N. mckinleyi (UMNH-VP 16420). This material was not extensively described by Hedrick et al. (2015). The phalangeal formula, 2-3-4, is common for theropods ( Ostrom 1969).
Phalanx I-1
Asin Falcarius ( Zanno 2006) , phalanxI-1 isthemostrobustinthe manus ( Figs 13 View Figure 13 , 15 View Figure 15 ). Phalanx I-1 is shorter relative to phalanges II-1 or III-1 than is described for Gallimimus ( Osmólska et al. 1972) . There is a modest amount of dorso-ventral distortion in the shaft. The shaft of phalanx I-1 maintains a nearly constant diameter along its length, but the ventral surface exhibits evidence of crushing. There is a wide, shallow, almost triangular excavation in the proximal ventral surface. Some constriction is present proximal to the distal condyles, as in Falcarius ( Zanno 2006) . It is more slender than in Allosaurus , but more robust than Deinonychus . Dorsally, the extensor pit is shallow and extends a short distance up the shaft. The ventral flexor pit is deeper than the extensor and extends farther proximally. In contrast to the almost triangular facet of Deinonychus ( Ostrom 1969) or Falcarius , the proximal facet of phalanx I-1 is nearly rectangular and very deep. Unlike Gallimimus ( Osmólska et al. 1972) , the facet is perpendicular to the shaft, with no lateral inclination. The base of the facet is horizontal and it forms a right angle with the medial edge. The dorsal margin is more convex than the ventral surface. There is a well-developed notch in the lateral margin that corresponds with a similar feature in Deinonychus . A similar feature present on the medial side of the phalanx in Deinonychus is poorly developed or absent in Nothronychus . This groove is interpreted as accommodating a well-developed lateral collateral ligament, so it is referred to as the collateral notch on either side. Within the proximal facet is a wide, oblique ridge on the lateral side, in contrast to Deinonychus , making the proximal facet asymmetrical, unlike the more symmetrical configuration in Allosaurus ( Madsen 1976) . The distal articulation is rotated clockwise, but this orientation may, in part, be a result of taphonomic distortion. The condyles are damaged. The distal condyles of phalanx I-1 of Nothronychus are oblique to the long axis of the bone, with the articular surface extending more ventrally than dorsally. In Nothronychus , the distal condyles are roughly symmetrical and divided by a deep intercondylar groove that extends into well-developed dorsal and ventral pits. In contrast to the oval condyles of Falcarius ( Zanno 2006) , those of Nothronychus are more circular and symmetrical, as in Deinonychus ( Ostrom 1969) . Like Deinonychus , the condyles of Nothronychus extend farther ventrally than dorsally on the shaft. There are deep collateral pits on each side. The lateral collateral pit is nearly in the middle of the circular condyle, whereas the medial pit is displaced closer to the dorsal margin. As a result, at full extension, digit I diverges from the long axis of metacarpal I at an angle of 25°, similar to Coelophysis ( Galton 1971) . The collateral pits are well developed in both distal condyles.
Phalanx II-1
Phalanx II-1 of N. graffami is laterally crushed ( Figs 13 View Figure 13 , 15 View Figure 15 ). The shaft is taller than wide, with a clear vertical constriction proximal to the distal condyles. In lateral view, the shaft gradually narrows distally until the distal condyles are reached. The proximal articulation is deformed, but the facets still fit well with the condyles of the second metacarpal. It is oval with a vertical major axis. The two facets making up the proximal articulation are separated by a well-defined, nearly vertical ridge. The ridge is slightly displaced medially, resulting in some asymmetry in the proximal facets. A long ventral intercondylar tuberosity extends proximally beyond the joint surface. This structure is much more pronounced than the faint dorsal intercondylar tuberosity. The distal condyles are very circular in lateral view. The lateral condyle is smaller than the medial, but both are very narrow and wedge shaped. They are separated by a distinct intercondylar sulcus that continues proximally into a deep pit in the ventral surface of the shaft. There is no corresponding dorsal pit. The distal condyles extend farther onto the shaft ventrally than dorsally. The collateral pits are wide and shallow. They are located somewhat dorsally on either side, unlike Falcarius ( Zanno 2006) and Beipiaosaurus ( Liao et al. 2021) . The lateral pit is larger than the medial. The distal condyles of phalanx II-1 are very circular and extend farther ventrally than dorsally. The lateral condyle is markedly smaller than the medial one. They are divided by a narrow, pronounced intercondylar sulcus. Ventrally, the sulcus expands into a deep, wide pit.
Phalanx II-2
Phalanx II-2 of N. graffami is laterally crushed ( Figs 13 View Figure 13 , 15 View Figure 15 ). The shaft is straight and narrow until it reaches the distal condyles. It is taller than wide. There is no evidence of a pronounced extensor pit proximally, but the ventral surface is marked by a deep flexor pit. The proximal articular surface is triangular and taller than wide. It fits closely with the distal condyles of phalanx II-1. The ventral surface of the shaft is flattened distally. A long ventral intercondylar tuberosity extends proximal to the joint surface, but the dorsal tuberosity is reduced. There is a vertical ridge extending dorsally from the tuberosity on the medial side of the articular surface, leaving only a lateral facet. In Falcarius , the articular surface is more symmetrical with a medially located vertical ridge. The distal condyles are circular and rounded, but the medial condyle is broken ventrally. The intercondylar sulcus is deep. The collateral pits are dorsally placed, wide, and shallow.
Phalanx III-1
Phalanx III-1 of N. graffami is short ( Figs 13 View Figure 13 , 16 View Figure 16 ). It is dorsoventrally crushed, but the proximal articular surface is well preserved. The shaft is wider than tall. It retains a longitudinal crest on the dorsal surface, similar to that described for Velociraptor , but more pronounced. In medial view, it is markedly concave, with a nearly constant width, in contrast to Falcarius , where it narrows proximal to the distal condyles ( Zanno 2006). The ventral surface bears a long, deep excavation that deepens distally. A dorsal excavation is also present, but may be the result of taphonomic distortion. The proximal articular surface is triangular and lacks dorsal or ventral intercondylar tuberosities. It is marked by an oblique ridge, resulting in the medial facet being larger than the lateral, in contrast to the condition reported for Falcarius , where it is absent. Like Falcarius , the ventral margin is expanded to form a buttress, but there is no tubercle in Nothronychus . Distally, the condyles are asymmetrical and divided by a deep intercondylar sulcus. The medial condyle is larger than the lateral. Neither condyle extends significantly beyond the shaft in either direction. The collateral pits are shallow and poorly preserved.
Phalanx III-2
Phalanx III-2 of N. graffami is fragmentary ( Figs 13 View Figure 13 , 16 View Figure 16 ), with only the distal end preserved. The shaft is wider than tall. It is marked by excavations dorsally and ventrally. The distal condyles are oblique and nearly symmetrical. They are divided by a wide intercondylar sulcus that widens ventrally. The collateral pits are shallow and dorsally placed on both sides. Neither condyle extends beyond the shaft dorsally or ventrally.
Phalanx III-3
Phalanx III-3 of N. mckinleyi ( Figs 13 View Figure 13 , 16 View Figure 16 ) is more slender than the other phalanges of digit III, as reported for Falcarius ( Zanno 2006) . The shaft narrows distally, so is straight and nearly circular in cross-section along its length. The proximal articulation is oval in outline, such that it is taller than wide. Both the dorsal and ventral margins are expanded to form buttresses in contrast to the morphology in Falcarius . The ventral buttress is expanded to form a short ventral intercondylar tuberosity. There is an oblique ridge, making the facets asymmetrical. The distal condyles are round, and are separated by a well-developed intercondylar sulcus. The lateral condyle is somewhat larger than the medial and slightly oblique to it. Both condyles extend beyond the ventral surface of the shaft. The collateral pits are wide and shallow, but well-developed. Both are centrally located. The lateral pit is more extensive than the medial one.
Manual unguals
Three large unguals are preserved for N. graffami ( Figs 13 View Figure 13 , 17 View Figure 17 ), a derived character. They are transversely flattened and blade like, similar to the manual unguals of Erliansaurus ( Xu et al. 2002) , Fukuivenator (Hattori et al. 2022) , and the recently described unguals from Japan Paralitherizinosaurus japonicus ( Murakami et al. 2008, Kobayashi et al. 2022). As previously noted ( Hedrick et al. 2015), the available manual unguals are laterally compressed and very trenchant. They are all more recurved than the manual unguals described for Terizinosaurus ( Barsbold 1976). All have large flexor tubercles. The proximal articulation conforms closely with the first phalanx. The flexor tubercle is enlarged relative to the lip-like extensor tubercle, the latter of which is a derived character.
Ilium
Both ilia of N. graffami (UMNH-VP 16420) are preserved ( Figs 18 View Figure 18 , 19 View Figure 19 ; Tables 8 View Table 8 , 9 View Table 9 ). The ilia and sacral vertebrae of Nothronychus were modified into a fused, apneumatic synsacrum ( Fig. 19 View Figure 19 ). The ilia are characterized by well-developed acetabulae and blades. The observed morphology is modestly dorso-ventrally flattened; however, the form is considered autapomorphically superficially similar to a bird, with a fused synsacrum and ilia, but anteroposteriorly shorter. The ilia show some degree of modest taphonomically induced lateral flare, whereas in life, they would have exhibited an unknown degree of minor ventral deflection. CT data, general right-to-left bilateral symmetry of the pre- and post-acetabular blades, and a lack of apparent damage to the neural canal, acetabulum, sacral ribs, and blades indicate minimal taphonomic distortion. Additional evidence is summarized in the geological setting above. The ilia consist of extensive bony alae that are separated by a deep notch above the acetabulum. Laterally flaring ilia are convergent with at least some alvarezsaurs, including Shuvuuia ( Chiappe et al. 2002) and Parvicursor ( Karhu and Rautian 1996) . Li et al. (2008) describe the preacetabular blade as vertical in the therizinosaur Suzhousaurus . Such an orientation would result in the ventral margin of the preacetabular blade extending beneath the acetabulum. The development of a fused synsacrum and ilium of Nothronychus may have been approached in Jianchangosaurus ( Pu et al. 2013) , Suzhousaurus ( Li et al. 2008) ( Fig. 19 View Figure 19 ), and Neimongosaurus ( Zhang et al. 2001) , where the dorsal margin is almost 30° to horizontal and at the same level as the sacral vertebrae and the pre-acetabular alae flare laterally.
In contrast to Velociraptor ( Norell and Makovicky 1997) , the preacetabular and postacetabular blades of Nothronychus are roughly the same anteroposterior length ( Liao et al. 2021). Both alae are nearly rectangular in dorsal view. The pre-acetabular blade is transversely wider than the post-acetabular one. The anterior margin of the preacetabular blade is very convex with a medial notch separating the blade from the first sacral rib. The anteroventral hook that Norell and Makovicky (1999) described for Velociraptor is expanded into a larger flange in Nothronychus . There is no trace of a medial antiliac shelf in front of the acetabulum that would serve as an origin for the M. cuppedicus ( Rowe 1986), like Neimongosaurus ( Zhang et al. 2001) . In Unenlagia , the medial antiliac shelf constitutes the boundary of the cuppedicus fossa ( Novas 2004). This ridge is also present in ornithomimids, troodontids, and Microvenator . Norell and Makovicky (1997) indicate that its absence is shared between dromaeosaurs and birds. Laterally, the ventral margin of the ilium overlies the pubic peduncle. The pubic peduncle is longer, but thinner and more concave than the ischiadic peduncle, similar to Segnosaurus ( Barsbold and Perle 1980) and Suzhousaurus ( Li et al. 2008) and in contrast to dromaeosaurs ( Norell and Makovicky 1997). As in Velociraptor , the lateral surface is slightly concave. The widest part of the acetabulum is posterior to the pubic peduncle. It remains wide ( Hedrick et al. 2015), but gradually narrows anteriorly and posteriorly. The ischiadic peduncle is rounder in cross-section and larger than the pubic peduncle, with a rugose surface. Hedrick et al. describe the ilium making up the dorsal 50% of the acetabulum. The antitrochanter is an articular surface on the lateral side of the ischiadic peduncle similar to Suzhousaurus and Erliansaurus ( Li et al. 2008) .
The acetabulum is deeply concave and oriented ventrally ( Figs 18 View Figure 18 , 19 View Figure 19 ). The pubic peduncle is more extensive, but thinner than the ischial peduncle. The cuppedicus fossa is reduced. There is a well-developed, laterally oriented supra-acetabular ridge making up the lateral margin in Nothronychus , which is convergent with sauropods ( Tsai et al. 2018). It is continuous with a ventrally oriented labrum, reversing the reduction observed in other therizinosaurs ( Tsai et al. 2018). The height of this labrum diminishes posteromedially to form a small ridge at the level of the alae. Rotation of the synsacrum increases the role of the labrum as a weight-bearing structure during the support phase ( Smith and Gillette 2023), at least partially subsuming that function of the supracetabular ridge in other non-avian theropods ( Tsai and Holliday 2015). The anterior surface of the labrum lacks any scarring to indicate the articulation with the pubis, the synovial capsule, or ligamentous attachment, so the apex is considered continuous with the capsule and ligaments. The internal surface of the labrum is smooth and deeply excavated. The posterior margin of the acetabulum consists of a short, robust process borne on the post-acetabular blade terminating with a wide, flat rugose surface.
A small transverse lamina ( Fig. 19 View Figure 19 ) within the acetabulum may mark the presence of an ossified fibrous enthesis ( Thomopoulos et al. 2012). Such structures can develop in high stress areas ( Benjamin et al. 2006) as was proposed for this region in Nothronychus ( Smith and Gillette 2023) . The anterodorsal part of the socket extends over this lamina. The excavated bone in this area is rugose, in contrast to the smoother bone over the acetabulum proper, indicative of a fibrocartilage pad ( Tsai et al. 2018). This change in texture would have marked the synovial capsule. The capsule extends posterior to the enthesis to articulate with the femoral head. The ossified pre-acetabular labrum continues ventrally past the enthesis. It made part of the iliofemoral ligament, but this attachment extended onto the enthesis.
A shallow lateral excavation marks the contact with the ischium ( Figs 18–20 View Figure 18 View Figure 19 View Figure 20 ). The antitrochanter expressed at the ischioilial contact of some theropods discussed by Tsai et al. (2018) is present in Nothronychus . The contact is an expanded joint behind the acetabulum. The acetabular soft tissue includes both intracapsular and extracapsular ligaments whose presence is indicated by scarring on the bone, osteological correlates of their existence in extant birds. A rugose surface extends anteriorly where it meets the smoother surface of the acetabulum at a clearly defined line. This line is interpreted as the posterior limit of the synovial capsule, so the rugose region is regarded as being external to the capsule. There is a textural change within this region that may have formed a boundary between a hyaline cartilage core and fibrocartilage ( Smith and Gillette 2023) as Tsai et al. (2018) described in other archosaurs. A faint antero-posterior lineation within the acetabulum is correctly anteroposteriorly oriented and is in the right location on the dorsal surface, so is suggestive of an attachment of the inner acetabular membrane onto the acetabulum. In birds, this membrane prevents medial displacement of the femoral head ( Tsai and Holliday 2015). The combined effects of the cartilage pads and acetabular tendons ( Smith and Gillette 2023) would reduce the size of the synovial capsule and leave little room for any rotational movement of the femoral head. Therefore, the femur was almost locked in place, such as observed in Numida , where protraction/retraction is limited to about 7° when walking ( Gatesy 1999).
The postacetabular blade is rotated laterally above the ischiadic peduncle such that the lateral face is oriented dorsally in Nothronychus . A thickened dorsal rim extends anteromedially from the ischiadic peduncle towards the second sacral rib. There is a well-developed supratrochanteric process dividing the pre- and postacetabular blades, which is convergent with birds ( Hutchinson 2001). The brevis fossa is marked by a ridge on the ventral surface, as seen in Alxasaurus ( Russell and Dong 1993) . The dorsal surface of the postacetabular blade is notably rugose, as in Suzhousaurus ( Li et al. 2008) and Neimongosaurus ( Zhang et al. 2001) , corresponding to the cubic projection that Barsbold and Perle (1980) described for Segnosaurus .
Pubis
Historically, Nothronychus has been regarded as possessing an opisthopubic pelvis ( Zanno et al., 2010), as defined by Barsbold (1979). This trait was recently elaborated upon by Macaluso and Tschopp (2018), as part of their argument that ventilation was a more important factor in its evolution than herbivory. The pubis is only partially retroverted, a derived character, resulting in a mesopubic condition ( Fig. 1 View Figure 1 ). The retroverted pubis in these theropods is analogous with the postacetabular bar of the pubis of ornithischians. Such a pelvis evolved multiple times, as it is not present in Falcarius ( Zanno 2010) or Jianchangosaurus (Pu et al. 2012) .
The pubis of Nothronychus ( Figs 18 View Figure 18 , 20 View Figure 20 ) is about 120% longer than the ischium, similar to Jianchangosaurus ( Pu et al. 2013) and in contrast to Gallimimus ( Osmólska et al. 1972) or Falcarius , where it is shorter. It does not contact the ischium along its entire length, but both border a large obturator foramen, with non-ankylosed contacts proximally and distally. The pubis forms an angle of roughly 130° with the ilium, less than the 155° reported for Velociraptor ( Norell and Makovicky 1997) . This agrees with the retroversion angle proposed by Macaluso and Tschopp (2018) for a mesopubic pelvis. Like Velociraptor , the proximal end of the pubis is laterally concave ( Norell and Makovicky 1997). However, the cross-section of the shaft is flattened medially, a derived character, and laterally convex, resulting in a semicircular shape longer anteroposteriorly than mediolaterally. It curves gently anteriorly, making the anterior margin concave, a derived character. The iliac peduncle is long and narrow, in contrast to the shorter, wider ischiadic peduncle. Both are quite similar to Segnosaurus ( Hedrick et al. 2015) . They combine to form the anteroventral margin of the acetabulum, as in Segnosaurus and Suzhousaurus . It is modestly thickened medially. A medial ridge, probably the attachment of the pubofemoral ligament ( Manafzadeh and Padian 2018, Smith and Gillette 2023), extends a short distance along the anterior iliac peduncle. It extends distally to form the medial margin of the acetabulum. Unlike Gallimimus ( Osmólska et al. 1972) , the iliac contact is smaller than the ischial, as is common in therizinosaurs ( Barsbold 1983). A pronounced notch is present in the ventral margin of the ischiadic peduncle, as is typical for therizinosaurs ( Hedrick et al. 2015). In contrast to the triangular contact with the ischium in Velociraptor , the contact in Nothronychus is sigmoidal. In between the peduncles, the acetabular margin is excavated, forming a groove in a transversely thickened border. It is reinforced ventrally by the craniolateral tubercle described by Hedrick et al. (2015) and Zanno (2010) for Falcarius . The pubis contributes more to the acetabulum than does the ischium. There is a longitudinal crest on the posterior margin of the shaft leading to the obturator process. In contrast to Fukuivenator (Hattori et al. 2022) , the medial surface lacks any remnant of a pubic apron, so it must have been very short, only contacting distally at the boot, resulting in an expansive pubic canal characteristic of derived theropods ( Norell and Makovicky 1997). The pubic boot extends more anteriorly than posteriorly, a derived character, similar to other therizinosaurs ( Hedrick et al. 2015), except Jianchangosaurus (Pu et al. 2012) . It is ventrally convex, with a sharp margin. The posterior portion is short and almost triangular.
Ischium
The ischium ( Figs 18 View Figure 18 , 20 View Figure 20 ) is about 80% of the tibia length, so it is relatively longer than in Deinonychus , and more similar to other theropods ( Ostrom 1969). It is shorter than the pubis, similar to Jianchangosaurus (Pu et al. 2012) . Like Velociraptor ( Norell and Makovicky 1997) there is no evidence of fusion of the adjacent ischia.
The proximal end is anteroposteriorly expanded. It makes up the posterior margin of the acetabulum. The margin is damaged, but was apparently thin, with no weight-bearing aspect. The iliac peduncle is short and rugose, and is nearly triangular in lateral view. The articular surface is rugose. It is expanded laterally to contribute to the antitrochanter, as in Tyrannosaurus ( Brochu 2003) . There is a medial flange behind the acetabulum, similar to Tyrannosaurus , that Brochu (2003) suggests may have served as the attachment for the teres ligament. A more specific attachment, however, may be related to the ischiofemoral ligament as in modern birds ( Manafzadeh and Padian 2018, Smith and Gillette 2023). The pubic peduncle is longer than the ischiadic, extending to the proximal margin of the obturator foramen. The contact with the pubis is proximally convex, but increasingly concave more distally. A rugose excavated region is present distal to the peduncle and is laterally bounded by an expanded lip. The shaft is transversely flattened and straight, both of which are derived characters. A lateral ridge extends from the base of the iliac peduncle to the level of the ischial tuberosity ( Hedrick et al. 2015) similar to that described for Tyrannosaurus ( Carrano and Hutchinson 2002) , a derived character. The posterior margin is marked by an expanded ischial tuberosity about midway along the shaft, similar to Tyrannosaurus ( Brochu 2003) , but it is more distal in position, a derived character. Barsbold (1983) described a similar flange in Segnosaurus , but it was not illustrated for Nanshiungosaurus ( Dong 1979) . The obturator process, a derived character, contacts the pubis, a derived character, resulting in a long, narrow obturator foramen, similar in position to Deinonychus ( Ostrom 1969) , and in contrast to the shorter condition in Sinraptor ( Currie and Zhao 1993) or Tyrannosaurus ( Brochu 2003) . The olecranon process is notched proximally and distally unlike Deinonychus ( Ostrom 1969) , that Hedrick et al. (2015) describe as nearly circular. They regard the distal notch as synapomorphic for Nothronychus . It is enlarged and displaced somewhat distally, as in Falcarius ( Liao et al. 2021) and in contrast to Jianchangosaurus ( Pu et al. 2013) . It is thicker at the margins than in the middle, forming a rugose surface, adjacent to the contact with the pubis. The distal end of the ischium is straight and ends abruptly, with little distal expansion, a derived character.
Femur
The femora of both N. graffami ( Figs 21 View Figure 21 , 22 View Figure 22 ; Table 10 View Table 10 ) and N. mckinleyi are anteroposteriorly flattened suggesting that they are not significantly taphonomically deformed. This morphology is reminiscent of the condition in titanosaurs, including Titanosaurus ( Wilson and Upchurch 1993) . The shaft is long and straight, in contrast to the curved femur seen in some, such as Suzhousaurus ( Li et al. 2008) . The femoral head extends from the shaft at an oblique angle of about 70°, a derived character, similar to Suzhousaurus and Alxasaurus and in contrast to other therizinosaurs (Hedrick and et al. 2015), such as Neimongosaurus ( Zhang et al. 2001) and Jianchangosaurus (Pu et al. 2012) , where it is perpendicular to the shaft. It is separated from the greater trochanter by a distinct excavation ( Zanno 2010) similar to Beipiaosaurus ( Liao et al. 2021) . Both the head and proximal surface of the metaphysis are somewhat roughened, with rugosity or striations, suggesting the presence of articular hyaline cartilage ( Tsai et al. 2018). The head is smaller than the diameter of the acetabulum, indicating considerable soft tissue between the bones. There is a modest fovea capitis, a derived character, in the anterior face of the head, indicating penetration of the ligamentum capitis. In extant birds, this architecture is related to the thickness of the articular cartilage.
The Bissekty therizinosaur material possesses an inclined head like Nothronychus , but there is no excavation separating it from the greater trochanter ( Sues and Averianov 2015). The head is almost triangular in medial view, such that it is taller than wide. In Deinonychus , the head is ovoid ( Ostrom 1976), whereas in Velociraptor it is nearly spherical ( Norell and Makovicky 1999). Unlike Deinonychus , there is no ventral lip separating the head from the neck, a derived character. The metaphysis is constricted, a derived character. A groove extends from the head along the posterior surface of the neck under the greater trochanter ( Zanno 2010), identified as the ligament groove in Falcarius and similar to that described for Velociraptor ( Norell and Makovicky 1999) . An additional groove extends perpendicular to the base of the head that could be taken as artifactual, but its presence in other theropods, including Tyrannosaurus ( Hutchinson 2001) and Velociraptor ( Norell and Makovicky 1999) indicates that it is real. The facies articularis antitrochantericus is inclined and narrower transversely than the head and greater trochanter, a derived character; it is marked by anterior and posterior excavations. The neck is longer and more distinct than in Velociraptor .
The lesser trochanter is a low cylindrical crest similar to Jianchangosaurus (Pu et al. 2012) , a derived character, in contrast to the expanded D-shaped structure described in Tyrannosaurus ( Brochu 2003) . The proximal end is displaced from the shaft on the anterior side by a short notch that is deeper than that seen in Deinonychus , and more similar to that of Sinraptor ( Currie and Zhao 1993) or Tyrannosaurus ( Brochu 2003) , a derived character. The development of this feature might be allometric. As in Sinraptor , it does not extend to the proximal end of the greater trochanter in contrast to other coelurosaurs ( Hutchinson 2001). The trochanteric shelf between the greater and lesser trochanters is reduced and indistinct, as is common in maniraptorans ( Hutchinson 2001). The accessory trochanter is represented by a rugose surface ( Hedrick et al. 2015), similar to the Bissekty therizinosaur ( Sues and Averianov 2015). It is indicated by a shallow lateral excavation on the base of the lesser trochanter ( Hutchinson 2001).
A low posterior trochanter extends from the posterolateral corner of the shaft distal to the greater trochanter similar to that in Velociraptor ( Norell and Makovicky 1999) . This structure is only preserved in the left femur. Ostrom (1976) suggests that it represents the insertion of the M. ischiofemoralis. A well-developed semicircular fourth trochanter is present more distally on the femoral shaft. The apex is about 34% of the way down the shaft. There is an excavation along the medial side, whereas the lateral side is flat. The fourth trochanter is more pronounced than in Erliansaurus ( Xu et al. 2002) .
In contrast to the curved femora of Deinonychus ( Ostrom 1976) , Velociraptor ( Norell and Makovicky 1999) , Falcarius ( Zanno 2010) , Alxasaurus ( Russell and Dong 1993) , or Suzhousaurus ( Li et al. 2008) , the shaft is very straight, as in Segnosaurus ( Perle 1979) . The shaft is marked by long, indistinct anterior and posterior grooves. There may be an oblique posterior intermuscular line extending from the base of the posterior trochanter to the medial condyle ( Baumel and Witmer 1993), as is common in archosaurs ( Hutchinson 2001). An additional anterior intermuscular line ( Baumel and Witmer 1993) extends distally from the base of the lesser trochanter becoming indistinct distally. In tyrannosaurs ( Hutchinson 2001) and extant birds ( Baumel and Witmer 1993), this line wraps to the medial condyle, so its apparent orientation in Nothronychus is taken as artifactual.
The distal end has well-developed distal condyles separated posteriorly by a deep intercondylar fossa, but the fossa is not so deep as in Neimongosaurus ( Zhang et al. 2001) . In lateral view, the lateral condyle is larger than the medial one, as in ostriches.In contrast to the Bissekty therizinosaur ( Sues and Averianov 2015), Nothronychus lacks the proximal step bordering the intercondylar fossa. A shallow excavation divides the condyles anteriorly. The condyles are not connected across the intercondylar fossa, but there is a medial boss that may correspond with the rugosities described for Tyrannosaurus ( Brochu 2003) . The medial condyle is more robust than the wedge-shaped lateral condyle, as observed by Hedrick et al. (2015). Like the Bissekty therizinosaur femur ( Sues and Averianov 2015), the lateral condyle of Nothronychus is confluent with the shaft. The medial condyle is more expanded posteriorly, but anteriorly is almost continuous with the shaft, as in Tyrannosaurus ( Brochu 2003) . The lateral condyle supports an oval crista tibiofibularis. A deep groove, referred to as the fibular trochlea in Tyrannosaurus ( Brochu 2003) , separates the crista from the lateral epicondyle and accommodates the head of the fibula in Gallimimus ( Osmólska et al. 1972) . An enlarged mediodistal crest extends proximally from the anteromedial corner of the medial condyle as in Carnotaurus ( Bonaparte et al. 1990) , a derived character. Hutchinson (2001) states that this character is typical in the femora of non-maniraptoran theropods, but it may be more widespread in nonavian theropods ( Brochu 2003). The lateral epicondyle is larger than the medial epicondyle.
As in the ostrich ( Chadwick et al. 2014), four stabilizing tendons are inferred in the knee of Nothronychus ( Smith and Gillette 2023) . The medial collateral ligament extended from the medial epicondyle to the ankle. The lateral collateral ligament originated at the lateral epicondyle and inserted on the fibular epiphysis and lateral side of the meniscus. The anterior cruciate ligament attached at a knob within the popliteal fossa and extended to the tibial eminence. There is a deep pit on the medial side of the lateral condyle. In ostriches, it marks the attachment for the posterior cruciate ligament. The ligament crosses over to insert on the posteromedial corner of the eminence. The medial meniscus is thicker than the lateral meniscus in ostriches ( Chadwick et al. 2014), and this was probably the case for Nothronychus .
Tibia
The left tibia ( Figs 21 View Figure 21 , 22 View Figure 22 ; Table 10 View Table 10 ) is preserved in N. graffami (UMNH-VP 16420). The proximal end is distorted so that the crests are rotated relative to the shaft and the cnemial crest is folded into the shaft. The tibia is about the same length as the femur in contrast to Fukuivenator (Hattori et al. 2022) and Jianchangosaurus (Pu et al. 2012) , where it is longer. It is flattened anteroposteriorly, but all specimens exhibit the same condition, so it probably not taphonomic. The longest metatarsal is about 30% of the length of the tibia, a lower ratio than that reported for Deinonychus , decisively implying short hindlimbs and that Nothronychus was not a cursorial animal ( Ostrom 1969, 1976).
The proximal end of the tibia is very similar to that of Velociraptor ( Norell and Makovicky 1999) . This specimen of Nothronychus apparently possesses a much narrower proximal endthandescribedfortheBissektytherizinosaur, butisdeformed ( Sues and Averianov 2015). Retrodeformation as proposed here makes the proximal end of the tibia in Nothronychus almost triangular and similar to Neimongosaurus ( Zhang et al. 2001) . The proximal end is flared with a cnemial crest extending proximally beyond the shaft to form the apex, similar to Velociraptor and in contrast to Gallimimus ( Osmólska et al. 1972) and the Bissekty therizinosaur ( Sues and Averianov 2015). Therefore, this, and not the lateral condyle, constitutes the apex of the tibia, as in derived therizinosaurs ( Hedrick et al. 2015). Like the Bissekty therizinosaur, the cnemial crest is oblique, rather than vertical. It possesses the lateral boss separated from the anterior margin as described for Velociraptor . Nothronychus lacks the notches adjacent to the crest described in Tyrannosaurus ( Brochu 2003) , but it is excavated anterolaterally constituting a pronounced incisura tibialis to accommodate the proximal end of the fibula, but like Velociraptor , there is no lateral hook on the cnemial crest. Continuing anterolaterally, the fibular contact is concave, like the Bissekty therizinosaur ( Sues and Averianov 2015). The lateral and medial condyles are separated by an oblique sulcus. As observed by Hedrick et al. (2015), the medial condyle is larger than the lateral one.
Like Velociraptor ( Norell and Makovicky 1999) , the shaft is straight, but with a slight medial curvature. The posterior groove in the shaft described by Hedrick et al. (2015) was observed in both specimens. The fibular crest is long, extending most of the way down the shaft, a derived character, but low and thin, similar to the condition in ornithomimids ( Osmólska et al. 1972), and in contrast to Tyrannosaurus , where it is laterally expanded, but much shorter ( Brochu 2003). It retains the rugose margin associated with the interosseous ligament.
The tibia of Nothronychus flares distally, very similar to the Bissekty therizinosaur ( Sues and Averianov 2015). Hedrick et al. (2015) observed that the medial malleolus is somewhat larger than the distal one. There is a triangular facet anteriorly associated with the astragalus. Nothronychus lacks a distal groove or fibular fossa laterally, as in Gallimimus ( Osmólska et al. 1972) or Tyrannosaurus ( Brochu 2003) , indicating the fibula did not significantly support the ankle.
Fibula
The fibula of N. graffami ( UMNH 16420) is nearly complete ( Figs 21–23 View Figure 21 View Figure 22 View Figure 23 ; Table 10 View Table 10 ). It is very thin and broken distally. This element did not significantly contribute to weight bearing at the ankle. The proximal end is wider anteriorly in proximal view, gradually narrowing posteriorly. The articular surface generally slopes posteriorly as in Alxasaurus ( Russell and Dong 1993) and Erliansaurus ( Xu et al. 2002) . It does not expand as abruptly as in Velociraptor ( Norell and Makovicky 1999) and Erliansaurus . The lateral proximal end is convex, whereas the medial end is shallowly concave, a derived character. A small anterior fossa is present on the proximal end. The medial surface is more deeply excavated distally, but a deep fibular fossa, as in ornithomimids ( Norell and Makovicky 1999) and Deltadromaeus ( Sereno et al. 1996) is absent. Anteriorly, a flange accommodates the tibia ( Hedrick et al. 2015), but this appears to be individually variable in size. The posterior head is not significantly elevated over the anterior in contrast to Erliansaurus ( Xu et al. 2002) .
The shaft is straight. It contacts the tibia in three places, at the proximal end, at the distal end, and again midlength at the fibular crest of the tibia, resulting in avian-like proximal and distal interosseous foramina ( Baumel and Witmer 1993, Brochu 2003). A rugose medial crest, that becomes sharper distally, suggests the presence of an interosseous ligament between the fibula and tibia. There is a pronounced anterolateral crest about midlength that has been described as the insertion for the M. iliofibularis ( Hedrick et al. 2015, Smith 2021), a derived character. From there, the shaft gradually thins. A deep sulcus extends along the anteromedial side in contrast to the circular cross-section described for Deinonychus ( Ostrom 1969) and probably corresponds with the cleft described in Tyrannosaurus ( Brochu 2003) , but extends more proximally in Nothronychus . Therefore, this cleft cannot be considered diagnostic for tyrannosaurs ( Mader and Bradley 1989, Brochu 2003) In other therizinosaurs, the fibula extends to the end of the tibia (Hedrick and et al. 2015), and this morphology may be present in Nothronychus .
Astragalus
There is no indication of fusion of any of the tarsal bones. The astragalus ( Figs 21–23 View Figure 21 View Figure 22 View Figure 23 ; Table 10 View Table 10 ) bears a short, wide ascending process similar to Deinonychus ( Ostrom 1969) . In contrast to Velociraptor , the astragalus does not wrap around to the posterior face of the tibia. The process is separated from the body by a shallow oblique groove laterally, but is more confluent medially in contrast to the deeper sulcus in Velociraptor ( Norell and Makovicky 1999) . In contrast to Velociraptor , the process is deeply notched medially and laterally. The lateral notch results in the reduction of the lateral condyle ( Zanno et al. 2009). In Gallimimus , it is only notched laterally ( Osmólska et al. 1972). The articular surface is composed of a medial and lateral condyle, with the lateral being larger. They are divided by a shallow intercondylar sulcus.
Distal tarsal IV
The element referred to a metatarsal V ( Zanno et al. 2009) is re-interpreted here as distal tarsal IV. This element is a small semicircular bone ( Fig. 23 View Figure 23 ) similar to Deinonychus ( Ostrom 1969) and it is not fused to the astragalus. It is generally convex laterally and gently excavated medially. The distal tarsal is thickest anteriorly, becoming progressively thinner posteriorly and distally. Overall, it compares well with the distal tarsals of Fukuivenator (Hattori et al. 2022) and ornithomimids ( Nottrodt and Farke 2021). A short anteroposterior spine extends proximally.
Pes
Most of the pes is preserved with minimal distortion ( Table 11 View Table 11 ).
Metatarsals
Metatarsals I–IV are all present ( Figs 24 View Figure 24 , 25 View Figure 25 ; Table 11 View Table 11 ). Generally, the metatarsals of Nothronychus are relatively shorter than those of Velociraptor ( Norell and Makovicky 1997) , a derived character. The metatarsals become successively longer from metatarsal I to IV ( Hedrick et al. 2015). They are tightly packed ( Fig. 25A View Figure 25 ) with metatarsal III which is the posteriormost element, broadly similar to Sinraptor ( Currie and Zhao 1993) and Velociraptor ( Norell and Makovicky 1997) . The articulation figured by Smith and Gillette (2023) is probably incorrect, indicating MT III is anterior to II and IV. This revised configuration would result in a pronounced avian-like hypotarsal sulcus ( Baumel and Witmer 1993). Metatarsal V may have been present, but is not preserved.
Metatarsal I
Metatarsal I ( Figs 24 View Figure 24 , 25 View Figure 25 ) is much shorter than II–IV ( Hedrick et al. 2015). It contacts metatarsal II only proximally at a triangular facet, significantly contributing to the proximal articular surface, similar to Neimongosaurus ( Zhang et al. 2001) and in contrast to the condition in Velociraptor and most theropods, notably including Falcarius ( Zanno 2010) and Beipiaosaurus ( Liao et al. 2021) , where it attaches midway down the metatarsal II shaft ( Norell and Makovicky 1997). Barsbold and Perle (1980) figure metatarsal I articulating with metatarsal II somewhat distal from the articular cotyle in Segnosaurus , but this configuration is unlikely for Nothronychus , as the articular facet on metatarsal II shows a proximal contact. Therefore, metatarsal I would fit tightly, behind the medial proximal end of metatarsal II. Metatarsal I would have diverged medially from metatarsal II ( Hedrick et al. 2015), as in Segnosaurus ( Barsbold and Perle 1980) and Terizinosaurus ( Perle 1982). As a result, it might not normally have been weight bearing, therefore Nothronychus might not have had a tetradactyl pes. This topic merits further study, especially as more material is uncovered.
The proximal articular surface is laterally inclined and excavated, similar to Neimongosaurus ( Zhang et al. 2001) . A long flange on the posterior articular surface suggests an additional insertion of the M. gastrocnemius pars medialis transferred to metatarsal I on an accessory medial hypotarsal crest. A lateral crest extends distally from the base of the facet, suggesting a ligamentous connection with metatarsal II. The proximal end is anteriorly convex and posteriorly flat, becoming excavated distally. The shaft narrows distally, so the thinnest point is proximal to the distal trochlea. The extensor pit is poorly defined. Distally, the trochlea is expanded and well developed. It is ginglymous, with a wedge-shaped medial condyle separated from the larger lateral condyle by a wide intercondylar sulcus only posteriorly ( Hedrick et al. 2015) and in contrast to the bulbous condyles described for Neimongosaurus ( Zhang et al. 2001) . The medial condyle extends posteromedially from the lateral condyle. In contrast to the Bissekty therizinosaur ( Sues and Averianov 2015), the lateral condyle does not extend distally past the medial one in Nothronychus . The lateral collateral pit is deeper than the medial one, in contrast to its absence in Velociraptor ( Norell and Makovicky 1997) . The intercondylar sulcus extends proximally into the posterior face of the shaft, presumably incorporating a deep supratrochlear plantar fossa.
Metatarsal II
The second metatarsal ( Figs 24 View Figure 24 , 25 View Figure 25 ) is crushed along the posterior surface of the shaft. Like the condition in Velociraptor (Norell and Makovicky 1977) , it is parallel to metatarsal III, but not tightly appressed to it along its length. The shaft is very straight and almost the same length as metatarsal III, which is unusual for theropods ( Ostrom 1969). In Segnosaurus , metatarsal IV is about the same length as III ( Perle 1979).
The proximal end is anteroposteriorly wide, extending posteriorly to much of the proximal third metatarsal ( Hedrick et al. 2015). There are no well-defined intermediate hypotarsal crests. There is a wide posterior crest representing the medial hypotarsal crest serving as a partial insertion for the M. gastrocnemius pars medialis on the proximal end of metatarsal II in Nothronychus . A medially directed flange extending in front of the contact for metatarsal I is interpreted as a proximally located medial plantar crest. Medial to the intermediate hypotarsal sulcus, there is a clear facet for the closely appressed metatarsal III. Anterolaterally, the proximal end overhangs the shaft, as in Velociraptor ( Norell and Makovicky 1997) . The surface beneath the overhang is excavated and faintly rugose.
The shaft narrows below the proximal end, such that the narrowest portion is at the midlength of the shaft. The shaft has an oval cross-section. A longitudinal sulcus extends along the anterior face of the shaft which is associated with a pronounced lateral crest, a derived character. The crest extends distally to join a wide extensor pit. The crest is well developed and also described for Velociraptor ( Norell and Makovicky 1997) as an elongate tubercle. They suggest that it served as the insertion for the M. tibialis anterior, as in birds ( George and Berger 1966, O’Connor et al. 2014). There is a transverse anterior extensor pit above the distal end of the articular surface that is probably real. The distal ginglymous articulation, a derived character, is wider than long, similar to some birds ( O’Connor et al. 2014). This ratio may be partially a result of damage, but Segnosaurus has a similar ratio ( Barsbold and Perle 1980). There are two distal condyles present ( Hedrick et al. 2015) that merge anteriorly. They are asymmetrical, larger laterally than medially, which is similar to Falcarius ( Zanno 2010) . The medial condyle is wedge shaped and oblique to the shaft, whereas the lateral condyle is more bulbous. The intervening sulcus is very shallow and is only seen posteriorly. The sulcus extends into the posterior surface of the shaft to form part of a supratrochlear plantar fossa homologous to that in birds ( Baumel and Witmer 1993). The collateral pits on either side are shallow, but the lateral pit is deeper and better defined than the medial one. The medial pit, however, is longer. A shallow sulcus extends from the collateral pit distally past the medial condyle. This excavation is abraded, but this fossa is seen in at least some birds (O’Connor and et al. 2014).
Metatarsal III
The third metatarsal ( Figs 24 View Figure 24 , 25 View Figure 25 ) is longer than metatarsal II, unlike the condition in Segnosaurus , where the major metatarsals are all roughly equal in length ( Perle 1979). The anterior side is well-preserved, but the posterior face is crushed. The proximal end is not arctometatarsalian (Holtz 1994). It reaches its greatest dimension oblique to the tarsometatarsal articular surface. The anterior surface is flat, with shallow, longitudinal striations. The anterior tubercle described for dromaeosaurs by Norell and Makovicky (1997) is also present in Nothronychus , dominating the proximal anterior face. The cotyle overhangs the shaft medially, but not laterally. A ridge extends distally from this structure. Another ridge separates the anterior face from the metatarsal II facet. There is a long contact with metatarsal II. The posterior margin makes up much of the hypotarsal sulcus. Similar to Velociraptor ( Norell and Makovicky 1997) , the proximal metatarsal III thins laterally so that the posterior face is level with metatarsal II, continuing the hypotarsal sulcus. It is quite flat in this area, similar to some birds ( O’Connor et al. 2014). There is a wide, concave facet marking the contact with metatarsal IV, stopping below the proximal end, in contrast to the condition in Velociraptor ( Norell and Makovicky 1997) where it nearly meets the distal end, but the distal excavation is subdivided by a longitudinal crest. A lateral crest extends behind metatarsal IV a short distance.
In contrast to Velociraptor ( Norell and Makovicky 1997) , the distal shaft of Nothronychus is straight. The shaft gradually expands mediolaterally to meet the trochlea, so that it is thinnest midway down the shaft. The trochlea is longer transversely than anteroposteriorly, but this may be partially a result of taphonomic deformation. There is an extensive, triangular extensor pit on the anterior face. Distally, there is no indication of a ginglymoid trochlea, as in many theropods other than dromaeosaurs ( Norell and Makovicky 1997). It completely lacks an intercondylar sulcus and there are no separated distal condyles ( Hedrick et al. 2015) in contrast to Segnosaurus ( Perle 1979) . The lateral collateral pit is wider, but shallower, than the medial one. A notch extends from the lateral collateral pit to the distal end.
Metatarsal IV
Metatarsal IV is well preserved ( Figs 24 View Figure 24 , 25 View Figure 25 ), but the posterior face is crushed. It is longer than the other metatarsals. The proximal end is widest mediolaterally. There is a large tubercle overhanging the anterior face, corresponding with the rugosity described in Tyrannosaurus , which is possibly associated with the tibiocranialis tubercle ( Brochu 2003). The tubercle is more prominent in Nothronychus than Tyrannosaurus and, in contrast, is more prominent on IV than II. The cotyle is gently concave medial to the tubercle. This excavation extends distally down the shaft, gradually diminishing, to end above the extensor pit. The facet for metatarsal III is anteriorly bounded by an extended crest that overlays much of the third metatarsal articular surface. Posteriorly, the proximal end forms a wide, low lateral hypotarsal crest. A lateral crest extends distally from the hypotarsal crest and stops above the midpoint of the shaft similar to Velociraptor that Norell and Makovicky (1997) referred to as the lateral plantar crest of birds. There is a reduced faint excavation between the lateral hypotarsal crest and the proximal anterior tubercle that probably represents the contact with a reduced metatarsal V. The shaft of metatarsal IV is parallel to that of metatarsal III, but they do not contact past the proximal articulation. Distal to the proximal end, the shaft gradually narrows such that it is thinnest about midway down. At this point it expands to meet the distal trochlea. This morphology is in contrast to that described for Velociraptor , where the narrowest point is just proximal to the trochlea ( Norell and Makovicky 1997). The trochlea is not ginglymous, as it consists of a single condyle with no intercondylar sulcus, similar to some birds ( O’Connor et al. 2014). It is longer transversely than anterolaterally, as in Segnosaurus ( Barsbold and Perle 1980) . The extensor pit is wide and triangular, but it extends medially to the distal end of the trochlea. It is separated from the medial collateral pit by a short longitudinal crest. The medial collateral pit is deeper and better defined than the lateral one. A short, posteriorly directed crest bounds the lateral collateral pit, described by Hedrick et al. (2015) as an incipient lateral condyle.
Phalanx I-1
Phalanx I-1 is quite robust ( Figs 24 View Figure 24 , 26 View Figure 26 ). This element exhibits the greatest dimensions proximally. The proximal articular facet is oval and is gently concave. The ventral surface is rugose proximally with two longitudinal ridges that curve to meet about a third of the way down the shaft. A wide, shallow flexor pit is present above the trochlea. The shaft is oval in cross-section, wider than high. In dorsal view, the shaft gradually narrows distally to meet the distal trochlea, but in lateral view it is constricted about mid-shaft. The medial side is straight, but the phalanx is somewhat concave medially. There is a shallow extensor pit on the dorsal surface above the distal trochlea. The distal trochlea is ginglymus. Both condyles are enlarged and completely separated by a deep intercondylar sulcus. The lateral condyle is inclined about 10° relative to the shaft. The medial condyle is bulbous. Both collateral pits are deep.
Phalanx II-1
Phalanx II-1 ( Figs 24 View Figure 24 , 26 View Figure 26 ) is well preserved.The shaft is minimally constricted above the pronounced ginglymus with enlarged condyles, as in Falcarius ( Zanno 2010) . In contrast to troodontids ( Norell and Makovicky 1997), the distal end is wider than tall. The intercondylar sulcus is very deep, as in Deinonychus ( Ostrom 1969) . It extends dorsally and ventrally relative to deep pits in the shaft. The dorsal pit is partially separated from the sulcus by a transverse ridge. Zanno (2010) ascribed it to a large extensor ligament pit in Falcarius . The resulting ventral groove is deeper than the dorsal one. Ventral and dorsal crests adjacent to the grooves medially and laterally probably connect the proximal articular region to the respective condyles. The medial crest is deeper than the lateral one. The condyles are very circular, permitting considerable extension and flexion of phalanx II-2. Ventral flexion would greatly exceed that modelled for Deinonychus , where minimal flexion is seen ( Ostrom 1969). The medial condyle is somewhat more bulbous than the lateral one. Wide, shallow collateral pits are present in the middle of each condyle, in contrast to the more dorsal location in Velociraptor ( Norell and Makovicky 1997) .
Phalanx II-2
Phalanx II-2 is wide ( Figs 24 View Figure 24 , 26 View Figure 26 ). The shaft is dorso-ventrally constricted, but less so than Velociraptor ( Norell and Makovicky 1997) and Deinonychus ( Ostrom 1969) . The dorsal and ventral surfaces are gently excavated, but the extensor pit is indistinct. The proximal articular surface is roughly tetragonal, with a shallow excavation in the base. Intercondylar tuberosities are present in Nothronychus , but not exaggerated as described for dromaeosaurs and troodontids ( Currie and Peng 1993). The articular surface is subdivided by an oblique ridge between the tuberosities. The resulting medial facet is taller than the lateral one. Phalanx II-2 fits tightly with phalanx I-1, such that extension and flexion are permitted, but little or no lateral or medial movement. The distal condyles are circular and enlarged over the shaft. The lateral condyle is inclined about 10° towards to the medial one. Both condyles extend further ventrally than dorsally. They are marked by collateral pits on either side positioned closer to the dorsal surface than the ventral. An intercondylar sulcus wider ventrally, narrowing dorsally is present.
Phalanx II-2
Phalanx II-2 ( Figs 24 View Figure 24 , 26 View Figure 26 ) of Nothronychus is roughly the same length as II-1, but somewhat more slender. The shaft is more dorsoventrally constricted than transversely. There is a dorso-ventral ridge subdividing the proximal articulation, but it is displaced laterally so the medial facet is larger than the lateral one. Short dorsal and ventral intercondylar buttresses extend proximally past the joint surface. The distal condyles are nearly circular and marked by shallow collateral facets. The lateral facet is of a smaller diameter than the medial one. The two condyles are inclined such that they are closer dorsally than ventrally. A well-developed intercondylar sulcus partially separates the condyles dorsally, distally, and ventrally. It expands dorsally and ventrally into extensor and flexor fossae.
Phalanx III-1
Phalanx III-1 ( Figs 24 View Figure 24 , 26 View Figure 26 ) of Nothronychus is thicker than Velociraptor ( Norell and Makovicky 1997) , but still longer than phalanx III-2. The shaft is dorso-ventrally but not medio-laterally constricted. The proximal articulation is somewhat concave, lacking any subdividing ridge. There is no basal excavation, so the base is very flat. Distally, the trochlea is well developed with enlarged medial and lateral condyles. The intercondylar sulcus is transversely wide. It continues into deep extensor and flexor pits in the shaft. The lateral condyle is oval and extends more proximally ventrally than dorsally, whereas the medial condyle is nearly circular. The collateral pits are wide and shallow, situated in the middle of the condyles.
Phalanx III-2
This element is not preserved.
Phalanx III-3
This element is not preserved.
Phalanx IV-1
Phalanx IV-1 is thick and straight ( Figs 24 View Figure 24 , 27 View Figure 27 ), in contrast to the medially bowed element of Velociraptor ( Norell and Makovicky 1997) . The shaft is modestly dorsoventrally constricted. The proximal articulation is concave and lacks any subdividing vertical ridge. There are two low tubercles laterally and medially and the base is faintly excavated. The distal trochlea is well developed. The lateral condyle is inclined and more wedge shaped than the medial one. The condyles are separated by a wide intercondylar sulcus. Both condyles are marked by well-developed collateral pits positioned somewhat dorsally. There are shallow extensor and flexor pits in the shaft.
Phalanx IV-2
There is some indication that digit IV diverged laterally from digit III at the base of IV-2. There is, however, no suggestion of a medial distal curve, in contrast to Velociraptor ( Norell and Makovicky 1997) . Phalanx IV-2 ( Figs 24 View Figure 24 , 27 View Figure 27 ) is considerably shorter than IV-1. The shaft is almost non-existent, but is wider than tall. The proximal articulation is subdivided by a dorsoventral ridge into two nearly symmetrical facets. The ridge extends to the highest point in the facet. The base is marked by short lateral and medial tubercles. The apex extends proximally to constitute a short dorsal intercondylar tuberosity. Distally, the trochlea is well developed. The lateral condyle is more wedge shaped than the medial one. They are separated by a wide intercondylar sulcus that extends to the extensor and flexor pits. The condyles are semicircular and marked by shallow collateral pits. The pits are located in the middle of the respective condyles.
Phalanx IV-3
Phalanx IV-3 ( Figs 24 View Figure 24 , 27 View Figure 27 ) is shorter than IV-2. The shaft is almost non-existent. The proximal articulation is nearly circular. It is subdivided by a low vertical ridge into two nearly symmetrical facets. The apex, but not the base, bears a dorsal intertubercular buttress extending proximally. The shaft is wider than tall. Distally, the lateral condyle is larger than the medial one. The condyles are separated by a wide, shallow intercondylar sulcus. The extensor and flexor pits are faint. The associated collateral pits are shallow and faint in contrast to Falcarius ( Zanno 2010) and are located in the middle of the respective condyles.
Phalanx IV-4
The last pre-ungual phalanx of digit IV is very short ( Figs 24 View Figure 24 , 27 View Figure 27 ). The proximal articular surface is subdivided into lateral and medial facets by a vertical ridge. The apex bears a very large dorsal intertubercular buttress, as in Velociraptor ( Norell and Makovicky 1997) . Distally, the trochlea consists of two poorly-developed condyles separated by a wide, shallow intercondylar sulcus. Extensor and flexor pits are very faint. The collateral pits are unclear.
Table 8. Dimensions (mm) of sacrum of N. graffami
Sacrum | |
---|---|
Craniocaudal length of sacral vertebrae in articulation | 525 |
Maximum transverse diameter, anterior iliac wings | 1080 |
Maximum transverse diameter, posterior iliac wings | 813 |
Transverse diameter of sacrum at outer edges of acetabulae | 564 |
Craniocaudal length of left acetabulum | 101 |
Transverse diameter of left acetabulum | 107 |
Transverse diameter of right acetabulum | 125 |
No known copyright restrictions apply. See Agosti, D., Egloff, W., 2009. Taxonomic information exchange and copyright: the Plazi approach. BMC Research Notes 2009, 2:53 for further explanation.
Kingdom |
|
Phylum |
|
Class |
|
Order |
|
Family |