Reconstruction of the pelvic girdle and hindlimb musculature of the early tetanurans Piatnitzkysauridae (Theropoda, Megalosauroidea)

Abstract Piatnitzkysauridae were Jurassic theropods that represented the earliest diverging branch of Megalosauroidea, being one of the earliest lineages to have evolved moderate body size. This clade's typical body size and some unusual anatomical features raise questions about locomotor function and specializations to aid in body support; and other palaeobiological issues. Biomechanical models and simulations can illuminate how extinct animals may have moved, but require anatomical data as inputs. With a phylogenetic context, osteological evidence, and neontological data on anatomy, it is possible to infer the musculature of extinct taxa. Here, we reconstructed the hindlimb musculature of Piatnitzkysauridae (Condorraptor, Marshosaurus, and Piatnitzkysaurus). We chose this clade for future usage in biomechanics, for comparisons with myological reconstructions of other theropods, and for the resulting evolutionary implications of our reconstructions; differential preservation affects these inferences, so we discuss these issues as well. We considered 32 muscles in total: for Piatnitzkysaurus, the attachments of 29 muscles could be inferred based on the osteological correlates; meanwhile, in Condorraptor and Marshosaurus, we respectively inferred 21 and 12 muscles. We found great anatomical similarity within Piatnitzkysauridae, but differences such as the origin of M. ambiens and size of M. caudofemoralis brevis are present. Similarities were evident with Aves, such as the division of the M. iliofemoralis externus and M. iliotrochantericus caudalis and a broad depression for the M. gastrocnemius pars medialis origin on the cnemial crest. Nevertheless, we infer plesiomorphic features such as the origins of M. puboischiofemoralis internus 1 around the “cuppedicus” fossa and M. ischiotrochantericus medially on the ischium. As the first attempt to reconstruct muscles in early tetanurans, our study allows a more complete understanding of myological evolution in theropod pelvic appendages.

However, in alternative phylogenies (Rauhut & Pol, 2019;Schade et al., 2023), Piatnitzkysauridae is considered to be an early divergent clade of Allosauroidea.Evolutionary implications related to the locomotor skeletal system of Tetanurae based on the alternative position of Piatnitzkysauridae were discussed by Lacerda et al. (2023).
Piatnitzkysauridae is a key clade for understanding the evolution of tetanuran theropods because they are the earliest and oldest known members of this clade (Carrano et al., 2012;Rauhut et al., 2016).
Piatnitzkysaurid species can be diagnosed, for example, by the following morphological features: (1) short or absent anterior maxillary ramus, (2) presence of two parallel rows of foramina on the maxilla, (3) vertically striated paradental plates, and (4) anteriorly inclined neural spines of the posterior dorsal vertebrae (further details in Carrano et al., 2012).The first cladistic studies that phylogenetically positioned and characterized these species as a clade were Benson (2010) and Carrano et al. (2012), who included the piatnitzkysaurids within the clade Megalosauroidea, differing from other approaches.Historical classifications generally had assigned Marshosaurus and Piatnitzkysaurus as members of allosaurids or megalosaurids (e.g., Bonaparte, 1979Bonaparte, , 1986;;Russell, 1984; for a summary see Carrano et al., 2012).

, whereas
Condorraptor is known from a probably subadult specimen (Rauhut, 2005).The estimated typical body length of the three species is 4.5 m, with a body mass of about 200 kg for Marshosaurus and Condorraptor; whereas the body mass of Piatnitzkysaurus was estimated as 275 kg (Paul, 1988(Paul, , 2016)).Hendrickx et al. (2015) estimated longer body lengths, between 5 and 6 m, and Foster (2020) estimated a slightly greater body mass for Marshosaurus (250 kg).Nevertheless, one estimate of body mass, which differs from the others, as it is based on femoral morphometrics, suggests that the Argentinean taxa Condorraptor and Piatnitzkysaurus could have reached ~360 and 750 kg in mass, respectively; exemplifying the origin of medium-sized tetanurans during the Jurassic and thus suggesting an increase in theropod macropredatory habits (Benson et al., 2014).

| MUSCLE RECON S TRUC TI ON IN E X TIN C T VERTEB R ATE S
Reconstruction of muscles and estimation of their architecture and functions is an important approach in palaeobiology (e.g., Bates & Falkingham, 2018;Bishop et al., 2021;Cuff, Demuth et al., 2023;Witmer, 1995).Even with intrinsic limitations to these reconstructions for fossil organisms, biomechanical models and simulations; and other useful methods; have been developed with the aid of computational techniques (e.g., Hutchinson, 2012).Advances in morphofunctional and ecomorphological studies in extinct vertebrates, together with advances in evolutionary biomechanics applied to locomotion, for example, are essential for understanding broader macroevolutionary aspects such as paleoecology and potential selective pressures (e.g., Jones et al., 2021).
A methodology that has been widely used in recent decades is the Extant Phylogenetic Bracket (EPB), formalized by Witmer (1995).
The EPB is based on the phylogenetic relationships of the extinct clade under study, with at least two evolutionarily outgroups having extant representatives.The EPB method represents a rigorously explicit method that aims to minimize speculations in muscle reconstruction, allowing tissue reconstruction to be performed and then judged through inference levels (see Section 4 below).Additionally, the inclusion of fossil taxa facilitates interpretations about muscular homology and evolution, because extinct relatives of the study taxon may present evidence for transitional character states or even novel states; either of these being absent in extant taxa (Bishop et al., 2021;Dilkes, 2000;Hutchinson, 2001aHutchinson, , 2001bHutchinson, , 2002;;Maidment & Barrett, 2011).

| WHY S TUDY MUSCUL ATURE IN
NON -AVIAN THEROP ODS? Hutchinson and Allen (2009) listed at least four questions considered fundamental for the understanding of macroevolution and morphofunctional adaptations that support and motivate researchers to reconstruct the musculature and locomotor aspects in theropod dinosaurs: (1) how did the bipedal stance and gait of birds evolve?
(2) what myological/locomotor traits are novel for birds?(3) how far down the phylogenetic tree is it possible to trace ancestral traits in theropods (or other archosaurs), and what are the plesiomorphic traits?and ( 4) how did novelties such as bipedalism and flight arise and/or were modified, or even how did the performance of terrestrial/aerial locomotion change over evolutionary time?
In addition to the studies cited above, there is an ongoing effort to understand the main evolutionary features related to bipedalism in theropod dinosaurs (e.g., Allen et al., 2021;Bishop, Hocknull, Clemente, Hutchinson, Barret, et al., 2018;Bishop, Hocknull, Clemente, Hutchinson, Farke, et al., 2018;Cuff, Demuth, et al., 2023).However, the earliest tetanuran clades studied generally include only allosauroids; whereas there have not been detailed studies of Megalosauroidea, the earliest-diverging branch of tetanu- Although piatnitzkysaurids are important representatives for understanding theropod evolution (Carrano et al., 2012;Rauhut, 2003), as well as tetanuran diversity and the acquisition of larger body size in terms of locomotor function and body support, little is known about these paleobiological issues (Lacerda et al., 2023).Our aim here is to begin addressing these deficiencies by reconstructing the hindlimb muscles (origins and insertions) of the three piatnitzkysaurid species (Condorraptor, Marshosaurus, and Piatnitzkysaurus), and to compare our findings with the myological reconstructions of other extinct and extant archosaurs.We chose these taxa not only for: (1) future usage in biomechanical models, and ( 2) comparisons with existing myological reconstructions of other theropods and resulting evolutionary implications, but also (3) addressing how similar their musculature might have been, ( 4) determining if any show unusual apomorphies, and (5) assessing how differential taphonomic preservation affects these inferences.We considered a total of 32 muscles, focusing on the major muscles (not the many, small, complex pedal muscles).

| Myological reconstruction, homology and character mapping
We used the EPB method (Witmer, 1995)  represents an equivocal reconstruction, when the ancestral condition for two or more taxa is ambiguous, such as the presence of a particular soft tissue and the osteological correlate only in one of the extant taxa; (III) represents an unequivocal absence of a particular feature, that is, when the ancestral condition favoured by the EPB involves not having the soft tissue and its osteological evidence (i.e., inferring an absent feature; with no or contrary evidence).In addition, if soft tissue inferences lack conclusive data from their osteological correlates, they are qualified as level I′, II', and III' inferences (Witmer, 1995).Using the EPB, our comparisons mainly were based on Crocodylia and Aves, but not restricted to these groups; Lepidosauria and Testudines were also considered (Bishop et al., 2021;Hutchinson, 2002).The pelvic and thigh musculature of extant taxa was evaluated from the following literature on Crocodylia (e.g., Hattori & Tsuihiji, 2021;Otero et al., 2010;Romer, 1923a;Suzuki et al., 2011;Wilhite, 2023), Avialae (e.g., Clifton et al., 2018;Hattori & Tsuihiji, 2021;Hudson et al., 1959;Meso et al., 2021;Patak & Baldwin, 1998;Picasso, 2010;Romer, 1923c;Rowe, 1986;Suzuki et al., 2014), and other Tetrapoda/Reptilia (e.g., Dick & Clemente, 2016;Gregory & Camp, 1918;Hattori & Tsuihiji, 2021;Romer, 1942).Dissection of one Crocodylus niloticus and one Numida meleagris specimen during this study further enhanced our musculoskeletal comparisons and delineations of the locomotory muscle positioning.
The phylogenetic framework adopted here was provided by Carrano et al. (2012), where Piatnitzkysauridae is an early Megalosauroidea clade composed of Marshosaurus as the earliest piatnitzkysaurid taxon to diverge, being sister taxon of a subclade composed of Piatnitzkysaurus and Condorraptor (Figure 1b).However, see Rauhut and Pol (2019) and Schade et al. (2023) for an alternative hypothesis; as discussed above (see also Lacerda et al., 2023).
The nomenclature of Baumel and Witmer (1993) is followed in the descriptions of osteological correlates and muscle scars.
Piatnitzkysaurus was scored for character states of 86 characters related to the pelvic musculature (character ranges 1-71, 78-88, and 97-100; see Appendix A), to replace the "basal Tetanurae" lineage (which previously was a rough composite of transitional character states from this and other lineages) from Hutchinson (2001aHutchinson ( , 2001bHutchinson ( , 2002) ) and Bishop et al. (2021) in a new taxon-character matrix.As usual for the EPB, we used the maximum parsimony criterion for our reconstructions, similar to previous studies (e.g., Bishop et al., 2021;Molnar et al., 2018;Witmer, 1995).By doing so, we refine character scoring for early Tetanurae in general, which will be useful for future studies.We only scored Piatnitzkysaurus, as it has more osteological correlates preserved than the other taxa do, and consequently, a greater number of muscles could be inferred for this species (see Section 5).However, we sought to test if any muscles reconstructed differed in any details across the three taxa.To score and trace evolutionary changes in locomotor muscles, as well as to assess the most parsimonious states in our reconstructions, we used Mesquite software version 3.6 (Maddison & Maddison, 2015), using an informal composite "consensus" tree of Reptilia based on the recent phylogenetic framework used by Bishop et al. (2021) and references therein.iliotibiales is a large and superficial sheet that generally is composed of three heads over the dorsal and anteroposterior rim of the ilium, superficially positioned in relation to the other pelvic and thigh muscles (Clifton et al., 2018;Hudson et al., 1959;Hutchinson, 2001bHutchinson, , 2002;;Otero et al., 2010;Patak & Baldwin, 1998;Picasso, 2010;Romer, 1923a).In other Reptilia, the homologous muscle presents one or two weakly separated heads (Dick & Clemente, 2016;Hutchinson, 2002;Romer, 1942).The IT1-3 muscles attach to the dorsal rim of the ilium and are dorsally delimited by the crista dorsolateralis ilii, marking the border between the dorsal and lateral surfaces of the supraacetabular iliac blade (Baumel & Witmer, 1993).
In the Piatnitzkysaurus ilium MACN-Pv-CH 895, the anteriormost margin of the preacetabular process is not preserved, so the anterior limits/extension of the IT1 are not possible to infer; however, a great part of the supraacetabular rim is preserved.On the anteriormost part of the preacetabular blade, an expanded area is evident.This area is posteriorly delimited by an invagination present over the dorsalmost part of the supraacetabular rim (Figure 2a,b).Furthermore, immediately ventral to the dorsal rim of the ilium, there is a rough osteological delimitation, which posteriorly becomes more dorsally positioned (Figure 3a).Because these osteological correlates are topologically compatible with the positions (and similar osteological correlates) noted in extant archosaurs (e.g., Carrano & Hutchinson, 2002;Hudson et al., 1959;Otero et al., 2010;Picasso, 2010;Romer, 1923a), the rough delimitation and the dorsal invagination seem to be the posterior edge of the IT1 (level I), as well as the anterior demarcation of the IT2 (Figures 2a and 3a).Concerning the IT2, we infer the anterior limits to be at the same position as the main axis of the pubic peduncle, on a dorsal invagination of the dorsal rim of the preacetabular blade (level I) (Figure 2a), as aforementioned.Although not clearly preserved, the posterior limits of this muscle head seem to be demarcated by a small protuberance on the dorsal postacetabular blade (Figure 2a), which is posterior to the posterior facet of the ischial peduncle.This protuberance also probably delimited the anterior origin of the IT3; the attachment area of the IT3 is on the posterior dorsal rim of the postacetabular blade of the ilium.
A rough scar which becomes posteriorly large is on the ilium of MACN-Pv-CH 895, seeming to be dorsally delimited by the crista dorsolateralis ilii.This area of the IT3 is delimited by a faint osteological protuberance (level I).Most of the origination region of the IT1-3 muscles is not preserved in Condorraptor-the supraacetabular crest is highly damaged anterior and dorsal to the acetabulum in the only preserved ilium MPEF-PV 1687 of this taxon (Figure 2c,d).Although fragmentary, this region has an osteological correlate indicating that the anterior boundaries of the IT2 origin were from an invagination preserved at the same axis of the pubic peduncle (level I).However, as a consequence of this poor preservation of the Condorraptor ilium, our reconstructions of the origins of IT1 and IT3, as well as the extent of IT2, are uncertain, although these origins should have been similar to those reconstructed for In Crocodylia, Aves and other Reptilia, those three heads of Mm.
iliotibiales converge with M. ambiens and Mm.femorotibiales into at least one extensor tendon and fascial sheet, which inserts on the tibial cnemial crest or crista cnemialis cranialis (Baumel & Witmer, 1993) of the proximal metaphysis of the tibia (Dick & Clemente, 2016; TA B L E 1 Muscular homologies in extant archosaurs, considering the musculature of the pelvic girdle and hindlimb (modified from Carrano & Hutchinson, 2002).The EPB uses the state in each most recent common ancestor of Crocodylia and of Aves as its bracket, informed by further data from outgroups Lepidosauria and Testudines (not shown here; see Hutchinson, 2002).
The tibiae of both Piatnitzkysaurus specimens, b), have an expanded and rough area on the tibial cnemial crest with an anterior protuberance, in lateral view, that is distal to the cnemial crest.On this basis, we infer the same condition that is observed in extant archosaurs, with the cnemial crest as the osteological correlate for the insertion of IT1-3 (and the remainder of the triceps femoris: AMB and FMTE, FMTI) (level I) (Figure 4a,b).In the Condorraptor holotype MPEF-PV 1672, the cnemial crest is rounded and presents a small ridge (Figure 4c,d) when compared with Piatnitzkysaurus, and similar to other archosaurs, this was probably the same attachment area for the main tendon(s) of IT1-3 and other triceps femoris muscles (level I) (Figure 4c,d).
No tibia associated with Marshosaurus has been formally described so far, to our knowledge.
The pubes of both Piatnitzkysaurus individuals (left and right in MACN-Pv-CH 895 and left in PVL 4073) have a pubic tubercle that is well-developed (Figure 5a-d), as in Aves and other theropods (Carrano & Hutchinson, 2002;Gregory & Camp, 1918;Grillo & Azevedo, 2011;Hudson et al., 1959;Hutchinson, 2001b;Romer, 1923b).However, this tubercle slightly differs from other piatnitzkysaurid species in position-being more laterally and distally positioned instead of anterior as in Condorraptor, and more distally positioned than the condition in Marshosaurus (Madsen, 1976) (Figure 5).Nonetheless, the pubic tubercle is an osteological correlate of the presence and origin of the single head of the AMB in Piatnitzkysaurus (level I), as previously noted by Bonaparte (1986).
gastrocnemius externuslateralis near the proximal fibula.Although this shared tendon might have been present in early tetanurans such as piatnitzkysaurids, as is ancestral for Archosauria, there is no evidence of it (Level I′).
As previously noted by Bonaparte (1986), the lateral surface of the iliac blade in Piatnitzkysaurus has a large and deep depression.
This lateral depression is subdivided by a swollen vertical ridge, positioned just above the acetabulum (Carrano et al., 2012;Lacerda et al., 2023).This ridge has been suggested as the anterior limit of the ILFB (Carrano & Hutchinson, 2002;Hutchinson, 2001b).Anterior to the vertical ridge and anterodorsal to the acetabulum, the lateral depression is large and deep; whereas the posterior depression is shallow and positioned just above the ischial peduncle (Figure 2a,b).
Topographically, this posterior concavity is equivalent to the ILFB origin, as in other extinct theropods and extant archosaurs (Carrano & Hutchinson, 2002;Grillo & Azevedo, 2011;Hutchinson, 2001a;Otero et al., 2010;Picasso, 2010).The ventral limit of the ILFB origin is indicated by the brevis shelf, and its anterior limits seem to be related to the vertical iliac ridge (Hutchinson, 2001a), whereas the posterodorsal limits appear to have been demarcated by a semi-circular scar just below the IT3 origin (level I).In Condorraptor, although the supraacetabular crest is fragmentary, the left ilium MPEF-PV 1687 bears a small and shallow concavity dorsal to the ischiadic peduncle and posterior to the supraacetabular vertical ridge, on the postacetabular blade (Figure 2c,d), which may be the osteological correlate for the anterior limits of the ILFB origin.As noted by Carrano and Hutchinson (2002), the scars made by ILFB are difficult to discern; however, a well-developed iliac ridge lies just above the acetabulum in most megalosauroids (Carrano et al., 2012;Lacerda et al., 2023) and abelisaurids (Cerroni et al., 2022), indicating the anterior edge of the ILFB origin and the posterior edge of the M. iliofemoralis externus.Ventrally, the concavity related to the ILFB origin is delimited by the brevis shelf.Although the anterior, posterior and ventral limits of the ILFB origin are discernible (level I), the dorsal limit of this muscle origin is unclear, because the supraacetabular rim is not preserved in the only known ilium of Condorraptor.The ilia of Marshosaurus seem to lack the supraacetabular vertical ridge, or at least taphonomic issues preclude scoring this character in this taxon (Carrano et al., 2012;Lacerda et al., 2023); however, the dorsal, ventral and posterior boundaries of the ILFB origin can be inferred in this species based on the presence of a concavity and its posterior,  Baumel & Witmer, 1993; also see Hutchinson, 2001a), which inserts onto M. gastrocnemius externus/lateralis near its origin (Dick & Clemente, 2016;Carrano & Hutchinson, 2002;Clifton et al., 2018;Hutchinson, 2001aHutchinson, , 2002;;Otero et al., 2010;Picasso, 2010;Romer, 1923a).
The right fibula of Piatnitzkysaurus PVL 4073 preserves the fibular tubercle (Lacerda et al., 2023), which also presents a small scar (Figure 8), as sometimes seen in other archosaurs.As in other  not divided, with an origin located just above the acetabular aperture and deep to IT2, on the lateral surface of the ilium (Gregory & Camp, 1918;Hutchinson, 2002;Otero et al., 2010;Romer, 1923aRomer, , 1923b)).In Aves, the "M.iliofemoralis" is split into two muscles (i.e.
Nonetheless, the area of origin of M. iliofemoralis does not present scars indicating these subdivisions between the IFE and ITC (Carrano & Hutchinson, 2002;Hutchinson, 2001a).We consider the semi-circular concavity of the Piatnitzkysaurus preacetabular ilium (MACN-Pv-CH 895; Bonaparte, 1986) anterior to the iliac ridge as the origin of both of these muscular divisions (level I) (Figure 2a,b).The dorsal limits of both muscle origins are quite visible, indicated by striations located just ventral to the origins of the IT1-3 (Figures 2 and 3).The anterior limits of the ITC are undefined in this specimen due to the lack of the anteriormost and anteroventralmost preacetabular blade (Bonaparte, 1986).Following avian myology (e.g., Hutchinson, 2002;Picasso, 2010;Rowe, 1986), the ITC origin presumably would be anterior to the IFE head (level As commented by Bonaparte (1986), the femur of Piatnitzkysaurus has a well-developed lesser trochanter in the shape of a proximodorsally positioned blade (Figures 6a-d and   9); as in other megalosauroids, it rises past the ventral margin of the femoral head (Carrano et al., 2012;Lacerda et al., 2023).A rough area on the trochanteric shelf is not discernible; however, this structure is quite elevated and distinct (Figure 9), being pos- (Figure 7).The femur of Marshosaurus is not preserved.
The IFI origin in Aves is on the lateral surface of the ilium, between the anterodorsal region of the pubic peduncle and the posteroventral extremity of the preacetabular blade (Hudson et al., 1959;Hutchinson, 2002;Picasso, 2010;Romer, 1923a;Rowe, 1986;Suzuki et al., 2014).In many extinct theropods, there is evidence of the muscle origin (in a state intermediate between the ancestral reptilian and derived avian condition) from a preacetabular "cuppedicus" fossa (Hutchinson, 2002[or preacetabular notch-Carrano et al., 2012;Lacerda et al., 2023]) in that same region, suggesting a shift of the muscle origin from the medial to lateral pelvis (Carrano & Hutchinson, 2002;Hutchinson, 2002;Romer, 1923a;Rowe, 1986).
This inference is complicated by the fact that homologs of the PIFI2 in Crocodylia also originate from a similar area in Aves, so there is some ambiguity about which PIFI1 or PIFI2 muscle(s) may have shifted into this fossa and when (Carrano & Hutchinson, 2002;Hutchinson, 2001bHutchinson, , 2002)).
In Piatnitzkysaurus, even though the anterior margin of the preacetabular iliac blade is not entirely preserved on the specimen MACN-Pv-CH 895, the "cuppedicus" fossa is evident in the ventromedial surface of the iliac blade (Figure 2a The PIFI1/IFI insertion in extant archosaurs is located on the anteromedial surface of the femoral shaft.In Crocodylia, the insertion is on a keel that separates the site of insertion of PIFI2 laterally; and the origin of FMTI; medially, anteromedial to the fourth trochanter (Hutchinson, 2001b(Hutchinson, , 2002;;Otero et al., 2010;Romer, 1923a).In Aves, IFI inserts onto a rounded mark on the proximomedial portion of the femur (Hudson et al., 1959;Hutchinson, 2001bHutchinson, , 2002;;Suzuki et al., 2014).

M. pubo-ischio
In Crocodylia, the PIFI2 is a triangular and broad "fan-shaped" muscle that originates from the centra of the last 6-7 dorsal vertebrae and the ventral surfaces of their transverse processes (Otero et al., 2010;Romer, 1923a;Suzuki et al., 2011).In Aves, the homologous ITCR and ITM are small muscles that originate from the anteroventralmost part of the lateral portion of the preacetabular iliac blade (Rowe, 1986;Patak & Baldwin, 1998;Picasso, 2010).As above, this evident evolutionary shift of muscle origins is related to the expansion of the preacetabular blade and the origination of the preacetabular notch (Hutchinson, 2001b(Hutchinson, , 2002;;Romer, 1923a).
The right femur of Piatnitzkysaurus PVL 4073 preserves a well-developed blade-shaped lesser trochanter (Bonaparte, 1986) with a clear anterolateral and distal projection (the accessory trochanter) which is inferred as the insertion of the PIFI2 muscle (level I) (Figures 6a-f and 9).In Condorraptor, although the best-preserved femur MPEF-PV 1690 has the base of a prominent lesser trochanter (Rauhut, 2005), the most proximal part of it is not preserved and the accessory trochanter is not discernible, so the PIFI2 insertion cannot directly be reconstructed.
The posteromedial surface of the proximal region of the tibia in Piatnitzkysaurus has a broad depression below the medial condyle (Figure 12a-d), mainly visible in the PVL 4073 specimen (Figure 12a).
On the ischium of Piatnitzkysaurus MACN-PV-CH 895, which is better preserved proximally, a prominent ischial tuberosity that is triangular in shape is present near the proximoposterior edge of the ischium, ventral to the iliac peduncle; we infer this location as the FTI3 origin (level II) (Figure 11a,b).The delimitation of the FTI3 origin in Condorraptor is less evident than in Piatnitzkysaurus, but is similarly positioned (level II) (Figure 11c,d).In Marshosaurus is not possible to determine the FTI3 origin due to a lack of osteological correlates (level II'), so the muscle origin was not reconstructed in any detail, but it should have been in the same location.
The FTI3 in extant archosaurs inserts onto the posterior surface of the proximal portion of the tibia together with the FTE and other FTI head(s), which may form a slightly roughened and rounded structure made by the "tibiocalcaneal tendon" (or ligament) (Otero et al., 2010;Romer, 1923a;Suzuki et al., 2011).

A region topologically related to the FTI3 insertion in
Piatnitzkysaurus is positioned on the posteromedial surface of the proximal tibia, just below the medial and lateral condyles, and some scarring is proximally located here (level II).Again, in Condorraptor there is no scar (level II').

M. flexor tibialis internus 4 (FTI4):
The M. flexor tibialis internus division called FTI4 is only present in the Crocodylia clade (though it may have been lost in Caiman; Otero et al., 2010), being the division equivalent to the superficial portion of FTI2 of other non-archosaurian Reptilia (Romer, 1942).It is a small and thin muscle that originates from the fascia around the posteroventral ilium and posterodorsal ischium (Romer, 1923a(Romer, , 1923b;;Suzuki et al., 2011).Since this muscle leaves no evident scars and is absent in Aves (Carrano & Hutchinson, 2002;Hutchinson, 2002), the presence in Piatnitzkysaurus and Marshosaurus is equivocal (level II') so we do not infer this muscle here.The condition is even more ambiguous in Condorraptor, as the posterior portion of the ilium is not well-preserved.Following prior studies, we assume that the FTI4 is a crocodylian autapomorphy, absent in theropods.
On the posterolateral region of the Piatnitzkysaurus ilium MACN-Pv-CH 895, above the brevis shelf and below the IT3 origin, there are some linear scars topographically equivalent to the position of the FTE origin in extant archosaurs.We thus infer that the FTE origin was located here (level I), as in other dinosaurs (e.g., Bishop et al., 2021;Carrano & Hutchinson, 2002;Grillo & Azevedo, 2011;Langer, 2003;Russell, 1972;Smith, 2021).Yet as noted by Bonaparte (1986), the posterior edge of the postacetabular blade of the Piatnitzkysaurus ilium is not entirely preserved, thus, the posterior limits of the FTE origin remain unclear.In Marshosaurus left ilium UMNH VP 6372, it is also possible to infer the FTE origin due to some anterior delimitations located posterior to the ILFB (level I) (Figure 3b).Because the postacetabular blade of Condorraptor is not preserved, is not possible to directly infer the FTE origin.
As the FTE inserts very close to the FTI3, or shares a common tendon in extant archosaurs (e.g., Otero et al., 2010;Romer, 1923aRomer, , 1923b)), this applies to Piatnitzkysaurus (level I), but more is equivocal in Condorraptor (level I′) and not possible to infer in Marshosaurus.
The incomplete obturator process in Piatnitzkysaurus extends proximally, almost to the anterior line of the pubic peduncle (Bonaparte, 1986).Some scars are visible on the most anterodorsal portion of the ischial apron (mainly in the PVL 4073 specimen), which could be related to the puboischiadic membrane (Hutchinson, 2002).
Based on relative positions, the ADD1 in Piatnitzkysaurus probably originated from the anteroventral obturator process of the ischium in the ventral portion of the ischial apron (level I′) (Figure 10a,b).
In Condorraptor MPEF-PV 1689, the obturator process is damaged, making it difficult to determine its anterior contact with the pubis.However, even in its broken state it obviously is a developed structure (Rauhut, 2005)  The ADD1 in extant archosaurs has a small, somewhat tendinous insertion located on the posterior shaft of the distalmost femur (Hutchinson, 2001a;Otero et al., 2010;Picasso, 2010;Romer, 1923a;Suzuki et al., 2014); medial to the ADD2; with both insertions located between the la and lip (Hutchinson, 2001a).
A small depression is evident on the posterodorsal rim of the right ischium of Piatnitzkysaurus MACN-Pv-CH 895, distally delimited by a bump.This position is topographically equivalent to the inferred ADD2 origin in other theropods (e.g., Bishop et al., 2021;Carrano & Hutchinson, 2002;Grillo & Azevedo, 2011;Smith, 2021).Although no roughened scars are discernible, this depression is interpreted as the origin of ADD2 (level II) (Figure 11a,b).The osteological correlate of the ADD2 origin on the ischium of Condorraptor is less evident, but can be delimited in a position similar to that of Piatnitzkysaurus, but extending further distally (level II) (Figure 11c,d).In Marshosaurus, the ADD2 boundaries were not observed, therefore this muscle's origin was not reconstructed.
However, in some Aves (presumably autapomorphically), the origin of this muscle has two parts, originating both from pubis and ischium surrounding the obturator foramen (e.g., Gangl et al., 2004).Only one head (or weak subdivision) of the PIFE is present in other nonarchosaurian Reptilia (Hutchinson, 2002;Romer, 1942).
Even though the distal regions of the pubes in Piatnitzkysaurus The PIFE1-3 in extant archosaurs have a common tendon of insertion that attaches to the proximolateral femur on the greater trochanter (Hutchinson, 2001a(Hutchinson, , 2002;;Otero et al., 2010;Romer, 1923a).
The greater trochanter in Piatnitzkysaurus has a straight angle to the femoral long axis (Bonaparte, 1986) and we infer it to represent the PIFE1 insertion (level I) (Figures 6 and 9).This structure is not preserved in Condorraptor.
Considering the well-developed pubic apron in the piatnitzkysaurids studied here, a level II inference allows us to infer that these taxa had a PIFE2 origin from the posterior portion of the pubic apron (Figure 5).The insertion with PIFE1-3 is described above (Figures 6   and 9).
The retention of the obturator process in Piatnitzkysaurus, as Staurikosaurus (Grillo & Azevedo, 2011) and Coelophysis (Bishop et al., 2021).In the right ischium of Piatnitzkysaurus MACN-Pv-CH 895, the probable origination site is more evident, being positioned between the ADD1 origin and the ischial ridge (level II) (Figure 11a,b).
In the poorly preserved ischia of PVL 4073, as well as in Condorraptor and Marshosaurus, the PIFE3 boundaries are not possible to reliably estimate, but the PIFE3's general position is (level II) (Figure 11c-f).
See above for details on the PIFE 1-3 insertion (Figures 6 and 9).
Among the three piatnitzkysaurids studied here, the Piatnitzkysaurus right ischium of the MACN-Pv-CH 895 specimen is the best preserved proximally; followed by the Marshosaurus left ischium UMNH VP 6380, which has both, iliac and pubic peduncles, but lacks the ventral part of the obturator process (Madsen, 1976); and the Condorraptor left ischium MPEF-PV 1689, which although lacking most of the proximal articulation, preserves a partial, well-developed obturator process (Rauhut, 2005).None of the ischia of the three taxa presents evidence of the apomorphic condition of lateral origin of the ISTR muscle; in both Piatnitzkysaurus and (Madsen, 1976), the medial surface of the ischium/ obturator process is covered by fine striations.In Condorraptor, such striations are not discernible (there might be small ventral marks on the obturator process, if not taphonomic artifacts), but Condorraptor probably had the same origin.Therefore, the ISTR origin was on the medial surface of the ischium, including the entire area of the obturator process (level II for Piatnitzkysaurus and Marshosaurus, and level II' for Condorraptor).
The best-preserved femur of Piatnitzkysaurus (right femur PVL 4073) has a clear posteriorly projected structure proximal to the fourth trochanter and distal to the greater trochanter (Figure 9), similar in position and shape to other tetanurans (e.g., -Carrano & Hutchinson, 2002).Although the bony surface is not well-preserved, this projection is preceded anteroposteriorly by a groove considered here as the insertion of ISTR (level I) (Figures 6 and 9).Due to the fragmentary nature of the proximal femur of Condorraptor MPEF-PV 1690 and the lack of femora for Marshosaurus, no inferences were made about the ISTR insertion for these taxa.
The posterior width of the brevis fossa varies in non-avian theropods (Carrano et al., 2012).In some tetanurans, the brevis fossa is posteriorly wide in Marshosaurus (Figure 14) and some spinosaurids; whereas it is subequal in width in Piatnitzkysaurus and some megalosaurids (Lacerda et al., 2023).The Piatnitzkysaurus ilium MACN-Pv-CH 895 has a large and relatively deep brevis fossa, and presumably the CFB in this taxon originated entirely in the fossa (level II), although the posterior edge of the postacetabular blade in this specimen is incomplete (Figure 14a).14b).
The CFB muscle of extant archosaurs inserts by a tendon on the posterolateral surface of the proximal region of the femur, positioned between the lip and the fourth trochanter (Hutchinson, 2001b(Hutchinson, , 2002;;Otero et al., 2010;Picasso, 2010;Suzuki et al., 2014).
Among the femora of Piatnitzkysaurus, the better-preserved fourth trochanter is on the right femur of PVL 4073 specimen; however, the surface between the fourth trochanter and the lip is not well-preserved and the insertion of CFB is not discernible based on scars, even though the well-developed fourth trochanter allows us to infer the insertion of this muscle safely (level I) (Figures 6 and 9), based on that in extant and extinct archosaurs.The Condorraptor left femur MPEF-PV 1690 preserves a well-developed fourth trochanter.
As noted by Rauhut (2005) it is a low but robust ridge, allowing us to infer the position and extent of the CFB insertion (level I) (Figure 7).
In most Aves (where present), the CFC origin is restricted to the last free caudal vertebra and the uropygium (Clifton et al., 2018;Hutchinson, 2001aHutchinson, , 2002;;Suzuki et al., 2014).no sign of a transverse process (Rauhut, 2005).These characteristics of the proximalmost and the proximal mid-caudal vertebrae allow inferring part of the CFL origin (level I).

Although the tail in
The CFL in non-avian Reptilia inserts onto the proximal femur, in the pit and on the medial surface of the fourth trochanter; a secondary tendon continues downwards to the fibula, contributing to the M. gastrocnemius externus (=lateralis of Aves) origin (Gatesy, 1990;Hutchinson, 2001aHutchinson, , 2002;;Otero et al., 2010;Romer, 1923a).Once birds reduced their tail, the CFC muscle reduced as well as the fourth trochanter which reduced to a roughed area (Gatesy, 1990).
Dinosauromorphs and theropods have a large crest-shaped fourth trochanter (e.g., Hutchinson, 2001a); this also being the condition in both Piatnitzkysaurus and Condorraptor; thus indicating the CFL insertion (level I) and exemplifying that it was a large muscle in early tetanurans (Figures 6 and 7).As proposed by Hutchinson (2001a) and Carrano and Hutchinson (2002), the secondary tendon of the CFL may have been lost in early theropods, as the fourth trochanter became less "pendant" (distally angled) than in many other archosaurs.Considering that the fourth trochanter of both studied taxa is well-developed but not pendant, this secondary tendon would probably have been absent (level II').
Generally, reconstructions of the TA muscle origin in theropods consider both muscular heads, as aforementioned originating from the lateral condyle of the femur and the proximal tibia (e.g., Carrano & Hutchinson, 2002;Smith, 2021Smith, , 2023)).In the femora of both Piatnitzkysaurus and Condorraptor, there is no evidence of the TA origin; the lateral condyles do not have the distally positioned fovea tendinis m. tibialis cranialis as in Aves (e.g., Baumel & Witmer, 1993;Hattori & Tsuihiji, 2021;Picasso, 2010) or proximal to the lateral condyle as in Crocodylia (e.g., Hattori & Tsuihiji, 2021;Suzuki et al., 2011).Therefore, the specific origin of this muscle head on the femur in both piatnitzkysaurids is ambiguous, but as the origin of this muscle is a conservative feature in Reptilia (Hattori & Tsuihiji, 2021), and probably present in theropods (Carrano & Hutchinson, 2002), we tentatively reconstruct this muscle on the anterior lateral condyle (level I′) (Figures 6 and 7).Considering the second TA head, the tibiae of both Piatnitzkysaurus specimens, MACN-Pv-CH 895 and PVL 4073, distal to the insertion of the triceps femoris on the cnemial crest, have an elliptical and well-demarcated depression (Figure 15).This depression is topologically equivalent to the TA reconstruction in early theropods such as Coelophysis (Bishop et al., 2021) and later coelurosaurs such as Tyrannosaurus (Carrano & Hutchinson, 2002), Nothronychus (Smith, 2021), as well as Aves (e.g., Hattori & Tsuihiji, 2021); thus, it is considered here as the second head of TA origin (level I) (Figures 4a,b and 15).In the Condorraptor tibia MPEF-PV 1672, this depression is not clearly noticeable (level I′) (Figure 4c,d).

M. extensor digitorum longus (EDL):
The EDL (previously termed as M. tibialis cranialis in non-avian Reptilia; see Hattori & Tsuihiji, 2021) in extant Reptilia originates from the proximal shaft of the tibia; in Crocodylia from a rugose surface in the proximalmost portion and in Aves from a broad surface located between the cranial and the lateral cnemial crests and distal to the insertion of the triceps femoris tendon (Hattori & Tsuihiji, 2021;Hutchinson, 2002;Suzuki et al., 2011Suzuki et al., , 2014)).
In the tibiae of Piatnitzkysaurus PVL 4073 and Condorraptor MPEF-PV 1672, a clear demarcation is visible on the anterolateral shaft, located in the sulcus intercnemialis.The proximal limits are not well-defined and may reach the cnemial crest (as in Aves-Suzuki et al., 2014;Hattori & Tsuihiji, 2021), but the anterior limits are bordered by a muscular line, and the posterior limits proximally by the fibular crest and distally by a posterior muscular line.As noted for Aves (e.g., Hattori & Tsuihiji, 2021), the distal part of the EDL origin is tapered (level I) (Figure 4).
The EDL insertion in non-avian Reptilia is limited to a bulge(s) on the dorsal surface of metatarsals I-II (Hattori & Tsuihiji, 2021;Hutchinson, 2002), whereas in Aves this insertion is on the proximal processes of the distal pedal phalanges, in the hyperextensor fossae (Hattori & Tsuihiji, 2021;Hutchinson, 2002;Picasso, 2010;Suzuki et al., 2014).The EDL insertion in early dinosaurs is inferred as a distally positioned when compared to non-avian Reptilia, due to the presence of large extensor fossae and rugosities on the dorsal surfaces of the pedal phalanges (Bishop et al., 2021;Carrano & Hutchinson, 2002;Hutchinson, 2002;Smith, 2021).Although the condition in piatnitzkysaurids is probably the same as in other dinosaurs, this insertion has not been reconstructed as the specimens elements (with some variability existing in crocodylians and other taxa), inserting onto the dorsal surface of the pedal phalanges; however, in Aves, this muscle is absent (Dilkes, 2000;Hattori & Tsuihiji, 2021;Hutchinson, 2002).The EDB is conjectured to have fused with the EDL in dinosaurs (Carrano & Hutchinson, 2002;Dilkes, 2000).We did not reconstruct this muscle following this hypothesis (a level II' reconstruction) as such elements are not preserved in piatnitzkysaurids.
M. extensor hallucis longus (EHL): The EHL (also termed M. flexor perforatus digiti II; homologue to the M. extensor hallucis brevis of crocodilians-Hattori & Tsuihiji, 2021) in non-archosaurian Reptilia is conservative in position.being a small and short muscle originating from the distal shaft of the fibula, inserted onto the hallucal phalanges (Hattori & Tsuihiji, 2021;Hutchinson, 2002).In Aves, related to loss of the distal fibula, the EHL muscle origin has moved distally to the anteromedial portion of the proximal tarsometatarsus (Hutchinson, 2002;Moreno, 1990;Patak & Baldwin, 1998), or the EHL is absent in species that have lost the hallux (e.g., Suzuki et al., 2014).
Condorraptor and Marshosaurus do not have a fibula preserved.
The EHL muscle insertion is onto the anterior portion of the hallucal phalanges in Reptilia; whereas in Aves it becomes more posterior due to changes in hallux position (Hutchinson, 2002).
Reconstructions of this muscle in non-avian theropods consider this insertion as onto the anterior side of the hallucal ungual (Bishop et al., 2021;Carrano & Hutchinson, 2002;Smith, 2023).The lack of the well-preserved pes in piatnitzkysaurids precludes any inferences about this muscular insertion.
In Crocodylia and Aves, this muscle originates from a lateroventral distinct depression on the posterior portion of the distal femur.In Aves, it is delimited by a rough depression and has deep and superficial layers, sometimes with an extra lateral muscle head (e.g., Hattori & Tsuihiji, 2021;Suzuki et al., 2011).In extant archosaurs, the GL and M. gastrocnemius medialis (GM) muscles or their homologues insert via a shared tendon (and aponeurosis/plantar fascia) to metatarsal V (plantar surface of pes) in Crocodylia, and the medial and plantar margins of the hypotarsus of Aves (Hattori & Tsuihiji, 2021;Hutchinson, 2002;Mckitrick, 1991;Otero et al., 2010;Picasso, 2010;Romer, 1923a).It is thought that the plantar aponeurosis was reduced in dinosaurs (see Hutchinson, 2002), yet with the GM + GL maintaining robust scars on the posterior metatarsal shafts (Bishop et al., 2021;Carrano & Hutchinson, 2002;Dilkes, 2000;Hutchinson, 2002;Smith, 2021).

Both femora of
The M. gastrocnemius pars medialis (GM): In Reptilia, the GM muscle (=M.gastrocnemius internus, GI in Crocodylia) originates from the medial surface of the proximal tibia, occupying a large area on the cnemial crest in Aves (Hattori & Tsuihiji, 2021;Hutchinson, 2002;Otero et al., 2010;Patak & Baldwin, 1998;Picasso, 2010;Romer, 1942;Suzuki et al., 2011Suzuki et al., , 2014)).Although Mm. gastrocnemii is composed of the lateral and medial head ancestrally in Reptilia, lepidosaurs, and Aves evolved a third head independently (M.gastrocnemius pars intermedia in Aves), the third head in lepidosaurs probably deriving from a subdivision of the lateral head and in Aves deriving from a subdivision of the medial head.At least in the lineage of Aves, the timing of the derivation of this extra head is difficult to determine (Hutchinson, 2002).
A large depression is on the medial surface of the proximal tibia are from the fibula, in some cases with contribution from the tibia (Dick & Clemente, 2016;Dilkes, 2000;Hattori & Tsuihiji, 2021;Hutchinson, 2002).Generally, the FL origin is on the lateral fibula, distal to the ILFB insertion; whereas the FB is more distally and anterolaterally positioned on the fibula (Hattori & Tsuihiji, 2021;Hutchinson, 2002;Suzuki et al., 2011Suzuki et al., , 2014)).With the distal region of the fibula lost in early Aves, the origin of Mm. fibulares became superficial on the lateral sides of the proximal tibiotarsus and fibula (Hutchinson, 2002;Patak & Baldwin, 1998;Picasso, 2010).
Similar to other dinosauriform reconstructions (e.g., Bishop et al., 2021;Carrano & Hutchinson, 2002;Dilkes, 2000;Piechowski & Tałanda, 2020;Smith, 2021), the presence of a distally unreduced fibula suggests that the FL and FB origins in Piatnitzkysaurus, were mainly from the middle to distal fibular shaft.Although there is no clear demarcation or scarring, we reconstructed these muscles (level I′) based on the PVL 4073 fibula.
In general, ancestral Reptilia have the FL and FB insertions on the proximolateral tarsals, metatarsal V, and 5th digit aponeurosis; and near the proximal end of metatarsal V, respectively (Hattori & Tsuihiji, 2021;Hutchinson, 2002).Some modifications occurred in the avian lineage, especially the reduction/loss of the plantar aponeurosis and the 5th digit, concentrating these muscular insertions onto the lateroproximal side of the tarsometatarsus in Aves (Hattori & Tsuihiji, 2021;Hutchinson, 2002).Reconstructions of these insertions in Dinosauriformes usually are onto the tarsal/metatarsal elements, particularly metatarsal V (Bishop et al., 2021;Carrano & Hutchinson, 2002;Dilkes, 2000;Piechowski & Tałanda, 2020;Smith, 2021).The lack of preserved tarsals and metatarsal V in piatnitzkysaurids prevents the reconstruction of the FL and FB insertions in detail, but presumably, they were the same as in other non-avian Dinosauriformes.

| Summary of muscle reconstructions
The muscle reconstructions inferred for Piatnitzkysaurus, Condorraptor, and Marshosaurus are summarized in Tables 2-4, respectively.Figure 17 presents a "muscle map" reconstruction for each of the studied species.Overall, we infer 29 muscles' origins for Piatnitzkysaurus, which is the best-preserved specimen; and among these 29 muscles, it was possible to infer the insertions of 25 (Figure 17a).In Condorraptor, 21 muscles were reconstructed; among these, 12 were inferred for both origin, and insertion (Figure 17b).
On the contrary, the following muscles were inferred as absent in this taxon: (1) M. puboischiotibialis (PIT), which is present in non-avian Reptilia, arising from a muscle scar on the anterolateral ilium (Hutchinson, 2002;Otero et al., 2010;Suzuki et al., 2011) that is absent in Avialae (Hutchinson, 2002); and ( 2) M. pubotibialis (PUT), which originates from the pubis, near the pubic tubercle TA B L E 2 Pelvic and hindlimb musculature inferred for Piatnitzkysaurus, and required inference level based on the EPB.Refer to Table 1 or the main text results for muscle abbreviations.and puboischiadic ligament, in some Reptilia (Hutchinson, 2002;Romer, 1942) but was lost in Archosauria (Bishop et al., 2021;Dilkes, 2000;Hutchinson, 2002;Romer, 1923a).Other muscles such as FTI2 and 4, GIM, and EDB (probably fused with EDL) are equivocal and were not reconstructed; thus, their presence or absence was not inferred (see Table 2 for Piatnitzkysaurus reconstruction; Hutchinson, 2002 andBishop et al., 2021 for further discussions).

| Myological comparisons among theropods
Here we have associated morphological structures with the myology of the locomotor apparatus in piatnitzkysaurids, combining our work with previous descriptions (Bonaparte, 1979(Bonaparte, , 1986;;Madsen, 1976;Rauhut, 2005).This is the first detailed myological study of the pelvic appendages in early Tetanurae, as far as we know.While the reconstructions we found are similar to others performed for different theropods, we present some key comparisons here.
The division of the IF of non-avian Reptilia into the ITC + IFE of Aves is ubiquitous in recent pelvic musculature reconstructions of theropods (e.g., Bates, Benson, & Falkingham, 2012;Bishop et al., 2021;Carrano & Hutchinson, 2002;Hutchinson & Gatesy, 2000), based mainly on inferred insertions of these muscles, and we infer the same division in piatnitzkysaurids (mainly Piatnitzkysaurus; Figures 2 and   17).However, similar to Staurikosaurus (Grillo & Azevedo, 2011), it is not possible to distinguish the origins of these two muscles, except that, based on the EPB (Figure 1a), that the ITC was anteriorly positioned and IFE immediately posterior (see Hutchinson, 2002).This reconstruction, like others, differs from the Falcarius reconstruction TA B L E 3 Pelvic and hindlimb musculature inferred as present in Condorraptor and required inference level based on the EPB.Refer to Table 1 or the main text results for muscle abbreviations.figured in Smith (2023), in which the IFE was positioned posteriorly on the ilium (between ILFB and FTE); a condition not known in Aves.
However, in our reconstructions, based on the position of the pubic tubercle, the AMB origin in piatnitzkysaurids appears to have been more laterodistal (Figures 5 and 17).Similarly, a slightly more distal AMB is suggested for Albertosaurus (Rhodes et al., 2021) and even more distally in the dinosauriform Silesaurus (Piechowski & Tałanda, 2020).
In early theropods such as Staurikosaurus (Grillo & Azevedo, 2011) and allosauroids (Bates, Benson, & Falkingham, 2012), the inferred ADD2 origin is restricted to the most posterior part of the ischial shaft, whereas in Coelophysis (Bishop et al., 2021), this muscle was reconstructed in a slightly more distal position.In our reconstruction of piatnitzkysaurids (mainly in Condorraptor; Figures 11 and 17), the ADD2 origin extends more distally, similar to Crocodylia (e.g., Suzuki et al., 2011) and reconstructions of some other theropods (e.g., Carrano & Hutchinson, 2002;Rhodes et al., 2021).Some of these differences might relate to subjective interpretations of the ADD2 scar location, but as Hutchinson (2001b) showed, this scar is fairly conservative and conspicuous in archosaurs.
As in Crocodylia (e.g., Otero et al., 2010;Romer, 1923b;Suzuki et al., 2011) as well as Staurikosaurus (Grillo & Azevedo, 2011) and Coelophysis (Bishop et al., 2021), we reconstructed the PIFE3 origin on the lateral aspect of the obturator process of the ischium, between the ADD1 + 2 (Figures 11 and 17).This origin extends more posteriorly in Crocodylia, but the anterior region lies between ADD1 + 2. In a Tyrannosaurus reconstruction (Carrano & Hutchinson, 2002), the PIFE3 originates slightly distal to ADD1, and is even more distally positioned in some maniraptoran reconstructions (Rhodes et al., 2021).These differences likely relate not only to subjective interpretations, but also to relative position of the ischial obturator process (Hutchinson, 2001b).
Our reconstruction of the ISTR origin differs from other myological studies'.In general, reconstructions in theropods consider the ISTR origin as dorsal and occupying a small medial part of the ischium (e.g., Carrano & Hutchinson, 2002;Grillo & Azevedo, 2011;Smith, 2021).
However, here we consider the condition in Crocodylia and other nonavian Reptilia (e.g., Romer, 1923a;Suzuki et al., 2011) to be more plausible, reconstructing the ISTR originating entirely on the medial surface TA B L E 4 Pelvic and hindlimb musculature inferred as present in Marshosaurus and required inference level based on the EPB.Refer to Table 1 or the main text results for muscle abbreviations.Note that some muscles are not shown here, and some bones have been mirrored to illustrate the reconstructions of the three piatnitzkysaurid species.Muscle abbreviations are provided in Table 1; see text for inference levels and other comparisons.CFL and PIFI2 origins are much smaller than expected; simply shown for relative positions.Medial muscle origins (e.g., FMTI) and insertions (e.g., PIFI1) are not shown.Mm. gastrocnemii are labelled as "Flexores" due to their action around the knee; lower leg (FB, FL, etc.) as "Flexores" for ankle dorsiflexion.Not to scale.
of the ischium, probably occupying the entire area of the obturator process (similar to that hypothesized for Coelophysis; Bishop et al., 2021).
In theropod reconstructions, in general, the GM origin is on the medial portion of the proximal tibia.However, it variably is reconstructed somewhat distally, either more anteromedially (e.g., Coelophysis; Bishop et al., 2021), posteromedially (e.g., Tyrannosaurus;Carrano & Hutchinson, 2002), or anteriorly (e.g., Falcarius;Smith, 2023).Our reconstruction of the GM origin, based on the medial concavities on the cnemial crest (Figure 12), more closely resembles the condition in Aves, of a more anterior and proximal origin occupying the entire medial side of the cnemial crest, immediately distal to the triceps femoris tendon (e.g., Suzuki et al., 2014).This finding is cause to reinvestigate the GM origin in other theropods (e.g., Carrano & Hutchinson, 2002).
Overall, as in other recent studies of earlier (e.g., Grillo & Azevedo, 2011) and later (e.g., Carrano & Hutchinson, 2002) theropods, we infer that the myology of the pelvic appendage more closely resembled that of Aves than Crocodylia, thus characterizing the evolution of locomotor musculature in the avian lineage (e.g., Hutchinson, 2001aHutchinson, , 2001bHutchinson, , 2002)).There seems to have been much conservatism across non-avian Theropoda until Maniraptora.For example, many theropods (and some other Dinosauriformes) are inferred to have had a division of IF into IFE and ITC, an origin of PIFI1 (and possibly some of PIFI2) from the "cuppedicus fossa," a CFB origin largely from the "brevis fossa," a fused EDL and EDB, and the putative absence of some apomorphic muscles of Crocodylia such as FTI4 and a second AMB head; or loss of plesiomorphic muscles

| Myological comparisons among piatnitzkysaurids
As a consequence of osteological similarities, the myology of the pelvic girdle in these taxa presents several similarities (Figure 17).
2. Position and area of AMB origin among the three piatnitzkysaurids-the pubic tubercle is well-developed in piatnitzkysaurids; even more robust in Condorraptor (Rauhut, 2005).Thus, in our muscle reconstructions, the AMB origin appears to have occupied a larger area in Condorraptor (and perhaps greater force potential), followed by Piatnitzkysaurus, with the origin also more distally positioned in both (Figures 5 and 17  and 15).Although the cnemial crest is moderately developed, it is an autapomophic feature in Condorraptor (Rauhut, 2005), and it implies a reduction of the triceps femoris insertion (Figure 17).men is a sub-adult individual (Rauhut, 2005), this difference might be an ontogenetic feature.

| Taphonomic limitations
The degree of preservation of the pelvic appendage varies in each of the three species studied here.Piatnitzkysaurus represents the bestpreserved specimen.This better preservation led to the larger number of successful inferences noted above (Table 2; [8][9][10][11][12][14][15][16][17].Furthermore, two individuals of Piatnitzkysaurus are known (Bonaparte, 1979(Bonaparte, , 1986)), allowing comparisons between individuals (e.g., Figures 12 and 15), thus increasing the reliability of our muscular reconstructions for this taxon.However, the absence of most of the pedal bones prevents the analysis of many lower limb muscles.
An illustration of Piatnitzkysaurus in Figure 18 summarizes the most superficial thigh muscles.
In taphonomic terms, Condorraptor has the second best-preserved piatnitzkysaurid pelvic appendages, although only one skeleton is known, probably from the same individual (Rauhut, 2005).The lack of preservation of many distal hindlimb elements such as the fibula and pes, as well as the fragmentary state of the femora of this taxon (compare Figures 6 and 7) caused the weaker inferences (Table 3; Figures 2,4,5,7,[11][12][13]16,17), especially for muscle insertions.
Although Marshosaurus is a better-known taxon in terms of the number of bones, which includes the cranium (Carrano et al., 2012;Madsen, 1976), it has fewer preserved pelvic appendage elements than other piatnitzkysaurids, which prevents a more robust evaluation of the musculature in this species; only the muscle origins (Table 4; 5,11,14,17).Description of new material (e.g., Chure et al., 1997) should provide additional clues on the locomotor musculature.

| CON CLUS ION
Here, we reconstructed the hindlimb musculature of the Jurassic Piatnitzkysauridae clade.We find a great anatomical similarity within Piatnitzkysauridae with minor differences, such as the origin of M. ambiens and size of M. caudofemoralis brevis.The similarities with Aves were the division of the M. iliofemoralis externus and M.
iliotrochantericus caudalis, as well as the broad depression of the M.
gastrocnemius pars medialis origin on the cnemial crest of the tibia.
Our results shed some light on palaeontological issues regarding megalosauroids and contribute to knowledge about the evolution of locomotor muscles in Theropoda.
ran evolution.Recently, Lacerda et al. (2023) mapped the evolution and reconstructed the ancestral states of the morphological characters of the pelvic appendage in Megalosauroidea, characterizing potential variations related to muscle attachments; and tested whether different homoplastic signals in different regions of the locomotor system are present in theropods.That study provides a stronger basis for the muscle reconstruction performed here (see below).
for our reconstructions (Figure 1a).Three levels of inference are established by EPB to characterize the confidence in reconstructing a particular soft tissue for an extinct species: (I) represents an unequivocal structure of a particular feature, that is, when the two (or more) extant taxa have the homologous soft tissue and its osteological correlate; (II) F I G U R E 1 (a) Simplified example of the Extant Phylogenetic Bracket (EPB) application in Theropoda.(b), Theropod phylogeny (up to Coelurosauria on the right side of the phylogeny) highlighting the phylogenetic position of Piatnitzkysauridae.(a), adapted from Grillo & Azevedo, 2011; (b), adapted from Carrano et al., 2012.M, muscle; O, osteological correlate.Silhouettes are from phylo pic.org; see Acknowledgements.

Piatnitzkysaurus.
In the studied ilia of Marshosaurus (UMNH VP 6372 and UMNH VP 6374) and the holotype UMNH VP 6373 [=UUVP 2826] specimen (Madsen, 1976), the best-preserved part is the postacetabular process of the ilium.Although the subdivisions of the IT heads are not as discernible as in Piatnitzkysaurus, the origin of the IT3 in both UMNH VP 6372 and UMNH VP 6373 is clearly discernible by several scars on the dorsal edge of the postacetabular blade and the presence of the crista dorsolateralis ilii (level I) (Figures 2e,f and 3b).
It thus is plausible, based on the osteological correlate of the right pubis MPEF-PV 1696 and a small fragment of the left pubis MPEF-PV 1688, that the AMB had a robust attachment to the pelvic girdle of Condorraptor (level I).The best-preserved pubis of Marshosaurus (right pubis UMNH VP 6387) also osteologically concurs with the single head of the AMB; as previously noted, the anterolateral part of the proximal portion of the pubis presents a visibly rough area (Madsen, 1976) topographically equivalent to the AMB origin (level I) (Figure 5i,j).
Mm. femorotibiales (FMTE and FMTI): The Mm. femorotibiales of Crocodylia has two heads (i.e.M. femorotibialis externus-FMTE and M. The three femora of the two Piatnitzkysaurus skeletons lack well-preserved shaft surfaces.Regardless, the left femur of PVL 4073 preserves the most distal parts of both la and lip on the posterior shaft of the femur, and lia on the distal femur, arising medially and becoming anteriorly positioned along the proximal shaft of the femur (Figure 6c,d).Furthermore, the right femur of PVL 4073 preserves the distal base of the la (Figure 6g,h).Although not entirely preserved, the presence of the la, lia and lip allows inference of the FMTE and FMTI origins without precise boundaries (Figure 6).The FMTE and FMTI in Piatnitzkysaurus, as well as in other theropods (e.g., Staurikosaurus-Grillo & Azevedo, 2011; Coelophysis-Bishop et al., 2021; allosauroids-Bates, Benson, & Falkingham, 2012; Tyrannosaurus-Carrano & Hutchinson, 2002; Nothronychus-Smith, 2021; and Skorpiovenator-Cerroni et al., 2022) seem to have had the same origins from the lateral and the anteromedial surfaces of the femoral shaft, respectively (level I).In Condorraptor, both femora are quite fragmentary, lacking the proximal portions.The right femur MPEF-PV 1690 is better preserved, with a great portion of the femoral shaft (Figure 7); however, the three longitudinal ridges/ lineae (lia, lip and la) are not completely preserved.It remains possible to reconstruct the FMTE and FMTI origins in positions similar to our Piatnitzkysaurus reconstruction (level I), but their proximal extent remains indeterminate.Rauhut (2005) noted the presence of the cdc in both Condorraptor femora (Figure 7e-h); this being a structure related to the distal divisions between the FMTE and FMTI

F I G U R E 3
Osteological correlates of M. iliotibiales 1-3 observed on the ilia of Piatnitzkysauridae (left ilia, lateral view).(a), Piatnitzkysaurus (MACN-Pv-CH 895).(b), Marshosaurus (UMNH VP 6372).Anatomical/muscular abbreviations: cdi, crista dorsolateralis ilii; IT1-3s, Mm. iliotibiales scars; IT1-2l, M. iliotibialis 1 and 2 limits.Arrows indicate muscle scars.Scale bar = 20 mm.The FMTE and FMTI heads converge into a main tendon and fascia inserting onto the tibial cnemial crest deep to IT1-3 and AMB (level I) (Figure 4), as noted above.M. iliofibularis (ILFB): In extant Reptilia, the ILFB originates from the lateral surface of the ilium in the postacetabular blade, positioned posterior to the IFE (IF in Crocodylia), anterior to the FTE dorsal and ventral delimitations (level I) (Figure 2e,f).The insertion of the ILFB in Reptilia is located on the fibular tubercle; a scarred or rounded and prominent structure on the F I G U R E 4 Osteological correlates of the triceps femoris insertion and origins of lower leg muscles from the tibiae of Piatnitzkysauridae (left tibiae, lateral view).(a, b) Piatnitzkysaurus (PVL 4073).(c, d) Condorraptor (MPEF-PV 1672).Anatomical/muscular abbreviations: cc, cnemial crest; EDLs, M. extensor digitorum longus scar; fc, fibular crest; it1-3 + amb + fmt, insertion of the tendon from the iliotibiales + ambiens + femorotibiales muscles; lc, lateral condyle; mc, medial condyle; pas, proximal articular surface; si, sulcus intercnemialis; TAd, M. tibialis anterior depression.Scale bar = 50 mm.proximal region of the fibular shaft; typically most prominent in archosaurs.Furthermore, a secondary tendon is present in extant taxa (in Aves, constrained by a loop termed ansa m. iliofibularis- theropods (e.g., Tyrannosaurus-Carrano & Hutchinson, 2002), there is no osteological evidence for a secondary tendon in early tetanurans based on Piatnitzkysaurus, although this structure is predicted to have been present (level I′).The fibula is not preserved in Condorraptor and Marshosaurus.5.1.2| Deep dorsal group M. iliofemoralis or M. iliofemoralis externus (IFE) and M. iliotrochantericus caudalis (ITC): The M. iliofemoralis in Crocodylia is a single muscle, F I G U R E 5 Osteological correlates observed on the pubes of Piatnitzkysauridae (right pubes, lateral and anterior views).(a-d) Piatnitzkysaurus (MACN-Pv-CH 895).(e-h) Condorraptor (MPEF-PV 1696).(i, j) Marshosaurus (UMNH VP 6387).Anatomical/muscular abbreviations: ac, acetabulum; AMBt, M. ambiens tubercle; ap, apron; ilc, iliac peduncle; ip, ischial peduncle; of, obturator foramen; pb, pubic boot; PIFE1s, M. puboischiofemoralis externus 1 scar; PIFE2s, M. puboischiofemoralis externus 2 scar; pt, pubic tubercle.(a, b, e, f, i, j) in lateral view; (c, d, g, h) in anterior view.Scale bar = 50 mm.F I G U R E 6 Osteological correlates observed on the femur of Piatnitzkysaurus (right femur, PVL 4073).(a, b) lateral view; (c, d) anterior view; (e, f) medial view; (g, h) posterior view.Anatomical/muscular abbreviations: add1 + 2 s, II').Even though the dorsal rim of the iliac blade is not preserved in Condorraptor, a large, deep, almost circular concavity is anterodorsal to the acetabulum (Figure 2c,d); again suggesting the origins of the IFE and ITC (level I).Otherwise, due to the fragmentary nature of the specimen, it is not possible to delimit the boundaries of these muscle origins in this taxon; only to suggest relative general positions.Although the ITC and IFE origins in Marshosaurus must have been in a similar pattern, it is not possible to reconstruct this musculature because the anterior part of the ilium is not preserved and the figured ilium (Figure 2e,f) represents a plaster reconstruction of the preacetabular process.
terodistal to the lesser trochanter and anterodistal to the greater trochanter of the femur.It is thus possible to infer the subdivision of M. iliofemoralis in this taxon; IFE should have inserted onto the femoral trochanteric shelf (level II) and ITC onto the lesser/ anterior trochanter (level II) (Figures 6 and 9).The left femur of Condorraptor MPEF-PV 1690 has the base of the lesser trochanter anterolaterally located, also indicating a quite well-developed lesser trochanter in Condorraptor (and perhaps a fragment of the trochanteric shelf) and IFE and ITC muscle subdivisions (level II) ,b), being dorsally delimited by the preacetabular ridge, suggesting the PIFI1 origin (level I).In Condorraptor and Marshosaurus (Madsen, 1976), despite the fragmentary nature of the ilium of MPEF-PV 1687 (Figure 2c,d) and UMNH VP 6372 (Figure 2e,f), respectively, the same "cuppedicus" fossa is evident and inferred as the PIFI1 origin (level I).F I G U R E 8 Osteological correlate observed on the left fibula of Piatnitzkysaurus (PVL 4073).(a, b) lateral view.Anatomical/muscular abbreviations: ift, iliofibularis (fibular) tubercle; ilfbs, M. iliofibularis insertion scar.Scale bar = 50 mm.
-femoralis internus 2 (PIFI2) or M. iliotrochantericus cranialis (ITCR) and M. iliotrochantericus medius (ITM): The PIFI2 muscle in Crocodylia is considered to be the homologous to the PIFI3 in non-archosaurian Reptilia instead of the homonymous muscle; however, it is uncertain whether, in the avian lineage, PIFI2 was completely lost (in this hypothesis IF split into four muscles: IFE, ITC, ITCR, and ITM) or whether PIFI2

Figure 10 )
Figure 10)  and could be part of the PIFI2 origin (level II) as in Crocodylia, which also potentially originated near the PIFI1 on the ilium (I′).Only two posteriormost presacral vertebra are preserved in Condorraptor (MPEF-PV 1680 and 1700), with massive vertebral centra(Rauhut, 2005) similar to those of Piatnitzkysaurus and also possessing a wide and well-demarcated shallow concavity that could have been part of the PIFI2 origin (level II).No vertebrae associated

F I G U R E 9
Osteological correlates observed on the femur of Piatnitzkysaurus (right femur, PVL 4073).(a, b) lateral view.Anatomical/ muscular abbreviations: at, acessory trochanter; cfbs, M. caudofemoralis brevis insertion scar; ft, fourth trochanter; gt, greater trochanter; ifes, M. iliofemoralis externus insertion scar; istrs, M. ischiotrochantericus insertion scar; itcs, M. iliotrochantericus caudalis insertion scar; lt, lesser trochanter; pifes, Mm. puboischiofemorales externi insertion scar; pifi2, M. puboischiofemoralis internus 2 insertion scars; ts, trochanteric shelf.Scale bar = 20 mm.Avetheropoda (Allosauroidea + Coelurosauria;Paul, 1988) evolved a large accessory trochanter, as a "blade-like" structure that, although small, is also present in some ceratosaurs as well as early Tetanurae (BrissónEgli et al., 2016;Carrano et al., 2012;Cerroni et al., 2022; speculated that the ischial tubercle might be the origin of the "M.ischiofemoralis" (or homologous M. ischiotrochantericus, ISTR).However, we interpret the ischial tubercle on the distal ischial shaft of the PVL 4073 left ischium as the origin for the FTI1 in Piatnitzkysaurus, as a level II inference (Figure11a,b) (see below for rationale for the ISTR origin).The distalmost portion of the ischial shaft in the ischium of Condorraptor MPEF-PV 1689 is not well-preserved, with no sign of the ischial tubercle; thus, we made no inference of the FTI1 origin in this taxon.In the left ischium of Marshosaurus UMNH VP 6380, although not as discernible as in Piatnitzkysaurus, the ischial tubercle appears to be positioned on the medial shaft of the ischium (Figure 11d,e), similar in position to Piatnitzkysaurus and topographically equivalent to the FTI1 origin (level II).
(Figure 10c,d), and the same is noted in Marshosaurus UMNH VP 6380 (Madsen, 1976) (Figure10d,e).A few scars are evident in the obturator process of the ischium; thus, this region probably was the site of origin of the ADD1 (level I′).
Condorraptor, and Marshosaurus (Figure 11), as well as in other nonavian theropods (e.g., Triassic Coelophysis-Bishop et al., 2021; and F I G U R E 1 3 Possible Mm. adductores femores division on the Condorraptor distal right femur (MPEF-PV 1691, posterior view).Anatomical abbreviations: cdc, craniomedial distal crest, fg, flexor groove; lc, lateral condyle.Arrows indicate these scars that may or may not pertain to ADD 1 + 2 (see text).Scale bar = 50 mm.Cretaceous Tyrannosaurus-Carrano & Hutchinson, 2002) is indicative of the PIFE3 origin.However, variations in the size and shape of the theropod puboischiadic plate throughout evolution indicate some variation in the size of the musculature associated with this region (Lacerda et al., 2023).Although the PIFE3 origin's exact limits are undefined, it probably was located anteroventral to the ischial ridge on the posterior portion of the obturator process, similar to the position in Crocodylia and other theropod species, such Piatnitzkysaurus and Condorraptor is poorly known, and any vertebral elements from Marshosaurus lack formal description, it is possible to reconstruct the CFL origin in the South American piatnitzkysaurids.Two caudal vertebrae are preserved in Piatnitzkysaurus PVL 4073 specimen, probably the 2nd and 4th (Bonaparte, 1986); both feature robust centra and transverse processes related to the CFL origin (level I).Three caudal vertebrae are known for Condorraptor; the proximalmost, MPEF-PV 1702, has a tall centrum with posterodorsal oval depression and a dorsolaterally, slightly posteriorly directed transverse process; the proximal mid-caudal vertebra, MPEF-PV 1682, has a more elongated centrum with a shallow depression and a prominent laterally/slightly dorsally and posteriorly directed transverse process; and the distal mid-caudal vertebra, MPEF-PV 1683, has a low and elongated centrum with
tarsal shafts in Piatnitzkysaurus specimen MACN-Pv-CH 895 have a longitudinal crest and a proximal excavation.We thus infer the proximal parts of metatarsals II-IV (mainly metatarsal II) as the TA insertions (level I).Only the left metatarsal IV (MPEF-PV 1692) is preserved in Condorraptor; although it does not have the evident ridge present in Piatnitzkysaurus, the proximal portion preserves an excavation, considered here a TA insertion (level I).
Piatnitzkysaurus PVL 4073 preserve a depression on the posterolateralmost part of the distal femoral shaft; the right femur has some degree of rugosity on the lateral base of the tibiofibular crest.This posterolateral depression is topologically located in a position similar to extant Reptilia; thus interpreted here as the GL origin (level I) (Figure 6g,h).A depression similar in position and shape is present on the Condorraptor left femur MPEF-PV 1690, allowing reconstruction of the GL origin (level I) (Figure 7g,h).
left metatarsals II-IV in the Piatnitzkysaurus MACN-Pv-CH 895 specimen are well-preserved.The metatarsal of Piatnitzkysaurus has a ridge on the posterior/plantar surface related to the insertion of GL + GM (level II) (Figure 16a,b), although not so prominent as in other theropods (e.g., Majungasaurus-Carrano, 2007; Skorpiovenator-Cerroni et al., 2022; Tyrannosaurus-Carrano & Hutchinson, 2002).The left metatarsal IV (MPEF-PV 1692) of Condorraptor, as previously noted by Rauhut (2005) has a posterolateral semilunate ridge on the posterior/plantar surface (Figure 16c,d), which is the insertion site of the GL + GM heads (level I), presumably also inserting onto metatarsals II and III; not preserved (level I′).
on the tibiae of both Piatnitzkysaurus specimens, MACN-Pv-CH 895 and PVL 4073, covering almost the entire cnemial crest (except the anteroproximalmost part where the triceps femoris tendon should have attached) (Figure 12a-d).Although this broad depression exists in both specimens, in the PVL 4073 tibia, a ridge subdivides this depression into two subconcavities (Figure 12a,b).It is not clear whether these subdivisions signify an "extra head" of the GM (as reported in Crocodylia, which originates from the triceps femoris tendon-Suzuki et al., 2011).Regardless, the origin of the GM muscle seems to have been in this position.The GM origin reconstructed here in Piatnitzkysaurus is the entire medial depression on the cnemial crest (level I), resembling the large area of GM origin in Aves (Figure 12a-d).Likewise, the medial side of the cnemial crest in the Condorraptor tibia MPEF-PV 1672 also has a broad and shallow depression, positioned distally in relation to the triceps femoris tendon, representing the GM origin (level I) (Figure 12e,f).The insertion site of GL + GM was described above (Figure 16c,d).Mm. fibulares longus et brevis (FL, FB): The FL and FB origins (also termed M. peroneus longus et brevis and Mm.peronei anterior et posterior) in Testudines, Lepidosauria and extant archosaurs
the lateral ilium (I), in a rough and dorsoventrally delimited area Tibial cnemial crest (I) IT2 Dorsal rim of the ilium (I); anterior limits over the horizontal axis of the pubic peduncle, posterior limit over the horizontal axis of the posterior facet of the ischial peduncle Tibial cnemial crest (I) IT3 Posterodorsal rim of the ilium (I); posterior to the IT2, in the posterodorsal end of the postacetabular ilium Tibial cnemial crest (I) AMB Pubic tubercle (I), on the lateral shaft of the pubis Tibial cnemial crest (I) FMTE Lateral surface of the femoral shaft, delimited by the lia and lip (I) Tibial cnemial crest (I) FMTI Anteromedial surface of the femoral shaft, delimited by lia and la (I) Tibial cnemial crest (I) ILFB Shallow depression on the postacetabular surface of the ilium, ventral to IT3 (I) Fibular tubercle (I) IFE Elliptical concavity on the dorsolateral surface of the ilium (I); posterior to ITC and ventral to IT2 (II) Femoral trochanteric shelf (II) ITC Elliptical concavity on the lateral surface of the ilium (I), anterior to IFE (II) Lesser trochanter (anterior) of the femur (II) PIFI1 Preacetabular ventrolateral 'cuppedicus' fossa (I) Anteromedial surface of the femur, distal to the lesser trochanter (I) PIFI2 Centra of vertebrae anterior to ilium (I), and potentially near PIFI1 on ilium (I′) Anterolateral surface of the femur, distal to the lesser trochanter ('accessory trochanter') (I) FTI1 Distal ischial tubercle (II) Proximal posteromedial surface of the tibia in a broad depression posterior to the ILFB (I) Posteromedial surface of the proximal tibia (I) ADD1 Obturator process of the ischium (ischial apron) (I′) Posterior shaft of the femoral diaphysis (I′) ADD2 Depression on the posterodorsal ischial shaft, slightly distal to the ischial tuberosity (II) Posterior shaft of the femoral diaphysis (I′) PIFE1 Anterior surface of the pubic apron (II) Femoral greater trochanter (I) PIFE2 Posterior surface of the pubic apron (II) Femoral greater trochanter (I) PIFE3 Obturator process of the ischium; between ADD1 and ADD2 (II) Femoral greater trochanter (I) ISTR Medial surface of ischium/obturator process (II) Posterolateral side of the proximal femur, between the greater and fourth trochanter (I) CFB Iliac brevis fossa (II) Lateral surface of the fourth trochanter (I) CFL Centra and haemal arches of the caudal vertebrae (I), continuing distally until the transverse processes are strongly reduced/absent (I′) Pit and crest of the medial to posterior surface of the fourth trochanter (posterolateral surface of the distal femoral shaft (I) Posterior/plantar surfaces of metatarsals II-IV (I) GM Depression on the anteromedial proximal tibia (I) Posterior/plantar surfaces of metatarsals II-IV (I) TA Anterolateral proximal side of femoral condyle (I′) and/or depression distal to the cnemial crest of the tibia ( the ilium (I); anterior limits over the horizontal axis of the pubic peduncle Tibial cnemial crest (I) AMB Pubic tubercle (I), on the lateral shaft of the pubis Tibial cnemial crest (I) FMTE Lateral surface of the femoral shaft, delimited by the lia and lip (I) Tibial cnemial crest (I) FMTI Anteromedial surface of the femoral shaft, delimited by lia and la (I) Tibial cnemial crest (I) ILFB Shallow fragmentary depression on the postacetabular surface of the ilium (I) ?IFE Fragmentary concavity on the dorsolateral surface of the ilium (I); posterior to ITC and ventral to IT2 (II) Femoral trochanteric shelf?(II) ITC Fragmentary concavity on the lateral surface of the ilium (I), anterior to IFE (II) Lesser trochanter (anterior) of the femur? the ischium (ischial apron) (I′) Posterior shaft of the femoral diaphysis (I′) ADD2 Depression on the posterodorsal ischial shaft, slightly distal to the ischial tuberosity (II) Posterior shaft of the femoral diaphysis (I′) PIFE1 Anterior surface of the pubic apron (II) ?PIFE2 Posterior surface of the pubic apron (II) ?PIFE3 Obturator process of the ischium; between ADD1 and FTI3? + ADD2 (II) ?ISTR Medial surface of ischium/obturator process (II') ?CFL Centra and haemal arches of the caudal vertebrae (I), continuing distally until the transverse processes are strongly reduced/absent (I′) Pit and crest of the medial to posterior surface of the fourth trochanter (I) EDL Anterolateral proximal shaft of the tibia (I) ?GL Depression on the posterolateral surface of the distal femoral shaft (I) Posterior/plantar surface of metatarsals II-IV (I) GM Depression on the anteromedial proximal tibia (I) Posterior/plantar surface of metatarsals II-IV (I) TA Anterolateral proximal side of femoral condyle (II') and/or depression distal to the cnemial crest of the tibia (II) Anteroproximal metatarsals II-IV (I) the ilium (I); posterior to the IT2, in the posterodorsal end of the postacetabular ilium ?AMB Pubic tubercle (I), on the lateral shaft of the pubis ?ILFB Shallow depression on the postacetabular surface of the ilium, ventral to IT3 (the ischium; between ADD1 and ADD2 (II) ?ISTR Medial surface of ischium/obturator process (II) ?CFB Iliac brevis fossa (II) ?F I G U R E 17 Pelvic and hindlimb 'muscle map' inferred for Piatnitzkysauridae (left lateral view).(a) Piatnitzkysaurus floresi.(b) Condorraptor currumili.(c) Marshosaurus bicentesimus.

F I G U R E 1 8
Restoration of pelvic and hindlimb muscles in Piatnitzkysaurus floresi (left lateral view).Artwork by Júlia d'Oliveira.such as FTI2 and PIT (e.g.,Hutchinson, 2002).Some features inferred for piatnitzkysaurids, such as the tibial triceps femoris insertion, three IT heads and two FMT heads, ILFB insertion onto the fibular tubercle, two ADD heads, and insertions of the PIFI1 + 2, FTI3, FTE, ADD1 + 2, PIFEs, ISTR, CFB and CFL are plesiomorphic muscular conditions (for Archosauria, Reptilia or earlier) that are relatively conservative even through evolution to Aves.Others evolved later on the avian stem, such as loss of the FTI1 and PIFE3, and shifts of the PIFI1 + 2, CFB, and ISTR origins from more medial to lateral.Lower limb muscle origins and insertions have more complex, and sometimes more ambiguous, evolutionary patterns, evident in piatnitzkysaurids.However, the insertions of Mm. gastrocnemii and M. extensor digitorum longus (although not reconstructed here, but piatnitzkysaurids probably shared a similar insertion to other theropods) moved distally, relative to crocodylians.This presumed change is related to the evolutionary transformations on the lineage to Aves/Neornithes (and bipedal locomotion).

3 . 4 .
).Consequently, the AMB moment arms about the hip joint (e.g.,Allen et al., 2021) should have varied among piatnitzkysaurids.Extent of ADD1 origin in Piatnitzkysaurus and Condorraptor-the shallow depression present on the ischium of both taxa (not observed in Marshosaurus), extends more distally in Condorraptor (Figures 11 and 17), and again these differences could change the maximal forces generated and moment arms about the hip joint (e.g., Allen et al., 2021).Triceps femoris tendon in Piatnitzkysaurus and Condorraptorthe cnemial crest in Condorraptor is only moderately developed (Figures 4b,c and 12e,f), differing from Piatnitzkysaurus, which has a well-developed and nearly rectangular crest (Figures 4a,b, 12a-d

5 .
TA origin in Piatnitzkysaurus and Condorraptor-the second TA head (tibial) appears to have been more robust in Piatnitzkysaurus (Figures 5a-d and 15); perhaps indicating greater force-generating capacity.However, considering that the Condorraptor speci-

Table 5
Reconstruction of character states for the Tetanurae node after scoring Piatnitzkysaurus floresi.States 01 and 012 represent ambiguous reconstructions.
TA B L E 5