A new massopodan sauropodomorph from Trossingen Formation (Germany) hidden as ‘Plateosaurus’ for 100 years in the historical Tübingen collection

A literature review showed that there is not a defined consensus on what specimens belong to Plateosaurus in current phylogenetic analyses, and after the assignation of SMNS 13200 as the neotype for Plateosaurus, the specimen composition of Plateosaurus as an operational taxonomic unit (OTU) needs to be addressed in further iterations of phylogenetic analyses. At least one of the specimens used to illustrate plateosaurian anatomy contains several characters identified in more derived sauropodomorphs commonly referred to as massopodans. This partial skeleton, traditionally known as specimen ‘GPIT IV’, was found in the lower dinosaur bone bed of the Obere Mühle, a Trossingen Formation outcrop, during an excavation in 1922 near the city of Tübingen, Germany. The holotype of Plateosaurus trossingensis and several other specimens referred to as this species were found in this level, which was initially interpreted as a synchronic deposit of animals. However, the current understanding of the Trossingen Formation indicates that this bed was probably a constant accumulation of carcasses through miring and transport down a river for hundreds of years. In this work, a framework to compare phylogenetic signals with morphological and histological data is provided to help in the species delineation of Plateosaurus, and support is found to refer the historic specimen ‘GPIT IV’ as a new genus and a new species.


Introduction
lected from localities near Tübingen, namely Trossingen, Bebenhausen, Pfrondorf, Kreßbach, and Steinenberg, but also from the beds near Aixheim and Löwenstein (Hinz and Werneburg 2019). Based on the records published so far, the material comes from several sections of the Upper Keuper Marls (Hinz and Werneburg 2019). The material has been referred to as the genus Plateosaurus Meyer, 1837, of which only four out of about 20 species are treated as valid today, i.e., P. 'engelhardti ' Meyer, 1837, P. longiceps Jaekel, 1913, P. ingens (Rütimeyer, 1856 and P. gracilis (von Huene, 1905) (see Tables 1 and 2 for references). Von Huene (1901Huene ( , 1905aHuene ( , b, 1907Huene ( , 1915Huene ( , 1926Huene ( , 1932Huene ( , 1956Huene ( , 1959 recognised as many as 22 species in Europe, and eight were found in Trossingen alone. Galton (2001b) interpreted this as a "sort of microcosm for plateosaurids of Germany". After revising the material, Galton (2001b) reduced the number of species to four, considering many as nomina dubia. The assignation was a consequence of the definition of Plateosaurus at this time, which was based on cranial and femoral characters. Several specimens housed in the GPIT collection lack cranial and femoral material. However, the material quality is similar to the conservation of distinct taxa elsewhere in South America and South Africa. Following Galton's (2001aGalton's ( , 2001b) taxonomic revision, many specimens and their variation were interpreted as either individual variation or as evidence of phenotypic plasticity. The different morphological characteristics of femora, and the overall distinction between gracile and robust categories, were explained as evidence of sexual dimorphism (Galton 1999) or ontogenetic stages (Galton 1973(Galton , 1985cGalton and Upchurch 2004). Although supported by morphometric analyses, the phenotypic plasticity is not consistent with endothermic animals (Weishampel and Chapman 1990); it was, therefore, suggested that the material from Trossingen corresponded to a transitional evolutionary stage towards dinosaurian endothermy (Sander and Klein 2005). Nevertheless, the histological evidence, namely the fibrolamellar complex in the long bones, corresponds to more endothermic animals (Ray et al. 2009).
The idea of ontogenetic changes has been supported by what has been found in other localities. For instance, in Argentina, several stages of Mussaurus Bonaparte and Vince, 1979, can be identified, showing a transition from biped to quadrupedality as they grew (Otero et al. 2019). On the other hand, in South Africa, Massospondylus Owen, 1854, also shows ontogenetic shifting, but, unlike Mussaurus, cranial evidence such as the inner ear anat- Figure 1. Summary of the taxonomic history of Late Triassic sauropodomorphs. The colour code corresponds to the decades in which the specimens used as holotypes for new genera were first collected, and, to the right, there is a list that indicates in which year a description of the specimen was first published, even if it was not a formal or detailed description. The taxa are grouped by regions, showing their alpha diversity. The time separating the discovery from the description does not mean that the specimen was not used as reference material in comparative anatomy. Although roughly half of the specimens used to erect the genera in this chart were discovered before or during the 1960s, only five genera were erected in the same interval. The infographic helps to illustrate that taxonomic revision of the material stored in collections around the world must be considered a constant work in progress. GPIT-PV-60296 These specimens were referred to as GPIT A, and it is unclear if this is meant to be one individual, a composite mounted to illustrate one species or several specimens collected over time that tentatively belonged to the same species.

Taxonomic opinions
Quenstedt used the name Zanclodon laevis Plieninger, 1846, but was later referred to as a new species, Zanclodon quenstedti Seeley, 1892 (see Koken 1900), then referred to as Plateosaurus quenstedti von Huene, 1905. According to Galton (1999Galton ( , 2001a, the lack of diagnostic characters in the sacrum and the skull material made this a nomen dubium. However, due to the uncertainty of its provenance, it is not clear whether this number refers to a specimen.

GPIT B GPIT-PV-60293
Found during a construction site at Roter Graben near Bebenhausen in 1870, when Forstrat Tscherning collected and sent them for study to Quenstedt (von Huene 1907: 127).

GPIT-PV-30790
It was discovered in 1881 by Jacob Hildenbrand from Ohmen when doing topographic prospections in Wüstenroth, southeast from Löwenstein.
Quenstedt acquired it for the palaeontological collection in Tübingen; it was until 1901 that the material, which Hildebrand put inside a cement block (von Huene 1907: 138, pl. 56-59). Von Huene mounted it after he prepared and isolated the elements.
It was discovered in 1864 and given the old number PV 11297, collected from a deep rift in Brandklinge, near Jächklinge. However, von Huene (1907: p. 146) stated this finding was done in the "late 1870s". Currently, in the collection, it is difficult to determine which material belongs to GPIT D and which belongs to GPIT A, both obtained from similar localities.
It is impossible to distinguish between the collection sites of Brandklinge and Jächklinge, and every collection element now has independent numbers. Von Huene (1907: p. 146) described them as "a large number of worthless fragments".

Taxonomic opinions
The large size of the metatarsals (mt. II being 52 cm long) was used as a condition to erect the species Pachysaurus giganteus von Huene, 1932.

Taxonomic opinions
Von Huene (1932) referred the specimen to Plateosaurus quenstedti von Huene, 1905. Galton (2001b placed it along with AMNH FARB 6810 as part of Plateosaurus longiceps Jaekel, 1913. omy and the locomotory apparatus suggests that Massospondylus carinatus Owen, 1854, retained bipedality through its life (Neenan et al. 2018). Finally, the discovery of what seems to be juvenile specimens of Plateosaurus from Frick in Switzerland could hint at there being some ontogenetic stages preserved in the Trossingen area (Hofmann and Sander 2014). However, as it is, the faunal composition of Germany is at odds with the faunal composition pattern identified in other well-studied tetrapod communities from the Late Triassic. For instance, in South Africa, Lesotho, and Zambia, there is an assortment of gracile bipedal animals, such as Nyasasaurus Nesbitt et al., 2012 (see Baron et al. 2017), and Plateosauravus von Huene, 1932(see McPhee et al. 2017, and large, robust quadrupeds like Meroktenos Peyre de Fabrègues and Allain, 2016, Melanorosaurus Haughton, 1924(see McPhee et al. 2017, and Eucnemesaurus van Hoepen, 1920(see McPhee et al. 2015a (Fig. 1). On the other hand, in western and southern Argentina, we have large-sized bipedal animals (Mussaurus Otero et al., 2019) and medium to large-sized quadrupeds (Ingentia Apaldetti et al., 2018); eastern Argentina shows a more homogenous composition in terms of morphology but is equally diverse in identified groups. Moreover, in the Bristol archipelago, in what is today the United Kingdom, which was facing southwards relative to Germany, France and Switzerland Lovegrove et al. 2021), a heterogeneous community existed both in morphological and phylogenetic terms, with very early diverging sauropodomorphs such as Thecodontosaurus Riley and Stutchbury, 1836, and Pantydraco Galton et al., 2007(Lovegrove et al. 2021, coexisting with the melanorosaurid Camelotia Galton, 1985d, which has a more robust and massive constitution (Galton, 1985d) (Fig. 1).
Furthermore, the recent redescription of a sauropodiform dinosaur from the Klettgau Formation of Switzerland, Schleitheimia Rauhut et al., 2020, which was previously referred to as Plateosaurus (Galton 1986), strengthens the argument that much of the diversity of sauropodomorphs of Germany has been hidden under the 'Plateosaurus' umbrella. The referral to Plateosaurus for all the material from Germany assumed that there are discrete taphonocoenoses and that all individuals represent a coherent population (Sander 1992;Galton 1999Galton , 2001a. The sauropodomorph diversity likely originated during the Late Triassic through sympatric speciation, such as niche partitioning (McPhee et al. 2015b). Stratigraphy alone cannot be a criterion to define early diverging sauropodomorph species, as the same taphonocoenosis can contain several lineages (e.g., the Rhaetic bonebed, Galton 2005). However, a recent study on cranial morphology of skulls referred to as Plateosaurus has found that the characters in 18 specimens show a high degree of variation without the consistent combination that may indicate different species -a pattern consistent with intraspecific variation (Lallensack et al. 2021). The sample in Lallensack et al. (2021) includes mostly material from the lower and middle bonebeds of the Gruhalde Quarry, Klettgau Formation in Frick, Switzerland, and also GPIT-PV-30784 (formerly designated as "GPIT/RE/09392", housed in Tübingen), and SMNS 12949, SMNS 13200, SMNS 5297, SMNS 52968, and SMNS 1950 (housed in Stuttgart) from Trossingen, as well as MB. R.1936 and MB.R.4430.1 (housed in Naturkundemuseum Berlin) from Halberstadt in Germany.
The material of Schleitheimia was collected in the 1950s by Emil Schultz . Galton (1986) referred the material to the species P. 'engelhardti'. The material was recently redescribed ) based on the ilium and femur morphology. The ilium is one of the most variable elements amongst early diverging sauropodomorphs. There was constant "experimentation" and innovation in the locomotion, with many animals evolving toward an obligate bipedal stance and others with a clear trend toward quadrupedalism (Fig. 1). Regalado Fernandez (2019) suggested quadrupedality may have originated at least twice. In some cases, such as in sauropodomorphs of a mussaurid morphotype, the quadrupedality may have evolved through paedogenesis.
In the present contribution, we provide a revision of the taxonomic history of Plateosaurus and its usage in the literature, and we do a preliminary assessment of several characters that have been identified as varying from spe-cies to species in other sauropodomorphs from other Late Triassic communities in the material that has been previously referred to Plateosaurus housed in the University of Tübingen collection (see Table 1 for an inventory of specimens with old and new catalogue numbers).

Usage of Plateosaurus in phylogenetic analyses
The specimens that have been included as part of the operational taxonomic unit (OTU) of Plateosaurus through time have not been constant. Two matrices with different taxonomic and character compositions were produced in 2007, namely by Upchurch et al. (2007) and Yates (2007) (Fig. 2). These matrices were subsequently modified through eight iterations until 2019. Table 2 summarises the different specimen compositions of the OTU Plateosaurus and the number of species considered valid. In 2019, the International Commission on Zoological Nomenclature (ICZN) resolved to replace the name of the type species, P. 'engelhardti', whose holotype is UEN 552, by the type species P. trossingensis Fraas, 1913, and its type specimen SMNS 13200 (ICZN 2019). Therefore, the current consensus on Plateosaurus seems to treat GPIT-PV-30784, and AMNH FARB 6810 as syntypes of P. trossingensis, with P. erlenbergiensis von Huene, 1905 being a junior synonym of P. trossingensis. P. longiceps is the name given to the Halberstadt material. P. gracilis corresponds to the name given to the material from the Untere Mühle, the Stromberg region quarry 'Weißer Steinbruch' (Pfaffenhofen), and the quarry 'Goessel' (Ochsenbach) [sic] (Moser 2003;Galton 2012).
Finally, P. ingens is given to the material from Niederschönthal near Füllinsdorf, in Switzerland (syntypes NMB NB 1582, 1584, 1585, 1875. Material from Frick (MSF1-13) was described as part of P. 'engelhardti', and thus P. ingens was considered as a junior synonym of P. 'engelhardti' (Galton 2012). In Novas et al. (2011), all these specimens were included in P. ingens, and the composition of this OTU does not seem to have changed through the literature. Nevertheless, additional material from the same locality has also been referred to as P. trossingensis (= P. 'engelhardti' in Galton 1986) rather than P. ingens, as it has been used in the iterations of phylogenetic analyses. Furthermore, P. ingens was initially the type species of Gresslyosaurus Rütimeyer, 1856, a genus that has been considered distinct from Plateosaurus (Moser 2003) and recently re-erected as valid . Therefore, the taxonomy of the new material from Frick described in Lallensack et al. (2021) needs to be revisited.

Taxonomy of Plateosaurus trossingensis (= Plateosaurus 'engelhardti')
The lectotype material of Plateosaurus 'engelhardti' used to comprise seven bones that did not belong to the same individual as they form part of an allochthonous assem- This lamina divides the sacral rib into an anterior portion that connects to the lateral portion of the first sacral and a posterior portion that connects to the centrum of the second sacral. Although the sacral vertebrae seem to be co-ossified, the suture between sacral 1 and sacral 2 does have a distinct fissure, but this fissure is not discernible between sacral 2 and the caudosacral. The morphology of the UEN 552 sacrum inspired coding new sacral characters to describe Plateosaurus taxonomy. However, many sacral characters have been found to vary from species to species. In Plateosaurus (= 'Sellosaurus') gracilis, it has been recognised that there are two types of sacra: type I, which involves a dorsosacral, primordial sacral 1 and primordial sacral 2, and type II, interpreted as primordial sacral 1, primordial sacral 2 and caudosacral (Galton 1999(Galton , 2000. The differences were interpreted as either evidence of sexual dimorphism (Gal- Figure 2. Iterations on the matrices originally built by Upchurch et al. (2007) and Yates (2007)  ton 2000) or homeotic transformation Upchurch 2000, Galton 2001b). Type I (dorsosacral, primordial sacral 1 and primordial sacral 2) is the most common condition among early-diverging sauropodomorphs, as in Ruehleia Galton, 2001b, andEfraasia Galton, 1973, taxa from Germany, and the persistent condition in Massospondylidae (Wang et al. 2017). Type II is only seen in Plateosaurus, and many of the specimens that Galton (2001b) regarded as type I have also been reinterpreted as type II (Moser 2003).
The specimen GPIT "Aixheim" corresponds to a type II sacrum, concordant with the UEN 552 holotype. However, there is no ventral keel in the sacrals, and the fossa lateral to the keel is not that pronounced, but the sutures between the sacrals 1 and 2 are present, and there is a centrodiapophyseal fossa present as well. The difference in size between UEN 552 and GPIT "Aixheim" could explain the changes in those characters as ontogenetic. In both specimens, the identity of the sacrum cannot be confidently asserted without seeing the morphology of the rib or the arrangement with the ilium. GPIT "Aixheim" seems different from the specimens with the current numbers GPIT-PV-60364, GPIT-PV-60446 and GPIT-PV-60448, specimens incorrectly associated with Aixheim in the literature (Fig. 3). See further discussion in the "Analysis 2" section.

The taxonomic history of other Central European genera of Late Triassic sauropodomorphs
The name Gresslyosaurus ingens was first given to material from Niederschönthal, canton Basel-Landschaft, Switzerland, which was discovered by the Swiss palaeontologist Amanz Gressly (Rütimeyer 1856). The holotype of Gresslyosaurus ingens (specimen numbers: NMB BM 1, 10, 24, 53, 530-1, 1521, 1572-74, 1576-78, 1582, 1584-85, 1591) includes a partial sacrum, four caudal vertebrae, a metacarpal, partial left and right tibiae, an almost complete fibula and pedal elements. During the late 19 th century and early 20 th century, Gresslyosaurus was applied to several remains in Switzerland and Germany. Gresslyosaurus robustus von Huene, 1905, includes only the specimen historically referred to as 'GPIT B', currently GPIT-PV-30786, from the Trossingen Formation . Bar chart summarising the frequency each specimen was used as either P. 'engelhardti' alone, as P. gracilis, as P. ingens, as P. longiceps or as 'P. erlenbergiensis'. Although P. trossingensis (P. 'engelhardti' at the time) was considered the only valid species, several species were also used as part of Plateosaurus. Most phylogenetic analyses base their character scores on the same specimens: SMNS 13200, BSP 1962 XLVI and GPIT-PV-30784. P. gracilis is the second species more frequently used in phylogenetic analyses. The skull collected from the younger Stubensandstein, former specimen GPIT 18318a, illustrated in von Huene (1915) and Galton (1985c), left the GPIT collection at some point in 2004, and it is not clear if several works refer to the skull of GPIT I as P. gracilis. The collection recovered the skull in 2022. Specimens GPIT 18064, GPIT 18318a and GPIT 18392 have been used as part of P. gracilis, but they are not given new catalogue numbers because it is not clear if they each correspond to one individual. Specimen GPIT-PV-30787 has not been explicitly used in phylogenetic analyses as part of the OTU for Plateosaurus 'engelhardti'.  , 10, 24, 53, 530-1, 1521, 1572-74, 1576-78, 1582, 1584-85, 1591 referred to 'Plateosaurus engelhardti' (Galton 1985a) referred to 'Plateosaurus longiceps' (Galton 2001a,b) Dimodosaurus poligniensis Pidancet and Chopard, 1862 POL 31-32, 57, 70, 76 postcrania are referred to as Gresslyosaurus ingens (von Huene 1907) removed from Plateosaurus and referred to as Camelotia borealis (Galton 1985d)
The specimen SMNS 13200 has effectively been used as the reference specimen for P. 'engelhardti' before being officially defined as the holotype for the species P. trossingensis (ICZN 2019). GPIT-PV-30784 ("GPIT I")

1930-1940
Pachysaurus giganteus von Huene, 1932 GPIT E referred to as Pachysauriscus giganteus (Kuhn 1959) referred to as Gresslyosaurus giganteus (Steel 1970) referred to as 'Plateosaurus engelhardti' (Galton 1985c) considered a nomen dubium (Galton 2001a) Pachysaurus wetzelianus von Huene, 1932 GPIT V referred to as Pachysauriscus wetzelianus (Kuhn 1959) referred to as Gresslyosaurus wetzelianus (Steel 1970) referred to as 'Plateosaurus engelhardti' (Galton 1985c) considered a nomen dubium (Galton 2001a) Plateosaurus fraasianus von Huene, 1932 SMNS 13200 the name replaced by Plateosaurus trossingensis and GPIT-PV-30785 ("GPIT II") have also been used as reference material for P. 'engelhardti', and although GPIT-PV-30785 is a composite of at least two similarly sized individuals, these two come from the same locality and could safely be considered the same species as SMNS 13200. Furthermore, AMNH FARB 6810 comes from the same site where GPIT-PV-30784 and GPIT-PV-30785 were excavated. Therefore, in cases where GPIT-PV-30784 and AMNH FARB 6810 are used instead of the neotype, these two specimens could be considered syntypes for P. 'engelhardti' given the large amount of literature on them. Nevertheless, in the literature, AMNH FARB 6810 has been referred to as P. longiceps or P. erlenbergiensis. The holotype of P. longiceps is MB.R.1937, and the holotype of P. erlenbergiensis is SMNS 6014, coming from the localities of Halberstadt and Ellenberg, respectively. Moreover, the specimen SMNS 13200 is the holotype of P. trossingensis, which made P. trossingensis a junior synonym of P. 'engelhardti' (Galton 1999(Galton , 2000(Galton , 2001a. In the case of the Halberstadt material, it has been considered as part of P. longiceps based on the association within the bonebed and because of two cranial autapomorphies. These autapomorphies are 1) a medially directed peg on the palatine, a sub-vertical lamina between the basipterygoid processes, and 2) the combination of characters of a diapophysis from sacral 1 forming a broad sheet with a semicircular outline with a narrow distal half on the adjacent edge of the first sacral rib, a diapophysis from sacral 2 posterolaterally directed and tapering gradually, a sigmoid femur, articular end surfaces of the anterior caudal centra are sub-parallel rather than wedgeshaped (Galton 2001b). The two cranial autapomorphies have also been found in AMNH FARB 6810 (Prieto-Márquez and Norell 2011). Nevertheless, several cranial characters have been interpreted as displaying a wide range of variability in Plateosaurus. For instance, GPIT-PV-30784 and AMNH FARB 6810 (Prieto-Márquez and Norell 2011;Lallensack et al. 2021) have five premaxillary teeth, whereas SMNS 13200 has six (Galton 1976;Lallensack et al. 2021) as well as the holotype of P. longiceps (MB.R.1937;Galton 1985). Therefore, a revision of the material from Halberstadt is needed to understand the combination of cranial and postcranial characteristics compared to the rest of the plateosaurian material from Central Europe. The holotype of P. erlenbergiensis, specimen SMNS 6014, from the Knollenmergel of Ellenberg, contains a neurocranium with a sub-vertical lamina between the basipterygoid processes. Furthermore, Galton (2001b) noted that specimen GPIT-PV-30785 was referred to as P. erlenbergiensis due to the postcranial anatomy. Therefore, P. erlenbergiensis could be considered a junior synonym of P. 'engelhardti' (Fig. 3).
The ilium of SMNS 80664 is incomplete and lacks most of the dorsal margin; however, its morphology is distinctively different from the one in GPIT-PV-30787. The preacetabular process in SMNS 80664 is short and with a triangular outline, whereas GPIT-PV-30787 is more quadrangular and more anteriorly expanded. On the postacetabular side, GPIT-PV-30787 bears a distinct brevis-fossa that gives the posterior margin an M-shaped outline, whereas the posterior margin in SMNS 80664 bears a reduced brevis fossa and a straighter-sided margin. SMNS 80664 has a more distinctive 'plateosaurian' morphology, i.e., like SMNS 13200, referred to as P. 'engelhardti' and one of the most preserved skeletons. Here, we describe specimen GPIT-PV-30787 as a new sauropodomorph that shows affinities to more derived non-sauropod sauropodomorphs.
The GPIT collection houses five historical catalogue numbers that have been used as specimens in the literature, although they refer to several individuals collected in the same expedition and not necessarily to one individual. These historical catalogue numbers are "GPIT A", "GPIT B", "GPIT C", "GPIT D", and "GPIT E", and a detailed breakdown of their new catalogue numbers is provided in Table 2. Specimens of GPIT A-D were collected during the late 19 th century. On the other hand, the specimens in the diorama are numbered GPIT I-IV. Except for GPIT VII, all the specimens in the diorama were collected from the Obere Mühle locality, as well as specimen GPIT E. GPIT C are fragments belonging to GPIT VII (see Table 2).

The specimens from Obere Mühle
Specimen SMNS 13200 was excavated from Obere Mühle in the summer of 1912 (Schoch 2011). The Obere Mühle outcrop has been the most productive site in the Trossingen locality, where 65 skeletons have been excavated (Schoch and Seegis 2014). The next round of excavations from 1921 to 1923 was organised by Friedrich von Huene (Tübingen University) and mainly financed by William Diller Matthew (American Museum of Natural History, New York) (Reinacher 2021). Half of the findings were sent to the AMNH, and the other half stayed at GPIT. This excavation yielded 12 skeletons (Schoch 2011;Schoch and Seegis 2014;Reinacher 2021), with seven staying in Tübingen. The specimen AMNH FARB 6810 was mounted and remained in the New York exhibition, and some other material was sent to the Museum of Comparative Zoology at Harvard University (Sander 1992). Of the material that remained in Tübingen, the complete skeleton was mounted, and two others were joined into a composite (Weishampel and Westphal 1986;Sander 1992). These two skeletons correspond to "GPIT I" (GPIT-PV-30784) and "GPIT II" (GPIT-PV-30785), respectively. Because the four specimens mentioned above, namely SMNS 13200, AMNH FARB 6810, GPIT-PV-30784, and tentatively GPIT-PV-30785, come from the same bonebed, known as the "Plateosaurus-bonebed", it is likely that these four specimens belong to the same species. These four specimens have been consistently referred to as P. 'engelhardti' in the literature. A recent study on the morphological variation clusters "GPIT I" (GPIT-PV-30784) and "GPIT II" (GPIT-PV-30785) and SMNS 13200 as part of the same group, neatly separated from Ruehleia and Efraasia (Lefebvre et al. 2020).
A third excavation in the quarry was organised by Reinhold Seemann in 1932, in an expedition that lasted for six months and recovered 65 bones, most of which are stored at the SMNS (Schoch 2011). The Plateosaurus collection at Stuttgart is also likely monospecific.
Specimen GPIT-PV-30787, historically known as "GPIT IV", was initially referred to as Plateosaurus 'plieningeri' (see Table 2). The holotype of P. plieningeri is specimen SMNS 80664 (von Huene 1907). It comprises 17 vertebrae, including the sacrum, ribs, ilium, pubis, distal ends of the femur, a fibula, metatarsal V, and the ends of other metatarsals found in 1847 by Theodor Plieninger in a bonebed at Degerloch, from the Knollenmergel, Trossingen Formation (now a district of Stuttgart) (Galton 2001a). The Degerloch specimen became the holotype of Gresslyosaurus plieningeri (von Huene 1907) and was referred to as Plateosaurus by von Huene (1932). Wellnhofer (1993), in a study on the stance of Plateosaurus, considered specimen SMNS 80664 as P. 'engelhardti', since it was a younger, more robust specimen compared to the older and more gracile specimens from Trossingen. A recent morphometric analysis shows strong evidence that SMNS 80664, along with GPIT-PV-30784 ("GPIT I") and GPIT-PV-30785 ("GPIT II", a composite -see Table 2), cluster together with P. 'engelhardti' and attribute the variability in the morphology to taphonomic effect (Lefebvre et al. 2020). However, the analysis did not include specimen GPIT-PV-30787.
According to Galton (2001a), P. 'plieningeri' included two Trossingen specimens: SMNS 80664, housed in Stuttgart, and specimen GPIT-PV-30787 ("GPIT IV"), housed in Tübingen. Nevertheless, specimen GPIT-PV-30787 has more preserved elements than the holotype of P. 'plieningeri', as it includes a complete pelvic girdle, the anterior and distal portion of the caudal vertebrae series, a fibula, a tibia, metatarsal I, and digits II and III and a femur. Galton (2001a) considered this specimen part of P. 'engelhardti' and identified it as a large female ( Table  2), based on comparisons with crocodiles and Tyrannosaurus Osborn, 1905, given the morphology of the femur and the small first chevron. Galton (2001a) included specimen SMNS 80664 as one of the historical referrals to P. 'plieningeri', but he concluded that the palatine peg and the vertical lamina between the basipterygoid process corresponded to another species, P. longiceps. The status of P. 'plieningeri' (von Huene 1907) is as a junior synonym of P. longiceps but restricted only to SMNS 80664.
The review of the literature presented here shows that there are two definitions of Plateosaurus: a phylogenetic definition with an inconsistent specimen composition, and a morphological definition that includes mostly the same specimens. Specimen GPIT-PV-30787 has been used as part of the morphological definition but has never been included in the phylogenetic or morphometric definition. In this work, we provide a framework to contextualise phylogenetic data with morphological data to help delimiting the specimen-composition of Plateosaurus as a taxonomic unit. Specimen GPIT-PV-30787 was added to a character-by-taxon matrix that included two species of Plateosaurus and early diverging sauropodomorphs from Central Europe Efraasia, Ruehleia, Schleitheimia and Ohmdenosaurus Wild, 1978 Otero et al. (2015) but amended the definition of P. trossingensis (there as P. 'engelhardti') to specimens AMNH FARB 6810, SMNS 13200 and SMNS 91310 (=F65). Rauhut et al. (2020) used this character-by-taxon matrix, making it a good candidate to add specimen GPIT-PV-30787 without a noisy signal. Other than the changes outlined in the following paragraphs, the characters that were considered as ordered by  are treated as such here, namely Chs. 8,13,19,23,40,57,69,92,102,117,121,129,132,145,148,150,151,158,168,170,171,178,210,213,232,237,245,254,263,268,282,295,316,322,330,352,365,368,370 and 375. Table 3 details the specimens and sources used for comparative purposes in this study.

Institutional
Several characters were changed (Appendix 1). Ch. 318 refers to a paramarginal ridge on the lateral surface of the cnemial crest. Since the character was not illustrated, it is not easy to know what structure this refers to and has been removed. It was initially scored by Yates (2007) for Lufengosaurus Young, 1940 (specimens IVPP V15, andillustrations in Young 1941;Barrett et al. 2005a), Yunnanosaurus huangi Young, 1940 (specimens IVPP V94, V505, and illustrations in Young 1942Young , 1951 and Jingshanosaurus Yang, 1995 (specimen LV 3, andillustrations in Zhang andYang 1994), however, upon first-hand examination of the tibia in Lufengosaurus (IVPP V15) and Yunnanosaurus huangi (IVPP V20, the one illustrated in Young 1941), it was not possible to identify the paramarginal ridge, neither based on the illustrations or the descriptions for Jingshanosaurus.
Ch. 380, which refers to the femoral length alone, was removed. It was not possible to replicate the character states using gap-coding, and several specimens could refer to juveniles, complicating the scoring of this character. Furthermore, there is already a character on the tibia:femur ratio in the matrix.
Two more characters must be reinterpreted, namely Ch. 379, 'Growth marks (LAGs or annuli) in the cortex' and Ch. 381, 'Relative abundance of parallel-fibered bone (PFB) and woven fibered bone (WFB)'. These characters were proposed by Cerda et al. (2017) and assessed in Riojasaurus Bonaparte, 1969, Coloradisaurus Galton, 1990, Massospondylus, Adeopapposaurus Martinez, 2009, Leyesaurus Apaldetti et al., 2011, Mussaurus, Leonerasaurus Pol et al. 2011, Lessemsaurus Bonaparte 1999, Volkheimeria Bonaparte, 1979 and Pa- Figure 4. Diagram outlining the workflow of the phylogenetic analyses carried out in this work. The character-by-taxon matrix has 67 taxa and 380 characters. The number of trees obtained from each iteration is shown next to the name with their length. A stand for the character-by-taxon matrix and the elements inside the curly brackets represent the OTUs removed before the analysis was run. The constrain refers to forcing GPIT-PV-30787 inside a clade with Plateosaurus. The Templeton Test in the two iterations only compared the total evidence trees (iteration 1.1 against iteration 1.5 and iteration 2.1 against iteration 2.5). The original character-by-taxon matrix by Rauhut et al. (2020) includes P. trossingensis as P. 'engelhardti', including several specimens (see text), and only two characters scored as polymorphisms: Ch. 17 (scored as 0 and 1), and Ch. 138 (scored as 0 and 1). In the second analysis, P. trossingensis is restricted to specimen SMNS 13200, where Ch. 17 is scored as 0, and Ch. 138 is scored as 1.
tagosaurus Bonaparte, 1979. Apaldetti et al. (2018 incorporated these two characters from Cerda et al. (2007) and assessed them on Ingentia. The scores on Plateosaurus were based on the histological analyses performed by Klein (2004) and Sander and Klein (2005). Several specimens were sampled as part of Plateosaurus, based in collections in SMNS, AMNH, MSF, NAA, PIMUZ, SMA, and GPIT (there referred to as the IFG, "Institut für Geowissenschaften"). For this second analysis, where all the other Plateosaurus material was removed, and the OTU was restricted to the neotype SMNS 13200 and GPIT-PV-30784, this character was rescored based on the results reported in those two works. Two specimens from the GPIT collection were mentioned, one referred to as 'IFG, exhibition', and the other as 'IFG, compactus' (archive). The femur and tibia belong to GPIT-PV-30787, the scapula and the humerus used in that study have not been located yet, and the trunk vertebrae used (Rückenwirbel [RW]: RW12 and RW14) are part of GPIT-PV-30790 ('GPIT C'). Coloradisaurus brevis Galton, 1990Apaldetti et al. (2011 PVL 5904 Efraasia minor (von Huene, 1908) Yates ( Giraffatitan brancai (Janensch, 1914) Janensch (1914) HMN SII Glacialisaurus hammeri Smith andPol, 2007 Smith and FMNH PR 1823, 1822 ( ‡) Herrerasaurus ischigualastensis Reig, 1963Novas (1994 PVSJ 373, PVL 2566 Jingshanosaurus xinwaensis Yang, 1994 Zhang andYang (1994) LFGT-ZLJ0113 Lessemsaurus sauropoides Bonaparte, 1999Bonaparte 1999Pol and Powell (2007)  Vulcanodon karibaensis Raath, 1972Cooper (1984 Yunnanosaurus huangi Young, 1940Young (1942 NGMJ 004546 Yunnanosaurus youngi, Lü et al., 2007Lü et al. (2007 CXMVZA 185 ( ‡) In Klein (2004), the 'IFG compactus' left femur corresponds to the left femur from GPIT-PV-30787 described above. This femur was the only one in her sample with highly vascularised fibrolamellar bone, higher than the normal fibrolamellar bone in other material referred to as Plateosaurus. In the femur of GPIT-PV-30787, there are also single vascular canals and thicker laminae. This tissue appeared after the fifth growth cycle, confirming that this individual was already an adult. In Klein (2004) and Sander and Klein (2005), it was interpreted that this growth pattern corresponded to an exceptional growth cycle due to favourable environmental conditions during a restricted period. However, based on our "section Systematic Palaeontology", this tissue confirms our suggestion that this animal corresponded to a different species with a faster growth rate. Moreover, SMNS 13200 and GPIT-PV-30784 were not included in the histological sampling, and we have removed the two scores that refer to the histological characters from their string. Nevertheless, GPIT-PV-30787 can be confidently scored for these two characters.
To explore the relationship of GPIT-PV-30787, several iterations of phylogenetic analyses were performed (see Appendix 2 for the added OTUs). In the first round of analysis (Fig. 4), five iterations were run using TNT 1.1 , employing the new searching techniques, with the sectorial, ratchet, drift and tree fusing algorithms run through 1.000 random addition sequences. Iteration 1.1 included all taxa; in iteration 1.2, the two Plateosaurus OTUs were removed and were then alternatively removed in iterations 1.3 and 1.4 (see Fig.  4). Finally, specimen GPIT-PV-30787 was constrained to form a clade with Plateosaurus, and a Templeton Test was performed to contrast the trees obtained in iteration 1.5 against iteration 1.1.
For the second round of analysis, P. trossingensis was restricted to the neotype. Characters were checked against the descriptions of the neotype (Moser 2003) and with notes on first-hand observations of SMNS 13200. The species P. gracilis was restricted to the material available in the GPIT collection (Fig. 3). The historical catalogue numbers GPIT 18064, GPIT 18318a and GPIT 18392 have not been given new catalogue numbers because documental evidence suggests the material in each catalogue number belongs to several individuals. Furthermore, the skull of specimen GPIT 18318a has been reported missing in the collection. To avoid confusion, in this iteration, we are referring to this OTU as the 'Sellosaurus' von Huene, 1907, complex, removing the scores for cranial characters. The same conditions as in Analysis 1 were repeated for Analysis 2 (Fig. 4).
A third phylogenetic analysis was performed, using the implied weighting on TNT 1.1, setting a 'gentle' concavity of k=12 as suggested in Goloboff et al. (2018).
Finally, a bivariate analysis of the morphology of the tibia following Ezcurra and Apaldetti (2011) was performed in this work. To the character-by-taxon matrix by Rauhut et al. (2020), the following specimens were added: GPIT-PV-30787, and Plateosaurus was represented by the specimens BSP 1962, MB.R.4405.1-67, SMNS 17928, SMNS 13200 (holotype of P. trossingensis) and GPIT-PV-30784 (referred to P. trossingensis), Mussaurus, Xingxiulong Wang et al., 2017, Yunnanosaurus huangi and Lufengosaurus. According to Ezcurra and Apaldetti (2011), the ratio between the total length and the anteroposterior depth at the mid-length of the tibia show a phylogenetic signal. Nevertheless, the character they proposed is not added to the version of the character-by-taxon matrix used in our recursive analysis.
The character-by-taxon matrices of Analysis 1 and Analysis 2 are stored in MorphoBank Project 4301 (http://morphobank.org/permalink/?P4301), along with high-definition pictures of the different elements that comprise GPIT-PV-30787.

Tuebingosaurus maierfritzorum sp. nov.
h t t p : / / z o o b a n k . o rg / B 6 0 C 4 F 7 6 -5 C FA -4 E 9 A -8 7 4 5 -E98E1CD6D8FC Diagnosis. As for the type and only species.
Etymology. The species name refers to Uwe Fritz and Wolfgang Maier. The former is the editor-in-chief of the journal Vertebrate Zoology, and, in his journal, he facilitated the Festschrift edited by Ingmar Werneburg and Irina Ruf in honour of Wolfgang Maier. The latter was a professor of evolutionary zoology in Tübingen from 1987 to 2007, and the Festschrift was published on the occasion of his 80 th birthday in 2022.
Diagnosis. Sauropodomorph with a unique combination of features: a fused pair of primordial sacrals; a robust and rugose expansion in the postacetabular process of the ilium; a pentagonal outline in the distal surface of the tibia, characterised by an additional posterior projection; a deep lateroventral fossa on the anterior margin of the astragalus; a ventrally directed heel with a lateral projection on the lateral articulation of the astragalus supporting the reduced calcaneum.
Description and comparison. The anatomic terminology adopted in this work follows Galton and Upchurch (2004) for general anatomy, Wilson (1999) for vertebral laminae, Wilson et al. (2011) for vertebral fossae, and Wilson (2011) for the sacrum. Stacked photographs of the bones produced the plates Figs 6-18, and the scale is an approximate reference. Every object has a scale in a different plane, roughly scaled up to the same size. However, for accurate measurements, please refer to the tables or the raw photographs stored in Morphobank.
Specimen GPIT-PV-30787 was referred to as P. 'longiceps' by Galton (2001b). The specimen was first illustrated by von Huene (1932) in his plate 38 and includes elements of the left forelimb (radius, metacarpal IV, phalanges from the fingers I, II and III), a sacrum with a pelvic girdle (including left and right ilia, left and right pubes, and left and right ischia), the first five anterior caudal vertebrae, and the left hindlimb (femur, tibia, fibula, and pes) (von Huene 1932;Galton 2001a).

Sacrum (Figs 6-7, Table 4)
The sacrum of Tuebingosaurus maierfritzorum is composed of two sacrals and one caudosacral. Sacral 1 and sacral 2 have co-ossified neural spines, whereas the caudosacral is broken at the anterior corner of the neural spine, and it is not possible to know if the co-ossification extends all the neural height of the caudosacral (Fig.   6). Most early-diverging sauropodomorphs possess three sacral vertebrae, unlike early sauropods with four. Due to the distortion, it is possible to see more fine details through the left-hand side (Fig. 6). The neural spines of sacral 1 and sacral 2 have an expanded spinal table. The intercostal fenestrae are small and face ventrally (Fig. 6); the alar process of the sacral rib is extensive, articulating with most of the medial iliac surface, and the acetabular process is thinner and posteriorly displaced compared to the alar process (Fig. 7). The sacral ribs between sacrals 1 and 2 form a large dorsal intercostal foramen (icf1) (Fig.  7). Sacral 1 has an intercostal foramen that is anteriorly facing. The anterior articular surface of sacral 1 is slightly concave, with parallel lateral margins and a rounded ventral margin. The centrum of sacral 1 is fused to sacral 2, but the suture between them is still visible. The prezygapophyseal centrodiapophyseal fossa (prcdf) is discernible, with very thick centroprezygapophyseal lamina (cprl) and a thicker prezygodiapophyseal laminae (prdl). Sacral 1 has a spinoprezygapophyseal laminae (sprl) with a rounded margin that meets at the base of the neural spine, where they merge into a prespinal lamina (prsl). Sacral 1 also has a thick strut in the spinopostzygapophyseal lamina (spol) position. It is unclear if there is an intracostal foramen in the second sacral rib. The centrum of sacral 2 is not completely fused into the caudosacral, with a pronounced suture. Through icf1, it is possible to see that sacral 2 also has a prezygodiapophyseal lamina (prdl) and a prezygodiapophyseal fossa. The second ala of the sacrum is thicker than the first one, with a developed dorsal shelf in the alar process.
The anterior corner of the neural spine of the caudosacral is broken. The neural spine of the caudosacral also has an expanded dorsal table. The morphology of the caudosacral rib is similar to that of sacral rib 2, with the alar process expanding towards the iliac surface after a median depression that expands into the intracostal foramen 2. The caudosacral rib is not separated into alar and acetabular projections. The caudosacral rib articulates with the medial side of the brevis fossa, and a suture between these two elements is quite clear (Fig. 6). A cavity sug-  Huene (1932, pl. 38) that have been replicated in the literature. The right fibula is marked in grey as it was found nearby with similar measurements to the left fibula and has been assumed to be part of the same individual.

Anterior caudal vertebrae
The first five anterior caudal vertebrae are preserved (Fig. 8). The morphology of the vertebrae is typical of non-eusauropodan sauropodomorphs, amphicoelous, and constricted mediolaterally with a deeply concave ventral surface in lateral view. The ventral margin of the articular surfaces has a thickened lip that serves as the articulation point for the chevrons (= haemal arches). In Lufengosaurus and Antetonitrus Yates and Kitching, 2003, the centrum is higher than long, a trait also observed in Tuebingosaurus, and as in many early sauropodomorphs, the lamination is reduced. The neural arches extend along the centrum length, starting at the anterior articular surface and ending short of the posterior centrum margin.
The first caudal vertebra is attached to the sacrum (Figs 6-7). The posteroventral corner of the centrum has been remodelled with plaster. Unlike the sacrals, the Abbreviations: bf, brevis fossa; Ca1, anterior caudal 1; CS, caudosacral; csr, caudal sacral rib; ib, iliac blade; ilp, ischiadic peduncle; isva1,2, intercostal space ventral aperture; paf, preacetabular fossa; pap, preacetabular process; par, iliac preacetabular ridge; pop, postacetabular process; prz, prezygapophyses; pup, pubic peduncle; S1, sacral 1; S2, sacral 2; sac, supracetabular crest; sr1, sacral rib 1; sr2, sacral rib 2. first caudal is apneumatic. The posterior articular surface is markedly convex on the ventral end, but a concavity extends below the neural arch. The neural spine ends in a knob-like structure, not a dorsal table like the preceding sacral vertebrae. The spinoprezygapophyseal lamina (sprl) meets at the midline at the base of the neural spine and forms an expanded pre-spinal lamina. The spinopostzygapophyseal laminae (spol) run from this knob-like structure to the postzygapophyses without forming a postspinal lamina. The diapophyses and the parapophyses are fused, but the former is longer than the latter allowing to distinguish both processes. The second anterior caudal vertebra is obliquely twisted, with a deeply concave anterior articular surface and a shallow concave posterior articular surface (Fig. 8). The prezygapophyses, the postzygapophyses and the distal tip of the neural spine are broken. The ventral margin of the anterior articular surface has a large ventral lip for the chevron articulation, but the morphology of this process is difficult to assess due to distortion and breakage (Fig.  8). On both sides of the central body, there is a shallow concavity corresponding to a shallow centrodiapophyseal fossa (Fig. 7). The anterior articular surface is also more prominent than the posterior articular surface. The anterior margin of the neural spine is set more posteriorly than the position of the prezygapophyses, lending it a saddle-shaped outline, possibly due to a prominent spinoprezygapophyseal lamina (sprl) (Fig. 8). A distinct and broad anterior centrodiapophyseal lamina (acdl) is discernible. The spinopostzygapophyseal laminae run along the posterior margin of the neural spine, but the cortical bone is lost towards the tip, and it is impossible to know if they meet (Fig. 8).
Two anterior caudal vertebrae are preserved fused, and according to the illustration made by von Huene (1932) documenting the specimen, they correspond to the third and fourth caudal vertebrae. The vertebrae are dorsoventrally compressed on the right side (Fig. 8). The posterior part of the third anterior caudal neural arch is damaged, but a hypanthrum was present (Fig. 8). The neural canal is circular, and the diapophyses are oriented laterally (Fig. 8). It is not possible to discern the shape of the articular surface's outline due to the distortion (Fig. 8). The ventral lips on the posterior surfaces are wider than the ventral lips on the anterior surfaces. The neural arch is twice as high as the central height in both vertebrae. The spinoprezygapophyseal laminae meet at the base of the neural spine, whereas the spinopostzygapophyseal laminae remain separated throughout the neural spine, and in the third anterior caudal, there is a somewhat deep sulcus separating the two laminae (Fig. 8). The neural spines retain a constant width and end in a rounded dorsal surface with no mediolateral expansion. The prezygapophyses of the third anterior caudal are broken at the tips, and they seem to be dorsolaterally oriented, whereas the prezygapo physes of the fourth anterior caudal end in a A Cross sections in sagittal planes along the first primordial sacral 1 ordered along the anteroposterior axis. B Cross sections in sagittal planes along the primordial sacral 2 ordered along the anteroposterior axis. C Cross sections in sagittal planes along the caudosacral vertebra are ordered along the anteroposterior axis. Abbreviations: icf1, intracostal fenestra 1; icf2, intracostal fenestra 2; icf3, intracostal fenestra 3; il, ilium; ip, ischidiac peduncle; ns, neural spine; rs1, sacral rib 1; rs2, sacral rib 2; rs3, sacral rib 3; sa, supraacetabular crest. point and are dorsally oriented. The anterior centrodiapophyseal lamina (acdl) is broad and short, only present in the second anterior vertebra (Fig. 8). The postzygapophyses are placed higher than the prezygapophyses in both neural arches.
The fifth anterior caudal vertebra is broken with part of the posterior half missing and has the same oblique twisting, even more markedly than in the preceding vertebrae (Fig. 8). As in the previous caudal vertebrae, the anterior articular surface is deeply concave, but the ventral margins are damaged, and the outline is unclear. The neural arch is slightly shorter than the centrum length, set posterior to the anterior articular surface and anterior to the posterior articular surface. The anterior margin of the neural spine is well set posterior to the anterior articular surface, aligned with the diapophysis. The prezygapophyses do not have the same tip morphology as the preceding vertebrae; instead, they are dorsoventrally widened in the proximal part.
Four chevrons are preserved and based on the description by von Huene (1932), corresponding to the three anterior-most chevrons. The chevrons have a closed Y-shaped chevron, with two proximal rami placed on each side of the haemal canal and distally composed of a laterally compressed blade. Proximally, the haemal ca-nal is closed by a bony bridge connecting both rami and closing the canal dorsally. The chevrons have a straight outline in lateral view (Fig. 9). All chevrons show signs of compressive deformation, and two are complete. In the two complete chevrons, it is possible to see a posterior grove extending until the blade's mid-shaft (Fig. 9). (Fig. 6, Table 6)

Ilium
The pre-acetabular process resembles other early-diverging sauropodomorphs in being a triangular projection rather than a vertically tall subtriangular plate seen in more advanced sauropodomorphs. The pre-acetabular process is facing anteriorly in both lateral and dorsal views, and on both ilia, there is a bulge on the lateral surface of the processes. In addition, a preacetabular ridge is present in both ilia (Fig. 6).
The dorsal margin of the ilium is different on the left and right sides, suggesting a diagenetic distortion of the specimen that has slightly compressed the right-hand side and expanded the left-hand side (Fig. 6). The dorsal margin in the lateral view is convex in the middle portion, with two slight inflexions at the pre-and post-acetabular processes. The ilium of P. trossingensis (GPIT-PV-30784) has a sigmoid dorsal margin in lateral view, and in Melanorosaurus and Riojasaurus, the ilium is stepped. A stepped ilium is also present in the Ellingen material (Moser 2003). The iliac blade is thinner dorsal to the acetabulum than the postacetabular process. The lateral surface of both ilia is concave along the anteroposterior and dorsoventral axes, but the degree of the concavity is different on both sides due to the distortion. This concave surface extends ventrally to a point close to the acetabular margin, as Meroktenos. In non-sauropod sauropodomorphs, this surface is restricted to the dorsal half of the iliac blade, as illustrated in Lufengosaurus (Young 1941), the Ellingen material (Moser 2003) and Riojasaurus (Bonaparte 1971). The iliac blade of Tuebingosaurus is very high, approximately two-thirds of the iliac height, a condition shared with Meroktenos.
The acetabulum is fully open like in most sauropodomorphs, except for Pantydraco (Galton and Kermack, 2010), Eoraptor Sereno et al., 1993 and Burio lestes Cabreira et al., 2016. The acetabular region is dorsoventrally high with a pronounced medial wall, similar to the morphology described for Anchisaurus Marsh, 1885 (Galton andCluver 1976), and Yunnanosaurus youngi Lü et al., 2007. The pubic peduncle is prominent and projects anteriorly to the anterior tip of the pre-acetabular process (Fig.  6). The transverse cross-section through the pubic peduncle is laterally expanded and medially narrow, giving it a D-shape in distal view. The lateral margin is expanded into a supraacetabular ridge extending well above the acetabulum, similar to P. trossingensis (GPIT-PV-30784). The pubic peduncle is distally expanded lateral view when compared to its base.
The ischial peduncle is prominent and anteroposteriorly wide, and the articular surface extends posteriorly, forming a 'heel' (Fig. 6), and this character is present in the Ellingen material (Moser 2003), Riojasaurus (Heerden 1979) and Melanorosaurus . However, the ischial heel in P. trossingensis (GPIT-PV-30785) and BSP 1962 (Moser 2003) is more acute than in Tuebingosaurus. In addition, the ischial peduncle  is positioned at the mid-length of the ilium, producing an elongated postacetabular process.
The postacetabular process comprises about 48% of the ilium length and is widened transversely towards the posterior-most corner of the postacetabular process, in contrast to the narrow dorsal margin like in Ruehleia (pers. obs.) and Lufengosaurus (pers. obs.). The ventral margin of the postacetabular process is ventrally deflected at the most posterodorsal corner and does not meet the posterodorsal margin of the postacetabular process (Fig. 6). The lateral profile of the postacetabular process is square-ended with rounded margins, as also occurs in more derived sauropodomorphs and contrasts with the acute lateral outline seen in other early-diverging sauropodomorphs such as Ruehleia (pers. obs.) and Jingshanosaurus. The base of the postacetabular process and the base of the ischial peduncle are connected by a strongly developed brevis fossa with an M-shaped posterior margin (Fig. 6). The brevis shelf is lost in sauropods but present in most dinosaurs as a plesiomorphic state (Gaton and Kermack, 2010). The postacetabular process of the ilium is 1.08 times longer than the distance between the ischiadic and the pubic peduncle (Table 6). (Fig. 10, Table 7) Both pubes are preserved, although the left pubis has the obturator plate medially broken and is deformed in the proximal end (Fig. 10). As in most early-diverging sauropodomorphs, the pubis is long and slender, whereas sauropods have broad pubes (Galton and Upchurch 2004). The proximal end is slightly twisted and laterally expanded in the anterior view, followed by a plate-like shaft that continues towards the distal end (Fig. 10). The overall morphology is similar to that of Plateosaurus (SMNS 13200) and Antetonitrus (BP/1/4952), but in the medial view, the proximal end is anterodorsally expanded, as the condition in the Ellingen material (BSP 1962, in Pl. 30, Moser 2003. The iliac peduncle is not laterally expanded, giving the pubis straight medial and lateral margins in the anterior view, unlike the more derived condition in Antetonitrus and Vulcanodon Raath, 1972, that have a waisted outline, and the 'intermediate' slightly concave condition in Lufengosaurus and Massospondylus carinatus.

Pubis
In P. trossingensis (SMNS 13200), the ischiadic articular surface is separated by an ample nonarticular surface from the iliac articular surface (Moser 2003). In contrast, in Tuebingosaurus, the ischiadic articular surface is separated from the iliac articular surface by a deep borrow, giving the distinctive sauropodomorph morphology of inflexion in the proximal anterior pubic profile. This borrow is present on both pubes, although the iliac articular surface in the left pubes is broken off (Fig. 10). The obturator foramen is relatively large and fully visible in lateral view, unlike in Antetonitrus, where the iliac peduncle obscures the obturator foramen. The iliac pedicel of the pubis partially occludes the obturator foramen in anterior view, a character shared with Saturnalia Langer et al., 1999, andGuaibasaurus Bonaparte et al., 1999. The pubic plate is approximately one-quarter of the total pubic length, measured from the proximal articular surface of the iliac peduncle to the distal surface of the pubic apron, a condition also observed in the Ellingen material (BSP 1962, Pl. 30, Moser 2003, Adeopapposaurus, Lufengosaurus (pers. obs.), Antetonitrus and Meroktenos. In Lessemsaurus and Vulcanodon, the pubic plate is closer to a third of the pubic length and is almost half the length in eusauropods, e.g., Giraffatitan brancai (Janensch, 1914) (MB.R.2180). As in Coloradisaurus and Plateosaurus, the distal end of the pubic apron is markedly anteroposteriorly expanded, and unlike Antetonitrus, Riojasaurus and Meroktenos. There is a pubic tubercle present on the left pubis, very prominently and directly ventral to the obturator foramen, but the same area on the right pubis is damaged. This pubic tubercle is present in Efraasia, Plateosaurus and Plateosauravus (Yates 2003b(Yates , 2007. The conjoined width of the pubes represents 38% of the total length of the pubis (Table 7), unlike in more derived sauropodomorphs where the conjoined width of the pubes is larger than 75% of the pubic length. In addition, the minimum transverse width of the apron is 28% larger than the distance between the pubic and ischiadic peduncle of the ilium; a condition shared with most early-diverging sauropodomorphs and like more derived sauropodomorphs, where the width of the pubic apron is smaller than 40% of the distance between the iliac peduncles (Table 7).  (Fig. 11, Table 8)

Ischium
Both ischia in Tuebingosaurus are preserved and fused along the midline (Fig. 11). As in many early-diverging sauropodomorphs, the ischia are almost rod-like and subtriangular structures where the ischiadic plate is the thinnest region and occupies the proximal third of the bone. As in Colorodisaurus, Plateosaurus and Lufengosaurus, the proximal plate is medially concave and laterally convex, but unlike Coloradisaurus and Plateosaurus, the distal end is not as dorsoventrally expanded and lacks their sub-ovoid morphology. The cortical end on the distal end is not preserved, and only the general morphology can be discerned, and it is not possible to know if there was a posteriorly directed heel as seen in SMNS 13200.
Tuebingosaurus has a more strongly dorsoventrally expanded axis than the mediolateral axis, a condition shared with Lufengosaurus.
The pubic process is widest transversely at the acetabular margin and tapers ventrally, giving it a V-shaped outline (Fig. 11). The mediolateral expansion corresponds to a medial projection that makes the internal border of the acetabular foramen. The ventral margin is expanded in the proximal part of the ischium forming an obturator plate but drastically decreases where the ischiadic shaft starts, with a notch separating the posteroventral end of the ischial obturator plate and the ischial shaft, which then retains a constant dorsoventral width up to the distalmost third where the distal expansion starts (Yates and Kitching, 2003) (Fig. 11). The iliac peduncle has a distinctive morphology, as posterior to the iliac articular surface, there is a concavity followed by a posteriorly oriented projection, which is not seen in the original illustration (von Huene 1932) nor other sauropodomorphs. Along the proximal part of the ischium is a well-developed and deep longitudinal dorsolateral sulcus, a common condition in  sauropodomorphs (Fig. 11). In Tuebingosaurus, the ventral margins of the pubic process meet up to a third of the length of the ischiadic shaft, unlike in P. trossingensis (GPIT-PV-30784 and GPIT-PV-30785, as illustrated in Yates 2003a), and Lufengosaurus, where the margins met by the beginning of the ischium.
In Tuebingosaurus, the distal end of the ischiadic shaft ends in a distal expansion (Fig. 11). In the distal view, the medial margin that meets the antimere is higher than the lateral margin. Only the left ischium has the distal surface of the distal expansion preserved, showing a subquadrangular outline (Fig. 11). In Plateosaurus (SMNS 13200), the distal expansion has a more subtriangular outline, where the medial expansion is four times larger than the lateral margin, and the lateral margin ends more in a point. In Tuebingosaurus, the anteroposterior length of the medial margin is slightly shorter than the lateromedial length of the distal surface.
In contrast, in Plateosaurus (SMNS 13200), the anteroposterior length of the medial margin is almost three times as big as the lateromedial length, giving the distal expansion in Plateosaurus a more gracile shape. The morphology of the distal expansion is similar, then, to Lufengosaurus (pers. obs.) and Mussaurus (Otero and Pol 2013). However, in the lateral view, the distal expansion has a small anterior projection, that in Plateosaurus (SMNS 13200) has more of a heel-like morphology, in Tuebingosaurus is more triangular; this triangular expan- 3.1.6. Femur (Fig. 12, Table 9, 10) In Tuebingosaurus, only the left femur is partially preserved, missing most of the medial condyle, and the medial condyle is reconstructed in the distal end with plaster (Fig. 12). The femur has the general morphology seen in early sauropodomorphs, straight in an anterior view and curved in a lateral view (Galton and Upchurch 2004, Fig. 12). In the earliest forms, like Buriolestes (Müller et al. 2018), the femur is curved in the anterior and lateral view, whereas in more derived forms, like in Barapasaurus Jain et al., 1975(Bandyopadhyay et al. 2010, it is straight in both views. In specimen SMNS 13200, the femoral head is slightly visible in lateral view, not fully medially inturned, whereas, in Tuebingosaurus, the femoral head is completely inturned and hidden in lateral view. The femoral head is broken on its anterior half, missing the anteromedial and anterolateral features. The sulcus for the ligamentum capitis femoralis is flat, compared to the marked concavity in Buriolestes (Müller et al. 2018), followed by a markedly concave but narrow facies articularis antitrochanterica. No proximal groove is on the proximal surface, like the one seen in Buriolestes (Müller et al. 2018).
The lesser trochanter is prominent, a feature shared with specimen SMNS 13200 (Moser 2003), but unlike in SMNS 13200, the dorsolateral trochanter (= trochanter major) is only a small bump, whereas in SMNS 13200 is a more developed protuberance. In the fourth trochanter in SMNS 13200, the dorsal margin and the ventral are parallel and similarly slope dorsoventrally (Moser 2003). The femur in Tuebingosaurus has the fourth trochanter with a dorsal margin running dorsoventrally and a ventral margin running more horizontally, giving the fourth trochanter a somewhat trapezoid shape (Fig.  12), quite similar to the morphology seen in Riojasaurus (Bonaparte 1972). In Coloradisaurus, the dorsal margin Abbreviations: fh, femoral head; ft, fourth trochanter; lc, lateral condyle; lt, lesser (= anterior) trochanter; mc, medial condyle; tfc, tibiofibular crest; ts, trochanteric shelf. runs dorsoventrally but not so steeply, whereas the ventral margin runs ventrodorsally with a more pronounced slope, giving the fourth trochanter a distinctive trapezoidal shape in an inverted orientation compared to Riojasaurus (Apaldetti et al. 2013). In Buriolestes (Müller et al. 2018) and Anchisaurus (Galton 1976), the fourth trochanter has a tra pezoidal shape, with the dorsal margin running dorsoventrally and the ventral margin running ventrodorsally, both with similar slopes, giving the fourth trochanter a regular trapezoid shape. In Mussaurus (Otero and Pol 2013), the dorsal and ventral margins run somewhat parallel, close to the horizontal, but have markedly curved edges. In Buriolestes (Müller et al. 2018), the fourth trochanter is closer to the medial margin along the mediolateral axis. In the medial view, the medial surface expands continuously onto the fourth trochanter; in the lateral view, there is an inflexion separating the lateral surface from the fourth trochanter. This same condition is observed in SMNS 13200, where the medial surface is continuously expanded onto the fourth trochanter plate in anteromedial view but separated from the lateral surface by an inflexion. The fourth trochanter in Tuebingosaurus has the same morphology, and this condition can be found in other early sauropodomorphs, e.g., Riojasaurus (Bonaparte 1972), Anchisaurus (Galton 1976), and Coloradisaurus (Apaldetti et al., 2014). In Mussaurus (Otero and Pol 2013), the fourth trochanter is closer to the lateral side, and the lateral surface continuously expands onto the fourth trochanter, whereas a marked inflexion separates the posterior surface from the fourth trochanter.
In the distal view, the median portion of the femur is reconstructed by plaster, but the outline seems more Table 9. Measurements (in mm) of the left femur of Tuebingosaurus maierfritzorum. Eccentricity index is expressed as a ratio of mediolateral width at midshaft/anteroposterior width at midshaft. The robustness index is expressed as a total length/circumference ratio under the fourth trochanter.

Measurements in Tuebingosaurus maierfritzorum
Length ( (Fig. 12). For instance, in SMNS 13200 and Coloradisaurus, the mediolateral axis is considerably longer than the anteroposterior one, giving the distal surface a more flattened elliptical shape. In Mussaurus, the distal end is hourglass-shaped, with a marked popliteal fossa posterior and a deep extensor depression anteriorly. Due to the plaster, it is impossible to know which of these two morphotypes is present in the femur. An extensor depression is present in most sauropodomorphs, except for the earliest forms, such as Buriolestes, Saturnalia, Pantydraco and Efraasia (SMNS 12216, pers. obs.). The plaster in the specimen does not outline an extensor depression, imitating the plesiomorphic condition. In Buriolestes (Müller et al. 2018), the medial condyle and the anterolateral tuber are similar, with a very lateromedially reduced lateral condyle. In Coloradisaurus, the medial condyle is the most prominent of the three condyles, and the tibiofibular condyle has a triangular outline, unlike the quadrangular one seen in earlier forms. The lateral condyle is laterally projected and separated from the tibiofibular condyle by a significant inflexion. The condylar morphology of Coloradisaurus is also seen in SMNS 13200 and Anchisaurus (Galton 1976). Despite the plaster, the medial condyle is larger in the lateromedial axis than the tibiofibular condyle. The lateral condyle is laterally projected and separated from the tibiofibular condyle by an inflexion, although not as marked as in Coloradisaurus.
The morphology in Mussaurus is unclear, as this portion is broken off (Otero and Pol 2013). Furthermore, a Ward clustering of the measurements in sauropodomorph femora in Table 10, showing Tuebingosaurus is placed in a cluster with Mussaurus and Lessemsaurus (Appendix 3, Figure A1). (Fig. 13, Table 11)

Tibia
The tibia is approximately 0.85 times the length of the femur (Tables 9 and 11), a proportion similar to all non-eusauropod sauropodomorphs (Apaldetti et al. 2013). The anteroposterior axis of the proximal end is horizontal in lateral view as in Mussaurus and Anchisaurus, whereas the anteroposterior axis in specimen SMNS 13200 and Coloradisaurus is dorsoventrally skewed. In the proximal view, the proximal end of the tibia has a scalene shape, with the medial condyle posteriorly expanded relative to the medial condyle and the cnemial crest facing laterally (Fig. 13). In specimen SMNS 13200, the medial and lateral condyles are roughly aligned, and the cnemial crest is more anteriorly oriented. In Coloradisaurus, the medial and lateral condyles are roughly aligned, but the cnemial crest is laterally oriented, whereas, in Mussaurus, the proximal outline is similar to that of specimen Tuebingosaurus. In earlier forms, such as in Buriolestes, the cnemial crest is laterally oriented, forming a 90 degrees angle with the anteroposterior axis of the proximal end of the tibia. In Tuebingosaurus, the fibular articular surface has a large protuberance, similar to the outline in SMNS 13200, although this protuberance is less pronounced. In Mussaurus and Coloradisaurus, this protuberance is more like a small tuber (Fig. 13).
The shaft of the tibia is straight with a sub-elliptical cross-section. The distal end has a quadrangular outline, with two lateral processes, the anterolateral and posterolateral processes, and a posteromedial and an anterolateral condyle (Fig. 13). The anterolateral process is twice as wide as the posterolateral process. In Tuebingosaurus, the medial surface has an additional projection not seen in other sauropodomorphs. In Coloradisaurus, the anteromedial process is medially expanded relative to the posteromedial condyle, a feature seen in Adeopapposaurus, Mussaurus and SMNS 13200. The distal end is lateromedially elongated, and in the posterior view, the posterolateral process (= posteroventral process, = caudoventral process) is distally expanded relative to the anterolateral process and reaches the lateral margin of the distal tibia, a condition shared with SMNS 13200, Riojasaurus, Adeopapposaurus and Coloradisaurus, but unlike Mussaurus, Anchisaurus and Aardonyx Yates et al., 2010, and other advanced sauropodomorphs, where the posterolateral process does not reach the lateral margin. As in most early sauropodomorphs, the posterolateral process distally exceeds the limits of the anterolateral process. The distal surface of the posterolateral process is horizontally oriented, whereas the distal surface of the anterolateral process is distolaterally oblique for the articulation of the ascending process of the astragalus.
Tuebingosaurus sits between the morphospaces outlined for Massospondylidae and "Melanorosauridae" in a bivariate plot of the ratios between the total length and anteroposterior depth of the proximal end of the tibia (L/ Pw) concerning the ratio between the total length and anteroposterior depth at mid-length of the tibia (L/Mw) (Fig. 14). Noteworthy, Plateosaurus has a large morphospace, compared to the other sauropodomorphs in the sample. First, this could represent that the morphospace of Plateosaurus captures better the intraspecific variability in the tibiae given the larger sample compared to other sauropodomorphs; however, the gradual increase in the robustness through time is quite clear. Furthermore, a restricted definition of Plateosaurus (SMNS 13200 and GPIT-PV-30785) occupies a similar space in the bivariate plot. Specimen BSP 1962 is, on the other hand, nested within "Melanorosauridae", close to Tuebingosaurus and Mussaurus, and supports the idea that the similarity between Tuebingosaurus and BSP 1962 outlined above in the pubis, ischia and the tibiae are better explained by considering BSP 1962 as a massopodan as well (Appendix 3 - Figure A2). (Fig. 15, Table 12)

Fibula
The two fibulae are preserved and have similar sizes. The fibula is a slender and long bone with an anteroposteriorly expanded proximal end and, to a lesser degree, the distal end (Fig. 15), similar to the condition in Riojasaurus, unlike in Anchisaurus and Mussaurus, where the distal end is not expanded. The proximal articular surface has a concave medial margin and a convex lateral margin, forming a crescent-shaped outline as in SMNS 13200, Adeopapposaurus and Mussaurus. However, in Tuebingosaurus, the medial margin is concave only in the anterior portion and straight in the posterior one. The shaft is straight in lateral and anterior views, unlike in Adeopapposaurus and Mussaurus, where the fibula is curved in anterior view. The distal end in lateral view is anteriorly slanted but horizontal in posterior view, as in Mussaurus and Adeopapposaurus.
There is a small protuberance on the anteromedial surface, a feature that has not been reported for sauropodomorphs and is present on both fibulae, discarding a noticeable pathological feature (mfp in Fig. 15). The medial condyle is larger than the lateral condyle, and a shallow triangular fossa is visible on the medial face of the distal end of the fibula. The lateral face of the fibula is, in turn, flat. (Fig. 16, Table 13)

Astragalus
The astragalus has the classic non-eusauropod sauropodomorph morphology, with a somewhat kidneyshaped outline (Fig. 16). In dorsal view, the medial margin is about 50% larger than the lateral margin; the lateral margin has a sigmoidal articulation, and the medial margin is posteriorly curved. The posterior margin is convex, as in Mussaurus (Otero and Pol 2013), Blikanasaurus van Heerden, 1985 (Galton andvan Heerden 1998), Vulcanodon (Cooper 1984), and Tazoudasaurus Allain et al., 2004 (Allain andAquesbi 2008); nevertheless, a con-  (To1-380). The outlines are not set to scale. Abbreviations: alp, anterolateral process, aspa, articular surface for the ascending process, cn, cnemial crest, lc, lateral condyle, mc, medial condyle, plp, posterolateral process. vex posterior margin is also present in P. trossingensis (as illustrated in von Huene 1926). The posterior margin is straight in Unaysaurus Leal et al., 2004(McPhee et al. 2019 and Macrocollum Müller et al., 2018. The medial margin in Unaysaurus, Macrocollum and Blikanasaurus has a prominent triangular process anteromedially projected, similar to the outline in Tuebingosaurus. At the midlength of the posterior margin, there is a prominent bulge like that present in Mussaurus, and a bulge is present in Blikanasaurus and P. trossingensis but not as pronounced (Fig. 16). The anterior margin has its highest point on the medial side, whereas the posterior margin has its highest point on the lateral side. The proximal surface is divided into two distinct articular facets: a lateral facet, with a deep socket-like concavity for the articulation with the distal end of the fibula, and a flat medial facet occupying most of the proximal surface, where the distal end of the tibia articulates with the as-tragalus. These two facets are divided by a rounded ridge that continues to form the posterior margin of the ascending process. As in many other early sauropodomorphs, the ascending process is not as prominent, and in anterior view, the ascending process rises slightly above the posterior bulge. Towards the lateral end of the anterior margin, there is a deep depression similar to those in early saurischians, e.g., Herrerasaurus Reig, 1963, Eoraptor (Sereno et al., 2012), and Saturnalia Langer (2003, but faces anterolaterally rather than lateroventrally, and it is placed right beneath the anterior margin of the ascending process. This fossa occupies a prominent space of the anterior margin, and it is not a feature seen in other early sauropodomorphs. Ventral to this fossa, a ventrally directed projection with a heel-like morphology supports the calcaneum by a laterally oriented projection. The distal surface has the characteristic rugose roller-shaped articulation in other sauropodomorphs. Figure 14. Bivariate plot showing the ratio between the total length and anteroposterior depth of the proximal end of the tibia (L/ Pw) concerning the ratio between the total length and anteroposterior depth at mid-length of the tibia (L/Mw). Data was taken from Ezcurra and Apaldetti (2011) and first-hand assessments obtained by ORRF. The convex hulls with solid lines show the morphospace generated by the groups 'Guaibasauridae', Massospondylidae, 'Melanorosauridae' and Sauropoda. The name 'Melanorosauridae' is here used to refer to sauropodomorphs that are not traditionally considered as sauropods. The triangles represent taxa traditionally considered sauropods, and the stars represent non-sauropod sauropodomorphs. The colours of the points represent the age of the taxa, with purple for the Late Triassic and blue for the Early Jurassic. The dashed convex hull represents the morphospace corresponding to Plateosaurus as currently defined. The yellow convex hull represents the taxa placed in a polytomy before the diversification of Massospondylidae and Sauropodiformes. 3.1.10. Calcaneum (Fig. 16-A, D, Table 13) The calcaneum in Tuebingosaurus is significantly reduced, albeit conserving the early-diverging sauropodomorph triradiate morphology. The calcaneum is lateromedially flattened, but the anterior end is thicker than the posterior end and lacriform in dorsal view. The anterior end is not straight but bears a distinct anterior projection in the anterolateral margin. The ventral process rests on the anterolateral projection of the lateral margin of the astragalus. The medial margin of the calcaneum is concave and articulates along the sigmoidal lateral margin of the astragalus. This articulation generates a pocket between the two elements that were probably filled with cartilage. The mediolateral length of the calcaneum represents 21% of the astragalar mediolateral length. In early sauropodomorphs, such as Saturnalia, the calcaneal length is roughly 50% of the astragalar length, and in Coloradisaurus, it is 40%, and towards the more derived sauropodomorphs, we have values lower than 30%, such Abbreviations: asa, astragalar articular surface, caa, calcaneum articular surface, mfp, anteromedial fibular process, tia, tibial articular surface. as in Anchisaurus, Vulcanodon, Shunosaurus Dong et al., 1983 andCamarasaurus Cope, 1877. This reduced calcaneum is consistent with what we expect in more obligated quadrupedal animals, such as sauropods. (Fig. 17, Table 13)

Metatarsal IV
According to the early drawings by von Huene (unpublished), an almost complete pes was recovered from the block as part of specimen "GPIT IV". The drawings show metatarsals I, II, III, and IV articulated to their respective phalanges. Currently, only metatarsal IV, the complete digits II and III, and one phalange of digit I are preserved in the collection (Fig. 17).
The only metatarsal element preserved is metatarsal IV. The metatarsal IV is a robust element with a constriction along the mid-section. Its proximal end is expanded lateromedially and flattened dorsoplantarly, whereas the distal end is expanded not as lateromedially but expanded dorsoplantarly, with a morphology similar to Massispondylus carinatus (BPI/I/4377) and Mussaurus (MLP 61-III-20-22) (Fig. 17). A well-developed crest on the proximal end extends proximodistally along the dorsal surface of the proximal end. This crest delimits a concave medial surface where metatarsal III articulates. The dorsoventral length at the crest level represents 30% of the lateromedial length of the proximal end of metatarsal IV, as in Mussaurus and Massospondylus, whereas this ratio reaches 50% in Saturnalia, Coloradisaurus, Plateosaurus, and Blikanasaurus.
The dorsal and plantar edges of the lateral half are parallel through the metatarsal length (Fig. 17), as in Mussaurus. The cross-section is ovoid, where the lateral margin is narrower than the medial one. The medial margin of metatarsal IV has a bulge close to the proximal end, a feature in Massospondylus, Mussaurus and Plateosaurus. In other early-branching sauropodomorphs, this bulge fits in a slight depression on the lateral margin of the shaft of metatarsal III. Distally, there is another bulge along the distal end of the shaft of metatarsal IV, a condition shared with Mussaurus.
The distal articular surface is quadrangular in distal view with an undivided and marked convexity, similar to Riojasaurus (PVL 3526). The lateral margin on the distal end has two processes that project laterally in distal view, whereas the medial margin has a marked expansion in the medio-plantar corner. On the lateral margin, the two projections are separated by a well-developed concavity; the medial margin is roughly straight.  (Fig. 18, Table 13)

Pedal digits
Only two digits are preserved, pedal digit II and pedal digit III, with two and three phalanges, respectively, and the first phalanx of digit I. Phalanx I.1 is identified due to the morphology, with a proximomedial projection (Fig.  18) like the morphology reported in Plateosaurus (von Huene 1926) but more developed. The proximal lateromedial width corresponds to 84% of the total proximodistal length of the phalange (Table 14). The proximal articular surface is concave, reniform and undivided with a concave dorsal edge and a convex ventral edge. The major axis of the proximal articular surface is twisted 5 o to the lateromedial axis of the distal surface. The shaft has subparallel lateral and medial margins, with a flat dorsal surface and a deeply concave plantar surface. The distal margin has two well-developed condyles separated by an intercondylar groove. The dorsoplantar length of the lateral and medial condyles is roughly the same, but the medial collateral ligament pit is more deeply concave.
Pedal digit II has two non-terminal phalanges and a well-developed ungual. Phalanx II.1 is robust, where the proximal lateromedial length is 78% of the proximodistal length. In phalanx II.1, the distal lateromedial length is similar to the lateromedial length, with a distinctive shaft with concave lateral and medial margins. The dorsal margin of the proximal articular surface of phalanx II.1 is shorter than the ventral margin. On the distal end, there is a distinctive dorsal depression (Fig. 18). The collateral ligament pits are not deeply excavated. A markedly concave intercondylar groove separates the lateral and medial condyles. Phalanx II.2 is shorter than II.1, with similar robustness to phalanx II.1 (proximal lateromedial length is 79% of the proximodistal length). The lateral and medial margins of phalanx II.2 are more concave, and the shaft is comparatively shorter than the one in phalanx II.1 (Fig. 18). The collateral ligament pits of phalanx II.2 are more deeply marked and seem to face dorsally, although this could be the product of deformation. Phalanx II.2 has a distinctive dorsal flange. The ungual pedal digit II is lateromedially flattened and distinctively curved. The articular surface is undivided.
Pedal digit III has three non-terminal phalanges. Phalanx III.1 is robust, with the proximal lateromedial length being 72% of the proximodistal length. The shaft of phalanx III.1 is defined by markedly concave lateral and medial margins (Fig. 18). The dorsal surface is slightly concave, but the plantar one is strongly concave. Phalanx III.1 has a more hour-glass shape than phalanx II.1. The  Apaldetti et al. 2013, mirrored), and Blikanasaurus (Galton and Van Heerden 1998). Abbreviations: d, dorsal; p, plantar; dc, dorsal crest; mtI (I), metatarsal I; mtII (II), metatarsal II; mtIII (III), metatarsal III; mt III cs, contact surface of metatarsal III; mtIV (and IV), metatarsal IV; mtV cs, metatarsal V contact surface; V, metatarsal V. dorsal margin of the proximal articular surface of phalanx III.1 is shorter than the ventral margin, and this morphology is also seen in the other non-terminal phalanges of pedal digit III. Phalanx III.1 has a very developed dorsal flange and a very developed ventral flange. Phalanx III.2 is also robust (proximal lateromedial length is 73% of the proximodistal length), with a more open concave lateral margin. The dorsal flange of phalanx III.2 is more reduced than in phalanx III.1, but the ventral flange is still prominent. The collateral ligament pits are deeply excavated (Fig. 18). Phalanx III.3 is more robust than the preceding phalanges, with the proximal lateromedial length representing 83% of the proximodistal length of the pha-lanx. The medial and lateral condyles are more defined than the preceding phalanges, and the collateral ligament pits are more deeply excavated. The pedal ungual III is more curved than pedal ungual II (Fig. 18).

Other material previously associated with specimen "GPIT IV"
According to von Huene (1932), during the expedition of 1922 in the Trossingen Formation near Tübingen. It is impossible to know how nearby these elements were to the pelvis and hind limb of Tuebingosaurus, as von Huene (1932) did not provide details on this. In two separate blocks located near the semi-articulated specimen described above, there was a mandible (in block 169), a partially articulated forearm (block 185) and a cervical vertebra (block 159). The mandible has a similar outline to P. trossingensis, with 24 alveoli and 23 preserved (3 rd tooth is missing); the mandible is damaged due to post-excavation preparation. The forearm elements correspond to a radius, a metacarpal (possibly metacarpal III, and manual digits I to III (digit I is complete, digit II is probably missing one phalange, and digit III only has two phalanges). The radius is more straightened and less mediolaterally twisted than that of GPIT-PV-30785 and has a proximal outline that is more similar to Plateosauravus (based on the drawings in Remes 2007). The preservation of the bone is also slightly better than the preservation of the elements outlined above. There is no evident distortion; the cortical bone is not flaked like the other long elements in Tuebingosaurus, suggesting a faster burial and less environmental exposure. It could be possible that the forearm got buried earlier than the rest of the carcass. The cervical vertebrae could not be located in the collection. Furthermore, there are no relevant details or documentation regarding the excavation from 1922 available to us. The pelvis, the hind limb, and the caudal vertebrae articulate with each other, and it is possible to associate them with a single individual, whereas the other bones are associated with this based on their distance to the larger block. These specimens were embedded into a plastic matrix as part of the diorama display to simulate the mud-burial. When trying to remove the mandible, it was clear that the material was glued to the plastic, and its removal may endanger the specimens. Thus, the mandible and the forearm are removed from specimen GPIT-PV-30787 and, as such, from the holotype of Tuebingosaurus. However, further work should test whether these specimens can be referred to the holotype.

Analysis 1
For the first analysis, Tuebingosaurus was added to the character-by-taxon matrix by Rauhut et al. (2020). Five iterations were run on this character-by-taxon matrix: 1) including all 67 taxa, b) removing P. 'engelhardti' and P. gracilis, c) removing P. gracilis only, d) removing P. 'engelhardti' only, e) constraining GPIT-PV-30787 to the genus Plateosaurus. The tree space was explored using the 'Clade Frequency in Trees' function in Mesquite.
The key result from Analysis 1 is that Tuebingosaurus is consistently placed nested within Massopoda, alternatively paired with Meroktenos, Mussaurus, Isanosaurus Buffetaut et al. 2000 and Pulanesaura. The trees from iteration 1.5 are very similar to the topologies with a broad definition of Plateosaurus, which supports the idea that several specimens included in Plateosaurus have a combination of plesiomorphic and derived traits that are not seen when the definition is reduced. The phylogenetic definitions used in the following descriptions are given in Table 14.

Iteration 1.2 Exclusion of P. 'engelhardti' and P. gracilis
Massopoda is found in 87% of the MPTs. Massospondylidae is found in 100% of the MPTs. Lessemsauridae is found in 89% of the MPTs. In the 153 MPTs, Tuebingosaurus is nested closer to Eusauropoda than in iteration 1.1: 1) at the base of a pectinate arrangement towards Sauropoda in 72% of the MPTs, 2) paired with Mussaurus in 17%, 3) paired with Pulanesaura in 4%, 4) at the base of Sauropoda (the most inclusive clade that includes Saltasaurus but not Melanorosaurus sensu Yates 2007a) in 4%, 5) paired with Meroktenos in 5 MPTs, and 6) in a clade with Meroktenos and Pulanesaura only in 1 MPT.

Iteration 1.3 Exclusion of P. gracilis
Massopoda is found in 100% of the MPTs. Plateosauridae and Massospondylidae are recovered in 100% of

Iteration 1.5 Forcing Tuebingosaurus within Plateosauridae
Unlike the previous iterations, the tree becomes more unstable. Massospondylidae is recovered in 100% of the MPTs but is nested within a clade with Seitaad Sertich and Loewen, 2010, Yunnanosaurus and Jingshanosaurus in 60% of the MPTs. The clade Sauropodiformes is found in all the trees, but Xingxiulong is placed as the sister taxon to Sauropodiformes in 60% of them. In iteration 1.1, Xingxiulong is placed at the base of Massopoda.
A Templeton Test was performed to compare the trees between iteration 1.1 against 1.5, and although the trees are longer in iteration 1.1, there is no statistical significance.

Analysis 2
For this second analysis, P. trossingensis is restricted to the neotype (SMNS 13200), and P. gracilis was replaced by an OTU defined as 'Sellosaurus' complex, restricted to the material in the GPIT collection. The skull characters were removed from the 'Sellosaurus' complex. The material GPIT 18392 does not correspond to one individual, but there is not enough documentation to know which elements belong together. Further work is needed to determine the anatomic identity of GPIT 18392. The tree space was explored using the 'Clade Frequency in Trees' function in Mesquite.

Iteration 2.1
Addition of Tuebingosaurus (Fig. 19, Table 15) In most of the trees (63%), Tuebingosaurus is paired with Meroktenos, supported by the morphology of the lesser trochanter, where the lesser trochanter is closer to the near centre of the anterior face of the femoral shaft and not visible in posterior view, characters shared with Plateosaurus. However, they are bracketed by Lessemsauridae and Eusauropoda, where the lesser trochanter is close to the lateral margin of the anterior face of the femoral shaft and visible in posterior view. In 30%, Tuebingosaurus is placed within a gradient towards Eusaropoda due to a combination of derived traits. In 7% of the MPTs, Tuebingosaurus is paired with Schleitheimia, with a projecting heel at the distal end of the ischial peduncle, a Table 14. Phylogenetic names used to compare the different cladograms. The content refers to the taxa included in that name from the character-by-taxon matrix employed here and used to identify groups in the different trees.

Eusauropoda
The least inclusive clade containing Shunosaurus and Saltasaurus (Upchurch et al. 2004) Shunosaurus, Amygdalodon Cabreira, 1947, Volkheimeria, Spinophorosaurus, Cetiosaurus, Omeisaurus, Mamenchisaurus, Neosauropoda  In most of the trees (72%), Tuebingosaurus is at the base of a gradient towards Eusauropoda due to several derived characters: prezygodiapophyseal laminae in the anterior caudal vertebrae, strongly anterior posteriorly convex articular surface of the ischial peduncle of the ilium, a laterally facing fibular trochanter, and a convex posterior margin of the astragalus. The latter is present also in Mussaurus, Blikanasaurus and early sauropods, later reversing to a straight posterior margin in Eusauropoda.
In 17% of the Tuebingosaurus is paired with Mussaurus, the brevis fossa is well developed but absent in Sauropodiformes.

Iteration 2.3
Exclusion of P. gracilis (Table 15) This iteration produces trees similar to iteration 2.1, with Tuebingosaurus paired with Meroktenos in 59% of the MPTs, and at the base of Eusaropoda in 30% of the MPTs. Tuebingosaurus is also paired with Schleitheimia (in 9%) and Isanosaurus (in 11%). Something new is the appearance of a clade containing Meroktenos and Pulanesaura in 7% of the MPTs; this clade is supported by synapomorphies that are scored as "?" in either Pulanesaura or Meroktenos.

Iteration 2.4 Exclusion of P. 'engelhardti' (Table 15)
In 64% of the trees, Tuebingosaurus is placed at the base of a gradient towards Eusauropoda, supported by the same synapomorphies in the previous iterations. In 20% of the MPTs, Tuebingosurus has an even more derived position at the base of Eusaropoda, supported by derived characters such as deep bases of the diapophyses in the anterior caudal vertebrae, a much shorter ischial peduncle of the ilium than the pubic peduncle, and a wedge-shaped astragalar body. In 14% of the MPTs, Tuebingosaurus is paired with Meroktenos.

Iteration 2.5 Forcing Tuebingosaurus within Plateosauridae
In the trees of this iteration (228), Plateosauridae is supported by the following synapomorphies: medial margin of the supratemporal fossa (Ch60-1), basipterygoid processes and parasphenoid rostrum are roughly aligned (Ch81-1), symphyseal end of the dentary strongly curved ventrally (Ch99-1), ventrolateral twisting of the transverse axis of the distal end of the first phalanx of manual digit I is much less than 60 degrees (Ch245-1), concave lateral margin of the proximal surface of metatarsal II (Ch356-1). Interestingly, all these characters are scored as "?" for Tuebingosaurus. In all the trees, an Anchisauria clade is supported by one unequivocal synapomorphy, a lateral margin of the descending posteroventral process of the distal end of the tibia set well back from the anterolateral corner of the distal tibia (Ch327-1), which is also scored as 1 in Tuebingosaurus. Another clade found in 100% of the trees is Gravisauria sensu Allain and Aquesbi (2008), supported by several synapomorphies: shallow lateral depression in the cervical vertebral centra (Ch129-1), spinodiapophyseal lamina on dorsal vertebrae joining to create a composite lamina (Ch171-2), deep bases of the anterior caudal diapophyses, extending from the centrum to the neural arch (Ch190-1), absence of the longitudinal ventral sulcus on the anterior and middle caudal vertebra (Ch193-1), strongly anteroposteriorly convex articular surface of the ischial peduncle of the ilium (Ch. 266-1), ischial peduncle of the ilium much shorter than pubic peduncle (Ch. 268-1), a straight longitudinal axis of the femur in lateral view , intensely rugosely pitted articular surface of the long bones of the limbs , and growth marks in the cortex absent or restricted to the outer cortex (Ch. 379). Tuebingosaurus shares Ch190-1, Ch193-1, Ch266-1, and Ch268-1. Although these trees are longer (1632 steps) and have similar consistency and retention indexes (CI=0.325, RI=0.651), a Templeton Test shows that there is no statistical significance between the constrained trees and the trees from iteration 2.1, thus the position of Tuebingosaurus nested within Plateosauridae seems to rely on the impact of missing information.

Analysis 3
The final analysis took the character-by-taxon matrix from Analysis 2, and a standard implied weighting was performed using the same searching settings described in Analysis 1 and 2. The implied weighting used a 'gentle' concavity of 12 as recommended in Goloboff. Down-weighting the homoplasies, Tuebingosaurus moves closer to the base of Sauropoda than the previous topologies. Fig. 20 shows the strict consensus of 3 MPTs, and Tuebingosaurus is placed as the sister taxon to a lineage that leads to Sauropoda, and Schleitheimia is the earliest member of the rest of the clade. The groups recovered in the iterations from Analysis 1 and 2 are also found in

Phylogenetic position of Tuebingosaurus maierfritzorum
Based on our three phylogenetic analyses, there is support for Tuebingosaurus being one of the earliest massopodan sauropodomorphs (Figs 19-20). Tuebingosaurus shares several characters with early massopodans but simultaneously retains several plesiomorphic characters that show some derived features seen in sauropodiformes.  whereas in most sauropodomorphs, this tip is distal to the femoral head. The fourth trochanter is located on the medial margin of the femur, like in Mussaurus, Jingshanosaurus, Anchisaurus, Coloradisaurus, and Lufengosaurus, but unlike Massospondylus, Adeopapposaurus, and Plateosaurus, where it is centrally located along the mediolateral axis.
Several characters in Tuebingosaurus show a large degree of plesiomorphy, with reversals being very common in the phylogenetic analyses in this work. The position of the obturator foramen of the pubis is partially occluded by the iliac pedicel in anterior view, a plesiomorphic character present in Seitaad and Jingshanosaurus, unlike most of the sauropodomorphs where it is completely visible. The lateral margins of the pubic apron in the anterior view also retain the plesiomorphic straight morphology, like Seitaad, Anchisaurus, Leonerasaurus and Yunnanosaurus. It also retains an ilium shorter than the pubis and an ischial component larger than the pubic component of the acetabular rim, both plesiomorphic conditions found in non-sauropodiform sauropodomorphs. The femur has the plesiomorphic condition of being strongly bent with an offset between the proximal and distal axes; a plateosaurian-type femoral morphology shared with Anchisaurus, Jingshanosaurus, Xingxiulong, Yunnanosaurus, Coloradisaurus, Lufengosaurus, Massospondylus and Adeopapposaurus. In Tuebingosaurus, the fourth trochanter is on the proximal half of the femur, a plesiomorphic condition that changes in Sauropodiformes but is retained in some (i.e., Melanorosaurus, Isanosaurus, Patagosaurus and Shunosaurus), where the fourth trochanter is straddling around the midpoint. Furthermore, the fourth trochanter is asymmetrical, with a steeper distal slope and a plateosaurian-type femoral morphology, a feature also present in Anchisaurus, Jingshanosaurus, Yunnanosaurus and Mussaurus, and unlike Coloradisaurus, Lufengosaurus, Massospondylus and Adeopapposaurus, where the fourth trochanter is more symmetrical.
Unlike any other massopodan, Tuebingosaurus displays a prezygodiapophyseal lamina on the anterior caudal vertebrae and the anterior caudal diapophyses extending from the centrum to the neural arch, both of which are derived characters that are seen in sauropods, the former appearing in Pulanesaura and the latter in Schleitheimia. The length of the ischial peduncle of the ilium is much shorter than the pubic peduncle, a derived trait in sauropods and Sarahsaurus. The angle between the long axis of the femoral head and the transverse axis of the distal femur is close to 0 o , a derived trait that originated early in massopodan evolution but about 30 o in early sauropodomorphs, like Thecodontosaurus, Efraasia and Plateosaurus. The articular surface of the tibia has an anteroposterior length twice or larger than the transverse width, a derived trait shared with Sauropodiformes but absent in earlier massopodans. The lateral margin of the descending posteroventral process of the distal end of Figure 20. Strict consensus of three MPTs obtained from implied weighting (k=12). Down-weighting the homoplasies places Tuebingosaurus as having a common ancestor with the lineage that leads to Sauropoda and earlier than Schleitheimia. Numbers correspond to 1) Plateosauridae, 2) Massopoda, 3) Massospondylidae, 4) Sauropodiformes, 5) Lessemsauridae, 6) Eusauropoda. The ages of Plateosaurus and Tuebingosaurus are restricted to the Obere Mühle outcrop, which has been assigned to the Sevatian, an informal unit used in the stratigraphy of the Late Triassic in Central Europe, from 211 to 203.6 Mya (Olsen et al. 2011). The base of Massopoda is not clear in the topologies (see discussion in the text), but Tuebingosaurus is placed at this level with several other Early Jurassic sauropodomorphs and Mussaurus from the Late Triassic. This suggests that a very rapid diversification event occurred in the Carnian, and the groups that originated during this time experienced further diversifications during the Norian. the tibia is set well back from the anterolateral corner of the distal tibia -a derived trait shared with Anchisaurus, Mussaurus and most Sauropodiformes. The position of the fibular trochanter is laterally facing, as in sauropodsa trait shared with Mussaurus.
Tuebingosaurus and Schleitheimia are found very closely related in the equally-weighted topologies, and in 7% of the MPTs in Analysis 2, the two OTUs form a clade (Table 15). The ilium of Schleitheimia is represented by the acetabular region and the postacetabular blade (PIMUZ A/III 550, in Rauhut et al. 2020). The outline of the pubic and ischiadic peduncle is similar to that in Tuebingosaurus, with a broad and lateromedially twisted pubic peduncle (in lateral view) and a posteriorly projected 'heel' in the ischiadic peduncle (Fig. 6). The postacetabular process in Schleitheimia has a triangular outline, whereas in Tuebingosaurus the ventral margin of the postacetabular process is ventrally turned (Fig. 6). In Tuebingosaurus, the supracetabular crest is accompanied by a medially projected crest of similar size, seen in both ilia (Fig. 7a), whereas in Schleitheimia, the ilium only has the supracetabular crest projecting laterally. The partial left femur of Schleitheimia (PIMUZ A/III 551, in Rauhut et al. 2020) cannot be accurately compared to that of Tuebingosaurus (Fig. 12) because the preserved elements in Schleitheimia, i.e., the midshaft and the distal end, correspond to some of the reconstructed parts of the femur in Tuebingosaurus. The fourth trochanter of Tuebingosaurus has a more pronounced asymmetrical outline, with a curved ventral margin and a straight dorsal margin, whereas in Schleitheimia, the fourth trochanter is more symmetrical, but the ventral margin has a slope in the opposite direction to the dorsal margin.
The astragalus recovered from Schleitheim-Santierge (PIMUZ A/III 4391, in Rauhut et al. 2020) was referred to as Plateosaurus sp. since it has a similar outline to the one of specimen SMNS 13200; however, the neotype SMNS 13200 does not have an anterior foramen, as is present in PIMUZ A/III 439. Furthermore, the lateral and posterior outlines are similar to the overall astragalar morphology of Tuebingosaurus, which suggests that this specimen is more appropriately referred to as cf. Tuebingosaurus than to as Plateosaurus sp.
Ohmdenosaurus is a sauropodomorph from the Early Jurassic found in the Posidonia Shale at Holzmaden, Germany (Wild 1978). It was considered a member of the 'family' Vulcanodontidae (McIntosh 1990) and has been compared to Rhoetosaurus, the earliest sauropods from the Early Jurassic characterised for having slender tibiae (Nair and Salisbury 2012). In our topologies, Ohmdenosaurus is consistently (93% of the MPTs) grouped with Isanosaurus, Tazoudasaurus, Vulcanodon and sauropods (iteration 2.1), consistent with previous diagnoses of Ohmdenosaurus. The proximal end of the tibia in Ohmdenosaurus has a circular outline, as lateromedially wide as it is anterodorsally long, with a cnemial crest oriented anteriorly; in Tuebingosaurus, the proximal outline is more elliptical, with a lateromedially width longer than the anteroposterior length, and a cnemial crest oriented anterolaterally. The tallest point of the cnemial crest in Ohmdenosaurus is halfway along the length of the crest, whereas in Tuebingosaurus, it is closer to the proximal end of the cnemial crest. In Ohmdenosaurus and Tuebingosaurus, in the proximal articular surface of the tibia, the posterior end of the fibular condyle is anteriorly placed to the posterior margin of the articular surface. Distally, the lateromedial width of the tibia is larger than the anteroposterior length in Ohmdenosaurus, but it is subequal in Tuebingosaurus. As in other Sauropodiformes, Ohmdenosaurus and Tuebingosaurus have a posteroventral process set well back from the anterolateral corner of the distal tibia.
Non-sauropod sauropodomorphs have a periodically interrupted growth, which translates into the formation of fibrolamellar bone interrupted by regularly spaced growth marks, unlike in sauropods, where the growth is continuous (Chinsamy 1993; Sander et al. 2004;Sander and Klein 2005;Cerda et al. 2017;Fig. 21). In sauropods, there is an uninterrupted deposition of fibrolamellar bone tissue during the early development, followed by periodical interruptions that form lines of arrested growth, LAGs, after sexual maturity was attained (Sander et al. 2004). The histological configuration of the femur in Tuebingosaurus is similar to that found in Mussaurus and Lessemsaurus (Cerda et al. 2017). In Mussaurus, the vascular canals have a plexiform arrangement alternating with regions of longitudinally oriented canals. In one specimen of Mussaurus (MLP 61-III-20-22), there is a large proportion of woven fibered bone relative to the parallel fibered bone, whereas another specimen (MPM-PV-1815) has more parallel fibered bone relative to the woven fibered bone. In Lessesaurus, however, specimen PVL 4822/64 has a larger proportion of woven fibered bone than parallel fibered bone. In our topologies, Tuebingosaurus is placed between the early non-sauropod sauropodomorphs, such as Riojasaurus and Coloradisaurus, and the more derived non-sauropod sauropodomorphs, such as Volkheimeria and Patagosaurus. Riojasaurus and Coloradisaurus have a larger proportion of parallel fibered bone than woven fibered bone; in Riojasaurus, there is an abrupt change of vascularisation pattern at the outer cortex (Cerda et al. 2017).
On the other hand, Volkheimeria and Patagosaurus have a matrix exclusively with woven fibered bone. In lamellar bone, successive thin layers form a plywood structure. Woven fibered bone consists of coarse and loosely packed collagen fibres with no spatial order and a high vascular density. Parallel fibered bone seems to be an intermediate between lamellar and woven fibered bones. Lamellar bone suggests a slow growth rate, whereas woven fibered bone suggests a fast growth. Therefore, animals like Riojasaurus and Coloradisaurus seem to have had a slower growth through growth cycles (Apaldetti et al. 2018), whereas animals like Mussaurus, Lessemsaurus (Apaldetti et al. 2018) and Tuebingosaurus have a faster growth rate but show growth cycles. Sauropods evolved fast and continuous growth rates.

Plateosaurian and massopodan characters
In Upchurch et al. (2007), Plateosauridae was restricted to Coloradisaurus, Plateosaurus and Riojasaurus, defined by three synapomorphies: 1) a deep, transverse wall of bone between the basipterygoid process, 2) a deltopectoral crest with a sigmoid outline in anterior view, 3) a 'heel'-like projection of the posterior margin of the ischial articulation. Of these three characters, the available material of Tuebingosaurus has a 'heel'-like projection in the ilium (3), justifying the referral to Plateosauridae. Massospondylus and Lufengosaurus were included by Upchurch et al. (2007) in the group Plateosauria, supported by six synapomorphies: 1) a triangular outline in the external naris, 2) a shelf-like area lateral to the external naris, 3) a prefrontal ventral process long and extending down the medial surface of the lacrimal, 4) a supratemporal fenestra obscured in lateral view by the supratemporal bar, 5) laterally expanded tables at the mid-length of the distal surface of cervical neural spines, and 6) a second distal carpal not wholly covering the proximal surface of metacarpal II. None of these characters can be traced to Tuebingosaurus.
The characters were mapped to compare the synapomorphies with Upchurch et al. (2007) using the character-by-taxon matrix and the topology reported therein. This character mapping was also repeated in all the topologies where these results are not reported in the paper. For example, in Yates (2007b), Massospondylidae was recovered as a group containing Massospondylus, Coloradisaurus and Lufengosaurus. Six synapomorphies supported this node: 1) dorsal profile of the snout with depression behind the naris, 2) the symphyseal end of the dentary is strongly curved ventrally, 3) the length of cervical 4 or 5 exceeds four times the anterior centrum height, 4) manual digit I is greater than the length of manual digit II, 5) a pyramidal dorsal process on the posteromedial corner of the astragalus, 6) the length of the pedal digit II is less than 90% of the length of the ungual of the pedal digit I. Again, none of these characters can be applied to Tuebingosaurus.  fig. 2b), and the histological sample of Tuebingosaurus was redrawn from the photographs published in Klein (2004, fig.3E), interpreted there as a fully grown Plateosaurus.
Regarding Massopoda, the mapping of characters in Yates (2007b) recovers that this node is supported by twelve synapomorphies: 1) a slot-shaped subnarial foramen, 2) an antorbital fossa on the ascending process of the maxilla delimited by a rounded rim or a change in slope, 3) the anterior process of the lacrimal is half or less as long as the ventral process of the lacrimal, 4) the ratio of the minimum depth of the jugal below the orbit to the length between the anterior end of the jugal and the anteroventral corner of the infratemporal fenestra is greater than 0.2 -a character shared with 'Plateosaurus engelhardti', 5) the dorsal margin of the postorbital in lateral view has a distinct embayment between the anterior and the posterior dorsal processes, 6) the quadratojugal sutures along the ventrolateral margin of the jugal, 7) the teeth have serrations along the mesial and distal carinae restricted to the upper half of the crown -although noting that Plateosaurus and Colorodisaurus share this character, 8) ventral keels on the anterior cervical vertebrae, but this is reversed later on in more advanced sauropodiformes, 9) the anterior margin of the scapula rises from the blade at an angle equal to or greater than 65 o from the scapular axis, 10) metacarpal I has a proximal width that represents 80% to 100% of the metacarpal I length, noting that this character then reverses several times in sauropodiformes, 11) notch separating the posteroventral end of the ischial obturator from the ischial shaft, 12) the position of the proximal tip of the lesser trochanter is levelled with the femoral head. Nevertheless, in Tuebingosaurus, the proximal tip of the lesser trochanter is distal to the femoral head.
In McPhee et al. (2015b) and the modification by Wang et al. (2017), in the synapomorphy-based definition of Massopoda, we found that the mapped characters are consistent with the definition of Yates (2007b), except for the synapomorphies of the scapula (9) and the metacarpal (10), and adds a new one, the medial peg of calcaneum fits into the astragalus. On the other hand, Plateosauridae seems to be supported by only four synapomorphies in Yates (2007b) when mapping their characters onto the topologies: 1) the medial margin of the supratemporal fossa bears a projection at the frontal/postorbital-parietal suture producing a scalloped margin, 2) the basipterygoid processes and the parabasisphenoid process are below the level of the basioccipital condyle and the basal tubera, a condition also reported in Coloradisaurus, 3) the symphyseal end of the dentary is strongly curved ventrally relative to the long axis of the dentary, a condition shared with Massospondylus and Coloradisaurus, 4) the lateral margin of the proximal margin in metatarsal II is straight, a character present in Tuebingosaurus. In McPhee et al. (2015b) and the modification by Wang et al. (2017), the synapomorphy-based of Plateosauridae is identical but adds the transverse width of the distal humerus is less than 33% of the humeral length.
Although we get a consistent definition of massopodan characters in the matrices derived from Yates (2007b), Massopoda is not recovered as a group in the matrices derived from Upchurch et al. (2007). Furthermore, based on the comparative cladistic analyses performed by Peyre de Fabrègues et al. (2015), only 80% of the characters are shared between Upchurch et al. (2007) and Yates (2007b), suggesting that the only way to obtain a consensus is by merging the two datasets. This was partially attempted by Sekiya et al. (2013), who merged both datasets but only on the taxa that they had in common, namely for a total of 27 sauropodomorphs. Sekiya et al. (2013) recovered the topology of Prosauropoda defined only by two synapomorphies, 1) a maxillary lamina that is twice as longer than it is high, a character found only in the matrix by Upchurch et al. (2007), and 2) a centrally located tubercle in the palatine, a character found only in the matrix by Yates (2007b). Since there are contradictory phylogenetic signals, all the taxa and the characters should be included in a single matrix to discern consistent plateosaurian and massopodan features. Alternatively, a character analysis is required to assess the character delineation, the impact of such delineation in the final topology, and a comparison of the character scores to resolve any disagreement between the authors.  Fig. 3). Specimen "GPIT II" (GPIT-PV-30785) has been less often used as a comparison point (Smith and Pol 2007;Bittencourt et al. 2012), and in one study, the skull specimen "GPIT 18318a" has been used to represent the anatomy of Plateosaurus (Cabreira et al. 2016). In a small number of studies, P. gracilis has been explicitly used to compare plateosaurian anatomy against other sauropodomorphs (Pol and Powell 2007;Claessens 2004;Otero and Pol 2013;Fechner and Gößling 2014;.

Plateosaurus in comparative anatomy
A study on the growth rings in long bones from material stored in SMNS, GPIT and MSF considered all individuals to belong to the same species (Klein 2004;Sander and Klein 2005). Klein (2004) included the specimens "GPIT I" (GPIT-PV-30785, the composite, no sample was taken), "GPIT 11921" (which refers to the humerus of GPIT-PV-30788, currently lost), "GPIT 192.1" (currently the femur of Tuebingosaurus), "GPIT 163" (currently the tibia of Tuebingosaurus), and "GPIT II" (GPIT-PV-30784, no histological sample was taken). The analysis of the medulla of long bones (Sander and Klein 2005) determined three groups of growth: "fast growth" in specimens where the fibrolamellar bone is the last tissue type to have been formed, "slow growth" in specimens in which growth cycles in fibrolamellar bone in the outer cortex become thinner and loses vascularity, and "fully grown" (i.e. "GPIT 163", "GPIT 11921", the femur and tibia of Tuebingosaurus), with a lamellar-zonal bone with closely spaced LAGs and poor to absent vascularisation (e.g. "GPIT II"). Sander and Klein (2005) explored three hypotheses to explain the disparity in the growth patterns found in Trossingen. The first hypothesis was that several biological species were represented by the material identified as P. 'engelhardti'. The hypothesis was rejected because, at the time, there was an agreement that there was only evidence for one species of sauropodomorph in the Triassic beds of Central Europe (Moser 2003;Yates 2003a;Galton and Upchurch 2004). Similarly, sexual dimorphism was rejected as an explanation since the body size shows a unimodal distribution (Sander and Klein 2005). Finally, the hypothesis stated that P. 'engelhardti' had strong developmental plasticity. The latter was supported because individuals from the Frick bone bed are smaller on average, whereas the Trossingen bone beds yield larger individuals. The phenotypic plasticity was explained as either a habitat difference with different plants in one region or a product of a change in habitat through time (Sander and Klein 2005). Nevertheless, as outlined in the section above (see section 'Taxonomic composition of Plateosaurus in phy-logenetic analyses'), after 2007, it became clear that the consensus on Plateosaurus as a monospecific genus was disputed, with four potential species being identified and used so far, namely P. 'engelhardti', P. gracilis, 'P. erlenbergiensis' (= P. longiceps) and P. ingens, with several specimens from Frick belonging to Gresslyosaurus and Schleitheimia . Then, after revising the taxonomy, the different growth patterns could support separating distinct species in Central Europe.

Conclusions
Based on our phylogenetic analysis, the new species Tuebingosaurus maierfritzorum is positioned as the earliest massopodan discovered in the Trossingen beds (Fig. 19). It displays some characters traditionally considered plateosaurian, like the heel-like projection in the posterior part of the ischiadic peduncle of the ilium and a straight lateral margin in metatarsal II. The fact that it has been illustrated since the early 20 th century as part of Plateo- GPIT-PV-41827). The cortical bone on the left side of the fossil is fractured into flakes, which can be explained if the carcass was exposed over a long time on the mud, two to four years, before being buried -in the reconstruction, the animal will fall to its right body side. The reconstruction shows the animal sinking in a mud trap, attacked by a rauisuchian, Teratosaurus Meyer, 1861, which has also been found in the Trossingen Formation in Baden-Württemberg (Brusatte et al. 2009). In the background, a herd of P. trossingensis runs away from the scene. The flora in the swamp is reconstructed based on fossils from the Germanic basin, with shoots of horsetails and ferns covering the swamp and a forest comprising cycads (Taeniopteris Brongniart, 1828), lycophytes (Lepacyclotes Emmons, 1856) and coniferous plants (Brachyphyllum Brongniart, 1828) (Kustatscher et al., 2018). saurus may suggest that some noise has been introduced into the phylogenetic analyses of the past decade by assuming all the medium to large-sized sauropodomorphs from Germany belonged to the same species. It is also clear that there is no consensus, in phylogenetic terms, on plateosaurian features and massopodan features since, through the literature, two incompatible overall topologies have been produced. Through comparative anatomy and the evidence from our phylogenetic analysis, Tuebingosaurus displays several derived features consistent with the position among massopodans and hints to an early diversification of Sauropodomorpha as they occupied the vacant niches in Pangaea left by rhynchosaurs and aetosaurs (Barrett et al. 2010). A rapid disparification event could explain the contradictory phylogenetic signals discussed in the literature. Many cranial characters that support one group could be a product of convergence as the animals adopted similar feeding strategies in different parts of Pangaea.
Furthermore, a thorough revision needs to be done to the material referred to as P. trossingensis or Plateosaurus that was not obtained from the Obere Mühle outcrop, and the hypothesis that these are different species needs to be tested with morphometric, specimen-level phylogenetic, and stratigraphic analyses. Nevertheless, restricting P. trossingensis to SMNS 13200, GPIT-PV-30784, AMNH FARB 6810, and all Seemann's material stored in Stuttgart should remove any noise that may have been added by using the literature in which all specimens were considered Plateosaurus. SMNS 13200, GPIT-PV-30784, AMNH FARB 6810, and all Seemann's material specimens come from the lower dinosaur bone bed in Obere Mühle and are likely to represent different individuals that died at different times, but that can be referred to as part of the same chronospecies.

Competing interests
The authors have declared no competing interests exist.

Author contributions
O.R.R.F and I.W. conceived the idea, analyzed, interpreted, and discussed the data, and contributed to the final version of the manuscript.

Competing interests
The authors have declared no competing interests exist.

Acknowledgements
We would like to thank Toru Sekiya and Emanuel Tschopp, whose valuable comments and suggestions helped us improve the quality of the manuscript.
We thank Henrik Stöhr for his help tracing the material in the collection and compiling a lot of the information known about the speci-mens in the collection as well as Agnes Fatz for taking the photographs of the material that help document this work. We also thank Marcus Burkhardt for the painting reconstructing Tuebingosaurus maierfritzorum.
Funding. Some of the first-hand data collected for this project was obtained as part of the funding from the Consejo Nacional de Ciencia y Tecnología (CVU 540250/Grant: 384614 granted to ORRF). In addition, IW was supported by DFG-grant WE5440/6-1.