bs_bs_banner Zoological Journal of the Linnean Society, 2014, 171, 623–655. With 18 figures The anatomy, phylogenetic relationships, and stratigraphic position of the Tithonian–Berriasian Spanish sauropod dinosaur Aragosaurus ischiaticus RAFAEL ROYO-TORRES1*, PAUL UPCHURCH2, PHILIP D. MANNION3, RAMÓN MAS4, ALBERTO COBOS1, FRANCISCO GASCÓ1, LUIS ALCALÁ1 and JOSÉ LUIS SANZ5 1 Fundación Conjunto Paleontológico de Teruel-Dinópolis/Museo Aragonés de Paleontología, Avda. Sagunto s/n, 44002 Teruel, Spain 2 Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, UK 3 Department of Earth Science and Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ, UK 4 Departamento de Estratigrafía, IGEO (CSIC), Universidad Complutense de Madrid, Ciudad Universitaria 28040 Madrid, Spain 5 Unidad de Paleontología, Departamento de Biología, Universidad Autónoma de Madrid, 28049 Madrid, Spain Received 16 October 2013; revised 16 January 2014; accepted for publication 30 January 2014 The first dinosaur to be named from Spain, the sauropod Aragosaurus ischiaticus Sanz, Buscalioni, Casanovas and Santafé 1987, is known from associated postcranial remains of one individual from the Las Zabacheras site in Galve, Teruel Province, Spain. Results of recent fieldwork confirm that the Las Zabacheras site represents a deltaic sediment complex in the Villar del Arzobispo Formation with a Tithonian–Berriasian (latest Jurassic– earliest Cretaceous) age. Description of the anatomy of Aragosaurus (including several previously undescribed elements) enables a re-evaluation of this taxon’s relationships. Aragosaurus ischiaticus has six autapomorphies in the axial and appendicular skeleton, including the presence of epipophysis-like protuberances on middle caudal postzygapophyses. Phylogenetic analyses, using three independent data sets, support the view that Aragosaurus is a basal macronarian sauropod, lying outside of Titanosauriformes. Aragosaurus possesses derived states that are shared with basal Titanosauriformes, indicating that some characters previously considered to represent titanosauriform synapomorphies have a slightly wider distribution. A tooth, previously described as Aragosaurus, cannot be referred to this taxon as it was recovered from a different locality, and there are no overlapping elements with the holotype; it is here regarded as representing an indeterminate titanosauriform. These results, combined with new data on the stratigraphic age of Aragosaurus, demonstrate that basal macronarian sauropods were present in Europe at the end of the Late Jurassic, alongside more derived titanosauriforms. Aragosaurus is one of four genera of sauropod recovered from the Villar del Arzobispo Formation in Spain, making the latter an important contributor to our understanding of Late Jurassic sauropod diversity alongside the well-known contemporaneous faunas of the African Tendaguru Formation and the North American Morrison Formation. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655. doi: 10.1111/zoj.12144 ADDITIONAL KEYWORDS: Berriasian – Cretaceous – Jurassic – Macronaria – Spain – Tithonian – Villar del Arzobispo Formation. *Corresponding author. E-mail: [email protected] © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 623 624 R. ROYO-TORRES ET AL. INTRODUCTION Sauropoda have been widely recorded in the Mesozoic, from their origin in the Late Triassic until their extinction at the end of the Cretaceous (McIntosh, 1990; Upchurch, Barrett & Dodson, 2004a). During their evolutionary history, the diversity of sauropod groups fluctuated, including a trough in the Oxfordian, and a peak in the Kimmeridgian–Tithonian, with more than 30 taxa recorded (Upchurch & Barrett, 2005; Mannion et al., 2011; Upchurch et al., 2011). Following an apparent extinction at the Jurassic–Cretaceous boundary, the earliest Cretaceous (Berriasian–Hauterivian) is characterized by particularly low diversity, with less than ten sauropod taxa known from each stage (Hunt et al., 1994; Upchurch & Barrett, 2005; Mannion et al., 2011; Upchurch et al., 2011). This reduction in the diversity of sauropod taxa seems to reflect some important faunal turnover events among dinosaur communities at this time (Sereno, 1997, 1999; Upchurch et al., 2011). The Villar del Arzobispo Formation in Spain reveals a high diversity of sauropods living on the Iberian Plate during the Jurassic–Cretaceous transition. To date, diplodocids, turiasaurs, and titanosauriforms have been recovered (Royo-Torres, Cobos & Alcalá, 2006; Royo-Torres et al., 2009; Suñer, Santisteban & Galobart, 2009). In this work we reevaluate the first dinosaur ever described in Spain: the sauropod Aragosaurus ischiaticus Sanz, Buscalioni, Casanovas and Santafé 1987. The site where this dinosaur was found, Las Zabacheras, is here placed in its correct stratigraphical context. Moreover, as well as re-describing previously considered elements of the Aragosaurus holotype, we present the first description and illustrations of additional elements (e.g. thoracic ribs, coracoid, humerus, carpal, four metacarpals, and pedal phalanges). Comparisons of Aragosaurus with other sauropods result in the recognition of several new autapomorphies for the Las Zabacheras taxon. We analyse the phylogenetic relationships of Aragosaurus using revised versions of the ‘traditional’ data sets presented by Wilson (2002) and Upchurch et al. (2004a) that include improved sampling of Iberian sauropods (Royo-Torres & Upchurch, 2012; Royo-Torres, Alcalá & Cobos, 2012). We also discuss its position based on a recent analysis of titanosauriform interrelationships (Mannion et al., 2013). INSTITUTIONAL ABBREVIATIONS CPT, Museo Aragonés de Paleontología (Fundación Conjunto Paleontológico de Teruel-Dinópolis), Spain; GMNH, Gunma Museum of Natural History, Gunma, Japan; IG, Museo Provincial de Teruel, Spain; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing, China; MCNV, Museo de Ciencias Naturales de Valencia, Spain; MG LNEG, Museu Geológico do Laboratório Nacional de Energia e Geologia, Lisboa, Portugal; MPG, Museo Paleontológico de Galve, Spain; MPZ, Museo Paleontológico de la Universidad de Zaragoza, Spain; USNM, National Museum of Natural History Smithsonian, Washington DC, USA; YPM, Peabody Museum, Yale, USA. COLLECTION HISTORY Las Zabacheras, the type locality of A. ischiaticus, is located 150 m north of Galve village (Fig. 1) in Teruel Province, Aragón, north-eastern Spain (grid coordinates Universal Transverse Mercator 679 266 and 4 503 362). In 1934, construction of the old road to Galve led to the discovery of the type locality, initially named ‘La Carretera’ (Fernández-Galiano, 1958; 1960). The first recorded discovery of dinosaur bones was in 1958, when a resident of the village, José María Herrero, reported dinosaur bones from Galve in the local newspaper Lucha, and then informed the service of excavations of the Diputación Provincial de Teruel (Instituto de Estudios Turolenses, IET; Millán, 2004; Ruiz-Omeñaca et al., 2004; Cobos, 2011). Dimas Fernández-Galiano and Purificación Atrián, members of IET, visited Galve and located elements of Aragosaurus held in private hands by members of the local population. After this, Fernández-Galiano returned to Galve in order to excavate new remains from the Las Zabacheras site, including some of the vertebrae and long bones (Fernández-Galiano, 1958). All of these elements, including those collected from villagers and the vertebrae and long bones collected later, were housed in the Museo Provincial de Teruel (Cobos, 2011). Albert F. de Lapparent (1905–1975), from the Institut Catholique, Paris, was the first palaeontologist to suggest that these remains represented a new taxon of sauropod in the scientific journal Teruel (Lapparent, 1960). This author was also responsible for the change of the site name from ‘La Carretera’ to ‘La Zabacheras’ (ZH; Lapparent, 1960; Ruiz-Omeñaca et al., 2004). Subsequently, in 1982, a team of palaeontologists comprising researchers from the Universidad Autónoma de Madrid and the Instituto de Paleontología de Sabadell (today the Institut Català de Paleontología) informed José María Herrero of the intention to excavate the Las Zabacheras site during the following year; however, it transpired that José María Herrero had already excavated part of the site, so an agreement was made to carry out joint fieldwork (Sanz, 2007). The resulting material, known as the ‘J.M. Herrero Collection’, enabled the first dinosaur from Spain to be named: Aragosaurus ischiaticus (Sanz et al., 1987, 1990). The holotype also included the material cited as ‘the sauropod from Las Zabacheras’ by Lapparent (1960), and a tooth was referred from a nearby unnamed © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 625 Figure 1. Geological map showing the locality where Aragosaurus was found in Spain. Modified from Díaz-Molina & Yebenes (1987), Soria de Miguel (1997), and with data from Peropadre, Meléndez & Liesa (2012): 1, Upper Triassic (Keuper facies); 2, Cortes de Tajuña, Cuevas Labradas and Barahona formations; 3, Turmiel Formation; 4, Chelva Formation; 5, Loriguilla Formation; 6, Higueruelas Formation; 7, Villar del Arzobispo Formation; 8, El Castellar Formation; 9, Camarillas Formation; 10, Artoles Formation; 11, Morella Formation; 12, Areniscas de las Parras de Martín Formation; 13, Utrillas Formation; 14, Upper Cretaceous; 15, Cenozoic; 16, Quaternary. locality (Sanz, 1982; Sanz et al., 1987). As a result of these events, the holotype specimens are distributed between two collections: (1) the material recovered by Dimas Fernández-Galiano in the 1950s, which is deposited at the Museo Provincial de Teruel with the acronym ‘IG’; and (2) the material excavated by J.M. Herrero and the team of author J.L.S. in the 1980s, with acronym ‘ZH’, now in the Museo Paleontológico de Galve. In 2001, on the occasion of the inauguration of the Dinópolis museum in Galve, two of us (R.R.T. and A.C.) recovered the material (fragments of vertebrae and ribs, coracoids, humerus, carpal, metacarpus, and ischia) deposited by Dimas FernándezGaliano in the Museo Provincial de Teruel (Royo-Torres, Canudo & Ruiz-Omeñaca, 1999; Cobos, 2011), moving it to a satellite museum of Dinópolis in Galve (today belonging to the Museo Aragonés de Paleontologia), with the other specimens remaining in the Museo Paleontológico de Galve. SYSTEMATIC BACKGROUND Of the material comprising the Aragosaurus holotype, Lapparent (1960) briefly described eight caudal vertebrae, ten fragments of ribs, a partial scapula, a radius, an ulna, a carpal, fragments of metacarpus, two phalanges, a proximal portion of the left ischium, a left pubis, and a left humerus (misidentified as a right femur). This material was suggested to represent a new sauropod genus related to ‘camarasaurids’, based on the forelimb appearing more similar to Camarasaurus than Apatosaurus, and was differentiated from © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 626 R. ROYO-TORRES ET AL. rebbachisaurids and ‘titanosaurids’ by the caudal vertebral morphology (Lapparent, 1960). Twenty-seven years later, a new taxon was erected based on this material (and newly recovered elements) in the Spanish journal Estudios Geológicos (Sanz et al., 1987). The genus name was dedicated to the region of Aragón, and the species name referred to the ischium, i.e. Aragosaurus ischiaticus. The diagnosis was based on the morphologies of the tooth, caudal neural spines, scapula, ischia, and the humerus/femur length ratio (Sanz et al., 1987, 1990). Those authors separated Aragosaurus from diplodocid taxa by the absence of pleurocoels and winglike lateral processes in the caudal vertebrae. They included Aragosaurus within Camarasauridae (sensu Steel, 1970), based on the presence of transversely expanded anterior caudal neural spines, a humerus/femur ratio of 0.82, a lateral bulge below the greater trochanter on the femur, and a ratio between the minimum and maximum widths of the scapular blade similar to that in Giraffatitan (= ‘Brachiosaurus’) (Sanz et al., 1987). Later, in his revision of the Sauropoda, McIntosh (1990) was in agreement, including A. ischiaticus within Camarasauridae, based on a number of overall similarities with Camarasaurus, as well as Lourinhasaurus alenquerensis, which was then considered a species of Camarasaurus. Subsequent studies, however, indicated that there are difficulties with the referral of Aragosaurus to the Camarasauridae. In particular, Upchurch (1995) and all subsequent phylogenetic analyses have shown that McIntosh’s (1990) Camarasauridae is a polyphyletic assemblage, including the mamenchisaurid Tienshanosaurus (Sekiya, 2011), the basal somphospondylan Euhelopus (Wilson & Upchurch, 2009), the derived titanosaur Opisthocoelicaudia (Upchurch, 1995), as well as Aragosaurus and Camarasaurus. Indeed, if Camarasauridae is defined phylogenetically as all taxa more closely related to Camarasaurus than to Saltasaurus, then the only member of this family is Camarasaurus itself (see Upchurch, 1995, 1998; Wilson, 2002; Upchurch et al., 2004a; Harris, 2006; Wilson & Upchurch, 2009; Mannion et al., 2013). After McIntosh (1990), only a few works have provided new information on Aragosaurus. Firstly, in a description of the pubis (Royo-Torres & Ruiz-Omeñaca, 1999; Royo-Torres et al., 1999), a position within Camarasauromorpha (sensu Salgado, Coria & Calvo, 1997), in the clade Camarasaurus + Titanosauriformes (sensu Wilson & Sereno, 1998), or in the clade ‘Brachiosauria’ (sensu Upchurch, 1998) was suggested. Canudo et al. (2001) proposed a slightly more precise placement of Aragosaurus, arguing that the presence of the lateral bulge on the proximal femur supported its inclusion within Titanosauriformes, a character considered diagnostic for this clade (Salgado et al., 1997). None of these studies included Aragosaurus in a cladistic analysis, however. Aragosaurus was incorporated into a phylogenetic data matrix for the first time in 2004, although the limited available anatomical data meant that many characters could not be scored and its position was highly unstable, leading Upchurch et al. (2004a) to consider it as Eusauropoda incertae sedis. In this sense, Sanz (2009) shows a position based on the matrix of Upchurch et al. (2004b), where Aragosaurus is included in a polytomy within Eusauropoda, but outside of Titanosauriformes. New anatomical data were presented in the thesis of the lead author (Royo-Torres, 2009), greatly increasing the number of characters that could be scored. This work, and a subsequent study exploring the evolutionary relationships of basal titanosauriforms (Mannion et al., 2013), included Aragosaurus in a phylogenetic analysis, in both cases recovering it as a basal macronarian, outside of Titanosauriformes. D’Emic (2012: table 9) also assessed the relationships of Aragosaurus by noting which synapomorphies of various clades were present and absent in the published material. Based on two features of the ischium, D’Emic (2012) provisionally included Aragosaurus as a non-titanosaur titanosauriform. GEOLOGICAL SETTING AND AGE OF THE LAS ZABACHERAS SITE The Las Zabacheras site is located in a syncline in the Galve area of Teruel Province, Aragón, north-eastern Spain, in the Central Iberian Range (Fig. 1). Here, there is a Mesozoic sedimentary succession infilling one of the sub-basins of the Maestrazgo Basin: the Galve subbasin (Salas et al., 2001). The Las Zabacheras section shows a complete sequence of units spanning from the Late Jurassic to the Early Cretaceous (Díaz-Molina et al., 1984; Díaz-Molina et al., 1985; Díaz-Molina & Yebenes, 1987), formed as a consequence of the upper synrift succession developed in the Iberian rift (Salas et al., 2001; Liesa et al., 2006). Originally, the Las Zabacheras site was included in the Early Cretaceous ‘Weald facies’ (Derréal, 1959; Lapparent, 1960). Detailed stratigraphic and cartographic work on the terrestrial sediments enabled Díaz-Molina & Yebenes (1987) to include the Las Zabacheras site in their ‘Unit 3’, suggesting a Hauterivian (?) age (Díaz-Molina et al., 1984, 1985; Díaz-Molina & Yebenes, 1987). Subsequently, Cuenca-Bescós et al. (1994) and Soria de Miguel (1997) included this ‘Unit 3’ in the ‘Areniscas y Calizas de El Castellar’ Formation, dated as Hauterivian–lower Barremian. This placement was based on the identification of: [sic] ‘Esta discontinuidad es identificable en el campo por un importante cambio litológico y sedimentológico marcada por la presencia de las arcillas rojas típicamente continentales de la base de la © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS Formación Castellar’ (‘This discontinuity is identifiable in the field by a major lithological and sedimentological change marked by the presence of the typically continental red clays of the base of the Castellar Formation’; Cuenca-Bescós et al., 1994: p. 51). This view has recently been defended by Canudo et al. (2012) based on the presence of synsedimentary faults on the discontinuity between the Villar del Arzobispo and El Castellar formations, and on charophyte and palynological data; however, as outlined below, there are a number of problems with these interpretations of the data, and we here provide a revised view of the geology of this area. To test whether ‘Unit 3’ is equivalent to the lower part of the El Castellar Formation, we have compared it with the stratotype of the latter, defined by Salas (1987) in the Peñagolosa sub-basin, in Barranco de Montoro, Teruel Province. The lower part of this formation is characterized by purple- and ochrecoloured shales, which are interbedded with channelized levels of tabular geometry and erosive bases of finegrained ochre-coloured sandstones, occasionally showing parallel lamination and cross stratification. The upper part of the El Castellar Formation is characterized by an alternation of marls and grey limestones, with abundant charophytes, gastropods, and fish teeth, showing frequent instances of plant bioturbation (Salas, 1987). The lower part of the El Castellar Formation is interpreted as having formed in an alluvial fan environment (Soria de Miguel, 1997; Liesa et al., 2006), whereas the upper part represents a lacustrine environment (Salas, 1987; Soria de Miguel, 1997; Liesa et al., 2006). Charophytes (Atopochara trivolvis triqueta) are present only in the upper part of the type section of the El Castellar Formation (Salas, 1987; Soria de Miguel, 1997; Ruiz-Omeñaca et al., 2004). The Galve ‘Unit 3’ of Díaz-Molina & Yebenes (1987) is an 85-m section comprising the following lithofacies: conglomerate with rounded carbonate pebbles; fine- to middle-grained quartz arenite sandstone, with glauconite grains showing lenticular and channelized geometry, wave rippleformed cross-lamination, sand dune-formed crossstratification; red clay, and silt with root traces associated with hydromorphic palaeosoils (Fig. 2). According to previous studies (Díaz-Molina et al., 1984, 1985; Díaz-Molina & Yebenes, 1987) and our own observations, these sediments can be interpreted as a deltaic coastal plain with tidal channels and intertidal to supratidal interchannel areas (with the latter showing hydromorphic palaeosoils) that were deposited during regressive–transgressive pulses in a tide-influenced coastal realm. These facies, and our interpretation of the sedimentary environment, are very different from the alluvial fan environment described for the lower unit of the El Castellar Formation (Soria de Miguel, 1997; Liesa et al., 2006). 627 Microfossils have not been found in ‘Unit 3’ (Díaz-Molina & Yebenes, 1987), and the only charophytes located by Canudo et al. (2012), Globator maillardi steinhauseri (latest Berriasian–Hauterivian in age) are stratigraphically situated 90 m up the Las Zabacheras section, at the top of the ‘local datum’ cited by Canudo et al. (2012). This level is a calcrete layer (Fig. 2), located just below ‘Unit 4’ of Díaz-Molina & Yebenes (1987). Palynological data are currently unavailable for the Las Zabacheras site, but preliminary analyses produced negative results (Bienvenido Díez pers. comm. 2013). Furthermore, the palynological data used by Canudo et al. (2012) comes from Piélago 0 (Díez et al., 1995), which is located more than 4 km away from Las Zabacheras. Piélago 0 has an unclear position in the column and even on maps, and is variably placed at the base or in the middle of their stratigraphic section of the lower El Castellar Formation (i.e. Ruiz-Omeñaca et al., 2004; Ruiz-Omeñaca, 2006; Canudo et al., 2012). As such, we suggest that it is not a reliable indicator for determining the age of the Las Zabacheras site. In summary, although Canudo et al. (2012) included ‘Unit 3’, and thus consequently the Las Zabacheras site, within the El Castellar Formation, the available evidence does not support this placement. We conclude that the El Castellar Formation in Galve is situated stratigraphically above ‘Unit 3’, and instead corresponds to the ‘Unit 4’ of Díaz-Molina & Yebenes (1987) (Figs 1 & 2). The contact between ‘Unit 3’ and ‘Unit 4’ is an angular nonconformity, and in the field it is possible to observe changes of strike (5–28°) and dip (5–20°) between both units (Figs 1 and 3; Table 1). These values were measured in the last layer of ‘Unit 3’ and in the first layer of ‘Unit 4’ (Figs 1 and 3). As Díaz-Molina & Yebenes (1987) indicated that a lithological transition between ‘Unit 3’ and ‘Unit 4’ is absent, the discordance described by Canudo et al. (2012) should be placed within the Villar del Arzobispo Formation, which is generally represented by sequences beginning with a subaerial exposure surface (incipient palaeokarst on top of carbonates), followed by a clear vertical facies change from shallow marine carbonates to coastal siliciclastic sediments (shales, sandstones, and conglomerates). Within the Villar del Arzobispo Formation, the transition between its dominantly carbonate part (‘Unit 2’) and its dominantly siliciclastic part (‘Unit 3’) is a characteristic feature (Mas, Alonso & Meléndez, 1984; Luque et al., 2005; Santisteban & Santos Cubedo, 2010a). In addition, some carbonated lithosomes from ‘Unit 2’ show a lateral change to siliciclastic facies, equivalent to those of ‘Unit 3’, in the south-west of Galve. Furthermore, in the Galve area, ‘Unit 4’ (El Castellar Formation) has, at its base, a massive to nodular calcrete level (Fig. 2). This level is particularly developed where this unit lies © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 628 R. ROYO-TORRES ET AL. Figure 2. General stratigraphic section indicating the stratigraphic settings for localities in the Villar del Arzobispo Formation from Galve and the section of the Las Zabacheras site. Modified from Díaz-Molina & Yebenes (1987). © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 629 Figure 3. The ‘Fault in the Zabacheras area’ – Mesozoic syn-rift faults affecting the Villar del Arzobispo Formation (‘Unit 2’ and ‘Unit 3’) fossilized by the Castellar Formation (‘Unit 4’). Abbreviations: U2, Villar del Arzobispo Formation dominantly carbonate facies; U3, Villar del Arzobispo Formation dominantly siliciclastic facies; U4, El Castellar Formation; U5, Camarillas Formation; ZH, Las Zabacheras site; black lines crossing U2 and U3, Mesozoic syn-rift faults; black lines perpendicular to the latter only in U2, Cenozoic faults; contact between U2 and U3, sedimentary contact; contact between units U3 and U4, unconformable contact; contact between units U4 and U5, sedimentary contact. Table 1. Measurements of the strikes and dips between the sedimentary contact of the Villar del Arzobispo Formation ‘Unit 3’ and El Castellar Formation ‘Unit 4’ (Fig. 1) in Galve, Spain Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 Point 7 Grid coordinates UTM Strike and dip Unit 3 Grid coordinates UTM Strike and dip Unit 4 E 0678711 N 4502099 E 0678895 N 4502534 E 0679164 N 4503138 E 0680625 N 4502996 E 0681338 N 4502098 E 0681807 N 4500199 E 0682050 N 4499748 206° 50E E 0678734 N 4502077 E 0678935 N 4502531 E 0678200 N 4503240 E 680618 N 4502988 E 0681338 N 4502098 E 0681791 N 4500197 E 0682020 N 4499762 217° 43E 198° 38W 230° 41S 315° 68S 325° 73S 336° 39W 005° 26W 226° 34W 225° 32S 338° 83 S 330° 75S 342° 40W 015° 24W UTM, universal transverse mercator. on the shaly facies of ‘Unit 3’ (the siliciclastic part of the Villar del Arzobispo Formation); however, when lying on sandstone bodies sandwiched between the shales of ‘Unit 3’, the process of development of the calcrete is less pronounced, and only develops a carbonated nodulization when the contact surface is sandstone. Moreover, it should also be noted that the faults affect ‘Unit 2’ and ‘Unit 3’, but not ‘Unit 4’, so it seems that the overlying El Castellar Formation (= ‘Unit 4’) deactivated these faults (Figs 1 and 3). © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 630 R. ROYO-TORRES ET AL. The thickness of the El Castellar Formation in Galve (= ‘Unit 4’) is about 24 m, which is reduced in comparison with other areas of the Galve sub-basin, such as those described by Liesa et al. (2006), or in comparison with the Peñagolosa sub-basin of the Maestrazgo Basin (Salas, 1987), where it has a thickness of 35 m. To sum up, in Galve, above the carbonated oolitic grainstones of the top of the underlying Higueruelas Formation (Gómez & Goy, 1979; Díaz-Molina & Yebenes, 1987), lies the Villar del Arzobispo Formation (‘Unit 2’ and ‘Unit 3’ of Díaz-Molina & Yebenes, 1987), with a thickness of approximately 260 m. ‘Unit 2’ is more carbonated (‘Baldovar facies association’ of Santisteban & Santos Cubedo, 2010a) and ‘Unit 3’ more detritical (‘Riodeva facies association’ of Santisteban & Santos Cubedo, 2010a), but both units share the same overall lithology and sedimentary interpretation with the holostratotype of the Villar del Arzobispo Formation in the South Iberian Basin and the parastratotype in the Peñagolosa sub-basin of the Maestrazgo Basin (Mas, Alonso & Meléndez, 1984; Mas et al., 1984, 2004; Salas et al., 2001). Therefore, the stratigraphical location of the Las Zabacheras site is confirmed as within the Villar del Arzobispo Formation (see also Luque et al., 2006–2007; Alcalá et al., 2009; Royo-Torres et al., 2009; Cobos, 2011; Cobos & Gascó, 2013). In addition, similar stratigraphic revisions have been reported for this formation in several other basins of the Iberian Range (Luque et al., 2005; Ipas et al., 2007; Santisteban & Santos Cubedo, 2010a, b). The Villar del Arzobispo Formation is generally considered as upper Tithonian–Berriasian in age (Mas et al., 1984, 2004). In the Galve sub-basin, the more carbonated part (‘Unit 2’ or Baldovar facies association) has been dated as Portlandian (= Tithonian), based on the presence of the Foraminifera Pseudocyclamina gr. lituus, Anchispirocyclina lusitanica, and Rectocyclamina arrabidensis (Díaz-Molina & Yebenes, 1987). It has also been dated as upper Tithonian–middle Berriasian, based on the first appearance of the foraminifera Anchispirocyclina lusitanica in the upper part of the underlying Higueruelas Formation (Aurell et al., 2010), and the presence of the charophyte Globator maillardii incrassatus close to the top of ‘Unit 2’ (Canudo et al., 2012). ‘Unit 2’ has also been considered as Tithonian based on the presence of the turtles Tropidemys sp., Plesiochelus sp., and aff. Plesiochelydae (Pérez-García, Scheyer & Murelaga, 2013). It is not currently possible to directly date the Las Zabacheras site in ‘Unit 3’, although there is a charophyte, Globator maillardi steinhauseri (latest Berriasian–Hauterivian age) located 90 m up, in the base of the El Castellar Formation (‘Unit 4’). In conclusion, the age of the Las Zabacheras site can be constrained between the Tithonian and the Berriasian, and is most likely Berriasian in age. SYSTEMATIC PALAEONTOLOGY DINOSAURIA OWEN, 1842 SAURISCHIA SEELEY, 1887 SAUROPODA MARSH, 1878 EUSAUROPODA UPCHURCH, 1995 NEOSAUROPODA BONAPARTE, 1986 MACRONARIA WILSON & SERENO, 1998 ARAGOSAURUS SANZ ET AL., 1987 Type species Aragosaurus ischiaticus Sanz et al., 1987. Generic diagnosis – see ‘Revised species diagnosis’. ARAGOSAURUS ISCHIATICUS SANZ ET AL., 1987 Holotype Aragosaurus ischiaticus is known from a partial, associated postcranial skeleton, comprising: seven dorsal rib fragments, IG 468 (V40), IG 481 (Cos), IG 487 (Phs1), IG 492 (Uls), V20; 14 caudal vertebrae, IG 453 (V8), IG 473 (V1), IG 474 (V2s), IG 450 bis (V5), IG 475 (V3), IG 476 (V4), IG 477 (V5), IG 479 (V7), IG 480 (V8), ZH-18, ZH-17, ZH-12, ZH-15, ZH-16; two caudal neural arches, IG 468 (V55 and X3); one caudal neural spine, IG 493 (V9s); eight chevrons, ZH-4, ZH-5, ZH7, ZH-8, ZH-9, ZH-11, ZH-13, ZH-14; a right scapula, ZH-1; a distal fragment of the left scapula, IG 482 (Oms); a left coracoid, IG 482 (Oms); a right humerus, IG 490 (Fes); a right ulna, IG 483 (Cu); a right radius, IG 484 (Ra); a right carpal, IG 485 (Cars); a right metacarpal I, IG 486 (rmc1); a right metacarpal II, IG 486 (rmc2); remains of metacarpals III and IV, IG 486 (rmc3), IG 486 (rmc4); a right pubis, IG 489 (Pus); a right ischium, ZH-3; a partial left ischium, IG 492 (Uls), IG 488 (Is); a left femur, ZH-2; two pedal phalanges, ZH-6, ZH10; and one ungual pedal phalanx, ZH-19. Type locality and horizon Las Zabacheras site, Villar del Arzobispo Formation (upper Tithonian–lower Berriasian). Revised species diagnosis Aragosaurus ischiaticus is diagnosed on the basis of the following new autapomorphies identified in the current study: (1) anterior caudal vertebrae possess neural spine summits with shallow, dorsolaterally facing concavities on either side of the midline; (2) small epipophysis-like protuberances on the dorsal surfaces of middle caudal vertebral postzygapophyses; (3) the long interosseous ridges on the ulna and radius represent at least a local autapomorphy, which is convergent with derived titanosaurs; (4) the interosseous ridge on the posterolateral face of the radial shaft terminates distally in a distinct tubercle; (5) carpal subquadrangular in shape in proximal and distal views, © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS with similar lateromedial and anteroposterior diameters; (6) proximal pedal phalanx (II-1 or III-1) possesses a midline ridge on its dorsal surface, where its lateral and medial faces meet each other to form an apex. Additional comments Despite the complicated history of the collection, all the bones belong to a single individual. All of them come from the same site, show the same preservation, there is no repetition of elements, and their relative sizes are consistent. A tooth described by Sanz (1982) and subsequently assigned to Aragosaurus (Sanz et al., 1987) came from a different locality, close to Las Zabacheras (Sanz et al., 1987; Canudo et al., 2001). As such, its isolation, and the lack of a preserved tooth in the holotype, precludes its referral to Aragosaurus. It was described and illustrated by Sanz (1982) as being more similar to the teeth of Giraffatitan than Camarasaurus. In labial view, the crown shows no constriction at its base, contrasting with non-neosauropods (Calvo, 1994; Upchurch, 1998), and it has a D-shaped cross section (with the greatest diameter oriented mesiodistally), differing from the cylindrical shape of diplodocoid and titanosaurian teeth (Wilson & Sereno, 1998). The slenderness index (apicobasal length of the tooth crown divided by its maximum mesiodistal width; Upchurch, 1998) of the crown is 2.45, comparable with the values seen in basal titanosauriforms, but higher than observed in most non-neosauropods and Camarasaurus, and lower than diplodocoids and most titanosaurs (Upchurch, 1998; Chure et al., 2010). The apical half of the lingual surface of the crown is concave between broad mesial (partly worn away) and distal ridges, although there is no lingual ridge. Serrations and denticles are absent. There is a V-shaped wear facet, contrasting with the planar wear facets present in diplodocoids and derived titanosauriforms (Wilson & Sereno, 1998). This combination of features suggests that this tooth crown should be regarded as representing an indeterminate titanosauriform. DESCRIPTION AND COMPARISONS AXIAL SKELETON Caudal vertebrae Portions of 14 anterior to posterior caudal vertebrae are preserved: IG 453 (V8) = Cd1, IG 493 (V9s) = the spine of Cd1, IG 474 (V2s) = Cd2, IG 473 (V1) = Cd3, ZH-18 = Cd4, IG 450 bis (V5) = Cd5, ZH-17 = Cd6, IG 475 (V3) = Cd7, IG 476 (V4) = Cd8, IG 477 (V5) = Cd9, ZH15 = Cd10, IG 479 (V7) = Cd11, IG 480 (V8) = Cd12, ZH-12 = Cd13, ZH-16 = Cd14, IG 468 (V55, X3) = fragments of neural arches. Several vertebrae are relatively complete (Fig. 4; Table 2). Although these do not 631 represent a continuous sequence, they are referred to here as Cd1–Cd14. Centra are amphicoelous/ amphiplatyan, with a concave anterior articular surface and a flat to gently concave posterior surface (Sanz et al., 1987; Royo-Torres, 2009). This type of articulation, termed ‘planiconcave’ by Tidwell, Carpenter & Meyer (2001), whereby the depth of the concavity of the anterior surface is greater than that of the posterior surface, is a feature common to most sauropods lacking procoelous anterior caudal centra (D’Emic, 2012), including Camarasaurus, Brachiosaurus, and Sauroposeidon (= ‘Paluxysaurus’) (D’Emic, 2013). Anterior caudal vertebrae of Aragosaurus differ from the procoelous anterior caudal vertebrae of titanosaurs, flagellicaudatans, mamenchisaurids, and turiasaurs (McIntosh, 1990; Upchurch, 1995, 1998; Royo-Torres et al., 2009; Mannion et al., 2013). All centra have a cylindrical cross section. Unlike many diplodocoids and titanosaurs (McIntosh, 1990; Upchurch, 1995, 1998; Wilson, 2002; Upchurch et al., 2004a; Curry Rogers, 2005; Mannion & Barrett, 2013), the ventral surfaces lack a midline sulcus or ventrolateral ridges, although very weak ridges support the posterior chevron facets. Anterior chevron facets are relatively weakly developed, but posterior chevron facets are present on all of the preserved centra. The lateral surface of the centrum merges smoothly into the ventral surface on the anteriormost caudal vertebrae, although this angle becomes slightly more abrupt further along the tail. There are no lateral pneumatic openings on the centra. The most anterior vertebrae bear a bulge on their lateroventral surface below the caudal rib (Fig. 4), probably in Cd1–Cd3. This is a derived character state shared with Giraffatitan, Abydosaurus (Chure et al., 2010), Tastavinsaurus, and the titanosaur Andesaurus (D’Emic, 2012; Mannion et al., 2013). This feature is absent in Cd4 and subsequent caudal vertebrae. This feature differs from the moderately well-developed, anteroposteriorly oriented ridge that is present on Cd5–8 at approximately two-thirds of the way up the lateral surface of the centrum. This ridge divides the lateral surface into two parts: the more ventral surface is the primary and the more dorsal surface is the secondary (sensu Salgado & Coria, 2002). The caudal ribs are simple, dorsoventrally compressed, laterally projecting processes, differing from the strongly posterolaterally directed ribs of most titanosauriforms (Mannion & Calvo, 2011; Mannion et al., 2013), but comparable with those of Brachiosaurus (Taylor, 2009). A prezygodiapophyseal lamina (PRDL) extends from the anterodorsal margin of the caudal rib (Fig. 4 of IG 473 and 474), a feature that Aragosaurus shares with a number of basal titanosauriforms, e.g. Tastavinsaurus (Canudo, Royo-Torres & Cuenca-Bescós, 2008), Venenosaurus (Tidwell et al., 2001), Abydosaurus (Chure et al., 2010), © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 632 R. ROYO-TORRES ET AL. Figure 4. Caudal vertebrae of Aragosaurus ischiaticus: A, the series of caudals in their relative positions from Cd1 to Cd14 in lateral view; B, C fragment (IG 453, Cd1) of anterior caudal vertebral centrum in anterior (B) and posterior (C) views; D, E, fragment (IG 474, Cd2) of anterior caudal vertebral centrum in lateral (D) and posterior (E) views; F, anterior caudal vertebral centrum (IG 473, Cd3) in anterior view from Lapparent (1960); G–I, anterior caudal vertebra (ZH18, Cd4) in left lateral (G), right lateral (H), and anterior (I) views; J–L, anterior caudal neural spine (IG 493, Cd1) in posterior (J), anterior (K), and lateral (L) views; LL, M, middle caudal vertebra (ZH-17, Cd 6) in left lateral (LL) and anterior (M) views; N, anterior caudal neural spine (IG 493, Cd1) in dorsal view; O, P, posterior caudal vertebra (ZH-15, Cd 10) in left lateral (O) and posterior (P) views; Q, distal caudal vertebrae (ZH-12 and 16, Cd 13 and 14) in left lateral view. Arrow 1: a bulge on the lateroventral surface below the caudal rib. Arrow 2: anterior caudal vertebrae possess neural spine summits with shallow, dorsolaterally facing concavities on either side of the midline. Arrow 3: small epipophysislike protuberances on the dorsal surfaces of middle caudal vertebral postzygapophyses. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 633 Table 2. Measurements of Aragosaurus ischiaticus caudal vertebrae Measurement Cd1 Cd3 Cd4 Cd5 Cd6 Cd7 Cd8 Cd9 Cd10 Cd 12 Cd 13 Cd 14 Centrum length Anterior centrum width Anterior centrum height Posterior centrum width Posterior centrum height Distance from front of centrum to arch Distance from back of centrum to arch Length of prezygapophyses Anteroposterior length of base of spine Transverse width of base of spine Anteroposterior length of top of spine Transverse width of top of spine – – – – – – – – – – 64 121 133 184 177 – – – – – – – – – 135 136 139 138 136 18ca 38 – – – – – 133 – 133 118 128 28ca 38 – – – – – 134 120 124 117 117 – – – – – – – 131 266 260 261 257 – – – – – – – 136 248 242 220 235 10 25 80 75 41 85ca 82 137 169 166 167 170 18 38 82 50ca 25 67 35 135 113 99 98 95 22 36 – – – – – 121 84 73 79 69 29 35 – – – – – 124 82 68 79 70 30ca 25ca – – – – – 111 73 65 70 64 25ca 35 42 – – – – Cd1–Cd14 is the ‘series’ shown in Figure 4. All measurements are in millimetres. Brachiosaurus, and Giraffatitan (Wilson, 2002), but that is also present in more basal eusauropods such as Mamenchisaurus (Young & Zhao, 1972), and most diplodocoids (Wilson, 2002), including the putative basal form Haplocanthosaurus (Hatcher, 1903). The neural arches are slightly biased towards the anterior margin of the centrum, but this is not the strong anterior shift seen in most titanosauriforms (Calvo & Salgado, 1995; Upchurch, 1995, 1998; Salgado et al., 1997), and middle caudal neural arches are situated on approximately the central half of the centrum (Fig. 4). The neural canal is circular in the most anterior caudal vertebrae, and is wider dorsoventrally than transversely in the middle caudal vertebrae (Sanz et al., 1987). Prezygapophyses project anterodorsally in the anteriormost caudal vertebrae (Cd4 and Cd6), but are directed mainly anteriorly in subsequent vertebrae. The flat prezygapophyseal articular surfaces face medially and slightly dorsally (Sanz et al., 1987). As a consequence of poor preservation in the postzygapophyseal region, it is not possible to determine whether a hyposphenal ridge was present in any of the anterior caudal vertebrae. Postzygapophyses are unusual in the middle caudal vertebrae in that they project as short processes posterolaterally from the underside of the spine (Cd10, Fig. 4N and O). These project beyond the posterior margin of the centrum but do not extend as far posteriorly as the tip of the spine. Some postzygapophyses also have epipophysis-like protuberances on their dorsal surfaces (Cd10, Fig. 4). This postzygapophyseal morphology is considered an autapomorphy of Aragosaurus. A single intraprezygapophyseal lamina (with a rounded anterior margin) floors the prespinal fossa, and this fossa is bounded by well-developed, rounded spinoprezygapophyseal laminae (SPRLs). The SPRLS merge into the anterolateral margins of the spine and diverge laterally at their anterior ends. This prespinal fossa differs from that seen in the caudal vertebrae of Tastavinsaurus (MPZ99/9) in that it is not slotlike or closed at its anterior end. Spinopostzygapophyseal laminae (SPOLs) rapidly fade into the posterolateral margins of the spine, meaning that there is no postspinal fossa along the dorsal two-thirds of the spine (Cd4, Fig. 4). The base of the neural spine is almost square in cross section, but this is given the appearance of a more transversely compressed cross section by the presence of the SPRLs. Neural spines project posterodorsally in the anteriormost caudal vertebrae, and are strongly directed posteriorly in middle caudal vertebrae. These caudal neural spine orientations represent the plesiomorphic condition, and thus differ from the derived anterodorsal orientation of the neural spines in some basal titanosauriforms such as Tastavinsaurus, Cedarosaurus, and Venenosaurus (D’Emic, 2012; Royo-Torres, Alcalá & Cobos, 2012). The ratio of the height of the neural spine to centrum height is 1.25 in anterior caudal vertebrae. Towards the dorsal end of the anteriormost neural spines (Cd1 and Cd4), the spine becomes slightly compressed anteroposteriorly and expands strongly transversely, giving the spine a club-shaped outline in anterior view, similar to the anterior caudal neural spines of Brachiosaurus, Camarasaurus, Losillasaurus, Lourinhasaurus, Tastavinsaurus, and Turiasaurus (Casanovas, Santafé & Sanz, 2001; Canudo et al., 2008; Taylor, 2009; Mannion et al., 2013; Mocho, Royo-Torres & Ortega, 2014). As a result of this expansion, the dorsal surface of the spine is strongly convex transversely and forms rounded projections that slightly overhang the lateral surfaces (Cd1, Fig. 4). The anterior, posterior, and dorsal surfaces are strongly rugose, whereas the lateral surfaces of the neural spine are smooth. Shallow © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 634 R. ROYO-TORRES ET AL. Figure 5. Dorsal ribs of Aragosaurus ischiaticus: A–D, anterior dorsal distal rib (IG 492 Uls) in lateral (A), posterior (B), medial (C), and distal (D) views; E–H, posterior dorsal distal rib (IG 481) in medial (E), posterior (F), lateral (G), and distal (H) views. dorsolateral concavities are present on either side of the expanded summits of anterior caudal neural spines (Cd1 and Cd4, Fig. 4), a probable autapomorphy of Aragosaurus. Subsequent caudal neural spines are laterally compressed plates with only slightly expanded dorsal ends (Cd5, Fig. 4), whereas middle–posterior caudal neural spines have cylindrical cross sections and taper to a transversely thin, acute dorsal ridge (Cd10, Fig. 4). Dorsal rib fragments Seven dorsal rib fragments [IG 468 (V40), IG 481 (Cos), IG 487 (Phs1), IG 492 (Uls), V20] have been recovered, representing portions of the middle and distal shafts. All of them have a solid bone texture and only in two distal ribs is it possible to identify a distal rugose surface. One of them is probably an anterior rib, with a plank-like structure (Fig. 5A–C), a feature originally optimized as a titanosauriform character (Wilson, 2002), and the second is a posterior rib (Fig. 5D–F). Haemal arches A total of eight anterior–middle chevrons are preserved (ZH-4, ZH-5, ZH-7, ZH-8, ZH-9, ZH-11, ZH13, ZH-14; Fig. 6; Table 3). All of them are open proximally, although in some cases (i.e. ZH-9) the proximal facets are very close (Sanz et al., 1987). Proximally open chevrons represent a derived state that occurs in most macronarians, but are also found in Shunosaurus and rebbachisaurids (Calvo & Salgado, 1995; Upchurch, 1998). Within Macronaria, several Chinese titanosauriforms (Mannion & Calvo, 2011) and Camarasaurus lewisi (McIntosh et al., 1996b) display proximally bridged chevrons, and it is possible that some (but not all) chevrons of Lusotitan are bridged (Mannion et al., 2013). The ratio of haemal canal depth to total chevron height varies along the tail: this ratio is approximately 50% in the second most anteriorly preserved chevron, but is only 29% in chevron ZH-4. Figure 6. Haemal arches of Aragosaurus ischiaticus: A, B, anterior chevron (ZH-4) in posterior (A) and anterior (B) views; C, D, anterior chevron (ZH-5) in anterior (C) and posterior (D) views; E, anterior chevron (ZH-7) in lateral view; F, G, middle chevron (ZH-9) in anterior (F) and lateral (G) views; H, posterior chevron (ZH-14) in lateral view. Wilson (1999, 2002) and Curry Rogers & Forster (2001) noticed that titanosaurs typically possess chevrons in which the ratio of haemal canal height to total chevron length is 0.5, whereas in other dinosaurs (including basal eusauropods and diplodocoids) this ratio is typically 0.3 or lower. The observation that Aragosaurus displays the derived state in some specimens and the plesiomorphic state in others could indicate that the derived condition first evolved in more anterior chevrons, and then spread to middle chevrons later in titanosauriform evolution. If this interpretation is correct, © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 635 Table 3. Measurements of Aragosaurus ischiaticus haemal arches, pectoral girdle, and forelimb elements Element Dimension Measurement Haemal arch ZH-4 Total length Depth of haemal canal Total length Depth of haemal canal Total length Depth of haemal canal Total length Depth of haemal canal Length Proximal plate dorsoventral height Dorsoventral height of blade at base (as preserved) Anteroposterior length of glenoid Transverse width of glenoid Anteroposterior width (as preserved) Transverse width of scapular articulation (measured at midheight of preserved portion) Maximum transverse width of glenoid Anteroposterior length of ventral concave margin Length Maximum transverse width of proximal end (as preserved) Transverse width of shaft (at broken end) 265 126 ∼272 137 ∼261 78 115 61 1260 826 245 271 159 354 143 Haemal arch ZH-5 Haemal arch ZH-9 Haemal arch ZH-14 Right scapula ZH-1 Left coracoid IG 482 Right humerus IG 490 Anteroposterior width of shaft (at broken end) Right ulna IG 483 Right radius IG 484 Carpal IG 485 Right metacarpal I IG 486 (rmc1) Right metacarpal II IG 486 (rmc2) Length Length of anteromedial process Length of anterolateral process Maximum diameter of shaft at midlength Anteroposterior diameter of distal end Transverse diameter of distal end Length Proximal end transverse width Proximal end anteroposterior width Transverse width of shaft at midlength Transverse width of distal end Anteroposterior width of distal end Transverse width Anteroposterior width Dorsoventral thickness at the thinnest margin Dorsoventral thickness at the thickest margin Length (on medial margin) Length (on lateral margin) Proximal end superopalmar height Proximal end lateromedial height Midshaft transverse width Distal end transverse width Length Proximal end maximum transverse width Midshaft transverse width Distal end transverse width 180 80 1120 (Lapparent, 1960) 330+ 131 (proximal portion), 128 (distal portion) 78 (proximal portion), 108 (distal portion) 820 298 269 95 164 140 792 175 126 87 172 134 171 160 55 87 289 291 125 75 53 96 ∼370 90 70 100 All measurements are in millimetres. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 636 R. ROYO-TORRES ET AL. Aragosaurus could be regarded as possessing an intermediate condition between the plesiomorphic and derived states. The proximal articular surfaces of the haemal arches are flat, lacking the anteroposteriorly convex surfaces of several titanosauriforms (Mannion & Calvo, 2011), including Tastavinsaurus (Canudo et al., 2008). The distal shafts of the two anteriormost preserved chevrons are anteroposteriorly compressed and slightly widened transversely, whereas from the third chevron onwards the distal shaft is transversely compressed. The most distally preserved chevron comes from a middle caudal vertebra, and is strongly curved in lateral view, with the distal shaft projecting backwards and downwards at an angle of approximately 45° to the horizontal. There is no anterior expansion of the shaft, suggesting that Aragosaurus lacked the ‘forked’ chevrons of most non-titanosauriforms (Berman & McIntosh, 1978; Upchurch, 1995, 1998; Wilson & Sereno, 1998). The haemal arches are generally too incomplete for measurements to be taken, although two measurements were obtained from the near-complete ‘chevron 4’ (Table 3). PECTORAL GIRDLE AND FORELIMB Scapula This element is known from the nearly complete right scapula (ZH-1) and the distal end of the blade of the left scapula (IG 482) (Fig. 7; Table 3). It will be described with the distal blade oriented horizontally, although in life it probably projected posterodorsally. The right scapula is well preserved, although it is missing portions from the anterodorsal part of the acromion and the dorsal margin of the distal end of the blade. This element is mounted with the medial surface exposed and the lateral surface accessible but facing a wall: consequently some aspects of the lateral surface are difficult to observe directly. The lateral surface of the acromion bears a well-developed acromial (‘deltoid’) ridge. Immediately posterior to this ridge, the lateral surface is shallowly concave as also occurs in most sauropods, except for basal forms (e.g. Shunosaurus) and some titanosaurs (e.g. Alamosaurus and Opisthocoelicaudia; Upchurch et al., 2004a; Harris, 2006). The angle between the acromial ridge and the blade appears to be close to 90° or slightly less. The stout glenoid region bears an articular surface that faces anteroventrally: thus, Aragosaurus possesses the plesiomorphic state rather than the derived medial deflection of the glenoid observed in somphospondylans and Apatosaurus (Wilson & Sereno, 1998; Upchurch, Tomida & Barrett, 2004b). The distinct bulge or triangular ventromedial process close to the proximal end of the ventral margin of the blade observed in forms such as Chubutisaurus (Carballido et al., 2011a), Figure 7. Pectoral girdle elements of Aragosaurus ischiaticus: A–D, left coracoid (IG 482) in lateral (A), ventral (B), medial (C), and anterior (D) views; E, right scapula (ZH-1) in medial view and partial left coracoid in lateral view; F, anatomical articulation between the coracoid and scapula in lateral view. Alamosaurus (D’Emic, Wilson & Williamson, 2011), and Lourinhasaurus (P.U. and P.D.M., pers. observ., 2009), cannot be observed in Aragosaurus, but this margin might have been damaged. In cross section, the base of the blade has a D-shaped profile, with a thinner dorsal margin and wider ventral margin: this is a derived state that occurs in most neosauropods, except for some derived titanosaurs such as Opisthocoelicaudia, where the cross section becomes subrectangular (Wilson, 2002). No ridges or rugosities are present on the medial surface of the blade, unlike some advanced titanosaurs such as Lirainosaurus (Sanz et al., 1999). It appears that the scapular blade of Aragosaurus expanded in dorsoventral height near its distal end, but the exact degree of expansion cannot be determined because of poor preservation; however, it is unlikely that the dorsal margin was prominently expanded as in rebbachisaurids (see Sereno et al., 2007; Mannion, 2009). Coracoid The left coracoid (IG 482) (Fig. 7; Table 3) will be described as if it were connected to a scapula, the distal blade of which is directed horizontally backwards (see above). Only the ventral half of the left coracoid is preserved. In lateral view, the anterior margin of the coracoid is gently convex dorsoventrally: this suggests that this margin and the dorsal margin might have merged © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS smoothly into each other (as occurs in most nonneosauropod eusauropods; Upchurch, 1998; Upchurch et al., 2004a), but the specimen is not well enough preserved for this character state to be established with confidence. The lateral surface of the coracoid is mildly convex and, as in other sauropods, lacks the tubercle on its anteroventral part that occurs in more basal sauropodomorphs (Upchurch, Barrett & Galton, 2007; Yates, 2007), and in the titanosaurs Lirainosaurus and Opisthocoelicaudia (Borsuk-Bialynicka, 1977; Díez Díaz, Pereda Suberbiola & Sanz, 2013). The anteroventral portion of the coracoid forms a slightly acute projection that is emphasized by the mild upward concavity of the ventral margin that lies just posterior to this corner. This concave, notch-like area is small (when compared with the size of the glenoid) relative to those observed in other sauropods. The very stout glenoid region lies immediately posterior to this ventral notch. The rugose glenoid articular surface faces posteroventrally, and is not deflected to face medially. The glenoid forms the transversely widest part of the coracoid (Table 3), largely because it expands prominently along its lateral edge to form a ‘lateral glenoid shelf ’. The lateral margin of this glenoid ‘shelf ’ curls upwards towards its edge, so that part of the rugose articular surface can be observed in lateral view. This feature is present in other neosauropods, e.g. Lourinhasaurus (Mocho et al., 2014), Camarasaurus grandis (Ostrom & McIntosh, 1966: pl. 46), Brachiosaurus (Riggs, 1903; 1904; Taylor, 2009), and Giraffatitan (Janensch, 1961: fig. Abb.1ab; Taylor, 2009). Posterior to the glenoid, the lower part of the articulation for the scapula can be observed. This is a robust area (although not as wide transversely as the glenoid), and faced either posteriorly or posterodorsally. Only the smooth posteroventral margin of the coracoid foramen is preserved, indicating that this opening lies close to the scapulocoracoid junction. The medial surface of the coracoid is shallowly concave. Humerus The right humerus (IG 490; Fig. 8, Table 3) appears to be nearly complete, but is preserved as separate proximal and distal halves. This bone was illustrated by Lapparent (1960), but was not described. The broken mid-shaft regions of each of these halves are nearly identical in terms of their cross-sectional shape and dimensions (Table 3), suggesting that little material has been lost from this region (Fig. 8). The medial portion of the proximal end is missing. In anterior view, the proximal articular surface and lateral margin of the humerus merge smoothly into each other to form a convex region: thus Aragosaurus retains the plesiomorphic state that occurs in most sauropods and lacks the derived ‘squared’ proximolateral corner observed in many somphospondylans (Upchurch, 1999; 637 Figure 8. Right humerus (IG 490) of Aragosaurus ischiaticus: A, humerus (in three portions) in anterior view; B, picture of the humerus, after Lapparent (1960), in anterior view; C, proximal portion in posterior view; D, proximal portion in lateral view. Wilson, 2002; Upchurch et al., 2004a; Mannion et al., 2013). Also, unlike derived titanosaurs such as Saltasaurus and Opisthocoelicaudia (Upchurch, 1998), the proximolateral corner of the humerus of Aragosaurus lacks a rounded supracoracoideus projection. The proximal part of the deltopectoral crest is preserved, but the more distal (and presumably more prominent) section of this crest is missing. The deltopectoral crest projects mainly anteriorly, but is also deflected slightly laterally: as a result, the lateral surface of the proximal part of the crest is mildly concave. At approximately midlength, where the shaft is broken, the horizontal cross section has a suboval outline, with the long axis of this oval oriented transversely. This cross-sectional profile is formed by relatively straight anterior, rounded lateral and posterior, and acute medial margins. Compared with the overall size of the humerus, the crosssectional area of the mid-shaft is relatively small, with a transverse width of shaft at midlength/humerus length ratio of approximately 0.12 (Table 3). Such gracile humeri (where the mid-shaft width/humerus length ratio is less than 0.15) also occur in some other sauropods, including the basal eusauropod Lapparentosaurus, several brachiosaurids, and the basal somphospondylan Ligabuesaurus (Curry Rogers, 2005; D’Emic, 2012; Mannion et al., 2013). The distribution of gracile humeri suggests that this feature evolved several times among different sauropod lineages, and it therefore does not provide strong support for any particular identification of the affinities of Aragosaurus. The profile of the lateral margin of the humeral shaft is concave rather © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 638 R. ROYO-TORRES ET AL. than the derived straight margin seen in some titanosaurs (Curry Rogers, 2005). The humerus lacks a pronounced tuberosity for the muscle latissimus dorsi, unlike some derived titanosaurs (Otero, 2010; D’Emic, 2012). The distal half of the humerus is damaged along its lateral margin. As in other sauropods, the distal part of the anterior surface is mildly convex transversely and lacks the deep concavity observed in basal sauropodomorphs and other dinosaurs (Upchurch et al., 2007; Yates, 2007). There are two small projections situated close to the midline, where the anterior surface meets the distal articular surface; as such, Aragosaurus retains the plesiomorphic state of a divided anterior condyle, differing from the condition in derived somphospondylans (D’Emic, 2012). The anteromedial part of the distal end is moderately expanded to form a rounded projection: as a result, the distal part of the medial surface faces posteromedially and is mildly convex. Posteriorly, there is a well-developed anconeal fossa that is defined medially and laterally by low rounded ridges that extend from the distal end and fade out proximally, but these are not the acute prominent ridges observed in many titanosaurs (Upchurch et al., 2004a). The rugose distal articular surface does not curve up onto the anterior or posterior faces of the humeral shaft, unlike the derived condition that occurs in several titanosaurs (Wilson & Carrano, 1999; Wilson, 2002). The ratio between humerus and femur lengths is approximately 0.82 (Sanz et al., 1987; Royo-Torres & Ruiz-Omeñaca, 1999), which is similar to most titanosauriforms (Mannion et al., 2013), as well as Lourinhasaurus (Mocho et al., 2014), but differs from the condition in brachiosaurids, whereby this ratio is closer to 1.0 (Wilson, 2002). In Aragosaurus, this ratio is slightly higher than in other basal macronarians such as Camarasaurus (0.75; Madsen, McIntosh & Berdman, 1995; Ikejiri, 2005) and Tehuelchesaurus (0.72; Carballido et al., 2011b). Ulna The right ulna (IG 483; Fig. 9; Table 3) is nearly complete and resembles those of other sauropods in most respects. The proximal end is triradiate, formed by long anteromedial and anterolateral processes and a weakly developed posterior process. There is a deep fossa between the anteromedial and anterolateral processes, for reception of the proximal end of the radius. The anteromedial and anterolateral processes are subequal in length (Table 3): this state also occurs in several other sauropods (e.g. Apatosaurus, Europasaurus, Malawisaurus, and Mamenchisaurus), whereas the former process is much longer [e.g. Camarasaurus, Giraffatitan, and Sauroposeidon (‘Paluxysaurus’ material)] or even twice as long (Cedarosaurus and Venenosaurus) as the latter process in several macronarians (Mannion et al., 2013). The articular surface of the anteromedial process is flat longitudinally and slopes strongly anterodistally at an angle of approximately 45° to the horizontal. A similar, potentially derived, condition occurs in the basal macronarian sauropod Lusotitan (Mannion et al., 2013). In contrast, the articular surface of the anterolateral process is flat longitudinally and moderately convex transversely. On the anterior surface of the shaft, there is a strong interosseous ridge that first appears on the proximal half of the ulna, increases in prominence distally, and finally disappears about 100 mm above the distal end. Such a ridge is common in other sauropods and probably represents the attachment site of ligaments that bound the ulna and radius together; however, this ridge is normally restricted to the distal part of the ulna, except in several titanosaurs where this structure can occupy most of the length of the element (Curry Rogers, 2005, 2009; see ‘Radius’ below for further discussion of this feature). Below the end of this ridge there is a triangular striated concavity. Although the distal end of the ulna is expanded relative to the shaft, this expansion does not seem to be particularly marked relative to that observed in most sauropods. The outline of the distal end is between elliptical and subcircular. Radius The right radius is complete (IG 484; Fig. 9; Table 3). The proximal articular surface is flat and has a suboval outline, with the long axis of this oval oriented mediolaterally. This element is relatively slender (proximal transverse width/radius length ratio = 0.22; Table 3), as in most sauropods: thus Aragosaurus lacks the apomorphically robust radius observed in many titanosaurs, where the proximal transverse width is often more than 30% of the radius length (McIntosh, 1990; Upchurch, 1995, 1998). In lateral view, the radial shaft is bowed slightly anteriorly. On the lateral part Figure 9. Right ulna (IG 483) of Aragosaurus ischiaticus in: A, posterior, B, ventral, C, anteromedial, and D, medial views. Right radius (IG 484) of Aragosaurus ischiaticus in: E, anterior, and F, posterior views. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS of the posterior surface, a well-developed ridge extends from the proximal end to approximately one-third of the radius length from the distal end, increasing in prominence distally, and corresponding with the ridge described on the anterior face of the ulna (see above). This ridge on the posterior face of the radius terminates in a rounded eminence or tubercle at a point approximately 25% of the shaft length from the distal end of the radius. Curry Rogers (2005: character 283) noted the presence of a well-developed interosseous ridge as a derived state in many titanosaurs (e.g. Aeolosaurus, Ampelosaurus, Neuquensaurus, Opisthocoelicaudia, and Rapetosaurus). In at least some of these taxa (e.g. Opisthocoelicaudia, Borsuk-Bialynicka, 1977: fig. 8C; Rapetosaurus, Curry Rogers, 2009: fig. 36D), the interosseous ridges extend along most of the length of both the ulna and radius, as occurs in Aragosaurus. As Aragosaurus is probably a basal macronarian (see below) that is not closely related to such derived titanosaurs, the elongated interosseous ridge is here provisionally considered to be a local autapomorphy. The tubercle at the distal end of this ridge is also potentially unique to Aragosaurus. The distal end has a subrectangular outline, similar to most sauropods (Wilson & Sereno, 1998), although the posterior margin has a slightly concave profile. The distal articular surface is rugose and strongly convex and, in anterior view, slopes proximally towards its lateral margin at an angle of 16–17° to the horizontal. This ‘bevelling’ of the lateral half of the distal articular surface occurs in a wide array of eusauropods (Mannion et al., 2013), although it is more prominently developed and continuous across the distal surface in derived titanosaurs such as Opisthocoelicaudia and Saltasaurus (Wilson, 2002; Mannion et al., 2013; Mateus, Mannion & Upchurch, in press). Carpal A single block-like carpal element is preserved (IG 485; Fig. 10; Table 3). It is similar to the carpal bone (cat. no. 445) of Camarasaurus grandis (GMNH-PV 101) that was described as a ‘radiale’ by McIntosh et al. (1996a). Given that the other forelimb and manual elements of Aragosaurus are all preserved from the right side, for the purposes of description it will be assumed that the carpal also belongs to the right manus. This element is subquadrangular in proximal and distal views, with rounded corners. The proximal articular surface is mildly concave, as is also seen in the carpals of Camarasaurus (McIntosh et al., 1996a), Turiasaurus, and Losillasaurus (R.R.T., P.U., and P.D.M., pers. observ. 2009). In contrast, the distal surface is mildly rugose and possesses only a mild projection on its posterolateral part, unlike Turiasaurus, where this projection is much more prominent (Royo-Torres et al., 2006). The Aragosaurus carpal also lacks the less prominent projection that 639 Figure 10. Right carpal (IG 485) of Aragosaurus ischiaticus in: A, anterior, B, posterior, C, medial, D, lateral, E, ventral, and F, dorsal views. Comparison between different carpals: G, Aragosaurus; H, Turiasaurus; and I, Losillasaurus. occurs on the posteromedial part of the distal surface in Turiasaurus. Neither the proximal nor distal articular surfaces are divided into two or more separate regions by an anteroposteriorly oriented ridge, differing from the proximal surface of the ‘scapholunar’ carpal described in Apatosaurus by Hatcher (1902: 367). In Aragosaurus, the anterior, posterior, lateral, and medial margins are generally proximodistally rounded, except at the posterolateral corner, where the carpal has its greatest proximodistal thickness. Here, the posterior part of the lateral surface is mildly concave, partly as a result of a slight outwards projection of the distal rim. This element maintains approximately the same proximodistal thickness over virtually all of its extent: thus the bone does not thin towards one or more of its edges, unlike the ‘scapho-lunar’ of Apatosaurus, where the anterior margin is much thinner than the posterior margin (Hatcher, 1902). The subquadrangular proximal outline of the carpal of Aragosaurus is potentially autapomorphic because the carpals of other sauropods are typically circular © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 640 R. ROYO-TORRES ET AL. or elliptical (e.g. Turiasaurus and Losillasaurus, Royo-Torres et al., 2006; Apatosaurus, Hatcher, 1902; Ostrom & McIntosh, 1966: pl. 54; and Camarasaurus, McIntosh et al., 1996a). Metacarpus Portions of four metacarpals (I, II, III, and IV) from the right manus are preserved [IG 486 (rmc1), IG 486 (rmc2), IG 486 (rmc3), and IG 486 (rmc4); Figs 11 and 12; Table 3]. In the following description, the orientations of these elements are based on the Camarasaurus manus (Ostrom & McIntosh, 1966; McIntosh et al., 1996a). In life, the metacarpals would have formed a U-shaped arrangement with the long axes of their shafts extending vertically, as in other eusauropods (McIntosh, 1990; Upchurch, 1998; Upchurch et al., 2004a); however, for the purposes of description, the metacarpals are treated as though they lie side-by-side with their long axes extending forwards and the long axis of the distal end of each metacarpal oriented transversely. The articular surface of the right metacarpal I [IG 486 (rmc1); Fig. 11; Table 3], in proximal view, has a trans- Figure 11. Right metacarpal I (IG 486, rmc1) of Aragosaurus ischiaticus in: A, lateral; B, dorsal; C, ventral; D, distal; E, medial; and F, proximal views. Figure 12. Right metacarpal II (IG 486, rmc2) of Aragosaurus ischiaticus in: A, proximal; B, medial; C, dorsal; D, ventral; E, distal; and F, lateral views. versely compressed D-shaped outline, with rounded dorsal and medial margins, and flat and concave ventral and lateral margins (Fig. 11). The proximal outline is similar to metacarpal I of Giraffatitan (Janensch, 1961), and differs from the more circular outline of Camarasaurus (Ostrom & McIntosh, 1966; McIntosh et al., 1996a). The proximal half of the lateral surface forms a subtriangular and longitudinally striated concavity that extends along the shaft of the metacarpal to approximately midlength. Just distal to midlength, the shaft has an elliptical cross section, with the long axis of this ellipse oriented transversely. There is a ventrolaterally facing longitudinal groove on this central region of the shaft, which is bounded by a distal ridge along the lateral margin. The dorsolateral margin of the shaft also forms a relatively acute ridge. In Aragosaurus the distal end of metacarpal I terminates in two prominent condyles that are separated along their midline by a wide groove. The medial condyle is larger than the lateral one, mainly because it projects further ventrally. In distal end view, these condyles appear to slant ventrally and slightly medially. There is a shallow pit on the lateral surface of the distal condyle below the distal ridge on the lateral margin. The medial surface of the distal condyle is flat and striated. In the right metacarpal II [IG 486 (rmc2); Fig. 12; Table 3], the proximal end is between subtriangular and D-shaped in outline, with a concave area on the proximal part of the lateral surface for articulation with metacarpal III, and a convex margin (with a slight proximodistal ridge) forming the medial area for articulation with metacarpal I. In general, the dorsal and medial surfaces are convex transversely and smooth, whereas the palmar area is flat and rugose. The shaft is triangular in transverse cross section and the distal end has two condyles separated by a groove. These condyles are most strongly developed on the palmar surface. The medial distal condyle is more prominent than the lateral one, but the lateral distal condyle projects further distally than the medial one. The distal articular surface does not appear to extend onto the dorsal surface of the metacarpal, a feature that Aragosaurus shares with most titanosauriforms (D’Emic, 2012). This metacarpal is at least 37% of the radius length (Table 3), and almost certainly much greater because of missing material: this suggests that Aragosaurus possessed the derived state (longest metacarpal/radius length ratio of 0.45 or greater) that occurs in most macronarians (Wilson & Sereno, 1998). Metacarpal II of Aragosaurus is different from those in Turiasaurus (Royo-Torres et al., 2006), and is similar to the metacarpal of Camarasaurus grandis (McIntosh et al., 1996a), because Aragosaurus and C. grandis (McIntosh et al., 1996a) have two distinct distal condyles in palmar view where the articular surfaces of the distal ends curve down onto the © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS palmar surface, whereas in Turiasaurus this surface is flat and distinct condyles are absent. Fragmentary remains of right metacarpal III [IG 486 (rmc3); Table 3] and right metacarpal IV [IG 486 (rmc4); Table 3] are preserved. The distal end of metacarpal III has two distinct condyles similar to MC II. The proximal end of metacarpal IV is triangular in outline, lacking the ‘chevron’ profile seen in many titanosauriforms (D’Emic, 2012; Mannion et al., 2013). PELVIC GIRDLE AND HINDLIMBS Right pubis Most of the right pubis is preserved [IG 489; Fig. 13; Table 4); however, much of the symphyseal margin (below the ischial articulation) is missing and has been restored, and the obturator foramen is also not preserved in this portion. Royo-Torres et al. (1999) initially identified a separate portion of bone (IG 488) as part of the pubis, but here it is reinterpreted as the pubic peduncle of the left ischium. The articular surface for the ilium is flat and strongly rugose. This articulation has a D-shaped outline in dorsal view, with the convex margin facing anteriorly and laterally. This articular area therefore lacks the extreme transverse compression seen in some titanosaurs (e.g. Andesaurus; Mannion & Calvo, 2011). Immediately below the iliac articulation, the anterior surface of the pubis forms a broad, slightly flattened triangular area that tapers to a point ventrally (Royo-Torres et al., 1999). This ventral tip extends forwards as a low, rounded projection. Thus, the ambiens process of Aragosaurus resembles those of other sauropods in most respects, especially those of Lourinhasaurus (Royo-Torres, 2009), Giraffatitan (Janensch, 1961), and possibly Haplocanthosaurus (Rauhut et al., 2005), but there is no prominent hooked ambiens process like the pubes of several flagellicaudatans (e.g. Diplodocus, Barosaurus, and Dicraeosaurus; McIntosh, 1990; Upchurch et al., Figure 13. Right pubis (IG 489) of Aragosaurus ischiaticus in: A, medial; B, anterior; C, lateral; D, dorsal; E, ventral; and F, posterior views. 641 2004a, b). The Aragosaurus pubis resembles those of derived eusauropods, with a broad distal shaft that probably met its partner on the midline along an embayed symphysis (i.e. the anterolateral margins of each pubis lie in front of the plane of the symphysis, and so form a V-shape in horizontal cross section; see Upchurch, 1995; Harris, 2006). The distal end of the pubis is expanded, especially medially. There is a flattened area on the posteromedial surface of the distal end: this probably met a similar area on the left pubis to form a strong ventral end to the midline symphysis. As in most other sauropods, the distal end surface is strongly convex and rugose. The distal end of the pubis is expanded anteriorly, with the anterior margin of the very distal end forming an acute angle with the remainder of the anterior margin, similar to that seen in Camarasaurus (Ostrom & McIntosh, 1966), but differing from the rounded ‘boot’-like process that occurs in some titanosauriforms, e.g. Giraffatitan and Tastavinsaurus (Canudo et al., 2008; Mannion et al., 2013). Ischium The ischium is represented by the nearly complete right element (ZH-3), the separate iliac peduncle and distal end of the shaft (IG 492), and a fragment of the pubic peduncle (IG 488) of the left element (Fig. 14; Table 4). There is some reconstruction in the ventral region of ZH-3, where the blade meets the proximal expansion, and at the ventral tip of the pubic articulation. The length of the ischium is a little shorter than the pubis (the ischium/pubis length ratio is 0.9). This value is thus close to 1.0, as occurs in basal macronarians such as Camarasaurus grandis (ratio = 1.07; McIntosh et al., 1996a), Camarasaurus lentus (ratio = 0.98; Gilmore, 1925; Royo-Torres, 2009), Lourinhasaurus (ratio = 0.9–1.0; Mocho et al., 2014), and basal titanosauriforms, such as Giraffatitan (ratio = 1.0; Janensch, 1961). In more derived Titanosauriformes, the typical condition is to have an ischium that is somewhat shorter than the pubis, with a ratio of less than 0.9 (e.g. Tastavinsaurus, Isisaurus, Opisthocoelicaudia, and Saltasaurus; Calvo & Salgado, 1995; Salgado et al., 1997; Upchurch, 1998; Upchurch et al., 2004a; Canudo et al., 2008; Royo-Torres, 2009; Mannion et al., 2013). The proximal expansion of the ischium resembles those of most other sauropods in having mildly dorsoventrally convex and concave lateral and medial surfaces, respectively, posterior to the pubic articulation. The latter is transversely widest at its dorsal (acetabular) end, and tapers ventrally. The iliac peduncle is widest anteroposteriorly at its base and narrows slightly towards its articular surface: thus, Aragosaurus lacks the distinct ‘neck’ observed in the ischia of several rebbachisaurids (Sereno et al., 2007) and some derived titanosaurs (e.g. Muyelensaurus; Mannion & Calvo, 2011). There is no tubercle on the lateral © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 642 R. ROYO-TORRES ET AL. Table 4. Measurements of Aragosaurus ischiaticus pelvic girdle elements and hindlimb Element Dimension Measurement Right pubis IG 489 Length Transverse width of iliac articulation Anteroposterior width of iliac articulation Distance from proximal end to most prominent point of ambiens area Anteroposterior width of distal end Transverse width of distal end Length Length of shaft along ventral margin Estimated dorsoventral height of proximal plate Length of the pubic articulation Anteroposterior diameter of iliac articulation Transverse width of iliac articulation Transverse width of shaft at midlength Length of acetabular margin Distal end transverse width Distal end dorsoventral width Length Transverse width of proximal end Lateromedial width of proximal end Length from proximal end to top of fourth trochanter Length of fourth trochanter Anteroposterior width of proximal articular head Anteroposterior width of greater trochanter Anteroposterior width of tibial condyle Anteroposterior width of the fibulal condyle Transverse width of distal end Anteroposterior width of midshaft Transverse width of midshaft Proximal transverse width in dorsal Proximal superopalmar width Distal transverse width Distal superopalmar width Medial proximodistal width Lateral proximodistal width Length Proximal end dorsoventral height Proximal end transverse width 868–880 220 240 132 350 190 875 552 543 355 118 132 91 166 119 50 1373 426 230 605 132 223 187 370 310 436 136 266 83 74 64 63 7.5 5.5 166 95 57 Right ischium ZH-3 Left femur ZH-2 Pedal phalanx II-1 ZH-10 Right pedal ungual ZH-19 All measurements are in millimetres. surface of the iliac peduncle, unlike the ischia of the titanosaurs Gondwanatitan and Opisthocoelicaudia (Borsuk-Bialynicka, 1977; Kellner & de Azevedo, 1999; Upchurch et al., 2004a). The acetabular surface curves smoothly into the lateral surface of the proximal expansion, whereas the medial margin of the acetabulum meets the medial surface of the ischial plate at an abrupt ridge-like edge (Sanz et al., 1987), contrasting with taxa such as Giraffatitan and Tastavinsaurus, which have a shallower acetabulum (Sanz et al., 1987; Canudo et al., 2008; Royo-Torres, 2009). The acetabular surface does not display the morphology observed in several rebbachisaurids (e.g. Demandasaurus; Mannion et al., 2012), in which it is transversely wide at the dorsal end of the iliac peduncle and pubic peduncle, but very narrow in its central portion. The ischium has a dorsoventrally long pubic peduncle in relation to the element’s total length (Sanz et al., 1987). This character can be quantified by the next ratio: the distance from the upper corner of the pubic blade up to the posterior margin of the element, to the dorsoventral length of the pubic articular surface (Salgado et al., 1997: fig. 5). The derived state, characterizing Camarasauromorpha (sensu Salgado et al., 1997), Macronaria (Wilson & Sereno, 1998), and Apatosaurus (Royo-Torres, 2009), occurs when the © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 643 Figure 14. Right ischium (ZH-3) of Aragosaurus ischiaticus in: A, posterior; B, lateral views. Fragments of the left ischium (IG 492) of Aragosaurus ischiaticus: C, D, pubic peduncle in anterior (C) and lateral (D) views; E, F, iliac peduncle in dorsal (E) and lateral (F) views. distance from the upper corner of the pubic blade of the ischium to the posterior border of the bone is shorter than the pubic articulation. In Aragosaurus the distal blade is well developed, and the morphology of the midline symphysis of the distal blade suggests that there would have been a V-shaped notch between the left and right ischia in dorsal view when these elements were articulated in life. Consequently, Aragosaurus is likely to have possessed the plesiomorphic style of ischiadic symphysis observed in most sauropods, rather than the derived condition seen in titanosaurs such as Alamosaurus, Andesaurus, and Opisthocoelicaudia, in which the ventral emargination is absent and the midline symphysis extends forwards to the ventral end of the pubic articulation (McIntosh, 1990; Upchurch, 1998; Wilson, 2002). The distal shaft is directed mainly posteriorly and only slightly ventrally: when the long axis of the distal shaft is projected forwards, it passes through the centre of the pubic articulation, contrasting with the derived condition that occurs in Giraffatitan (Janensch, 1961) and Lapparentosaurus (Upchurch, 1995), where the shaft is directed strongly downwards. The distal ends of the ischia are subtriangular in outline and form a V-shaped arrangement in distal end view. Thus, Aragosaurus retains the plesiomorphic state that occurs in basal sauropods (e.g. Vulcanodon, Cooper, 1984; ‘Bothriospondylus madagascariensis’, Mannion, 2010) and flagellicaudatans, rather than the derived copla- Figure 15. Left femur (ZH-2) of Aragosaurus ischiaticus in: A, posterior; B, medial; and C, anterior views. Features observed in the femur: D, lateral bulge; E, head deflected medially; and F, bevelling of the distal end of the femur relative to the long axis of the shaft. nar arrangement that occurs in Haplocanthosaurus, rebbachisaurids, and most macronarians (Upchurch, 1998: fig. 16; Wilson & Sereno, 1998). Femur The left femur is complete (ZH-2; Fig. 15; Table 4). This element displays most of the features that are typical in sauropod femora, including a dorsomedially directed proximal articular head, absence of the lesser trochanter, a shaft that is straight in both lateral and anterior views, and well-developed distal condyles (McIntosh, 1990; Upchurch, 1995, 1998; Wilson, 2002; Upchurch et al., 2004a, 2007). There is no ‘neck’ or constriction between the proximal articular head and the greater trochanter. The articular surface of the greater trochanter, at its lateral end, curves convexly onto a posteriorly projecting, low, rounded bulge, forming a trochanteric shelf. A similar morphology has been reported in several rebbachisaurids (Sereno et al., 2007) and titanosaurs (Otero, 2010; Mannion et al., 2013). © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 644 R. ROYO-TORRES ET AL. In anterior view, the lateral margin of the proximal end is deflected medially relative to that of the shaft, with the long axis of the diaphysis deflected 6° from vertical (Fig. 15), and just below the level of the greater trochanter there is a lateral ‘bulge’. This combined morphology characterizes most macronarians, although the medial deflection is lost in some derived titanosaurs (Royo-Torres, 2009; Royo-Torres et al., 2012; Mannion et al., 2013). This 6° angle also occurs in Tastavinsaurus and Lourinhasaurus (Royo-Torres, 2009; Royo-Torres et al., 2012), so these Iberian taxa might have an intermediate state, with a more derived state represented by an angle of at least 10° in titanosaurs (Wilson & Carrano, 1999; Wilson, 2002). As in most other eusauropods, the fourth trochanter of Aragosaurus is a low rounded ridge that is situated on the posteromedial margin of the shaft at approximately midlength, and is not visible in anterior view. Just anterior to the fourth trochanter, there is a shallow concavity on the medial surface of the shaft. At midlength, the femoral shaft is strongly compressed anteroposteriorly (transverse width/ anteroposterior width ratio = 1.96). Wilson & Carrano (1999) noted that the femoral shafts of titanosaurs tend to be more compressed anteroposteriorly than those of other sauropods, and Wilson (2002) defined the derived condition as a transverse width/anteroposterior width ratio of greater than 1.85, although several more basal taxa (e.g. Brachiosaurus, Giraffatitan, and Ligabuesaurus) also possess the derived state (D’Emic, 2012; Mannion et al., 2013). The distal condyle for articulation with the fibula bears a well-developed vertical groove on its posterolateral margin, as occurs in other sauropodomorphs and most other dinosaurs. The intercondylar groove is well developed, especially on the posterior surface of the femur. The distal articular surface curves up onto both the posterior and anterior faces of the condyles: the latter feature is a derived condition that occurs in many titanosaurs (Wilson & Carrano, 1999) and Diplodocus (Wilson, 2002). Pedal phalanges The pedal phalanx ZH-10 most closely resembles the left pedal phalanges II-1 and III-1 of other sauropods, although III-1 is usually slightly longer than wide in dorsal view (Upchurch, 1993), and so we consider ZH-10 to be phalanx II-1 (Fig. 16; Table 4). It is subrectangular in dorsal view, with the medial distal condyle projecting slightly further distally than the lateral distal condyle (Sanz et al., 1987). This element is unusual in that it possesses a subtriangular transverse cross section at its midlength, with a dorsally directed apex, convex lateral and medial surfaces, and a mildly concave ventral (plantar) surface. The presence of the midline dorsal apex is provisionally regarded as an autapomorphy of Aragosaurus. Figure 16. Right pedal elements of Aragosaurus ischiaticus: A, ungual phalanx II (ZH-19) in dorsal view; B, ungual phalanx II (ZH-19) and proximal phalanx (ZH-10) in medial views; C, ungual phalanx II in lateral view. Pedal phalanx ZH-6 is only partially preserved and yields little anatomical data. This element was described by Sanz et al. (1987) as a bone with numerous faces for articulation. ZH-19 is a typical, recurved eusauropod ungual claw that probably belongs to digit II of the right pes (Fig. 16; Table 4). It is laterally compressed throughout its length, as occurs in other eusauropods (Wilson & Sereno, 1998; Upchurch et al., 2007; Yates, 2007). The proximal articular surface lies in a plane that is slightly oblique to the long axis of the ungual, so that in life the claw would have been directed forwards and slightly laterally. In dorsal view, the ungual also curves slightly laterally towards its distal tip. The medial surface of the ungual is more convex dorsoventrally than is the lateral surface. The ventromedial margin, in its distal part, is slightly bulbous but does not form the tuberosities seen in many titanosauriforms (Mannion et al., 2013), including Gobititan (IVPP 12579; P.U. and P.D.M., pers. observ. 2007) and Tastavinsaurus (MPZ 99/9; P.U. and P.D.M., pers. observ. 2009). The ungual has a smooth lateral groove that is less deep than in Tastavinsaurus. This feature has also been observed in a possible second ungual phalanx from Galve (MPG CA-1). It is a part of historic material from the Villar del Arzobispo Formation according to © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS Sánchez-Hernández (2005), and this phalanx appears in photographs of the elements of Aragosaurus taken by J.L.S. in the 1980s. This second ungual element could therefore be from Las Zabacheras, but it was never mentioned in the lists of bones from this locality (Lapparent, 1960; Sanz et al., 1987). PHYLOGENETIC AFFINITIES OF ARAGOSAURUS Although Mannion et al. (2013) incorporated Aragosaurus into a phylogenetic analysis focused on basal macronarians, recovering it as a basal member of this clade, outside Titanosauriformes, this study excluded a number of other contemporaneous Iberian taxa (i.e. Losillasaurus, Lourinhasaurus, and Turiasaurus) that might have some bearing on its relationships. In order to further understand the phylogenetic position and relationships of Aragosaurus with regard to other Iberian taxa, we have used the updated matrices of Wilson (2002) and Upchurch et al. (2004a), following Wilson & Upchurch (2009), Royo-Torres et al. (2012), and Royo-Torres & Upchurch (2012). In the case of the Upchurch et al. (2004a) data matrix, Lapparentosaurus, Nigersaurus, ‘Pleurocoelus-tex’ (a chimera comprising material now referred to Astrophocaudia, Cedarosaurus, and Sauroposeidon; D’Emic & Foreman, 2012; D’Emic, 2013), Andesaurus, and Argentinosaurus were omitted a priori. For each data matrix, we have: (1) split the Brachiosaurus operational taxonomic unit (OTU) into Brachiosaurus altithorax and Giraffatitan brancai, following Taylor (2009); and (2) added in or revised scores for Cedarosaurus (based on Tidwell, Carpenter & Brooks, 1999 and D’Emic, 2012; Wilson matrix only), Europasaurus (based on Sander et al., 2006), Lourinhasaurus (based on Mocho et al., 2014), Tehuelchesaurus (based on Carballido et al., 2011b), and Venenosaurus (based on Tidwell et al., 2001; D’Emic, 2012). These new codings are provided in the Appendix. Data matrices were analysed using TNT 1.1 (Goloboff, Farris & Nixon, 2003) in order to find the most parsimonious trees (MPTs). We implemented a heuristic tree search, performing 1000 replications of Wagner trees (using random addition sequences) followed by tree bisection reconnection (TBR) as the swapping algorithm, saving 100 trees per replicate. In order to test the support of the phylogenies, the Bremer support and the bootstrap values were obtained (absolute frequencies based on 5000 replicates; Figs 17 and 18). RESULTS WILSON 2002 DATA SET As in the original analysis, the multistate characters 8, 37, 64, 66, and 198 were ordered. This analy- 645 sis includes 41 taxa and 234 morphological characters. Analysis with TNT produces 66 MPTs with a length of 534 steps (consistency index, CI = 0.538; homplasy index, HI = 0.463; retention index, RI = 0.735). The strict consensus tree places Aragosaurus in a polytomy with Camarasaurus, Lourinhasaurus, Tehuelchesaurus, Europasaurus, and the Titanosauriformes (Fig. 17). The general topology obtained is similar to the proposal of Royo-Torres et al. (2012) and Royo-Torres & Upchurch (2012), and generally resembles the topology found by Wilson (2002). Aragosaurus is retained inside Macronaria but outside Titanosauriformes. UPCHURCH ET AL. 2004A DATA SET This analysis includes 49 taxa and 309 morphological characters. The result of the parsimony analysis yielded 18 MPTs with 703 steps (CI = 0.455; HI = 0.545; RI = 0.765). The strict consensus closely matches the main topology obtained in recent publications (e.g. Wilson & Upchurch, 2009; Royo-Torres & Upchurch, 2012; Royo-Torres et al., 2012). Aragosaurus is placed as a macronarian lying outside Camarasauromorpha, as the sister taxon to Europasaurus + Titanosauriformes (Fig. 18). DISCUSSION THE PHYLOGENETIC POSITION OF ARAGOSAURUS Aragosaurus possesses a suite of characters supporting its inclusion in Macronaria: (1) anterior caudal chevrons with open proximal articulations (Upchurch, 1995; Wilson & Sereno, 1998); (2) ischial distal shafts nearly coplanar (Wilson & Sereno, 1998) (although note that features 1 and 2 are both present in basal diplodocoids such as Haplocanthosaurus and rebbachisaurids, and thus it is equally parsimonious to conclude that these two features are neosauropod synapomorphies that were reversed in flagellicaudatans); (3) length of longest metacarpal at least 45% of radius length (Salgado et al., 1997; Wilson & Sereno, 1998; Wilson, 2002); and (4) puboischial contact dorsoventrally deep (Salgado et al., 1997; Wilson & Sereno, 1998). Aragosaurus also possesses an ischium with a dorsoventrally long pubic peduncle, similar to Camarasaurus and basal Titanosauriformes (Salgado et al., 1997; Wilson & Sereno, 1998; Wilson, 2002). Aragosaurus also displays several characters that have traditionally been optimized as supporting a more derived placement, within Titanosauriformes. For example, the proximal onethird of the femur of Aragosaurus is deflected medially (Wilson & Sereno, 1998), and there is a lateral bulge (Salgado et al., 1997). According to the phylogenetic results presented here, and in D’Emic (2012) and Mannion et al. (2013), these features have a wider distribution within Macronaria. Other characters © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 646 R. ROYO-TORRES ET AL. Figure 17. The strict consensus tree produced by phylogenetic analysis of the updated Wilson (2002) data set. Bootstrap values are shown as percentages (nodes lacking percentages have bootstrap values of less than 50%). Bremer supports are shown in square brackets. previously suggested as synapomorphic of Titanosauriformes, but developed in Aragosaurus, include a ridge on middle caudal centra, anterior (although not the middle) caudal vertebrae with neural arches set on the anterior half of the centrum (Salgado et al., 1997), and an ischium with a raised tubercle on the dorsolateral face (Wilson, 2002; D’Emic, 2012). In contrast, Aragosaurus lacks a number of features proposed to characterize Titanosauriformes, including: (1) the presence of small, vascular foramina piercing the lateral or ventral surfaces of anterior–middle caudal centra (Mannion et al., 2013); (2) a scapula with a tubercle on the ventral margin of the base of the scapular blade; (3) a humerus/femur length ratio greater than 85%; (4) a long anteromedial process of the ulna; and (5) a distally expanded ulna (Wilson, 2002; D’Emic, 2012). Royo-Torres (2009) proposed Laurasiformes as a clade comprising three taxa, Cedarosaurus, Tastavinsaurus, and Venenosaurus, and was supported in Royo-Torres et al. (2012), using the Wilson (2002) and Upchurch et al. (2004a) data matrices. This clade was not supported by recent phylogenetic analyses focusing on titanosauriform interrelationships, however (D’Emic, 2012; Mannion et al., 2013). According to the current study, Aragosaurus shares some features previously considered synapomorphies for Laurasiformes: (1) a well-marked ridge on the lateral face of the neural arch of middle–posterior caudal vertebrae; (2) the © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 647 Figure 18. The strict consensus tree produced by phylogenetic analysis of the updated Upchurch et al. (2004a) data set. Bootstrap values are shown as percentages (nodes lacking percentages have bootstrap values of less than 50%). Bremer supports are shown in square brackets. morphology of the articular surfaces of the centra of anterior caudal vertebrae (except the first), in which the anterior face is concave and the posterior face is nearly flat or slightly convex. The latter state is difficult to distinguish from the amphiplatyan condition of Eusauropoda (Wilson, 2002; Upchurch et al., 2004a; D’Emic, 2012), and the basal titanosaur Andesaurus also shares this morphology (Mannion & Calvo, 2011); as such, it is not unique to any potential clustering of laurasiaform taxa. Another putative Laurasiformes feature, the femoral diaphysis slanted at an angle of 6° to the vertical, is also present in most macronarians more derived than Camarasaurus (D’Emic, 2012; Royo-Torres et al., 2012; Mannion et al., 2013). Two features proposed as Laurasiformes synapomorphies – the presence of mild opisthocoely in the anteriormost caudal centra, and anteriorly deflected anterior– middle caudal neural spines (Royo-Torres, 2009; Royo-Torres et al., 2012) – are absent in Aragosaurus. The latter character state is present in the three genera originally proposed as members of Laurasiformes (i.e. Venenosaurus, Cedarosaurus, and Tastavinsaurus). D’Emic (2012) did not support the monophyly of Laurasiformes, but some of the character state scores for Tastavinsaurus require revision. For example, Tastavinsaurus shares with Cedarosaurus and Venenosaurus the following derived character states in the analysis of D’Emic (2012): (1) anterior caudal vertebrae with sporadically distributed, shallow fossae on the lateral faces of the centrum; (2) middle caudal © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 648 R. ROYO-TORRES ET AL. vertebrae with neural spines that lean anteriorly; and (3) metatarsal IV with distal end bevelled upwards medially. Finally, Aragosaurus is unlikely to be a somphospondylan sauropod because it lacks the medial deflection of the scapular glenoid (Wilson & Sereno, 1998), the humerus has a rounded rather than ‘squared’ proximolateral corner (Wilson, 2002), and the posterior articular surfaces of middle–posterior caudal centra are flat rather than convex (Mannion et al., 2013). In summary, Aragosaurus is a basal macronarian with some, but not all, of the character states that diagnose the Titanosauriformes. Therefore, as proposed by the cladistic analyses of Royo-Torres (2009), Mannion et al. (2013), and this article, Aragosaurus ischiaticus is a basal macronarian that is more closely related to Titanosauriformes than is Camarasaurus. TAXONOMIC STATUS OF ARAGOSAURUS 6. 7. 8. AND COMPARISON WITH CONTEMPORANEOUS IBERIAN SAUROPODS The autapomorphies originally listed by Sanz et al. (1987) to diagnose Aragosaurus have since been demonstrated to have a wider distribution amongst Sauropoda, and are summarized here. 1. Great widening of the neural spines of the anteriormost caudal vertebrae. This character state is shared with Losillasaurus, Camarasaurus, Tastavinsaurus, Venenosaurus, and a caudal series from Tendaguru (MBR 2091.1-30) originally referred to Janenschia (although see Bonaparte, Heinrich & Wild, 2000), but more recently suggested to have turiasaur (Royo-Torres & Cobos, 2009) or mamenchisaurid (Mannion et al., 2013) affinities. 2. Medium development of the distal expansion of the scapular blade. This is similar to the condition in Camarasaurus (Ostrom & McIntosh, 1966) and Lourinhasaurus (Mocho et al., 2014), as well as many other sauropods (e.g. Mannion et al., 2013). 3. Great dorsoventral development of the acromion– glenoid region of the scapula (the ratio of the minimum width of the blade to the acromion– glenoid width is 0.27). This feature is present in Omeisaurus and other neosauropods, whereby the acromion process is more than 150% the minimum width of the scapular blade (Upchurch et al., 2004a). 4. Iliac peduncle of the ischium is well developed. This character is defined in vague terms, but similar proportions seem to occur in other sauropods, such as Haplocanthosaurus and Lourinhasaurus (Royo-Torres, 2009). 5. Large pubic process of the ischium (ratio of the length of this process to the length of the ischium 9. is 0.63). A large pubic process is shared with basal titanosauriforms such as Andesaurus, Tangvayosaurus, Venenosaurus, and Tastavinsaurus (Royo-Torres, 2009). Ischial distal shaft terminates in a conspicuous expansion. This feature can also be observed in Lourinhasaurus and several other sauropods (McIntosh, 1990; Upchurch et al., 2004a). Humerus/femur length ratio is relatively high (0.82). The value of this ratio in Aragosaurus is higher than that in Camarasaurus grandis or C. lentus, but is similar to other basal macronarians such as Lourinhasaurus (McIntosh, 1990; P.U. and P.D.M., pers. observ. 2009). The presence of a relative prominence, located distally to the large femoral trochanter. This character state is the lateral bulge originally noted by Salgado et al. (1997) as a synapomorphy of Titanosauriformes, but considered to be present in most eusauropods by Mannion et al. (2013). The associated medial deflection of the proximal part of the femur (Wilson, 2002) characterizes most macronarians (Royo-Torres et al., 2012; Mannion et al., 2013). Following Salgado et al.’s (1997) method for quantifying the size of the lateral bulge, Aragosaurus has an intermediate value between basal macronarians (such as Camarasaurus and Lourinhasaurus, with small lateral bulges) and Titanosauriformes. The ratio of the maximum anteroposterior width of the distal tibial condyle of the femur to the transverse width of the distal end of the femur is 0.77. Similar ratios occur in other neosauropods, such as Ferganasaurus (0.78) and Apatosaurus (0.79) (Royo-Torres, 2009). Thus, none of the nine diagnostic character states of Aragosaurus, proposed by Sanz et al. (1987), provides strong evidence for the uniqueness of this taxon; however, we have been able to recognize six new autapomorphies of Aragosaurus (see ‘Diagnosis’ and ‘Description and comparisons’ above), supporting the validity of this genus. Comparisons between Aragosaurus and other contemporaneous Iberian taxa, such as Lourinhasaurus, Galveosaurus, Losillasaurus, and Turiasaurus, reveal numerous differences, some of which have been noted in the ‘Description and comparisons’ section. In general, the forelimb proportions of Aragosaurus differ from those of other Iberian taxa (see above). Aragosaurus differs from Turiasaurus because the latter possesses opisthocoelous distal caudals and a distal process on the carpal. Aragosaurus also differs from Losillasaurus in the articulation of the proximal caudal vertebrae (procoelous in the latter taxon) and in the strong development of the distal part of the neural spines. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS Aragosaurus is different from Galveosaurus in the shapes of the humerus and ischia, and the latter taxon has a smooth acromion ridge, whereas in Aragosaurus this ridge is prominently developed. Aragosaurus is distinct from Lourinhasaurus in that Lourinhasaurus possesses hooked-like processes emanating from the dorsal surface of the most anterior caudal neural spines. The femoral lateral bulge is also much more developed in Aragosaurus than in Lourinhasaurus, and the latter has a ridge and groove structure in the acetabular region of the ischium. Aragosaurus can also be distinguished from other, non-contemporaneous Iberian taxa. For example, Tastavinsaurus has anterodorsally oriented anterior and middle caudal neural spines and a relatively short ischium compared with pubis length (Canudo et al., 2008; Royo-Torres, 2009). The rebbachisaurid Demandasaurus is very different from Aragosaurus (e.g. the latter lacks the diagnostic characters of Demandasaurus in the anterior caudal vertebrae, such as deep pneumatic cavities in the caudal ribs and the development of complex laminae; Torcida Fernández-Baldor et al., 2011). These data corroborate the hypothesis, suggested in previous studies, that basal macronarian sauropods were present in the Villar del Arzobispo Formation in the Iberian Peninsula at the end of the Jurassic (Royo-Torres et al., 2009). This stratigraphic unit is important for the study of sauropods across the Jurassic/Cretaceous boundary, and for examining palaeobiogeographic patterns in Europe at this time. As well as basal macronarians, the Villar del Arzobispo Formation preserves turiasaurs (Royo-Torres et al., 2006), diplodocids (Cuenca-Bescós, Canudo & Ruiz-Omeñaca, 1997; Royo-Torres, Cobos & Alcalá, 2006; Royo-Torres et al., 2009), and basal titanosauriforms (Suñer et al., 2009), including the tooth previously referred to Aragosaurus. Thus, the diversity of this formation is similar to those of the Morrison Formation in North America and the Tendaguru Formation of Tanzania. Once these various Iberian taxa have been studied more fully and integrated into phylogenetic analyses, they should make a substantial impact on our knowledge of Late Jurassic sauropod biodiversity and biogeography. CONCLUSIONS The complicated collection history of Aragosaurus, spanning more than 50 years, has resulted in a holotype split between two institutions [the Museo Provincial de Teruel (Dinopolis Exhibition) and the Museo Paleontológico de Galve]. The Las Zabacheras site represents a deltaic sedimentary complex in the Villar del Arzobispo Formation. Consequently, the age of this site can be constrained to the Tithonian–Berriasian interval, with the latter stratigraphic stage being the most 649 likely. Aragosaurus ischiaticus, erected in 1987, is a valid taxon based on the recognition of six new autapomorphies: (1) anterior caudal vertebrae possess neural spine summits with shallow, dorsolaterally facing concavities on either side of the midline; (2) epipophysislike protuberances on the dorsal surfaces of middle caudal postzygapophyses; (3) long interosseous ridges on the ulna and radius; (4) the interosseous ridge on the posterolateral face of the radial shaft terminates distally in a distinct tubercle; (5) carpal subquadrangular in shape in proximal and distal views, with similar lateromedial and anteroposterior diameters; and (6) a pedal phalanx (II-1 or III-1) that possesses a midline ridge on its dorsal surface, where its lateral and medial faces meet each other to form an apex. Aragosaurus shares some, but not all, of the derived character states that have been considered to characterize the Titanosauriformes. Therefore, as supported by cladistic analyses, Aragosaurus ischiaticus is a basal macronarian that is more closely related to Titanosauriformes than is Camarasaurus. ACKNOWLEDGEMENTS This study is part of the palaeontological research projects of the Departamento de Educación, Universidad, Cultura y Deporte, Gobierno de Aragón, and has been supported by its Dirección General de Patrimonio Cultural, FOCONTUR (Grupo de Investigación Consolidado E-62, Departamento de Industria e Innovación, Gobierno de Aragón and Fondo Social Europeo), Instituto Aragonés de Fomento, Fundación Conjunto Paleontológico de Teruel-Dinópolis, and SIBELCO Hispania S.A. Collection visits to institutions in Spain by P.U. and P.D.M. were supported by research grants awarded by The Palaeontological Association and the Abbey International Collaboration Scheme. P.D.M.’s research is also funded by an Imperial College London Junior Research Fellowship. R.M. thanks the CGL2011-22709 project from the Ministerio de Ciencia y Educación. F.G. is the holder of a grant FPU (Formación de Profesorado Universitario) from the Ministerio de Educación (Ref. AP2008-00846). We thank the Museo Aragonés de Paleontología, Museo Provincial de Teruel, and the late José María Herrero (Galve) for allowing us to study the material in their care. The photographs used in this work are from «Archivo López Segura del Instituto de Estudios Turolenses», Museo Provincial de Teruel, and José Luis Sanz’s team. We are grateful to Mike D’Emic and an anonymous reviewer, whose critical reviews improved this article. REFERENCES Alcalá L, Cobos A, Delclòs X, Luque L, Mampel L, Royo-Torres R, Soriano C. 2009. Mesozoic terrestrial © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 650 R. ROYO-TORRES ET AL. ecosystems in Teruel. In: Alcalá L, Royo-Torres R, coord. Mesozoic Terrestrial Ecosystems in Eastern Spain. ¡Fundamental! 14: 94–130. Aurell M, Bádenas B, Ipas J, Ramajo J. 2010. Sedimentary evolution of an Upper Jurassic carbonate ramp (Iberian Basin, NE Spain). In: van Buchem F, Gerdes K, Esteban M, eds. Reference models of Mesozoic and Cenozoic Carbonate systems in Europe and the Middle East – stratigraohy and diagenesis. London: Geological Society of London, Special Publication no. 329. 89–111. Berman DS, McIntosh JS. 1978. Skull and relationships of the Upper Jurassic sauropod Apatosaurus (Reptilia: Saurischia). Bulletin of Carnegie Museum of Natural History 8: 1–35. Bonaparte JF. 1986. The early radiation and phylogenetic relationships of the Jurassic sauropod dinosaurs, based on vertebral anatomy. In: Padian K, ed. The beginning of the age of dinosaurs. Cambridge: Cambridge University Press, 247–258. Bonaparte JF, Heinrich WD, Wild R. 2000. Review of Janenschia Wild, with the description of a new sauropod from the Tendaguru beds of Tanzania and a discussion on the systematic value of procoelous caudal vertebrae in the Sauropoda. Palaeontographica A 256: 25–76. Borsuk-Bialynicka M. 1977. A new camarasaurid sauropod Opisthocoelicaudia skarzynskii gen. n., sp. n. from the Upper Cretaceous of Mongolia. Palaeontologica Polonica 37: 5–63. Calvo JO. 1994. Jaw mechanics in sauropod dinosaurs. Gaia 10: 183–193. Calvo JO, Salgado L. 1995. Rebbachisaurus tessonei sp. nov. a new Sauropoda from the Albian-Cenomanian of Argentina; new evidence on the origin of the Diplodocidae. Gaia 11: 13–33. Canudo JI, Barco JL, Royo-Torres R, Ruiz-Omeñaca JI. 2001. Precisiones sobre la posición taxonómica de Aragosaurus ischiaticus (Dinosauria, Sauropoda). In: Meléndez G, Herrera Z, Delvene G, Azanza B, eds. Los fósiles y la Paleogeografía. Zaragoza: Publicaciones del Seminario de Paleontología de Zaragoza, Vol. 5. 263–270. Canudo JI, Gasca JM, Moreno-Azanza M, Aurell M. 2012. New information about the stratigraphic position and age of the sauropod Aragosaurus ischiaticus from the Early Cretaceous of the Iberian Peninsula. Geological Magazine 149: 252–263. Canudo JI, Royo-Torres R, Cuenca-Bescós G. 2008. A new sauropod: Tastavinsaurus sanzi gen. et sp. nov. from the Early Cretaceous (Aptian) of Spain. Journal of Vertebrate Paleontology 28: 712–731. Carballido JL, Pol D, Cerda I, Salgado L. 2011a. The osteology of Chubutisaurus insignis del Corro, 1975 (Dinosauria: Neosauropoda) from the ‘middle’ Cretaceous of central Patagonia, Argentina. Journal of Vertebrate Paleontology 31: 3–110. Carballido JL, Rauhut OWM, Pol D, Salgado L. 2011b. Osteology and phylogeny relationships of Tehuelchesaurus benitezii (Dinosauria, Sauropoda) from the Upper Jurassic of Patagonia. Zoological Journal of the Linnean Society 163: 605–662. Casanovas ML, Santafé JV, Sanz JL. 2001. Losillasaurus giganteus, un nuevo saurópodo del tránsito JurásicoCretácico de la cuenca de ‘Los Serranos’ (Valencia, España). Paleontologia i Evolució 32-33: 99–122. Chure D, Britt BB, Whitlock JA, Wilson JA. 2010. First complete sauropod dinosaur skull from the Cretaceous of the Americas and the evolution of sauropod dentition. Die Naturwissenschaften 97: 379–391. Cobos A. 2011. Los dinosaurios de Teruel como recurso para el desarrollo territorial. Unpublished PhD thesis. Universidad del País Vasco. Cobos A, Gascó F. 2013. New vertebral remains of the stegosaurian dinosaur Dacentrurus from Riodeva (Teruel, Spain). Geogaceta 53: 17–20. Cooper MR. 1984. Reassessment of Vulcanodon karibaensis Raath (Dinosauria: Saurischia) and the origin of the Sauropoda. Palaeontología Africana 25: 203–231. Cuenca-Bescós G, Amo O, Aurell M, Buscalioni AD, Canudo JI, Laplana C, Pérez-Oñate J, Ruiz-Omeñaca JI, Sanz JL, Soria AR. 1994. Los vertebrados del tránsito Jurásico-Cretácico de Galve (Teruel). Comunicaciones de las X Jornadas de la Sociedad Española de Paleontología, Madrid: 50–53. Cuenca-Bescós G, Canudo JI, Ruiz-Omeñaca JI. 1997. Dinosaurios del tránsito Jurásico-Cretácico en Aragón. In: Gámez Vintaed JA, Liñan E, eds. V Jornadas Aragonesas de Paleontología ‘Vida y ambientes del Jurásico’. Zaragoza: Instititución Fernando el Católico, 193–221. Curry Rogers KA. 2005. Titanosauria: a phylogenetic overview. In: Curry Rogers KA, Wilson JA, eds. The Sauropods: evolution and paleobiology. Berkeley and Los Angeles: University of California Press, 50–103. Curry Rogers KA. 2009. The postcranial osteology of Rapetosaurus krausei (Sauropoda: Titanosauria) from the Late Cretaceous of Madagascar. Journal of Vertebrate Paleontology 29: 1046–1086. Curry Rogers KA, Forster CA. 2001. The last of the dinosaur titans: a new sauropod from Madagascar. Nature 412: 530–534. D’Emic MD. 2012. The early evolution of titanosauriform sauropod dinosaurs. Zoological Journal of the Linnean Society 166: 624–671. D’Emic MD. 2013. Revision of the sauropod dinosaurs of the Early Cretaceous Trinity Group, southern USA, with the description of a new genus. Journal of Systematic Palaeontology 11: 707–726. D’Emic MD, Foreman BZ. 2012. The beginning of the sauropod dinosaur hiatus in North America: insights from the Lower Cretaceous Cloverly Formation of Wyoming. Journal of Vertebrate Paleontology 32: 883–902. D’Emic MD, Wilson JA, Williamson TE. 2011. A sauropod dinosaur pes from the latest Cretaceous of North America and the validity of Alamosaurus sanjuanensis (Sauropoda, Titanosauria). Journal of Vertebrate Paleontology 31: 1072– 1079. Derréal Y. 1959. Etude géologique de la région comprise entre Camarillas et Galve (Province de Teruel, Espagne). Unpublished Diplôme Institute Catholique et Diplôme © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS d´Études Supérieures, Faculté des Sciences, Université de Poitiers. Díaz-Molina M, Yebenes A. 1987. La sedimentación litoral y continental durante el Cretácico inferior. Sinclinal de Galve, Teruel. Estudios Geológicos Volumen extraordinario GalveTremp: 3–21. Díaz-Molina M, Yébenes A, Goy A, Sanz JL. 1984. Landscapes inhabited by Upper Jurassic/Lower Cretaceous archosaurs. III Symposium Mesozoic Terrestrial Ecosystems, 67–72. Díaz-Molina M, Yébenes A, Goy A, Sanz JL. 1985. Le Jurassic superieur detritique-carbonate du synclinal de Galve (Teruel). Strata 2: 155–170. Díez BJ, Pons D, Canudo JI, Cuenca G, Ferrer J. 1995. Primeros datos palinológicos del Cretácico inferior continental de Pielago-0, Galve (Teruel, España). Libro de resúmenes de las XI Jornadas de la Sociedad Española de Paleontología, 79–80. Díez Díaz V, Pereda Suberbiola X, Sanz JL. 2013. Appendicular skeleton and dermal armour of the Late Cretaceous titanosaur Lirainosaurus astibiae (Dinosauria: Sauropoda) from Spain. Palaeontologia Electronica 16: 1–18. Fernández-Galiano D. 1958. Descubrimiento de restos de dinosaurios en Galve. Teruel 20: 1–3. Fernández-Galiano D. 1960. Yacimientos de dinosaurios en Galve. Teruel. Boletín de la Real Sociedad Española de Historia Natural 58: 95–96. Gilmore CW. 1925. A nearly complete articulated skeleton of Camarasaurus, a saurischian dinosaur from the Dinosaur National Monument, Utah. Memoirs of the Carnegie Museum 10: 347–384. Goloboff P, Farris J, Nixon K. 2003. T.N.T.: tree analysis using new technology. Program and documentation. Available at: http://www.zmuc.dk/public/phylogeny/tnt Gómez JJ, Goy A. 1979. Las unidades litoestratigráficas del Jurásico medio y superior en facies carbonatadas del Sector Levantino de la Cordillera Ibérica. Estudios Geológicos 35: 569–598. Harris JD. 2006. The significance of Suuwassea emilieae (Dinosauria: Sauropoda) for flagellicaudatan intrarelationships and evolution. Journal of Systematic Palaeontology 4: 185– 198. Hatcher JB. 1902. Structure of the fore limb and manus of Brontosaurus. Annals of the Carnegie Museum 1: 356–376. Hatcher JB. 1903. Osteology of Haplocanthosaurus with description of a new species, and remarks on the probable habits of the Sauropoda and the age and origin of the Atlantosaurus beds. Memoirs of the Carnegie Museum 2: 1–72. Hunt AP, Lockley MG, Lucas SG, Meyer CA. 1994. The global sauropod fossil record. Gaia 10: 261–279. Ikejiri T. 2005. Distribution and biochronology of Camarasaurus (Dinosauria, Sauropoda) from the Jurassic Morrison Formation of the Rocky Mountain Region. In: Lucas SG, Zeigler KE, Lueth VW, Owen DE, eds. Geology of the Chama Basin. Socorro, NM: New Mexico Geological Society, Guidebook, 56th Field Conference, 367–379. Ipas J, Aurell M, Bádenas B, Canudo JI, Liesa C, Mas R, Soria AR. 2007. Caracterización de la Formación Villar 651 del Arzobispo al sur de Zaragoza (Titónico, Cordillera Ibérica). Geogaceta 41: 111–114. Janensch W. 1950. Die wirbelsaule von Brachiosaurus brancai. Palaeontographica 3 (Suppl. 7): 27–93. Janensch W. 1961. Die Gliedmaszen und Gliedmaszengürtel der Sauropoden der Tendaguru-Schichten. Palaeontographica 3 (Suppl. 7): 177–235. Kellner AWA, de Azevedo SAK. 1999. A new sauropod dinosaur (Titanosauria) from the Late Cretaceous of Brazil. Tokyo National Science Museum Monographs 15: 111–142. Lapparent AF. 1960. Los dos Dinosaurios de Galve. Teruel 24: 177–198. Liesa CL, Soria AR, Meléndez N, Meléndez A. 2006. Estensional fault control on the sedimentation patterns in a continental rift basin: El Castellar Formation, Galve subbasin, Spain. Journal of the Geological Society 163: 487–498. Luque L, Cobos A, Royo-Torres R, Espilez E, Alcalá L. 2005. Caracterización de los depósitos sedimentarios con dinosaurios de Riodeva (Teruel). Geogaceta 38: 27–30. Luque L, Espilez E, Cobos A, Alcalá L, Royo-Torres R, Aberasturi A, Mampel L, González A. 2006–2007. Iguanodóntidos y otros vertebrados cretácicos de la Sierra (Teruel). Teruel 91: 57–87. Madsen JH, McIntosh JS, Berdman DS. 1995. Skull and atlas complex of the Upper Jurassic sauropod Camarasaurus Cope (Reptilia: Saurichia). Bulletin of the Carnegie Museum of Natural History 31: 1–115. Mannion P, Upchurch P, Barnes RN, Mateus O. 2013. Osteology of the Late Jurassic Portuguese sauropod dinosaur Lusotitan atalaiensis (Macronaria) and the evolutionary history of basal titanosauriformes. Zoological Journal of the Linnean Society 168: 98–206. Mannion P, Upchurch P, Mateus O, Barnes RN, Jones MEH. 2012. New information on the anatomy and systematic position of Dinheirosaurus lourinhanensis (Sauropoda: Diplodocoidea) from the Late Jurassic of Portugal, with a review of European diplodocoids. Journal of Systematic Palaeontology 10: 521–551. Mannion PD. 2009. A rebbachisaurid sauropod from the Lower Cretaceous of the Isle of Wight, England. Cretaceous Research 30: 521–526. Mannion PD. 2010. A revision of the sauropod dinosaur genus ‘Bothriospondylus’ with a redescription of the type material of the Middle Jurassic form ‘B. madagascariensis’. Palaeontology 53: 277–296. Mannion PD, Barrett PM. 2013. Additions to the sauropod dinosaur fauna of the Cenomanian (early Late Cretaceous) Kem Kem beds of Morocco: palaeobiogeographical implications of the mid-Cretaceous African sauropod fossil record. Cretaceous Research 45: 49–59. Mannion PD, Calvo JO. 2011. Anatomy of the basal titanosaur (Dinosauria, Sauropoda) Andesaurus delgadoi from the midCretaceous (Albian–early Cenomanian) Río Limay Formation, Neuquén Province, Argentina: implications for titanosaur systematics. Zoological Journal of the Linnean Society 163: 155–181. Mannion PD, Upchurch P, Carrano MT, Barrett PM. 2011. Testing the effect of the rock record on diversity: © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 652 R. ROYO-TORRES ET AL. a multidisciplinary approach to elucidating the generic richness of sauropodomorph dinosaurs through time. Biological Reviews 86: 157–181. Marsh OC. 1878. Principal characters of American Jurassic dinosaurs. Part I. American Journal of Science 16: 411– 416. Mas R, Alonso A, Meléndez N. 1984. La formación Villar del Arzobispo: un ejemplo de llanuras de marea siliciclásticas asociadas a plataformas carbonatadas. Jurásico terminal (NW de Valencia y E de Cuenca). Publicaciones de Geología de la Universidad Autónoma de Barcelona 20: 175– 188. Mas R, García A, Salas R, Meléndez A, Alonso A, Aurell M, Bádenas B, Benito MI, Carenas B, García-Hidalgo JF, Gil J, Segura M. 2004. Segunda fase de rifting: Jurásico Superior-Cretácico Inferior. In: Vera JA, ed. Geología de España. Madrid: Sociedad Geológica de España-Instituto Geológico y Minero de España, 503–510. Mateus O, Mannion PD, Upchurch P. In press. Zby atlanticus, a new turiasaurian sauropod (Dinosauria, Eusauropoda) from the Late Jurassic of Portugal. Journal of Vertebrate Paleontology. McIntosh JS. 1990. Sauropoda. In: Weishampel DB, Dodson P, Ósmolska H, eds. The Dinosauria, 1st edn. Berkeley, CA: University California Press, 345–401. McIntosh JS, Miles CA, Cloward KC, Parker JR. 1996a. A new nearly complete skeleton of Camarasaurus. Bulletin Gunma Museum of Natural History 1: 1–87. McIntosh JS, Miller WE, Stadtman KL, Gillette DD. 1996b. The osteology of Camarasaurus lewisi (Jensen, 1988). Bringhan Young University Geology Studies 41: 73–115. Millán FJ. 2004. Una relacción de complicidad: Dinosaurios y medios de comunicación. In: Alcalá L, Cobos A, coord. Dinosaurios de Teruel ¡Fundamental! 2: 65–94. Mocho P, Royo-Torres R, Ortega F. 2014. Phylogenetic reassessment of Lourinhasaurus alenquerensis. A basal Macronaria (Sauropoda) from the Upper Jurassic of Portugal. Zoological Journal of the Linnean Society 170: 875– 916. Ostrom JH, McIntosh JS. 1966. Marsh’s dinosaurs. The collections from Como Bluff. New Haven, CT: Yale University Press. Otero A. 2010. The appendicular skeleton of Neuquensaurus, a Late Cretaceous saltasaurine sauropod from Patagonia, Argentina. Acta Palaeontologica Polonica 55: 399–426. Owen R. 1841–1842. Report on British fossil reptiles. Reports of the British Association for the Advancement of Science 11: 60–204. Pérez-García A, Scheyer TM, Murelaga X. 2013. The turtles from the uppermost Jurassic and Early Cretaceous of Galve (Iberian Range, Spain): anatomical, systematic, biostratigraphic and palaeobiogeographical implications. Cretaceous Research 44: 64–82. Peropadre C, Meléndez N, Liesa CL. 2012. Nuevas unidades estratigráficas del Aptiense en la cuenca del Maestrazgo (este de España). Geo-Temas 13: 86–89. Rauhut OWM, Remes K, Fechner R, Cladera G, Puerta P. 2005. Discovery of a short-necked sauropod dinosaur from the Late Jurassic period of Patagonia. Nature 435: 670– 672. Riggs ES. 1903. Structure y relationships of opisthocoelian dinosaurs. Part. I. Apatosaurus Marsh. Field Columbian Museum, Geological Series 2 4: 165–196. Riggs ES. 1904. Structure and relationships of opisthocoelian dinosaurs. Part II. The Brachiosauridae. Field Columbian Museum, Geologica Series 2 6: 229–247. Royo-Torres R. 2009. El saurópodo de Peñarroya de Tastavins. Teruel: Instituto de Estudios turolenses-Fundación Conjunto Paleontológico de Teruel, Monografías Turonlenses 6. Royo-Torres R, Alcalá L, Cobos A. 2012. A new specimen of the Cretaceous sauropod Tastavinsaurus sanzi from El Castellar (Teruel, Spain), and a phylogenetic analysis of the Laurasiformes. Cretaceous Research 34: 61–83. Royo-Torres R, Canudo JI, Ruiz-Omeñaca JI. 1999. Nueva descripción del pubis de Aragosaurus ischiaticus Sanz, Buscalioni, Casanovas y Santafé, 1987 (Dinosauria, Sauropoda) del Hauteriviense superior (Cretácico inferior) en Galve (Teruel). Actas de las XV Jornadas de la Sociedad Española de Paleontología Madrid, 1: 325–330. Royo-Torres R, Cobos A. 2009. Turiasaur sauropods in the Tendaguru Beds of Tanzania. Journal of Vertebrate Paleontology 29: 173A. Royo-Torres R, Cobos A, Alcalá L. 2006. A giant European dinosaur and a new sauropod clade. Science 314: 1925– 1927. Royo-Torres R, Cobos A, Luque L, Aberasturi A, Espílez E, Fierro I, González A, Mampel L, Alcalá L. 2009. High European sauropod dinosaur diversity during JurassicCretaceous transition in Riodeva (Teruel, Spain). Palaeontology 52: 1009–1027. Royo-Torres R, Ruiz-Omeñaca JI. 1999. The pubis of Aragosaurus ischiaticus Sanz, Buscalioni, Casanovas & Santafé (1987), a camarasaurid sauropod from the upper Hauterivian (Lower Cretaceous) of Galve, Teruel, Spain. IV European Workshop on Vertebrate Palaeontology, Albarracin, Spain, 96–97. Royo-Torres R, Upchurch P. 2012. The cranial anatomy of the sauropod Turiasaurus riodevensis and implications for its phylogenetic relationships. Journal of Systematic Palaeontology 10: 553–583. Ruiz-Omeñaca JI. 2006. Restos directos de dinosaurios (Saurischia, Ornithischia) en el Barremiense (Cretácico Inferior) de la Cordillera Ibérica en Aragón (Teruel, España). Unpublished PhD thesis. Universidad de Zaragoza. Ruiz-Omeñaca JI, Canudo JI, Burell M, Bádenas B, Barco JL, Cuenca-Bescós G, Ipas J. 2004. Estado de las investigaciones sobre los vertebrados del Jurásico Superior y Cretácico Inferior de Galve (Teruel). Estudios Geológicos 60: 179–202. Salas R. 1987. El Malm i el Cretaci inferior entre el Massis de Garraf i la serra d’Espadà. Unpublished PhD thesis. Universidad de Barcelona. Salas R, Guimerá J, Mas R, Martín-Closas C, Meléndez A, Alonso A. 2001. Evolution of the Mesozoic Central Iberian Rift System and its Cainozoic Inversion (Iberian Chain). In: Cavazza W, Robertson AHF, Ziegler P, Crasquin-Soleau S, © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS eds. Peri-Tethyan Rift/Wrench Basin and Passive Margin. Vol. 186. Paris: Mémoires du Muséum National d’Histoire Naturelle, 145–185. Salgado L, Coria R. 2002. Variación morfológica en la secuencia de vértebras caudales de algunos saurópodos titanosaurios. Revista Española de Paleontología 17: 211–216. Salgado L, Coria RA, Calvo JO. 1997. Evolution of titanosaurid sauropods. I: phylogenetic analysis based on the postcranial evidence. Ameghiniana 34: 3–32. Sánchez-Hernández B. 2005. Galveosaurus herreroi, a new sauropod dinosaur from Villar del Arzobispo Formation (Tithonian-Berriasian) of Spain. Zootaxa 1034: 1–20. Sander MP, Mateus O, Laven T, Knötschke N. 2006. Bone histology indicates insular dwarfism in a new Late Jurassic sauropod dinosaur. Nature 441: 739–741. Santisteban C, Santos Cubedo A. 2010a. Patrones de variación de facies en relación con regresiones forzadas en los depósitos de la Formación Villar del Arzobispo (Cuenca Íbero-Levantina). Comunicaciones del V Congreso Jurásico de España. Museo Jurásico de Asturias (MUJA), Colunga, 142–148. Santisteban C, Santos Cubedo A. 2010b. Relación entre playas aterrazadas y cauces encajados, en depósitos deltaicos de la Formación Villar del Arzobispo (Cuenca Íbero-Levantina). Comunicaciones del V Congreso Jurásico de España. Museo Jurásico de Asturias (MUJA), Colunga, 135–141. Sanz JL. 1982. A sauropod dinosaur tooth from the Lower Cretaceous of Galve (Province of Teruel, Spain). Geobios 15: 943–949. Sanz JL. 2007. Cazadores de dragones. Barcelona: Editorial Ariel. Sanz JL. 2009. Aragosaurus, el primer dinosaurio español. In: Instituto de Estudios Turolenses, ed. II Jornadas Paleontológicas de Galve. 50 años del descubrimiento del Aragosaurus y 20 años de su definición. Teruel: Instituto de Estudios Turolenses, 11–17. Sanz JL, Buscalioni AD, Casanovas ML, Santafé JV. 1987. Dinosaurios del Cretacico Inferior de Galve (Teruel, España). Estudios Geológicos Volumen Extraordinario, GalveTremp: 45–64. Sanz JL, Buscalioni AD, Moratalla JJ, Francés V, Antón M. 1990. Los reptiles mesozoicos del registro español. Madrid: Monografías del Museo Nacional de Ciencias Naturales-CSIC. Sanz JL, Powell JE, Le Loeuff J, Martinez R, Suberbiola XP. 1999. Sauropod remains from the Upper Cretaceous of Laño (northcentral Spain). Titanosaur phylogenetic relationships. Estudios del Museo de Ciencias Naturales de Alava 14: 235–255. Sekiya T. 2011. Re-examination of Chuanjiesaurus anaensis (Dinosauria: Sauropoda) from the Middle Jurassic Chuanjie Formation, Lufeng County, Yunnan Province, southwest China. Memoir of the Fukui Prefectural Dinosaur Museum 10: 1–54. Sereno PC. 1997. The origin and evolution of dinosaurs. Annual Review of Earth and Planetary Science 25: 435–489. Sereno PC. 1999. The evolution of dinosaurs. Science 284: 2137– 2147. 653 Sereno PC, Wilson JA, Witmer LM, Whitlock JA, Maga A, Ide O, Rowe TA. 2007. Structural extremes in a Cretaceous dinosaur. PLoS ONE 2: e1230. doi:10.1371/ journal.pone.0001230. Soria de Miguel AR. 1997. La sedimentación em las cuencas marginales del surco Ibérico durante el Cretácico Inferior y su control estructural. Thesis of the Universidad de Zaragoza. Steel R. 1970. Saurischia. Handbuch der Palaoherpetologie 14: 1–87. Suñer M, Santisteban C, Galobart À. 2009. Direct evidence of Titanosauriformes found in a Upper Jurassic site of Alpuente (Los Serranos, Valencia, Spain). In: DelgadoBuscalioni A, Fregenal-Martínez M, eds. Tenth international symposium on Mesozoic Terrestrial ecosystems and Biota. Madrid: UAM ediciones, 139–140. Taylor MP. 2009. A re-evaluation of Brachiosaurus altithorax Riggs 1903 (Dinosauria, Sauropoda) and its generic separation from Giraffatitan brancai (Janensch 1914). Journal of Vertebrate Paleontology 29: 787–806. Tidwell V, Carpenter K, Brooks W. 1999. New sauropod from the Lower Cretaceous of Utah, USA. Oryctos 2: 21–37. Tidwell V, Carpenter K, Meyer S. 2001. New Titanosauriform (Sauropoda) from the Poison Strip Member of the Cedar Mountain Formation (Lower Cretaceous), Utah. In: Tanke DH, Carpenter K, eds. Mesozoic vertebrate life. Bloomington: Indiana University Press, 139–165. Torcida Fernández-Baldor F, Canudo JI, Huerta P, Montero D, Pereda Suberbiola X, Salgado L. 2011. Demandasaurus darwini, a new rebbachisaurid sauropod from the Early Cretaceous of the Iberian Peninsula. Acta Palaeontologica Polonica 56: 535–552. Upchurch P. 1993. The anatomy, phylogeny and systematics of the Sauropod Dinosaurs. Thesis of the University of Cambridge. Upchurch P. 1995. The evolutionary history of sauropod dinosaurs. Philosophical Transactions of the Royal Society of London, Series B 349: 365–390. Upchurch P. 1998. The phylogenetic relationships of sauropod dinosaurs. Zoological Journal of the Linnean Society 124: 43–103. Upchurch P. 1999. The phylogenetic relationships of the Nemegtosauridae. Journal of Vertebrate Paleontology 19: 106– 125. Upchurch P, Barrett PM. 2005. A phylogenetic perspective on sauropod diversity. In: Curry Rogers KA, Wilson JA, eds. The sauropods: evolution and paleobiology. Berkeley; Los Angeles: University of California Press, 104–124. Upchurch P, Barrett PM, Dodson P. 2004a. Sauropoda. In: Weishampel DB, Dodson P, Osmólska H, eds. The Dinosauria, 2nd edn. Berkeley, CA: University of California Press, 259– 324. Upchurch P, Barrett PM, Galton PM. 2007. The phylogenetic relationships of basal sauropodomorphs: implications for the origin of sauropods. In: Barrett PM, Batten DJ, eds. Evolution and palaeobiology of early sauropodomorph dinosaurs. Special Papers in Palaeontology, 77, London: The Palaeontological Association, 57–90. © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 654 R. ROYO-TORRES ET AL. Upchurch P, Mannion PD, Butler RJ, Benson RBJ, Carrano MT. 2011. Geological and anthropogenic controls on the sampling of the terrestrial fossil record: a case study from the Dinosauria. Geological Society, London, Special Publications 358: 209–240. Upchurch P, Tomida Y, Barrett PM. 2004b. A new specimen of Apatosaurus ajax (Sauropoda: Diplodocidae) from the Morrison Formation (Upper Jurassic) of Wyoming, USA. National Science Museum Monographs, Tokyo 26: 1–108. Wilson JA. 1999. The evolution and phylogeny of sauropod dinosaurs. Thesis of the University of Chicago. Wilson JA. 2002. Sauropod dinosaur phylogeny: critique and cladistic analysis. Zoological Journal of the Linnean Society 136: 217–276. Wilson JA, Carrano MT. 1999. Titanosaurs and the origin of ‘wide-gauge’ trackways: a biomechanical and systematic perspective on sauropod locomotion. Paleobiology 25: 252– 267. Wilson JA, Sereno PC. 1998. Early evolution and higherlevel phylogeny of sauropod dinosaurs. Society of Vertebrate Paleontology Memoir Series 5: 1–68. Wilson JA, Upchurch P. 2009. Redescription and reassessment of the phylogenetic affinities of Euhelopus zdanskyi (Dinosauria: Sauropoda) from the Late Jurassic or Early Cretaceous of China. Journal of Systematic Palaeontology 7: 199– 239. Yates AM. 2007. The first complete skull of the Triassic dinosaur Melanorosaurus Haughton (Sauropodomorpha: Anchisauria). In: Barrett PM, Batten DJ, eds. Evolution and palaeobiology of early sauropodomorph dinosaurs. Special Papers in Palaeontology, 77, London: The Palaeontological Association, 9–55. Young CC, Zhao XJ. 1972. Mamenchisaurus. Memoirs of the Institute of Vertebrate Paleontology and Paleoanthropology Monograph Series 8: 1–30. MODIFICATIONS TO APPENDIX THE WILSON (2002) DATA MATRIX Brachiosaurus is divided into two OTUs, Brachiosaurus altithorax and Giraffatitan brancai, after Taylor (2009). Brachiosaurus altithorax (Upchurch et al., 2004a; Taylor, 2009) W98 (from 1 to ?), W107 (we retain 0 rather than 1, as proposed by Taylor, 2009), W130 (we retain state 0 and not the ?, as proposed by Taylor, 2009), W198 (from 1 to 2), W200 (from 1 to 0). ??????????????????????????????????????????????????????????? ?????????????????1??????????????101011?111100100211010?? ??000001100100000????????11??????1?????000?010101????? ??????????????11101???????1121000?????????????????????????? ?????0 Giraffatitan brancai (Upchurch et al., 2004a; Taylor, 2009) W98 [we retain state 1 and not ?, as proposed by Taylor (2009), because the character state is present, see Janensch, (1950: 46, fig. 63)], W101 (we retain state 1 and not 0, as proposed by Taylor, 2009), W107 (we retain state 0 and not 1, as proposed by Taylor, 2009), W130 (we retain state 0 and not ?, as proposed by Taylor, 2009), W146 [from 0 (Wilson, 2002; Taylor, 2009) to 1, based on Janensch (1950), Upchurch et al. (2004a), and Royo-Torres (2009)],W198 (from 1 to 2), W200 [we retain state 1 (Wilson, 2002) and not 0, as proposed by Taylor (2009)]. 111100111111000101011000010111111100101111000000 01100110011111121110011210001103011001100931010111 111100100211010?100000001100100000000???1011??11??1 10101000001010111101002111111111011110101110111121 100110111111110011110111?11?1??11?0 Cedarosaurus weiskopfae (Tidwell et al., 1999; D’Emic, 2012) ????????????????????????????????????????????????????????????? ???????????????11??????????????01011010???010??????0?1?00 0000010110??10100??????10?11??11?10100001101?1110?10 02??1?1??????????????10?????1????1?????????????1?11????1?1 0111?0 Europasaurus holgeri (Sander et al., 2006) 1111?01111?1001101?01?10010111011100101???100000011 0011001111?1?1110?11110?101??0110000000?10101101111 00100210?001?000000011001000000000??11000910?0110 10000000101011110100???11??????111101?1110111110100 110111011110?11?1???1??1?1??11?? Lourinhasaurus alenquerensis (Mocho et al., 2014) ????????????????????????????????????????????????????????????? ???????????????01??011??0??1??1?1?110???1??1??2???10????0 0?001001100?00????????01??????1101010000010101111010 01?1??????????110001?101111101001101?1011?10?????????? ?????????0 Tastavinsaurus sanzi (Canudo et al., 2008; Royo-Torres et al., 2012) ????????????????????????????????????????????????????????????? ???????????????11?????????????10?0110110110100211010?12 0000000001101000100?????1?0911111???????????????????10 1????????????111010111011111110011111101?100011111110 11111001110 Tehuelchesaurus benitezii (Carballido et al., 2011b) ????????????????????????????????????????????????????????????? ???????????????01?????????????001?100101010100?1???????? ???????????????????????01??????110001001?0101011110100? ????????????1???01110111110000???????????????????????????? ???? Venenosaurus dicrocei (Tidwell et al., 2001; D’Emic, 2012; Royo-Torres et al., 2012) ????????????????????????????????????????????????????????????? ????????????????1???????????????????????????????????1?2000? © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655 ARAGOSAURUS ISCHIATICUS 000101101010000?????1???11??111001??????????1100100?? ?111111????????0111011??????????????????????1101??1?????? ???0 Aragosaurus ischiaticus ????????????????????????????????????????????????????????????? ??????????????????????????????????????????????????????????0?? ??000000001100000000???????10110?110101?00?01010111 101001?1111?00????????01?1?111111100???????????????????? ?????1???1?0 MODIFICATIONS TO THE UPCHURCH ET AL. (2004A) 655 0000111110221110110010111002111011100110000010001 011011101101111011001100[01]0000100000011010010011 1100011100011002000111010000100111001011111111010 1111110111000011111111111111011111111011111101111101 111??010111111 Europasaurus hogeri (Sander et al., 2006) 110011011110??111111?010100110??0?10110?0?00000?00? 0?00011001100111010011?1111?1011110010??1110????0?0 11110022??00110010?????0111101???1?0??10110010111010 011?1011011001??000000????00001?100100111100011?0?? 111?20?????011???000????????111?????10111111???10?011?? 11110101111?10111?1101111???1??1???????????????1? DATA MATRIX Brachiosaurus is divided into two OTUs, Brachiosaurus altithorax and Giraffatitan brancai, after Taylor (2009). Brachiosaurus altithorax (Upchurch et al., 2004a; Taylor, 2009) U126 (we retain state 0–2 and not ?, as proposed by Taylor, 2009), U127 (from 1 to ?), U131 (from 1 to 0), U135 (from 1 to 0), U137 (from 1 to ?), U149 (from 1 to 0), U152 (from 0 to 1), U155 (we retain state 0 and not 1, as proposed by Taylor, 2009), U157 (from 0 to 1), U172 (from 1 to 0), U208 (from 1 to 0), U249 (from 1 to 0), U271 (from 1 to 0). ??????????????????????????????????????????????????????????? ?????????????????????????????????????????????1??????????????? ???????110011001?000?0100010010111011011110110011?? 0????1???000?10101?001?????????????1002000111010000??? ????????????????10111110???????????111111111100?????????? ??????????????????????????1 Giraffatitan brancai (Upchurch et al., 2004a; Taylor, 2009) U124 (from ? to 0, because G. brancai has 12 dorsal vertebrae), U126 (we retain states 0–2 and not ?, as proposed by Taylor, 2009), U127 [we retain state 1 (Upchurch et al., 2004a) and not 0, as proposed by Taylor (2009)], U135 (from 1 to 0), U140 (from 0 to 1), U141 (from ? to 0), U152 (from 0 to 1), U157 (from 0 to 1), U172 (scored as 0&1), U187 (we retain state 1 and not 0, as proposed by Taylor, 2009), U208 (from 1 to 0), U267 (from 1 to 0), U271 [we retain state 1 (Upchurch et al., 2004a) and not 0, as proposed by Taylor (2009)]. 110011011110001111111011100111110110110100001001 000100001100100111101011111111111001111000110101?0 Lourinhasaurus alenquerensis (Mocho et al., 2014) ????????????????????????????????????????????????????????????? ????????????????????????????????????????011111??2????11???? ???00011100?1?110?????????1?0???1??????0110?1001???000 ??1???010???1001?0?1?????1110??111?00001100010001001 1??0???????????10001111011?0000101111101111110111111? 101?1111?????????????????????0 Tastavinsaurus sanzi (Canudo et al., 2008; Royo-Torres et al., 2012) – three characters have been changed: U105 (from? to 1), U142 (from 0 to 1), U249 (from 0 to 1) ????????????????????????????????????????????????????????????? ???????????????????????????????????????????1????????????????? ???0?110111?01011?2111011000010001??01110010011?011 000100?0011111011001111101??????????????1??????????01?? ?????????????1011111100100010101111101011110111101110 111110?11111001101?010111011 Tehuelchesaurus benitezii (Carballido et al., 2011b; Mocho et al., 2014) ????????????????????????????????????????????????????????????? ??????????????????????????????????????????11????????????????? ???01110011??1000??11?01?00000000110?01???1??????????? ???????????????????????111???011???????00010001011??????? ????????????????01110000101111101111100????????????????? ???????????????????0 Aragosaurus ischiaticus ????????????????????????????????????????????????????????????? ????????????????????????????????????????????????????????????? ???????????????????????????????????????????00???00000?1000 000000?01100?111000111000????????1?000100010011??0?0 1?11?????????????011?000010?111111111110???????????????? ???????????????????1? © 2014 The Linnean Society of London, Zoological Journal of the Linnean Society, 2014, 171, 623–655
© Copyright 2026 Paperzz