All numbers pertaining to specimens A–C are prefixed by CPP.
Cervical vertebrae. The material assigned to Uberabatitan ribeiroi includes 15 cervical vertebrae in different states of preservation, which correspond to distinct positions within the cervical series. The order inferred of the cervical sequence is as follows (from the anteriormost to the posteriormost element, and taking into account all of the specimens): 058-UrHo, 914-UrHo, 1091-UrHo, 1022-UrB, 1057-UrHo, 1023-UrHo, 994-UrB, 1070-UrB, 993-UrHo, 1108-UrB, and 991-UrB.
1058-UrHo, a small but relatively high neural arch, corresponds to an anterior vertebra, possibly to the third one. It is very similar to cervical 3 of the ‘Serie A’ from Peirópolis described by Powell (2003, pl. 13, fig. 2a–b) and identified as ‘Titanosaurinae’ indet.
The centrum of the cervical vertebrae is opisthocoelus, proportionally low and elongate in the anterior elements (914-UrHo, 1022-UrB), and proportionally short and high in the mid and posterior cervicals (Text-figs 4–7).
In the anterior cervicals, the lateral faces of the centrum are flat, but in the posteriormost vertebrae of the preserved sequence, these are occupied by a shallow pleurocoel. In the posterior cervical 1108-UrB, the eye-like pleurocoels are dorsally bordered by an edge or lip (Text-fig. 7A). The ventral face of the anterior cervical centra is flat, but in the posterior vertebrae of the cervical series, the ventral face becomes slightly concave anteriorly; this is most clearly seen in Specimen B (1022-UrB and 085-UrB).
The cervical parapophyses project outwards and a little downwards. Probably correlated with the sequential shortening of the cervical centra, the lateroventral rim existing between the parapophysis and the posterolateroventral corner of the centrum (Text-figs 6B–C, 7A, white arrow), becomes proportionally shorter with respect to the centrum length, much more than in Trigonosaurus pricei (Campos et al. 2005), and in Powell’s ‘Serie A’ (Powell 2003, pl. 14, fig. 6). An equally short postparapophyseal rim is observed in the cervical vertebrae of AlamosaurusGilmore 1922 (Lehman and Coulson 2002, fig. 2).
The cervical prezygapophyses of U. ribeiroi pass beyond the border of the centrum, unlike the representatives of Saltasaurini (sensuSalgado and Bonaparte 2007), where the articular facets of the prezygapophyses are located practically above the diapophyses (Powell 2003). In turn, in the Brazilian species, the postzygapophyses barely pass beyond the posterior margin of the cervical centrum (Text-figs 5–6).
The neural spines of the anterior and mid-cervicals of U. ribeiroi are relatively low and triangular in lateral view (Text-figs 5–6A), as in T. pricei. Throughout the cervical sequence, the neural spines become progressively higher.
The tips of the posterior cervical neural spines of U. ribeiroi seem to be transversely expanded, but not as much as in Powell’s ‘Serie A’ (Powell 2003, pl. 14, fig. 6). Laterally, immediately above the diapophyses, the neural spines become progressively deeper throughout the cervical sequence, as seen in 994-UrB, 1070-UrB, 993-UrHo, 1108-UrB (Text-fig. 7A) and 991-UrB (Text-fig. 7B); this is another character in common with T. pricei and Powell’s ‘Serie A’ (Powell 2003, pl. 14, fig. 6). Kellner et al. (2006) reported the existence in Maxakalisaurus topai of a lateral depression at the base of the neural spine, although they noted that this condition is not exclusive to the three titanosaurs from the Bauru Group, because it is present in the distantly related titanosaur Mendozasaurus neguyelapGonzález Riga 2003 (González Riga 2005, fig. 2).
The cervical neural spine of U. ribeiroi is formed by the spinoprezygapophyseal (sprl) and spinopostzygapophyseal (spol) laminae. In the anterior cervical 1091-UrHo, the sprl do not reach the prezygapophyseal articular facets; instead, they are gradually reduced to become virtually non-existent. In the posterior cervicals (e.g. 991-UrB), the sprl are conspicuous, pillar-like structures that enclose laterally a deep median fossa (Text-fig. 7C), as seen in cervicals 11–13 of T. pricei (Campos et al. 2005, fig. 3). In 1057-UrHo, well-developed spol enclose a deep posterior space above the neural canal. None of the preserved cervical vertebrae of U. ribeiroi shows signs of a prespinal (prsl) or a postspinal lamina (posl).
The anterior and, especially, the posterior centrodiapophyseal laminae (acdl and pcdl respectively) are well developed, mostly in the posterior cervicals (as seen in 1108-UrB) (Text-fig. 7A). The most notable feature of the cervical vertebrae of U. ribeiroi is the poorly developed postzygodiapophyseal lamina (podl) that is divided into two segments, one of which (the zygapophyseal) extends over the other (the diapophyseal). This character is observed in many vertebrae of the holotype specimen, 1091-UrHo (Text-fig. 4B–C), 1057-UrHo (Text-fig. 5A) and 1023-UrHo (Text-fig. 6A).
CPP-UrB does not comprise well-preserved anterior or mid-cervicals. However, incomplete posterior cervicals belonging to Specimen B (e.g. 1024-UrB and 1108-UrB) show a completely formed podl (Text-fig. 7A), which, according to our interpretation, means that the segmentation of the podl occurs only in the anterior and mid-cervicals. The incomplete development of the podl reported in the anterior and mid-cervicals of U. ribeiroi is in some way comparable to the condition present in Alamosaurus (Lehman and Coulson 2002).
Finally, the centroprezygapophyseal laminae (cprl) of the posterior cervicals are well developed, and the centropostzygapophyseal laminae (cpol) are pillar-like structures.
Cervical ribs. Many cervical ribs were preserved articulated to their respective centra (919-UrHo and 1057-UrHo), whereas others were isolated (Text-fig. 8). Not only were their heads preserved but also their delicate shafts. The cervical ribs seem to have extended at least up to the posterior end of the subsequent vertebra.
Dorsal vertebrae. The dorsal series of U. ribeiroi is represented by only two elements of the holotype specimen: a complete anterior vertebra (1077-UrHo) (Text-fig. 9), and a neural arch of, possibly, the fifth or sixth dorsal (1068-UrHo) (Text-fig. 10).
The centrum of the anterior dorsal is slightly wider than high (Text-fig. 9), and has a small pleurocoel; its neural arch is low and wide. The prezygapophyses are widely separated, and connected by a ridge. The articular facets of the prezygapophyses are wide and oval-shaped, with their longest axis transversely orientated.
The neural spine of 1077-UrHo, although incompletely preserved, is low. In this vertebra, a modest prespinal lamina is found, and the sprl, subtly insinuated, are lateral to the prsl. Likewise, a low crest, close to the podl, is interpreted as a spinodiapophyseal lamina (spdl) (Salgado et al. 2006). There is a deep space between the podl, the cpol and the pcdl. The parapophyses are placed between the centrum and the neural arch. The neural canal is oval and slightly wider than high. Both spols enclose a deep median space, within which there is no posl, although this may be owing to poor preservation. 1077-UrHo resembles CPP-036, as figured by Santucci and Bertini (2006b, fig. 7B–C).
The mid-dorsal neural arch (1068-UrHo) is robust by comparison with other titanosaurs from Peirópolis (e.g. T. pricei) (Text-fig. 10). There is a prespinal lamina, the basal half of which is laterally enclosed by two conspicuous laminae, as in CPP-494 (Santucci and Bertini 2006b) (Text-fig. 10A). However, in contrast to the opinion of Santucci and Bertini (2006b), who interpreted these structures as ‘accessory prespinal laminae’, we consider them as sprl, which, as noted above, are only weakly insinuated in the anterior dorsal (1077-UrHo). In T. pricei, the sprl never join the prsl in a single axial composite lamina, but they virtually disappear by the second dorsal (pers. obs.). In the case of U. ribeiroi (and possibly CPP-494), the sprl would have been maintained as individual structures in the mid- and (possibly) posterior dorsals.
In lateral view (Text-fig. 10B), a massive vertical diapophyseal lamina is observed (dl in Text-figs 10–11), which is thought to be homologous with the (cervical) podl. As occurs in T. pricei and Alamosaurus sanjuanensis, this lamina does not connect to the postzygapophysis (see Salgado et al. 2006). In 1068-UrHo, the notable development of this lamina hides the postzygapophyses in anterodorsal view.
In the left transverse process of 1068-UrHo, a nearly flat dorsal surface is observed that may be similar to the ‘ama’ (‘area for muscular attachment’) described by Santucci and Bertini (2006b, figs 5–6). In the same vertebra there is a postspinal lamina (posl), although very weakly developed.
Dorsal ribs. Some incomplete dorsal ribs from different areas of the ribcage were preserved (Text-fig. 12A–B). These elements do not show characters that are distinctive from other titanosaurs.
Sacral vertebrae. The incomplete centrum 1099-UrHo probably corresponds to the first sacral vertebra. It has a convex anterior articular surface, and a flat posterior face, which was certainly not fused to the rest of the sacrum.
Caudal vertebrae. Specimens A and B preserve vertebrae from different regions of the caudal series. All are strongly procoelus, a titanosaurian characteristic (Powell 2003), except the distalmost ones, which are procoelous only in saltasaurines (Curry-Rogers 2005) (Text-figs 13, 15).
The holotype has the only anterior caudal of the collection (1079-UrHo); it is probably caudal 2 or 3. 1079-UrHo has excavated lateral faces (although not to the extent seen in the mid-caudals, as discussed below), apparently more so than in the possible caudal 2 of T. pricei (Campos et al. 2005, fig. 33). In 1079-UrHo the neural spine is apparently compressed laterally, as is the case in the mid-caudals, unlike T. pricei (Campos et al. 2005, figs 27–30, 34–35, 38–39), Baurutitan (Kellner et al. 2005, figs 9–10, 13–14, 17–18, 20–22) and Adamantisaurus mezzalirai (Santucci and Bertini 2006a, figs 3–4, 6–7), in which these structures are thick and transversely expanded at their tips. The caudal prezygapophyses of U. ribeiroi are long and project up and forewards.
Five mid-caudals were recovered: 1020-UrB (Text-fig. 13A–C; two fused elements), 1018-UrB (Text-fig. 13D–F), 1019-UrB (Text-fig. 13G–J) and 1017-UrHo (Text-fig. 13K–N). This last vertebra is posterior to the others because it lacks the transverse processes and its neural spine is somewhat lower (it may be caudal 11 or 12).
The mid-caudals (as in 1018-UrB and 1019-UrB) have their ventral faces excavated, whereas in lateral view, the lower margin of the vertebral centrum is markedly concave, and the facets for the haemapophyses are prominent.
The lateral faces of the mid-caudal centra are more deeply excavated than in the anterior caudal. This seems to be the result of two different facts, namely that: (1) the lateral faces of the centrum incline laterally, as described by Salgado and García (2002) in Laplatasaurus araukanicusHuene, 1929 and other titanosaurs, and (2) the posterior articulation does not appear (in posterior view) heart-shaped, as it does in Gondwanatitan faustoi (Kellner and Azevedo 1999, p. 126) and Baurutitan britoi (Kellner et al. 2005, p. 549). In fact, in U. ribeiroi, the posterior articulation of the mid-caudals is subquadrangular in posterior view. The posterior articulation of 1018-UrB (Text-fig. 13E), 1019-UrB (Text-fig. 13I) and 1020-UrB is somewhat wider than high; in 017-UrHo it is practically as wide as high.
In lateral view, the mid-caudal neural spine of U. ribeiroi has the form of a parallelogram, which is higher than long in 1020-UrB (Text-fig. 13A) and 1019-UrB (Text-fig. 13G), and longer than high in 017-UrHo (Text-fig. 13K); this variation is probably a result of the different positions of these vertebrae in the caudal series. In U. ribeiroi (1020-UrB, Text-fig. 13A, and 1019-UrB, Text-fig. 13G), the neural spine is somewhat inclined anteriorly as in T. pricei (Campos et al. figs 26, 37); in B. britoi, by contrast, the inclination of the neural spine is always posterior (Kellner et al. 2005, figs 16, 19) (Text-fig. 14).
In U. ribeiroi, the caudal posl is moderately developed, unlike B. britoi, in which the notable expansion of the posl extends the posteriormost point of the neural spine up to the level of the posterior margin of the postzygapophyses (Kellner et al. 2005, fig. 19) (Text-fig. 14).
The mid-caudal prezygapophyses of U. ribeiroi are horizontally projected, and their bases are deep with respect to the apophyseal shafts. In contrast, the mid-caudal prezygapophyses of B. britoi are slightly directed upwards (Kellner et al. 2005, figs 16, 19).
The postzygapophyses of the mid-caudals of U. ribeiroi are posteriorly projected, with their articular facets lateroventrally orientated, as in T. pricei (Campos et al. 2005, fig. 26), and unlike B. britoi (Kellner et al. 2005, figs 16, 18–19, 21, 24), in which the postzygapophyses are very close to the base of the neural spine and the articular facets are practically parallel to the axial plane (Text-fig. 14).
In the mid-caudals 1020-UrB (Text-fig. 13B–C), 1018-UrB (Text-fig. 13F) and 1019-UrB (Text-fig. 13J) there is a lamina connecting both prezygapophyses above the neural canal that hides the dorsal margin of the anterior articulation (which is incomplete in 1019-UrB) in dorsal view. A similar lamina is present in Rinconsaurus caudamirus (Calvo and González Riga, 2003) and in an unnamed titanosaur from La Pampa Province in Argentina (González Riga et al. 2005).
Above the lamina that connects both prezygapophyses, at the base of the caudal neural spine of U. ribeiroi, there is a deep fossa. In 1017-UrHo this lamina disappears; thus, the dorsal border of the anterior articulation of the vertebral centrum is visible in dorsal view (Text-fig. 13N). In this vertebra, the fossa mentioned above is absent.
As noted above, 1020-UrB is composed of two solidly fused caudal vertebrae (Text-fig. 13A–C). The fusion is the result of an atypical ossification. Firstly, a great amount of extra bone, which may correspond to ossified tendons, is present on the lateral and ventral faces of the centrum. Secondly, on their right side, the vertebrae show an abnormal connection between the prezygapophysis (which is hypertrophied) and a posterior protrusion of the base of the neural spine of the preceding element; this connection is lateral to the postzygapophysis of the anterior vertebra. Between the right postzygapophysis and the posterior extension of the anterior neural arch, there is a deep gap. It is unclear if the hypertrophied prezygapophysis establishes a connection only with this structure or if it also connects with the corresponding postzygapophysis.
There are many distal caudals assigned to U. ribeiroi: 1008-UrB (Text-fig. 15E–F), 1009-UrHo (Text-fig. 15A–B), 1010-UrHo (Text-fig. 15C–D), 1011-UrHo, 1012-UrHo, 1014-UrB. The exact sequential order of these elements is doubtful: all have the same elongate, subcylindrical morphology, with their neural spines nearly horizontal and the prezygapophyses anteriorly projected. The vertebra 1008-UrB, undoubtedly the most posterior one, is biconcave (Text-fig. 15E–F), as is usual in non-saltasaurine titanosaurs (Curry-Rogers 2005).
Unlike U. ribeiroi, the holotypes of B. britoi and T. pricei do not comprise distal caudal vertebrae; only in Gondwanatitan faustoi (Kellner and Azevedo 1999, fig. 15; Costa-Franco et al. 2004, fig. 8) and Maxakalisaurus topai (Kellner et al. 2006, fig. 15) are these elements known. Apparently, the distal caudals of U. ribeiroi match well with the corresponding elements of these two species, although their posterior articulation is less pronounced. In turn, the biconcave distal caudal of U. ribeiroi resembles other vertebrae found in Peirópolis assigned to TitanosaurusLydekker, 1877 by Santucci and Bertini (2001, fig. 3A–B). Others, such as 1008-UrB and 1010-UrHo, are comparable to the distal caudals described by Trotta et al. (2002) from the Mombuca site in Peirópolis.
Chevrons. Five complete haemal arches of U. ribeiroi and fragments of many more are preserved (Text-fig. 16A–E). The anteriormost chevrons seem to have two articular facets, suggesting that each haemal arch probably articulated with two successive vertebrae. The ventral process of the chevrons is transversely flattened. Their shaft is longer than in B. britoi (Kellner et al. 2005, figs 25–27). The chevrons of the new species are generally comparable to those of M. topai (Kellner et al. 2006, figs 17–18), although in this latter species they seem to be thinner.
Sternal plate. A right sternal plate is preserved (1027-UrHo). The entire bone is thick, and it is 38 cm long. Its lateral border is concave, whereas the inner margin is straight to slightly concave. The anterior end of the plate has a crest on its ventral face, as in other titanosaurs.
Coracoid. The holotype of U. ribeiroi includes the right coracoid (1109-UrHo), which is apparently quadrangular (24 cm long, 33 cm high) (Text-fig. 17A). The scapular articulation is not well preserved, and its anterodorsal corner is missing. In contrast, the glenoid cavity is well preserved. The coracoid foramen is oval and completely closed.
Specimen B has the left coracoid preserved (1120-UrB). In this bone, the scapular articulation and the glenoid cavity are more robust that in the holotype (Text-fig. 17B–C).
Humerus. The proximal half of a left humerus is preserved (1030-UrHo) (Text-fig. 17D). The bone is subquadrangular and notably robust, with its deltopectoral crest extremely developed, and the anterior face deeply excavated.
Radius. The holotype specimen of U. ribeiroi comprises both radii (left radius, 1032-UrHo, Text-fig. 17G–H; right radius, 911-UrHo, Text-fig. 17I). The measurements of the elements are: length, 44.5 cm (left), 45 cm (right); proximal expansion, 10.5 cm (both); distal expansion, 11.5 cm (both). Distally, on the ulnar face of the bone, there is a strong rugosity or prominence. Also on the ulnar face, a sharp edge that extends diagonally through most of the length bone is present; the anti-ulnar face is flat. The distal end is very robust and expanded.
Metacarpal. The holotype of U. ribeiroi includes only the right metacarpal II (1080-UrHo) (Text-fig. 17E–F). Its length is 27.5 cm, which is 62 per cent of the radius length. Its ends are slightly expanded (7 cm at proximal end, 8.3 cm distally). Its proximal articulation is roughly triangular (Text-fig. 17E). The distal articulation is subquadrangular, without any indication of articular facets for phalanges.
On the lateral face of the bone, a series of crests, which constitute the contact with metacarpal III, are seen: one anterior, which extends along the entire bone; another, also anterior, which is placed on the distal third of the metacarpal; and one posterior, very sharp. Proximally, there is an anterior protuberance on the lateral face, which may also be for the contact with the metacarpal III. Unlike the lateral side, the medial face of metacarpal II is relatively flat to slightly convex.
Pubis. Both pubes of the holotype are preserved (left, 1029-UrHo; right, 1103-UrHo) (Text-fig. 18A–C). The bones are incomplete; the left has the part of the obturator foramen preserved (Text-fig. 18A).
One of the most noticeable characteristics of the pubis is its unusual thickness and robustness. The inner (dorsal) face is flat, but the external (ventral) face has a stout crest that divides the bone into two concave surfaces. A similar crest is observed in other titanosaurs, but in U. ribeiroi it is much more developed. There is a fragment of a right pubis (913-UrB) that is much larger than those of the holotype; the crest is not so marked here as in the holotype, although the fragment is, in overall terms, more robust and thick. It possibly corresponds to the distal fragment of the pubis.
Ischium. The ischium is a relatively thick laminar bone (1026-UrB). It preserves the base of the iliac process (Text-fig. 18D). Although this bone is poorly preserved, it is apparently broader than the ischia of Gondwanatitan faustoi (Kellner and Azevedo 1999, fig. 19) and Maxakalisaurus topai (Kellner et al. 2006, fig. 21).
Femur. Specimen C includes part of the diaphysis of a right femur (894-UrC) (Text-fig. 19A). In common with the other elements of its appendicular skeleton, the femur of U. ribeiroi is characterized by its robustness; it shows the diagnostic lateral bulge of the titanosauriform femur (Salgado et al. 1997), and a prominent fourth trochanter. The medial face of the femur at the level of this trochanter is slightly concave.
Additionally, there is a distal end of a much larger femur (898-UrB). The condyles of this element are badly eroded, with its fibular condyle virtually missing. The lateral face of the distal end is slightly concave above the tibial condyle.
Tibia. The left tibia of the holotype specimen of U. ribeiroi is preserved (912-UrHo) (Text-fig. 19B–F). The bone is very robust, particularly in its proximal third, unlike the tibia of Gondwanatitan faustoi (Kellner and Azevedo 1999, fig. 21), which is gracile. Proximally, 912-UrHo presents an unusual protuberance on its lateral face. Between this prominence and the cnemial crest, there is a deep recess, which received the anterior part of the proximal end of the fibula (Text-fig. 19B–C). In other sauropods, such as ApatosaurusMarsh, 1877 (Gilmore 1936, fig. 23D) and CamarasaurusCope, 1877 (Wilson and Sereno 1998, fig. 32E), and including other titanosaurs such as NeuquensaurusPowell, 1992 (Salgado et al. 2005, fig. 7i) and MendozasaurusGonzález Riga, 2003 (González Riga 2003, fig. 6A), the U-like recess between the cnemial crest and the main body of the tibia is wide. In U. ribeiroi, by contrast, this recess is narrower, which is the result of the extreme development of the aforementioned protuberance. In fact, this structure forms an acute angle (nearly 15 degrees) with the cnemial crest. The cnemial crest of U. ribeiroi is itself expanded and thickened on its distal border; its proximal border is slightly concave, unlike in other titanosaurs.
Distal to the cnemial crest, a ridge extends over the anterior margin of the tibia. Laterally, this ridge is slightly concave; the morphology of the anterolateral surface of the tibial shaft of U. ribeiroi has not been reported in other titanosaurs.
The cnemial crest and the distal expansion of the tibia are not perpendicular as in other titanosaurs, but form an angle of nearly 30 degrees. The inner face of the bone is slightly convex in this area.
Fibula. The holotype specimen of U. ribeiroi includes a left fibula (1107-UrHo) (Text-fig. 19B–C, G–J). The bone is notably robust, as for the other appendicular elements. Posterior to the lateral tuberosity there is a pronounced lateral process, beyond which there is a markedly concave surface on the posterior face of the bone (Text-fig. 19H, white arrow).
The medial face of the fibula is slightly concave. Proximally, this concavity is proximocaudally orientated, whereas distally it is in an anteroposterior direction. The proximal concavity of the fibula is preceded by a notable knob; this is the postive structure that articulates with the posterior side of the above-mentioned tibial protuberance (Text-fig. 19B). Thus, the development of the tibial protuberance and the fibular knob are thought to be mutually correlated, and are regarded here as a single autapomorphy (character 6 of the diagnosis).
The distal end of the fibula is semicircular: the straight side is for the contact with the tibia whereas the convex side is lateral.
Astragalus. The astragalus of U. ribeiroi (1082-UrHo, Text-fig. 20A–B) has the typical pyramidal shape of the astragali in other titanosaurs. Unlike non-titanosaurian sauropods (e.g. Camarasaurus; Wilson and Sereno 1998, fig. 31), it is relatively large: it reaches almost 69 per cent of the distal expansion of the tibia (in the titanosaur OpisthocoelicaudiaBorsuk-Bialynicka 1977, the same relationship is 54 per cent; in Neuquensaurus, 56 per cent; Salgado et al. 2005).
The posterior face of the astragalus is rugose and its ventral face smooth. The fibular articulation is broad; between this and the tibial articulation there is a planar surface, apparently broader than in other titanosaurs.