An overview of the appendicular skeletal anatomy of South American titanosaurian sauropods, with definition of a newly recognized clade

In the last two decades, the number of phylogenetically informative anatomical characters recognized in the appendicular skeleton of titanosaurian sauropod dinosaurs has increased dramatically with the discovery of new and comparatively complete specimens. here we provide an overview of the appendicular skeletal morphology of South American titanosaurs and discuss its significance for phylogenetic reconstruction. the appendicular skeletal diversity of South American titanosaurs is substantially greater than was initially appreciated. Moreover, some regions of the appendicular skeleton, such as the pes, exhibit remarkable variability in form. Multiple synapomorphies of titanosauria and the less inclusive clades Lithostrotia and Saltasauridae consist of characters of the girdles and limbs. Although the phylogenetic definitions of titanosaurian clades such as Saltasaurinae and Lognkosauria are stable, the taxonomic content of these clades has varied in recent analyses depending on the phylogenetic topology recovered. Within titanosauria, the results of four recent, largely independent analyses support the existence of a derived titanosaurian lineage distinct from the ‘Saltasaurinae line,’ which is herein termed Colossosauria. At present, this clade is mainly comprised by taxa within Lognkosauria and rinconsauria, and is useful in discussions of titanosaurian lower-level relationships.

Anatomical and phylogenetic analyses of titanosaurs are crucial for deciphering the evolutionary history and paleobiology of these most titanosaurs, and in some taxa, the presence or absence of ossified carpals and manual phalanges is controversial (see, e.g., Apesteguía 2005, Mannion and Otero 2012, Poropat et al. 2015).Similarly, the pes of titanosaurs is poorly known; until recently, complete, articulated pedes were known only in Epachthosaurus and Opisthocoelicaudia.Fortunately, knowledge of titanosaurian pedal osteology has recently been enhanced by the discovery of the giant titanosaur Notocolossus from the Late Cretaceous of Mendoza Province, Argentina (González riga et al. 2016) plus two other, as-yet unnamed species from Mendoza (the 'Agua del Padrillo taxon,' UNCUYO-LD 313, González riga et al. 2015) and Neuquén (the 'La Invernada taxon, ' MUCPv-1533, González riga et al. 2008a) provinces, respectively (Figure 1).Furthermore, nearly complete pedes are known for a few other titanosaurs as well, such as Bonitasaura (Gallina and Apesteguía 2015), Mendozasaurus (González riga et al. 2018), Rapetosaurus (Curry rogers 2009), and possibly Alamosaurus (the latter based on NMMNh P-49967, a questionably referred specimen from the Upper Cretaceous of the southwestern U.S.A., D'Emic et al. 2011).the pes is very incompletely known in most other titanosaurs.
Many aspects of the girdle and limb morphology of titanosaurs (plus additional taxa within the more inclusive neosauropod clade Macronaria, Wilson and Sereno 1998) have been interpreted as being related to the acquisition of wide-gauge posture, where the manus and pedes are located at a considerable distance from the sagittal midline.this interpretation has been supported by many Late Cretaceous sauropod trackways attributed to derived titanosaurs (e.g., Farlow 1992, Lockley et al. 1994, Calvo 1999, Wilson and Carrano 1999, Wilson 2006, González riga and Calvo 2009).however, several exceptions to this potential correlation between appendicular morphology and posture have also been recognized in the ichnological record (e.g., Lockley et al. 2002, Stevens et al. 2016), and as such, it is now customary for well-preserved trackways to be carefully analyzed, including studies of gait (e.g., Vila et al. 2008, González riga andtomaselli 2019).Additionally, computer-aided biomechanical studies are casting new light on the stance and locomotion of sauropods in general (Klinkhamer et al. 2018).
Another complex paleobiological aspect of titanosauria is the gigantism attained by some lineages and its relationship to appendicular osteology.Wilson and Carrano (1999) and Carrano (2005) argued that many appendicular features seen in titanosaurs and other sauropods-such as graviportal, columnar limb posture, increased limb bone robusticity, shortened distal limb segments, and increased femoral midshaft eccentricityappear intimately related to the acquisition of large body size.As in other quadrupedal dinosaurs, the manus and pes of sauropods exhibit phalangeal reduction (Osborn 1904, Coombs 1975, Upchurch 1995, 1998, Wilson and Sereno 1998).typically, such reduction occurred primarily in terms of length, with individual phalanges becoming compact and often disc-like.however, titanosaurs continued this trend, with many taxa reducing and in some cases apparently eliminating ossified manual phalanges.Furthermore, titanosaurs also exhibit the most reduced pedal phalangeal formulae seen within Sauropoda (Borsuk-Bialynicka 1977, Salgado et al. 1997, Wilson and Sereno 1998, Martínez et al. 2004, Apesteguía 2005, González riga et al. 2008a, 2016).
In the present contribution, we provide an overview of titanosaurian appendicular skeletal anatomy, focusing on the many representatives of this clade that have been recovered from the Cretaceous of South America.We also discuss several of the principal appendicular skeletal characters that have been used in previous phylogenetic analyses of titanosauria and its subclades, with the goal of identifying areas of agreement and conflict.Finally, we formally define a new clade of derived South American titanosaurs, recognized on the basis of results of recent cladistic analyses.
As in non-titanosaurian sauropods, the proximal (i.e., anterior, if the long axis of the scapula is oriented horizontally) end of the scapula is more expanded than the distal (i.e., posterior) end.In most South American titanosaurs (e.g., Drusilasaura, Mendozasaurus, Patagotitan, Pitekunsaurus, Saltasaurus), the proximal expansion is less than twice the dorsoventral breadth of the distal end, with the exceptions of Dreadnoughtus and Muyelensaurus, in which the proximal end is more than twice as broad as the distal.Nevertheless, the scapulae of most South American titanosaurs also exhibit a slight distal expansion, with the dorsal (i.e., acromial) edge of the blade being dorsally deflected and the ventral edge remaining nearly straight.In most titanosaurs, however, the tallest part of the dorsal margin of the scapular blade is still lower than that of the proximal expansion.An exception is seen in Patagotitan, in which the most elevated point of the blade is approximately the same height as that of the proximal end (Figure 2h).Among South American titanosaurs, a reversal to an unexpanded distal scapular blade occurs in Dreadnoughtus, in which the dorsal and ventral edges of the blade are subparallel to one another (Figure 2d).
In some South American titanosaurs (e.g., Patagotitan, Pitekunsaurus), as in other members of the clade (e.g., Rapetosaurus), the orientation of the scapular blade relative to the coracoid articulation is roughly 45°.In other forms (e.g., Dreadnoughtus, Muyelensaurus), however, the long axes of the blade and coracoid articulation are oriented roughly perpendicular to one another.In titanosaurs, the acromial ridge is generally not as developed as it is in non-titanosaurian neosauropods such as Diplodocus (hatcher 1901), Giraffatitan (Janensch 1961), or Camarasaurus (McIntosh et al. 1996), in which this ridge clearly delimits a proximal scapular fossa, widely regarded as the origin site of the M. supracoracoideus (Meers 2003, Otero 2018) (Figure 2m-o).In Patagotitan, Pitekunsaurus, and the recently described, probably basal titanosaur Choconsaurus, the acromial region grades much more smoothly into the scapular blade than is the case in some other taxa (e.g., Aeolosaurus rionegrinus, Dreadnoughtus, Elaltitan, Muyelensaurus).When examined in lateral view, the scapular blade is dorsoventrally deep in some taxa (e.g., A. rionegrinus, Patagotitan, Saltasaurus) but substantially shallower in others (e.g., Dreadnoughtus, Muyelensaurus).A ventromedial process on the ventral margin of the blade (sensu Carballido et al. 2011) 3a-c, e).In titanosauria, the proximodistal (i.e., anteroposterior, if the long axis of the scapulocoracoid is oriented horizontally) length of the coracoid may be up to twice the length of the scapular articulation.the coracoid was progressively modified through the evolution of Saltasauridae, increasing in proximodistal elongation, becoming 'squared' at its anteroventral margin, and ultimately extending to the height of the acromial process (i.e., attaining a flush suture with the scapula).these derived morphologies are reported in many South American titanosaurs, such as the saltasaurines Neuquensaurus and Saltasaurus and the lognkosaurian Quetecsaurus, and are shared with some Asian titanosaurs (e.g., Opisthocoelicaudia, ZPAL MgD-I/48, Borsuk-Bialynicka 1977) and the North American Alamosaurus (USNM 15560, Gilmore 1946) as well.An infraglenoid lip is present in most or all South American titanosaurs for which the coracoid is preserved.the position of the coracoid foramen varies among South American titanosaurs, being situated immediately adjacent to the scapular articulation in some taxa (e.g., Quetecsaurus, Figure 3c) but located further from this articulation in others (e.g., Saltasaurus, Figure 3b).Nevertheless, as the position of the coracoid foramen appears to change through neosauropod ontogeny (see, e.g., Ullmann and Lacovara 2016), these distinctions are probably of limited taxonomic and phylogenetic significance.

StErNUM
Sternal plates are reported with some frequency among South American titanosaurs, being known in Bonitasaura (Gallina and Apesteguía 2 0 1 5 ) , C h o c o n s a u r u s , D re a d n o u g h t u s , Epachthosaurus, Maxakalisaurus (Kellner et al. 2006) [Kellner et al. 2011], Bonatitan [Martinelli andForasiepi 2004, Salgado et al. 2015], Futalognkosaurus [Calvo, 2014], Elaltitan, Mendozasaurus, Muyelensaurus, Narambuenatitan, Neuquensaurus, Uberabatitan), the proximal margin of the humerus is straight, or nearly so, in anterior view (e.g., Figure 4a, c).In other titanosaurs, however (e.g., Notocolossus [González riga et al. 2016], Opisthocoelicaudia, Paralititan, Quetecsaurus, Saltasaurus), the proximal margin of the humerus is sinuous due to the marked proximal deflection of the humeral head relative to the remainder of the bone (Figure 4j). the lateral margin of the humerus is nearly straight through approximately the proximal half of the element.Futalognkosaurus has a relatively robust humerus with an expanded proximal end that reaches 40% of the total length of the bone, as in Saltasaurus, Neuquensaurus, and Opisthocoelicaudia (Calvo 2014).the humeri of many South American titanosaurs (e.g., the saltasaurines Neuquensaurus and Saltasaurus, the lognkosaurian Futalognkosaurus [Calvo 2014], Patagotitan, and Dreadnoughtus) possess a well-developed posterolateral bulge around the level of the deltopectoral crest, which is frequently regarded as the insertion site of the M. deltoideus clavicularis (Meers 2003, Otero 2018).Cienc (2019) 91(Suppl. 2)  e20180374 12 | 42 Unlike many other South American titanosaurs, Neuquensaurus and Saltasaurus also exhibit a deltopectoral crest that is markedly expanded distally, a feature that is shared with the Laurasian titanosaurs Alamosaurus and Opisthocoelicaudia and that consequently has been regarded as a synapomorphy of Saltasauridae (Wilson 2002, D'Emic 2012).South American titanosaurs also exhibit variability in features such as the distal extent and medial deflection of the deltopectoral crest, the development of a proximolateral process, and the concavity of the medial margin of the shaft in anterior view.In saltasaurines and other lithostrotians (e.g., the non-South American genera Alamosaurus, Isisaurus, Opisthocoelicaudia, and Rapetosaurus), the articular surfaces of the humeral distal condyles are exposed on the anterior surface of the bone.the humeral distal condyles of titanosaurs are generally flat, except in saltasaurids and Epachthosaurus in which they are divided.the longest humerus known for any titanosaur is that of the holotypic specimen of Notocolossus (UNCUYO-LD 301, Figures 1c, 4j, González riga et al. 2016).this bone is even longer and more proximally robust than the humerus of Patagotitan (Figure 4h), a titanosaur that was recently described as the largest dinosaur yet discovered (Carballido et al. 2017).the humerus of Notocolossus is also   Most lithostrotians exhibit a proximally elevated olecranon process of the ulna, which constitutes a reversal to the basal sauropodomorph condition (Wilson and Sereno 1998, Galton and Upchurch 2004, Mannion et al. 2013).the ulnae of saltasaurid titanosaurs are further characterized by their stout proportions (Wilson 2002); this condition is developed to an extreme in the saltasaurines Neuquensaurus and Saltasaurus.Most other titanosaurs have more gracile ulnae, a condition that is especially true for taxa such as Mendozasaurus and Narambuenatitan.In titanosaurs, and sauropods in general, the anteromedial process of the proximal ulna is usually longer than the anterolateral process.(Calvo andBonaparte 1991, Mannion andCalvo 2011).In many titanosaurs, the ratio of the proximodistal length of metacarpal I to that of metacarpal II or III is 1.0 or greater (Upchurch 1998, Mannion et al. 2013); also, the proximal end of metacarpal V is often subequal in size to that of metacarpal I (D'Emic 2012).Nevetheless, in taxa such as Choconsaurus, Epachthosaurus, and Quetecsaurus, the proximal end of metacarpal V is clearly smaller than that of metacarpal I, with the latter being much more anteroposteriorly elongate (Figure 5b).Multiple authors have argued that, in titanosauria or clades therein, most or even all manual phalanges were absent or unossified (e.g., Salgado et al. 1997, Wilson 2002, Apesteguía 2005, Curry rogers 2005).Among South American titanosaurs, this contention is supported by the condition in Argyrosaurus, which is known from a complete, articulated forelimb that nonetheless lacks direct evidence of manual phalanges (Mannion and Otero 2012) (but see Discussion below).Similarly, an exceptionally complete, articulated postcranium of Epachthosaurus (UNPSJB-PV 920) possesses only a single rudimentary phalanx on manual digit IV (Martínez et al. 2004); an identical condition occurs in the holotypic skeleton of the Mongolian titanosaur Opisthocoelicaudia (ZPAL MgD-I/48, Borsuk-Bialynicka 1977).Nevertheless, manual phalanges are well-documented in the Australian titanosaurs Diamantinasaurus and Savannasaurus (hocknull et al. 2009(hocknull et al. , Poropat et al. 2015(hocknull et al. , 2016)), indicating that the diversity of manual morphologies within titanosauria was almost certainly greater than is presently appreciated (see Discussion).I).Among these taxa, the largest complete ilia and sacrum belong to Futalognkosaurus.In titanosaurs, the pre-and postacetabular processes of the ilium are expanded anteroposteriorly and dorsoventrally (Otero andVizcaíno 2008, Otero 2010).Although lateral projection of the iliac preacetabular process characterizes titanosaurs as a whole, the orientation of this process (i.e., lateral or anterolateral) varies considerably among titanosaurian taxa (Figure 6a, b).For example, the lateral projection of the preacetabular process is more pronounced in Futalognkosaurus, Neuquensaurus (Figure 6a), and Saltasaurus than it is in Dreadnoughtus, Epachthosaurus, Overosaurus (Figure 6b), or Trigonosaurus. to more rigorously define this character, Salgado et al. (2005) proposed an estimate of the ratio between the distance of the lateralmost point of the pubic peduncle versus that of the preacetabular process.

An Acad Bras
A strongly laterally directed preacetabular process appears to be associated with elongation of the sacral ribs, as seen in, for example, Neuquensaurus (MCS-5/16, B.J.G.r.pers.obs.).
According to many phylogenetic studies (e.g., Salgado et al. 1997), a typical titanosaurian character is the presence of a pubis that is proximodistally longer than the ischium.this feature seems to be present in most or even all titanosaurs, but in taxa such as Futalognkosaurus, it is developed to an extreme, with the pubis being markedly longer and more robust than the ischium (Figure 6f).Furthermore, in this massive Patagonian titanosaur, the bone has a slightly subcircular and expanded distal end and is strongly thickened distally (MUCPv-323, B.J.G.r.pers.obs.). the pubis of the basal titanosaur Andesaurus possesses a proximodistally elongate ischial articulation (Figure 6e); in most other titanosaurian taxa, by contrast, this articular surface is shorter.
ISChIUM the ischium is known in many South American titanosaurs, being especially well-preserved in taxa such as Andesaurus, Bonitasaura, Dreadnoughtus, Futalognkosaurus, Muyelensaurus, and Saltasaurus (table I, Figure 6c-f).the titanosaurian ischium is a short bone with a relatively broad, plate-like blade (Salgado et al. 1997, Otero 2010); this morphology as well as the absence of emargination distal to the pubic articulation are typical of the clade (Wilson 2002, Díez Díaz et al. 2016).A mediolaterally compressed iliac articular surface also appears to be a titanosaurian character (Mannion and Calvo 2011).the pubic peduncle is anteroposteriorly elongate in forms such as Aeolosaurus rionegrinus, Antarctosaurus FEMUr the femur is at least partially preserved in 25 species of South American titanosaurs; after the humerus, it is the most frequently recovered appendicular bone of these sauropods (table I, Figure 7).the femur is remarkably slender in Atacamatitan (Figure 7a), and to a lesser degree in 'Antarctosaurus' giganteus (Figure 7j), Bonatitan (Figure 7k), Mendozasaurus (Figure 7g), Patagotitan (Figure 7h), Petrobrasaurus (Figure 7f), and Rinconsaurus (Calvo and González riga 2003, fig.3c), but considerably more robust in taxa such as Dreadnoughtus (Figure 7i), Futalognkosaurus (Calvo 2014), and the saltasaurines Neuquensaurus (Figure 7b, c), Rocasaurus (Salgado and Azpilicueta 2000, fig.9a), and Saltasaurus (Figure 7e); titanosaurs such as Epachthosaurus (Figure 7d) and Traukutitan (Figure 7l) exhibit an intermediate condition.As with the humerus (see above), the markedly differing robusticities of the femur in giant titanosaurs such as 'A.' giganteus, Dreadnoughtus, and Patagotitan are relevant in discussions of the overall body dimensions of these taxa, though the significance of these differences for mass estimation is not yet completely clear.the femoral head is canted strongly proximomedially in Aeolosaurus maximus (Santucci and de Arruda-Campos 2011), Bonitasaura (Gallina and Apesteguía 2015), and Rinconsaurus, but noticeably less so in taxa such as Bonatitan, Dreadnoughtus, Epachthosaurus, Patagotitan, Petrobrasaurus, and saltasaurines.In most South American titanosaurs, the proximal one third of the femur is angled strongly medially relative to the remainder of the shaft, but a few taxa (e.g., Epachthosaurus, Figure 7d) lack this morphology (Mannion et al. 2013).
Most South American titanosaurs (e.g., Bonatitan, Dreadnoughtus, Patagotitan, Petrobrasaurus, Traukutitan, saltasaurines) also have a prominent lateral bulge on the proximal third of the femur distal to the greater trochanter.this feature was initially recognized by McIntosh (1990) and regarded by Salgado et al. (1997) as a synapomorphy of their then-newly defined clade titanosauriformes; it is also present but less developed in some nontitanosauriform taxa (Mannion et al. 2013).Salgado et al. (1997, fig. 10) tentatively quantified the presence of the lateral bulge as being 30% the minimum mediolateral width of the femoral shaft.
In South American titanosaurs, the femoral shaft is anteroposteriorly compressed, rendering it elliptical in cross section.the fourth trochanter is positioned near midshaft in some taxa (e.g., Antarctosaurus wichmannianus, 'A.' giganteus, Bonitasaura, Dreadnoughtus, Epachthosaurus, Mendozasaurus) but more proximally in others (e.g., Elaltitan, Neuquensaurus, Patagotitan).It is also more prominent in taxa such as Dreadnoughtus (Ullmann and Lacovara 2016) and Patagotitan (Carballido et al. 2017) than it is in others such as saltasaurines (e.g., Figure 7b, e).Saltasaurines exhibit a longitudinal ridge on the anterior surface of the femoral shaft (Otero 2010, D'Emic 2012).Another important femoral character is the orientation of the long axis of the distal condyles in anterior or posterior view; in many South American titanosaurs, the distal condyles are beveled 10° dorsomedially (Wilson 2002, 2006, D'Emic 2012), though there are exceptions to this condition   8a-l).Like other appendicular elements, the tibia and fibula of South American titanosaurs vary in proportions from relatively gracile to robust.Saltasaurines (Figure 8e) have very stout tibiae with a prominent cnemial crest, whereas in other taxa such as A. wichmannianus, Epachthosaurus (Figure 8c), and Mendozasaurus (Figure 8a) the tibia is considerably slenderer and the crest is weakly developed; other forms such as Dreadnoughtus (Figure 8d), Futalognkosaurus (Calvo 2014), and Laplatasaurus (Figure 8b) exhibit an intermediate condition.the extent of the cnemial fossa on the proximal end of the tibia is also highly variable among South American titanosaurs (Gallina and Otero 2015).
In lithostrotian titanosaurs, the mediolateral width of the distal end of the tibia is at least twice the diameter of the bone at midshaft (Wilson 2002, Mannion et al. 2013).the fibulae of South American titanosaurs exhibit variability in aspects such as the anteroposterior width of the proximal end, the robusticity and straightness of the shaft, and the development, location, and morphology of the lateral trochanter.the proximal end is anteroposteriorly broad in taxa such as Dreadnoughtus (Figure   9, see also Curry rogers 2005, González riga 2011), and indeed, sauropods as a whole (McIntosh 1990).Intrinsic factors related to the large body dimensions of these dinosaurs coupled with the relatively small size and fragility of their skull bones, posterior caudal vertebrae, and manual and pedal elements evidently led to the early disarticulation, and therefore loss, of these comparatively diminutive bones during the biostratinomic stage of necrokinesis (González riga et al. 2008b).In fact, of the well over 70 valid titanosaurian taxa recognized at present (Wilson et al. 2016  In sauropods, a progressive reduction in both the number and length of the pedal phalanges has been previously documented and is the most apparent evolutionary trend in the structure of the hind foot in these herbivorous dinosaurs (Bonnan 2005, González riga et al. 2008a, 2016).For example, in the basal eusauropods Shunosaurus (Zhang 1988) and Omeisaurus (he et al. 1988), a total of 12 pedal phalanges are described.titanosaurs, by contrast, have fewer pedal phalanges.the possible basal titanosaur Epachthosaurus has a pedal phalangeal formula of 2-2-3-2-0 (nine phalanges total, Martínez et al. 2004), and an even more reduced formula of 2-2-2-2-0 (eight phalanges total) occurs in the Padrillo (UNCUYO-LD 313) and Invernada (MUCPv-1533) taxa, Notocolossus (González riga et al. 2016), and Mendozasaurus (González riga et al. 2018).Opisthocoelicaudia was originally described as having a pedal phalangeal formula of 2-2-2-1-0 (Borsuk-Bialynicka 1977), but the shape of the distal condyle of the first phalanx of digit IV suggests the presence of a second ossified phalanx in this digit.Because of this, we herein interpret that this taxon had a phalangeal formula of 2-2-2-2-0, as is the case in all other derived titanosaurs for which the pes is completely represented.
Another relevant aspect of sauropod pedal structure is the development of the distal articular facets of the metatarsals.In diplodocids, for example, the articular facets are strongly convex and extend onto the dorsal (= anterior) face of the metatarsals (Bonnan 2005).this condition is clearly visible in Barosaurus (AMNh 6341, B.J.G.r.pers.obs.), in which the distal end of metatarsal I is strongly convex, indicating a wide range of mobility of phalanx I-1.A similar but less pronounced case is observed in Apatosaurus (CM 3018, phalangeal formula 2-3-3-2-1, B.J.G.r.pers.obs.), suggesting that the elevated mass of this taxon may have led to a reduction in the mobility of its phalanges in comparison with the more lightly-built diplodocine diplodocids Barosaurus and Diplodocus.In the macronarian Camarasaurus, two distal articular facets are present in metatarsals I and II (YPM 1901, B.J.G.r. pers. obs.), in accordance with the well-developed pedal phalanges of this taxon (phalangeal formula 2-3-2-2-1, McIntosh et al. 1996).
In titanosaurs, by contrast, the distal articular facets of the metatarsals are less developed than in other sauropods.In Mendozasaurus, for instance, these facets are only slightly convex and only some of them extend onto the dorsal surface of the metatarsal in question (González riga et al. 2018).An extreme case is observed in Notocolossus, the metatarsals of which have nearly flat distal articular facets, indicating reduced mobility of the digits (González riga et al. 2016, B.J.G.r. pers. obs.).Interestingly, unlike other titanosaurs, the pedal unguals of this gigantic taxon are small, blunt, and amorphous; although there is some possibility that this condition is pathologic (González riga An Acad Bras Cienc (2019) 91(Suppl.2) e20180374 24 | 42 et al. 2016), it is also consistent with the limited development of the distal articular facets of the metatarsals.
González riga et al. ( 2016) preliminarily recognized two primary titanosaurian pedal skeletal morphotypes, which they termed 'longfooted' (where, as in non-titanosaurian sauropods, the first four metatarsals exhibit a significant increase in length and a decrease in robusticity from medial to lateral) and 'short-footed' (with metatarsals that are all roughly the same length).Further study has led us to recognize considerable variation within González riga et al.'s (2016) longfooted morphotype, revealing a broad diversity in form (Figure 9).Moreover, within titanosauria, there is no clear correlation between body size and pedal osteology; instead, the differing hind foot architecture of various titanosaurs is probably more intimately related to evolutionary trends seen within different lineages.
the short-footed pedal morphotype is a massive structure that has thus far been observed only in Notocolossus (Figure 9g).this giant sauropod exhibits the lowest differences between the lengths of the metatarsals of any titanosaurian taxon yet discovered (for instance, the ratio of the length of metatarsal III to that of metatarsal I is only 1.14, González riga et al. 2016).Moreover, in Notocolossus, differences in robusticity between metatarsal I and metatarsals II-V are less pronounced than in other titanosaurs.this may be quantified using the Metatarsal Robustness Index (MtRI), which is herein defined as the minimum mediolateral breadth of metatarsals II-V divided by that of metatarsal I.In Notocolossus, the MtrI is greater than 0.70 in metatarsals II-V.Moreover, the non-ungual phalanges are relatively long and wide in relation to the metatarsals, and both the size and shape of the unguals are unique, as noted above and as was described by González riga et al. (2016).
the other long-footed titanosaurs analyzed herein are Epachthosaurus (Figure 9b, Martínez et al. 2004), Mendozasaurus (Figure 9f, González riga et al. 2018), the unnamed Invernada and Padrillo taxa (Figure 9c, d, González riga et al. 2008a, 2015), and NMMNh P-49967, an isolated pes provisionally attributed to Alamosaurus (Figure 9e, D'Emic et al. 2011).In contrast to Opisthocoelicaudia, in all of these taxa, metatarsal V is longer than metatarsal I and metatarsals I-IV show a progressive increase in length; because of this, metatarsal IV is the longest (the length ratio of metatarsal IV/metatarsal I is 1.39-1.57in these taxa) (table II). the unguals are relatively large in relation to metatarsal length.Of all titanosaurians for which the pes is completely known, Epachthosaurus is unique in retaining nine phalanges.this accords with the basal position of this genus that is frequently recovered by phylogenetic analyses (e.g., Carballido et al. 2017) and the hypothesis of progressive reduction of the  pedal phalanges within titanosauria (González riga et al. 2008a(González riga et al. , 2016)).Among this second group of long-footed titanosaurs, there are, as yet, no definitive correlations between pedal structure and body size, though some possible trends are evident.In the smallerbodied taxa within this group (Epachthosaurus and the Invernada and Padrillo taxa, with body lengths of up to approximately 10 m, Martínez et al. 2004, González riga et al. 2008a), metatarsals IV and V are relatively slender (Figure 9b-d).In contrast, in ?Alamosaurus (NMMNh P-49967) and Mendozasaurus, metatarsal V is relatively robust and longer than metatarsals I and II (Figure 9e, f).Both of these latter animals were very large: in ?Alamosaurus, metatarsal IV is 29.1 cm in length, and the femoral length of this individual has been estimated at 1.6-2.0m (D'Emic et al. 2011, González riga et al. 2016).Similarly, metatarsal III of an undescribed specimen of Mendozasaurus (UNCUYO-LD 356) is 29.2 cm in length, and as such, the individual in question was probably comparable in size to that represented by NMMNh P-49967.Further discoveries of relatively complete titanosaurian pedes are needed to further evaluate the potential relationships between body size and pedal morphology discussed herein.

DISCUSSION
the skeletal structure of sauropods has traditionally been interpreted as being relatively conservative in comparison to that of other dinosaurs.this is documented in, for example, Wilson and Curry rogers' (2005) summary of the history of sauropod discoveries.In an early stage of the study of these iconic herbivorous dinosaurs, Romer (1968) lamented the difficulty in achieving a classification of sauropods due to their relatively incomplete fossil record.thankfully, however, the sauropod record has improved dramatically in recent decades, leading to significant advances in knowledge of the anatomy, evolution, and paleobiology of these animals.As Wilson and Curry rogers (2005) pointed out, "the improvement in our understanding of sauropod phylogeny is the result of an improved sauropod fossil record." Many sauropod species are primarily defined on anatomical characters derived from the presacral, sacral, and/or anterior caudal vertebrae, and therefore, many authors have justifiably focused much of their attention on the axial skeleton (e.g., Bonaparte 1999, Wilson 1999, 2012).Knowledge of other regions of the sauropod skeleton, especially the skull, manus, and pes, has lagged behind understanding of vertebral anatomy, and as such, the extent of morphological variability in these parts of the skeleton has not been as thoroughly characterized.Nevertheless, there is no definitive evidence that certain structures changed significantly more than others through sauropod evolutionary history.As more discoveries have been made, it has become apparent that some parts of the sauropod skeleton deserve more attention from researchers than they had previously been afforded.This is certainly the case as regards the appendicular skeleton of titanosaurs.At first glance, the appendicular anatomy of titanosaurs may appear fairly homogeneous.however, as more well-preserved specimens have come to light, it has become clear that there is considerable variation in the size and morphology of the girdle and limb elements within the clade.Accordingly, a significant number of appendicular skeletal characters have been incorporated into recent phylogenetic analyses of titanosauria and more inclusive clades such as titanosauriformes, Macronaria, and Neosauropoda (e.g., Curry rogers 2005, D'Emic 2012, Mannion et al. 2013, González riga et al. 2016, 2018, Gorscak and O'Connor 2016, Carballido et al. 2017, Sallam et al. 2018).Although the percentages of appendicular skeletal characters in these analyses have varied considerably, the absolute number of characters has generally increased over the years.For example, in a study of the evolutionary history of titanosauriformes, D'Emic (2012) included 51 appendicular characters out of a total of 119 (42%), whereas in their analysis of the relationships of Notocolossus, González Riga et al. (2016) modified the dataset of Carballido and Sander (2014) to include 119 appendicular features, comprising 34% of the total.
COMMENtS ON APPENDICULAr SYNAPOMOrPhIES OF SELECtED tItANOSAUrIAN CLADES here, we follow the node-based phylogenetic definition of Titanosauria proposed by Salgado et al. (1997) and subsequently modified by Wilson and Upchurch (2003).Many titanosaurian clades exhibit considerable diversity.One of these appears to be Lognkosauria, which is defined as the most recent common ancestor of Mendozasaurus neguyelap and Futalognkosaurus dukei and all descendants (Calvo et al. 2007a).Whereas prior studies (e.g., González riga andOrtiz David 2014, González riga et al. 2016) restricted Lognkosauria to these two species, more recent analyses by Carballido et al. (2017) and González riga et al. (2018) have suggested a more diverse clade that also includes Argentinosaurus, Patagotitan, and possibly Drusilasaura, Notocolossus, Puertasaurus, and/or Quetecsaurus.
In previous phylogenetic analyses, titanosauria has been diagnosed by various appendicular synapomorphies, some of which were initially proposed for the titanosaurian subclades titanosauroidea or titanosauridae (which have since been abandoned due to the invalid status of the genus Titanosaurus, Wilson and Upchurch 2003).A review of existing phylogenetic analyses of titanosauria demonstrates that the clade has not always been supported by the same suite of appendicular skeletal characters; in other words, there is no universal agreement among researchers as to which morphologies of the girdles and limbs are diagnostic of titanosauria.In recent years, phylogenetic studies have included additional taxa and characters, and as a result, some previouslyproposed titanosaurian synapomorphies are now thought to characterize either more inclusive or less inclusive clades.the appendicular skeletal character states that were proposed as synapomorphies of titanosauria and several of its subclades in the
In the phylogenetic study of González riga and Ortiz David (2014), titanosauria was supported by two unambiguous appendicular synapomorphies: absence of well-developed distal phalangeal articular facets on metacarpals (character 71, state 1), and humerus/femur length ratio less than 0.9 (character 77, state 1).Both of these traits are absent in the non-titanosaurian titanosauriform Ligabuesaurus (Bonaparte et al. 2006).In their revision of another non-titanosaurian titanosauriform, Chubutisaurus, Carballido et al. (2011:104) estimated a humerus/femur ratio of 0.86 for this taxon, proposing that a value of less than 0.8 was characteristic of titanosaurs.
the absence of ossified manual phalanges was proposed as a synapomorphy of Opisthocoelicaudiinae by Wilson (2002:character 181, state 2), although Opisthocoelicaudia possesses at least one vestigial phalanx on manual digit IV (Borsuk-Bialynicka 1977:31).Previously, Salgado et al. (1997) had proposed this morphology as diagnostic of their 'titanosauridae' (a clade that is largely similar to what is now known as Lithostrotia).Salgado et al. (1997) indicated that the absence of manual phalanges should be evaluated based on the morphology of the distal articular facets of the metacarpals, due to the likelihood that such phalanges could easily be lost due to the taphonomic process of necrokinesis.Similarly, Giménez (1992) proposed to examine the distal ends of metacarpals to assess the presence of manual phalanges, since in several derived titanosaurs the metacarpals exhibit roughened, flattened distal surfaces rather than convex articular   facets.As was described by Apesteguía (2005), there seems to have been a progressive reduction of the manual phalanges from basal titanosauriforms to derived titanosaurs.We observe at least two manual morphologies in titanosaurs that are pertinent to discussions of the presence or absence of ossified phalanges in these sauropods: (1) metacarpals with well-defined distal articular facets and curved metacarpal I, as evidence of manual phalanges, and (2) metacarpals with poorly-defined distal articular facets, with or without vestigial ossified manual phalanges.In the first case, the presence of manual phalanges is supported both by well-defined distal articular facets on the metacarpals and the discovery of manual phalanges associated with the specimens in question.this is the case for two early Late Cretaceous Australian taxa, Diamantinasaurus and Savannasaurus, described by hocknull et al. ( 2009) and Poropat et al. (2015Poropat et al. ( , 2016)), respectively.In Diamantinasaurus, the manual elements were not preserved in articulation but have been tentatively An Acad Bras Cienc (2019) 91(Suppl.2) e20180374 31 | 42 interpreted as indicative of a manual phalangeal formula of 2-1-1-1-1.In Savannasaurus, Poropat et al. (2016) recognized at least two manual phalanges, though these authors did not specify to which digits these bones pertained.these important discoveries hint at the existence of diverse manual structures within titanosauria, though articulated specimens are needed to confirm certain aspects of their anatomy (e.g., the proposed phalangeal formula of Diamantinasaurus).Additional titanosaurian taxa such as Andesaurus and Argyrosaurus exhibit indirect evidence of manual phalanges, though these bones have yet to be discovered in these taxa (Mannion andCalvo 2011, Mannion andOtero 2012).In these two large Patagonian titanosaurs, metacarpal I is curved and somewhat 'bananashaped,' similar to that of the basal titanosauriform Janenschia (Apesteguía 2005:334), thereby suggesting the possible presence of one or more diminutive manual phalanges (e.g., Figure 5c, Argyrosaurus).
In the second case, the metacarpals have poorly developed distal articular facets, and in two genera, Epachthosaurus and Opisthocoelicaudia, a rudimentary phalanx is present on digit IV. though both of these taxa are represented by fully articulated postcranial skeletons, there is no evidence of other ossified manual phalanges (Borsuk-Bialynicka 1977, Martínez et al. 2004).Similarly, in the unnamed Invernada taxon (González riga et al. 2008a), no manual phalanges were discovered, although much of the skeleton was exquisitely preserved, including the articulated left fore-and hind limbs with all metacarpals and the complete pes (B.J.G.r.pers.obs.).An important feature of these taxa is that the metacarpals are in contact distally, forming a structure that is more tubular than the metacarpus of other neosauropods (Figure 5c, Epachthosaurus).In this context, the absence of ossified manual phalanges cannot be used as positive evidence (i.e., that these bones were present and subsequently removed by taphonomic processes, as suggested by Poropat et al. 2016Poropat et al. :1012)), nor as negative evidence of genuine absence (e.g., Salgado et al. 1997, Wilson 2002).to more rigorously evaluate this character in a given titanosaurian taxon, one must consider the presence or absence of distal articular facets on the metacarpals and the taphonomic context of known specimens.
In the analysis of González riga et al. ( 2018), Lognkosauria was diagnosed by eight synapomorphies, although none of these were regarded as unique to the clade.two of these morphologies (a deep spinodiapophyseal fossa on the lateral surface of the base of the neural spine in posterior cervical vertebrae and laterally expanded posterior cervical neural spines resulting from expansion of the lateral lamina) are also present in the North American titanosaur Alamosaurus (tykoski and Fiorillo 2017).From their analysis of  Carballido et al. (2017), in their study of the giant titanosaur Patagotitan, analyzed a dataset of 405 characters and 87 sauropodomorph taxa (including 28 titanosaurs) that was modified from the matrix of Carballido and Sander (2014).In contrast to most previous studies, Malawisaurus was not recovered as a comparatively basal titanosaur, but instead was placed in a position more derived than Lognkosauria and rinconsauria.Because Malawisaurus is included in the definition of Lithostrotia (Upchurch et al. 2004), under this phylogenetic hypothesis, members of both Lognkosauria and rinconsauria would be considered to be non-lithostrotian titanosaurs (Figure 10b).In Carballido et al.'s (2017) topology, the colossal Argentinosaurus and Patagotitan are again nested within Lognkosauria, this time accompanied by the similarly gigantic Puertasaurus.Another enormous titanosaur, Notocolossus, is placed as the sister taxon of Lognkosauria, and two main lineages are recovered within Carballido et al.'s (2017) Eutitanosauria: a 'lithostrotian line' that includes Saltasauridae and a second lineage that includes Lognkosauria and rinconsauria.
the existence of a clade that includes undisputed members of Lognkosauria and rinconsauria was previously recovered by tykoski and Fiorillo (2017) based on the data matrix of González riga and Ortiz David (2014) (Figure 10c).Similarly, Gallina and Apesteguía (2011) also recovered this clade, termed 'node A' in their analysis (Figure 10a).these authors based their study on the dataset of Calvo et al. (2007a) and González riga et al. (2009), but they added new cranial and postcranial characters, as well as some taxa.
relevant taxa such as Futalognkosaurus, M e n d o z a s a u r u s , M u y e l e n s a u r u s , a n d Rinconsaurus have been excluded from other recent and pertinent phylogenetic analyses, and as such, it is difficult to further evaluate the existence of the new clade proposed herein.however, although Mendozasaurus and Rinconsaurus were not included in the phylogenetic study of the bizarre Australian titanosaur Savannasaurus (Poropat et al. 2016, dataset of 297 characters and 72 taxa), the new group is supported by the recovery of a Muyelensaurus + (Epachthosaurus + Futalognkosaurus) clade.In this case, the new clade is independent from Nemegtosauridae and Saltasauridae (Poropat et al. 2016, fig. 7).Similarly, although Muyelensaurus and Rinconsaurus were not included in González riga et al.'s (2016) phylogenetic analysis of Notocolossus (dataset of 350 characters and 33 taxa), a distinct, well-defined clade that includes the matrix was analyzed under equal character weighting using tNt (tree analysis using New technology) v. 1.1 (Goloboff et al. 2008).the multistate characters 11, 14, 15, 27, 40, 51, 104, 122, 147, 148, 177, 195, 205, and 259 were treated as ordered.In addition, eleven highly incomplete taxa (and therefore unstable) were excluded prior to the analysis (specimen AODF 836, Astrophocaudia, Australodocus, Brontomerus, Fukuititan, Fusuisaurus, Huanghetitan, ´Huanguhetitan´ ruyangensis, Liubangosaurus, Mongolosaurus and Tendaguria).First, the data matrix was analyzed using New technology Search with the functions 'sectorial searches', 'drift' and 'tree fusing'.It was also used 'get tree' from 'driven search' and 'find minimum length' three times.Second, the resultant trees were searching by traditional Search using the option ´tree bisection-reconstruction´. this process resulted in 660 MPts of 1741 steps and produced a fairly well-resolved strict consensus tree (Consistency Index, 0.248; retention index, 0.560).the strict consensus of these generates a polytomy of basal titanosaurians but recovers the phylogenetic relationships of lithostrotian taxa within two primary clades: a lineage containing Lognkosauria and rinconsauria, herein termed Colossosauria, and a clade containing Saltasaurus and other derived taxa.
PhYLOGENEtIC DEFINItION OF COLOSSOSAUrIA Four recent phylogenetic analyses based on largely independent datasets (Gallina and Apesteguía 2011, Carballido et al. 2017, tykoski and Fiorillo 2017, and González riga et al. 2018) recovered a clade that includes Lognkosauria and rinconsauria.Moreover, the results of an amended phylogenetic analysis based on the dataset of González riga et al. (2018) that includes three additional appendicular characters also supports the existence of this group.Accordingly, we herein propose a new taxon following the tenets of phylogenetic taxonomy (Sereno 2005, Cantino andde Queiroz 2010), as follows: Colossosauria new taxon Etymology.From the ancient Greek colossos, colossus, giant, in reference to the gigantic size of some genera within the clade; from the Greek saurus, lizard, reptile.Definition.Colossosauria is phylogenetically defined as the most inclusive clade containing Mendozasaurus neguyelap but not Saltasaurus loricatus or Epachthosaurus sciuttoi (stem-based).Specifiers.Mendozasaurus neguyelap González riga, 2003, Saltasaurus loricatus Bonaparte and Powell, 1980, Epachthosaurus sciuttoi Powell, 1990.Taxa.Following González riga et al. (2018)  discoveries of new specimens that preserve the appendicular skeleton in its entirety, or nearly so.these fossils have provided a wealth of new anatomical and paleobiological information on titanosaurs, a group that is characterized by marked variation in both body size and morphology.An overview of the appendicular skeletal morphology of titanosaurian taxa from South America indicates that this part of the skeleton exhibits greater anatomical diversity than was initially appreciated.Detailed comparative studies of each of these appendicular elements and skeletal regions would likely yield additional character information useful for phylogenetic analyses.From a systematic point of view, the present review shows that there is no definitive consensus on the appendicular character states that serve as synapomorphies for titanosauria and Lithostrotia.however, within titanosauria, recent phylogenetic analyses confirm the presence of a titanosaurian lineage that differs from the 'Saltasaurinae line.'Four recent cladistic studies based on largely independent datasets have recovered this new clade, which is herein termed Colossosauria.At present, this clade is mainly comprised by taxa belonging to rinconsauria and Lognkosauria, the latter including several exceptionally gigantic species.

Figure 11 -
Figure 11 -Strict consensus cladogram (limited to titanosauria) generated from a revised analysis of the data matrix of González riga et al. (2018) with the addition of three characters (this paper), showing the position and taxonomic content of the newlyrecognized stem-based clade Colossosauria.Abbreviations: Lo, Lognkosauria, ri, rinconsauria, Sa, Saltasauridae.

,
Mendozasaurus, Muyelensaurus, Wilson and Upchurch 2003)rus, and especially Muyelensaurus, in which the proximal and distal ends are not markedly expanded, to exceedingly robust as in Dreadnoughtus and saltasaurines (Figure4).For example, the humeral robustness index (sensuWilson and Upchurch 2003)of Muyelensaurus is 0.18, whereas in specimens of the saltasaurine Neuquensaurus (e.g., MLP-CS 1049)this index may reach values of up to 0.32.In some South American titanosaurs (e.g., Argyrosaurus, Atacamatitan Lacovara et al. 2014, Bates et al. 2015, González riga et al. 2016, Carballido et al. 2017ite the incomplete nature of the known remains, Notocolossus is likely among the largest titanosaurs discovered thus far.Nevertheless, we recommend caution when estimating the body size of one gigantic titanosaurian taxon versus another.Accurate estimation of body size (e.g., total length, mass, volume) is highly problematic in the largest titanosaurs because, with the exceptions of Dreadnoughtus, Futalognkosaurus, widely varying dimensions postulated in recent works (e.g.,Lacovara et al. 2014, Bates et al. 2015, González riga et al. 2016, Carballido et al. 2017).
An Acad Bras Cienc (2019) 91(Suppl.2) e20180374 14 | 42 longer than those of other giant titanosaurs such as Dreadnoughtus (Figure 4g, Lacovara et al. 2014), and Patagotitan, most of these taxa (e.g., 'Antarctosaurus' giganteus, Argentinosaurus, Paralititan, Puertasaurus, Notocolossus) are represented by very incomplete skeletons (Lacovara et al. 2014).Moreover, given the morphological disparity seen in relatively complete, smallerbodied titanosaurs (e.g., Diamantinasaurus, Epachthosaurus, Isisaurus, Mendozasaurus, Opisthocoelicaudia, Rapetosaurus, Saltasaurus), it is probable that different gigantic species also had markedly different anatomical proportions, such as the relative proportions and robusticity of the limb elements, the lengths of the cervical and caudal series, and the distance from the glenoid to the acetabulum.Indeed, this contention has already been borne out, at least to some degree, by the substantially different humeral proportions of the giant titanosaurs Dreadnoughtus, Notocolossus, Paralititan, and Patagotitan (Figure 4g-j), with Notocolossus and especially Dreadnoughtus possessing exceedingly robust humeri, that of Paralititan being more slender, and that of Patagotitan exhibiting an intermediate condition.Although issues surrounding body size estimation in the largest titanosaurs have been partly ameliorated by recent discoveries of relatively complete skeletons of Dreadnoughtus, Futalognkosaurus, and Patagotitan, additional, similarly complete giant titanosaur specimens will be needed to definitively assess the Quetecsaurus, Saltasaurus, and Tapuiasaurus, the former bone is additionally known in Uberabatitan and the latter in Narambuenatitan and Pitekunsaurus (table I).With the exceptions of a few taxa (e.g., Dreadnoughtus, Elaltitan) the distal mediolateral breadth of the radius is approximately twice its breadth at midshaft.In saltasaurid titanosaurs (and probably some other neosauropods, Upchurch et al. 2015), the distal radius is also beveled approximately 20° proximolaterally relative to the long axis of the shaft (Wilson 2002, D'Emic 2012).
Muyelensaurus compared to the condition in other forms such as Rinconsaurus and the saltasaurines Rocasaurus and Saltasaurus.