Sarmientosaurus, AS A

Poropat, Stephen F, Kundrát, Martin, Mannion, Philip D, Upchurch, Paul, Tischler, Travis R & Elliott, David A, 2021, Second specimen of the Late Cretaceous Australian sauropod dinosaur Diamantinasaurus matildae provides new anatomical information on the skull and neck of early titanosaurs, Zoological Journal of the Linnean Society 192 (2), pp. 610-610 : 610-

publication ID

https://doi.org/ 10.1093/zoolinnean/zlaa173

persistent identifier

https://treatment.plazi.org/id/E00087DE-FFBE-FF8A-FF5B-FD68DEECEAB0

treatment provided by

Felipe

scientific name

Sarmientosaurus
status

 

SARMIENTOSAURUS AS A DIAMANTINASAURIAN

All phylogenetic analyses conducted in this study returned Sarmientosaurus in a polytomy with Savannasaurus and the two specimens of Diamantinasaurus . Thus, based on our results and our definition of Diamantinasauria, Sarmientosaurus is a member of this newly erected clade. If correct, this would provide fresh support for previously hypothesized interchange between titanosaurians in Australia and South America during the mid- Cretaceous ( Poropat et al., 2016; see below).

Sarmientosaurus is represented by a virtually complete skull and a series of fragmentary anterior– middle cervical vertebrae. Consequently, the type and only known specimen of this taxon does not overlap anatomically with the type specimens of either Diamantinasaurus or Savannasaurus. However, it can be compared with the referred specimen of Diamantinasaurus (AODF 836) described herein. As outlined above, numerous features are shared between Sarmientosaurus and AODF 836: (1) the quadrate fossa faces posterolaterally; (2) the parietal occipital process is slightly taller than the foramen magnum; (3) the occipital fossa is restricted to the medial half of the parietal; (4) the basal tubera are narrowly divergent (~40°); (5) the laterosphenoid hosts more than one opening for the trigeminal nerve (CN V; two in Diamantinasaurus , three in Sarmientosaurus ); (6) the anterior semicircular canal is much larger than its posterior counterpart; and (7) the cervical centra possess prominent lateral pneumatic foramina [also seen in the sole preserved posterior cervical vertebra of Savannasaurus ( Poropat et al., 2020a)]. Some of these features are more widespread within Titanosauria (e.g. a posterolaterally facing quadrate fossa), whereas others appear to be plesiomorphic for that clade (e.g. prominent lateral pneumatic foramina in cervical centra). By contrast, at least one of these features (the presence of more than one opening on each side for CN V) has otherwise only ever been observed in Sarmientosaurus among neosauropods ( Martínez et al., 2016).

Despite the similarities between Sarmientosaurus and Diamantinasaurus listed above, there are also differences. For example, the braincase of Sarmientosaurus possesses two exits for CN XII, whereas Diamantinasaurus has only one; thus, the latter displays the morphology once considered ‘typical’ for Titanosauria ( Paulina Carabajal, 2012), albeit one that has since been shown to have a much more complex distribution within that clade ( Knoll et al., 2019). More notable, however, are the apparent similarities and differences between the cervical vertebrae. The axis of AODF 836 is short anteroposteriorly relative to its dorsoventral height, seemingly contrasting with the incomplete axis of Sarmientosaurus , which, as described, is long and low, with substantially longer postzygapophyses than observed in other sauropod axes ( Martínez et al., 2016). The axis of Sarmientosaurus was also described as lacking lateral pneumatic fossae ( Martínez et al., 2016), which, if true, would constitute a clear difference between Sarmientosaurus and Diamantinasaurus . However, the incompleteness of the axis in Sarmientosaurus might mean that these structures were simply not preserved, rather than genuinely absent. The axial neural spine of Sarmientosaurus is substantially shorter than that of Diamantinasaurus , although again this might be because it is incomplete.

The two most complete cervical vertebrae in Sarmientosaurus were interpreted as the sixth and seventh by Martínez et al. (2016). Each is reasonably well preserved and (in detail, at least) similar to the middle cervical vertebrae of Diamantinasaurus . A lateral pneumatic excavation, dorsally bounded by a horizontal lamina, is present on each side of the centrum of cervical six in both Sarmientosaurus ( Martínez et al., 2016) and Diamantinasaurus ( Fig. 21 View Figure 21 ). Martínez et al. (2016) regarded the presence of ‘centroprezygapophyseal pillars’, i.e. posteriorly unsupported CPRLs, as an autapomorphy of Sarmientosaurus . However, as discussed by those authors, this might be an artefact of preservation. In the middle cervical neural arch of Diamantinasaurus ( Fig. 22 View Figure 22 ), the formation of a PRCDF immediately posterior to the CPRL means that the bone in this region is extremely thin. It is not implausible that a similarly delicate sheet of bone was lost in the middle cervical vertebrae of Sarmientosaurus ( Martínez et al., 2016) . Finally, in the middle cervical vertebrae of both Sarmientosaurus and Diamantinasaurus , the apex of the neural spine is situated well posterior to the midlength of the centrum. Despite these similarities, there is one obvious difference between the middle cervical vertebrae of Diamantinasaurus and Sarmientosaurus : their degree of elongation. In cervical six of Sarmientosaurus , the average elongation index is 4.71, whereas in cervical six of Diamantinasaurus it is 2.67. If the serial positions of all cervical vertebrae concerned are approximately correct, this represents a substantial incongruence. Simply put, it would imply that Sarmientosaurus had a far longer neck, relative to body size, than Diamantinasaurus .

Assuming that Sarmientosaurus is a diamantinasaurian, it would be reasonable to expect some degree of morphological divergence from Australian relatives, given its spatial (and minor temporal) separation from them. However, it is also possible that these anatomical differences indicate a more distant relationship between Sarmientosaurus and diamantinasaurians than the results of the phylogenetic analyses presented herein suggest. Under this scenario, many of their seemingly shared features would be more widespread amongst earlydiverging titanosaurs (or somphospondylans slightly outside of Titanosauria), but the incompleteness of early titanosaur fossils, coupled with the absence of these features in better-preserved early-diverging somphospondylans and later-branching titanosaurs, results in their false recovery as diamantinasaurian synapomorphies. Consequently, we tentatively support Sarmientosaurus as a diamantinasaurian, pending the discovery of more complete postcranial material of this genus, in addition to increased sampling of early titanosaurs in general.

PALAEOBIOGEOGRAPHICAL IMPLICATIONS OF DIAMANTINASAURIA

The origins and biogeographical history of mid- Cretaceous Australian dinosaur faunas have proved to be controversial (e.g. Agnolin et al., 2010; Barrett et al., 2011; Benson et al., 2012; Novas et al., 2013). This topic was reviewed by Poropat et al. (2015b). Here, we focus our discussion on developments during the past 5 years. Although potentially an oversimplification, it is convenient to divide explanations of the origin of Australian Cretaceous dinosaur faunas into two broad categories. The first set of hypotheses argues for the existence of largely cosmopolitan dinosaur clades during the Jurassic, with regional differences developing in the Cretaceous as a result of extinction ( Barrett et al., 2011; Benson et al., 2012; Fitzgerald et al., 2012; Rich et al., 2014). Curiously, under this scenario it seemed that Australian faunas were more similar to those from Laurasia, especially East Asia, than they were to those from other parts of Gondwana (e.g. Benson et al., 2012). One potential explanation for apparent similarities between mid-Cretaceous Australian and East Asian faunas, despite separation of these areas by thousands of kilometres of ocean, is that these regions had more mesic climates, whereas other Pangaean fragments were more arid ( Benson et al., 2012). For example, if tyrannosauroids preferred higher-latitude, cooler, more humid environments, whereas abelisauroids preferred hotter and drier climates, this could explain observations such as the presence of the latter clade in South America, Africa, and so on, and their absence in Australia, in the mid- Cretaceous ( Benson et al., 2012). The second set of hypotheses focus on closer biotic similarities between Australia and other parts of Gondwana, especially South America ( Molnar, 1992; Upchurch et al., 2002; Smith et al., 2008; Agnolin et al., 2010; Herne et al., 2010; Novas et al., 2013; Poropat et al., 2015b, 2016; Bell et al., 2016). These explanations invoke trans-Antarctic dispersal in the mid-Cretaceous (often mediated by latitudinal shifts in climate) and vicariance.

The first point to note is that, despite occasionally strident debate in the literature, these two broad categories of hypotheses are not mutually exclusive. It is possible for a major clade, such as Titanosauria, to be cosmopolitan, while simultaneously containing less-inclusive clades with more restricted Gondwanan or trans-Antarctic distributions. Likewise, there is nothing in biogeographical theory to suggest that regional extinction and vicariance can never work together to create differences between biotas. The goal of palaeobiogeographers working on this issue is thus not to overturn one of these broad hypothesis sets, but to disentangle their combined role in producing Australian dinosaur faunas.

Recent Australian discoveries, although still often fragmentary, have tended to strengthen the case for strong biotic affinities between Australia and South America during the mid-Cretaceous. For example, Brougham et al. (2020) described a cervical vertebra from the Griman Creek Formation, near Lightning Ridge, which they identified as the first evidence for the presence of the theropod clade Noasauridae in Australia. Noasauridae is currently known only from Gondwanan landmasses ( Poropat et al., 2020b), and the Griman Creek cervical vertebra shares derived features that (at present) are seen elsewhere only in Noasaurus Bonaparte & Powell, 1980 from the Maastrichtian of Argentina ( Brougham et al., 2020). However, caution is required at this stage, because other noasaurids are so incomplete that it is difficult to evaluate the true phylogenetic significance of these potential synapomorphies. This illustrates a common problem: many of the Australian dinosaur specimens that lie at the heart of the current controversy are incomplete and thus have uncertain affinities [e.g. compare Agnolin et al. (2010) and Novas et al. (2013) with Benson et al. (2012) and Rich et al. (2014)]. For example, Barrett et al. (2011) proposed that an isolated cervical vertebra belonged to an Australian spinosaurid (see also Benson et al., 2012), whereas Novas et al. (2013) reassessed each of the relevant characters and concluded that it could not be identified beyond indeterminate Averostra or Tetanurae. More recently, Poropat et al. (2019) described new Australian megaraptorid remains, allowing them to demonstrate that the putative spinosaurid cervical was more plausibly identified as a megaraptorid.

Until relatively recent times, the mid-Cretaceous Australian dinosaur fossil record largely lacked wellpreserved specimens that could be incorporated into phylogenetic data sets and thus inform this biogeographical debate. Clearly, phylogenetic topologies themselves are subject to debate and change and therefore do not offer a panacea, but they can yield more secure grounds for testing biogeographical hypotheses than do isolated specimens that can be compared only on the basis of a handful of character states. Substantial progress has been made over the past 5–10 years, with a number of more complete dinosaur specimens being reported from Australia and added to phylogenetic data sets (e.g. Poropat et al., 2015b, 2016; Bell et al., 2016, 2019; Herne et al., 2018). These studies have typically demonstrated that although mid-Cretaceous Australian dinosaurs are indeed usually members of large clades with virtually global distributions, they also tend to have their closest relatives among Gondwanan taxa, notably often those from South America. In particular, Herne et al. (2019) found evidence in support of a Gondwanan clade of elasmarian ornithopods that included a cluster of small-bodied taxa from the Cretaceous of Australia and South America. This clade potentially also encompasses taxa from Antarctica ( Rozadilla et al., 2016; Herne et al., 2019). However, caution is again warranted because this elasmarian clade had only weak support, and the larger-bodied Australian iguanodontian Muttaburrasaurus Bartholomai & Molnar, 1981 did not display close affinities with exclusively Gondwanan taxa ( Herne et al., 2019). Bell et al. (2019) described a new early-branching iguanodontian, Fostoria dhimbangunmal Bell et al., 2019, from the Cenomanian Griman Creek Formation. Their phylogenetic analysis placed Fostoria Bell et al., 2019 as the sister taxon to a Gondwanan clade that included Muttaburrasaurus in addition to Anabisetia Coria & Calvo, 2002 from the Turonian of Argentina and Talenkauen Novas et al., 2004 from the Campanian–Maastrichtian of Argentina ( Rozadilla et al., 2019). Once more, the phylogenetic topology, although suggestive, was regarded as weakly supported by Bell et al. (2019), and these authors considered it premature to infer any biogeographical implications at that time. As an example of the labile nature of ornithopod relationships, a recent analysis recovered Muttaburrasaurus and Fostoria as the earliest-diverging members of Rhabdodontomorpha, otherwise known only from Europe, whereas only South American and Antarctic species (including Anabisetia and Talenkauen ) were recovered within Elasmaria ( Dieudonné et al., 2020).

Australovenator wintonensis Hocknull et al., 2009 is the most completely preserved non-avian theropod currently known from Australia ( White et al., 2012, 2013, 2015). It is generally accepted that this taxon is a megaraptoran ( Novas et al., 2013; Bell et al., 2016; Poropat et al., 2019), although its exact position within this group is debated (Lamanna et al., 2020). Older phylogenetic studies placed Australovenator Hocknull et al., 2009 as the sister taxon of Fukuiraptor (Azuma & Currie, 2000) from Japan ( Benson et al., 2010). However, more recently there has been support for a monophyletic Gondwanan Megaraptoridae that includes Megaraptor Novas, 1998 from Argentina and Australovenator , with Fukuiraptor as sister taxon to this clade ( Novas et al., 2013; Bell et al., 2016; Porfiri et al., 2018). Bell et al. (2016) applied quantitative biogeographical methods (Statistical Dispersal- Vicariance Analysis [S-DIVA] and Bayesian Binary Markov [BBM]) to their phylogenies and found that Megaraptora originated in Laurasia and then dispersed into Gondwana, where it gave rise to Megaraptoridae. Australian and South American megaraptorids are each other’s closest relatives and imply at least one trans-Antarctic dispersal event between ~102 and 92 Mya. By contrast, recent work on tyrannosauroids found that the mid-Cretaceous South American Santanaraptor Kellner, 1999 was not the closest relative of Australian forms such as Timimus hermani Rich & Vickers-Rich, 1994 , a result that is more consistent with an early Pangaean radiation of this clade in the Jurassic, followed by regional extinction ( Delcourt & Grillo, 2018). However, given that only one tyrannosauroid specimen from South America and two from Australia are currently known [assuming that megaraptorans are not tyrannosauroids ( Novas et al., 2016; Porfiri et al., 2018)], sampling failure could easily have obscured trans-Antarctic relationships.

In this context, the phylogenetic results of the present work contribute further support to the hypothesis that, at least at finer taxonomic levels, Cretaceous dinosaurian faunas of Australia often display affinities with those of South America. In preliminary phylogenetic analyses, Diamantinasaurus clustered with latest Cretaceous Asian taxa, such as Opisthocoelicaudia ( Hocknull et al., 2009) , and apparently added evidence in favour of cosmopolitanism or closer biotic affinities between Australia and Laurasia than with other Gondwanan areas (e.g. Barrett et al., 2011). This position shifted slightly as a result of the discovery of more material of Diamantinasaurus , the addition of Savannasaurus and updated phylogenetic work ( Poropat et al., 2015b, 2016; Mannion et al., 2017). In the study by Mannion et al. (2017), in particular, Australian titanosaurs formed a monophyletic group that was the sister taxon of a larger clade containing some Laurasian taxa and a large number of Gondwanan, taxa. Moreover, Poropat et al. (2016) applied the quantitative biogeographical method BIOGEOBEARS, a maximum likelihood approach that estimates ancestral areas ( Matzke, 2013, 2014). These analyses suggested that a large clade of somphospondylans had become widespread across much of Pangaea during the Jurassic and earliest Cretaceous, with subsequent faunas differentiating as a result of both regional extinction and dispersal. Poropat et al. (2016) combined data on fossil record sampling, climatic shifts and biogeographical history to infer dispersal from South America to Australia in the late Albian or later. This event coincided with global warming that resulted in a southward shift of more temperate conditions, potentially increasing the feasibility of the dispersal of sauropods across Antarctica.

The Australian Cretaceous sauropod record comprises footprints demonstrating that sauropods lived in north-west Australia during the Valanginian– Barremian ( Thulborn et al., 1994; Thulborn, 2012; Salisbury et al., 2017) and body fossils that evince the presence of somphospondylan titanosauriforms (including non-titanosaurian somphospondylans and early-branching titanosaurs) in north-east Australia (Queensland and New South Wales) throughout the late Albian–Cenomanian ( Longman, 1933; Coombs & Molnar, 1981; Molnar, 2001, 2010, 2011a, b; Molnar & Salisbury, 2005; Hocknull et al., 2009; Poropat et al., 2015a, b, 2016, 2017, 2020a). Sauropods are unknown in the Cretaceous of south-central and south-east Australia, despite abundant evidence of ornithopod, ankylosaurian and theropod dinosaurs in Barremianto Albian-aged deposits in this region ( Rich & Rich, 1989; Rich & Vickers-Rich, 1999; Barrett et al., 2010a, b; Benson et al., 2012; Herne et al., 2018, 2019; Poropat et al., 2018, 2019, 2020b). Consequently, sauropods are presumed to have been genuinely absent in this region at this time. This implies that the Albian–Cenomanian sauropods of north-east Australia descended either from lineages that persisted in northern Australia from the Barremian until the Albian [specifically, in areas not submerged by the Eromanga Sea ( Cook et al., 2013)] or from sauropods that entered Australia during or after the Albian. By that time, the only other continents to which Australia was connected were Zealandia to the east ( Mortimer et al., 2017) and Antarctica (via Tasmania) to the south-east ( Seton et al., 2012). The fact that the Cenomanian-aged Diamantinasaurus (represented by AODF 603 and AODF 836) and Savannasaurus (AODF 660) from Australia form a clade with the Cenomanian- to Turonian-aged Sarmientosaurus musacchioi from South America implies interchange of early-branching titanosaurians across Antarctica slightly before or during the Cenomanian stage ( Poropat et al., 2016). As with the phylogenies of ornithopods and theropods discussed above, our current topology offers only relatively weak support (Bremer support = 3) for a sister-taxon relationship between a South American form ( Sarmientosaurus ) and Australian titanosaurs ( Diamantinasaurus and Savannasaurus), reflecting the incompleteness of the postcranial skeleton of the former taxon, in particular. Nevertheless, our currently best-supported topology ( Fig. 28 View Figure 28 ) represents a further example of the shift towards discovery of close biotic affinities among the mid-Cretaceous dinosaurs of South America and Australia, and it directly supports the inferred presence of ancestors of the Australian titanosaurs in South America predicted by the BIOGEOBEARS results of Poropat et al. (2016).

In summary, the dichotomy between cosmopolitanism plus regional extinction and close biotic affinities with South America, as explanations of the origins of mid-Cretaceous Australian dinosaurian faunas, is a false one. Current evidence does suggest that, at higher taxonomic levels, Australian dinosaurs are often members of widespread clades. However, at lower levels (i.e. typically, relationships between small clusters of genera or species) a picture of close affinity with Gondwanan taxa, especially those in South America, is beginning to emerge. This matches the general pattern noted by Upchurch et al. (2002), who proposed that statistically significant continent-scale vicariance signals seem to manifest themselves at the generic level among dinosaurs. Although caution is required because of the perennial issues of sampling failure and phylogenetic instability in biogeographical reconstructions (e.g. Mannion et al., 2019b; Kubo, 2020), recent analyses have started to find evidence for South America– Australia sister-taxon relationships among ornithopods, megaraptorids and titanosaurs. This runs counter to the claims of a lack of such evidence in earlier studies that supported cosmopolitanism plus regional extinction (e.g. Barrett et al., 2011; Benson et al., 2012; Fitzgerald et al., 2012). The more recent phylogenetic results also tend to undermine the hypothesis that Australian and South American dinosaurian faunas differentiated in the Cretaceous as a result of developing more mesic and more arid climatic regimens, respectively (e.g. Benson et al., 2012). The interpretation proposed here is consistent with the recent biogeographical analyses of Kubo (2020), who used a phylogenetic network approach and a dinosaurian supertree to demonstrate that Cretaceous Australian dinosaur faunas have their closest links to South America, within a larger Gondwanan set of biotic affinities. It will be important to test these ideas further via discoveries of more complete specimens of key taxa, such as Sarmientosaurus , combined with the ongoing process of adding and revising phylogenetic characters.

THE TIMING AND SELECTIVITY OF TRANS- ANTARCTIC SAUROPOD DISPERSALS

In neither Australian nor southern South American (Patagonian) Cenomanian faunas were diamantinasaurians the only sauropods present. In addition to Sarmientosaurus , the lower Bajo Barreal Formation of southern Argentina has yielded the rebbachisaurid diplodocoid Katepensaurus goicoecheai Ibiricu et al., 2013 ( Ibiricu et al., 2012, 2013, 2015) and the early-branching titanosaur Epachthosaurus sciuttoi Powell, 1990 ( Martínez et al., 2004; Ibiricu et al., 2020). Although no teeth are presently known for Katepensaurus Ibiricu et al., 2013 , phylogenetic bracketing implies that it, like all diplodocoids, had narrow-crowned teeth ( Whitlock, 2011a). In support of this hypothesis, a narrow-crowned sauropod tooth from the lower Bajo Barreal Formation (UNPSJB-PV 847), originally assigned to Titanosauridae ( Powell et al., 1989) , has recently been reinterpreted as being from a rebbachisaurid ( Alvarez et al., 2019). Thus, sauropods with both narrow- and (relatively) broadcrowned (i.e. Sarmientosaurus ) teeth coexisted in the earliest Late Cretaceous of Patagonia. Lack of anatomical overlap between Epachthosaurus (which preserves most of the postcranial skeleton, except for the neck) and Sarmientosaurus (known only from the skull and anterior–middle cervical vertebrae) currently precludes comparison of these two taxa. However, as pointed out by Martínez et al. (2016), rare sauropod cranial elements from the lower Bajo Barreal Formation [e.g. the maxilla UNPSJB-PV 583 ( Sciutto & Martínez, 1994)] support the notion that at least one other titanosaurian taxon, with slightly narrower-crowned teeth than Sarmientosaurus , lived alongside it.

In contrast to the lower Bajo Barreal Formation of Patagonia, the Winton Formation (and, indeed, the Australian fossil record as a whole; Frauenfelder et al., 2020) preserves no evidence of rebbachisaurid sauropods, nor of any sauropods with narrowcrowned teeth. Although future discoveries might change this, based on current evidence the only nondiamantinasaurian sauropod known from the Winton Formation is the non-titanosaurian somphospondylan Wintonotitan wattsi ( Hocknull et al., 2009; Poropat et al., 2015a). Rigorous assessment of niche partitioning between Wintonotitan and diamantinasaurians is precluded by the incompleteness of the former and by the fact that no sauropod teeth or dentulous elements have yet been reported from the Winton Formation.

Whether or not we should expect to find sauropods with narrow-crowned teeth in the Winton Formation (specifically, diplodocoids) remains an open question. Before the Jurassic–Cretaceous transition, flagellicaudatan diplodocoids (dicraeosaurids and diplodocids) would have been able to enter Australia from the west, because Madagascar and India formed a contiguous land area between eastern Africa, Antarctica and south-west Australia ( Seton et al., 2012). Given that dicraeosaurids and diplodocids were both thriving in eastern Africa ( Janensch, 1914, 1929, 1935 – 1936, 1961; Remes, 2006, 2009; Schwarz- Wings & Böhm, 2014) and Patagonia ( Rauhut et al., 2005, 2015; Salgado et al., 2015b) during the latest Jurassic and persisted into the earliest Cretaceous of southern Africa ( McPhee et al., 2016) and Patagonia ( Gallina et al., 2014, 2019; Paulina Carabajal et al., 2018; Coria et al., 2019; Windholz et al., 2020), they would have had ample opportunity to enter Australia at this time. However, by the end of the Barremian, both diplodocids and dicraeosaurids appear to have gone extinct worldwide; the dicraeosaurids of the La Amarga Formation are the geologically youngest flagellicaudatans known ( Salgado & Bonaparte, 1991; Salgado & Calvo, 1992; Apesteguía, 2007; Gallina, 2016; Windholz et al., 2021). Thus, even if flagellicaudatans occupied Australia, they might not have persisted until the mid-Cretaceous. For rebbachisaurids, the scenario was somewhat different. The oldest putative rebbachisaurids date to the latest Jurassic of North America ( Carpenter, 2018) and the earliest Cretaceous of Europe ( Taylor, 2018), implying a northern origin for the clade. Although rebbachisaurids appear not to have persisted into the Cretaceous in North America, they occupied Europe until at least the early Aptian ( Dalla Vecchia, 1999; Pereda Suberbiola et al., 2003; Mannion, 2009; Mannion et al., 2011; Torcida Fernández-Baldor et al., 2011). Rebbachisaurids make their earliest appearances in Afro-Arabia in the late Hauterivian– early Barremian ( Histriasaurus Dalla Vecchia, 1998 from present-day Croatia; Dalla Vecchia, 2005) and South America in the Barremian ( Zapalasaurus Salgado et al., 2006 ), by which time Africa and Indo- Madagascar had detached from Antarctica and each other ( Seton et al., 2012). Based on the distribution of their earliest occurrences, rebbachisaurids are likely to have dispersed into Africa from Europe (via the ‘Apulian route’), then from north-west Africa into northeast South America before or during the Barremian ( Lindoso et al., 2019; Pereira et al., 2020). During the mid-Cretaceous, rebbachisaurids proliferated in northern Africa ( Sereno et al., 1999, 2007; Fanti et al., 2013, 2014, 2015; Mannion & Barrett, 2013; Wilson & Allain, 2015), north-east South America ( Carvalho et al., 2003; Medeiros & Schultz, 2004; Castro et al., 2007; Lindoso et al., 2019; Pereira et al., 2020) and Patagonia ( Calvo & Salgado, 1995; Bonaparte, 1996; Calvo, 1999; Salgado et al., 2004, 2006, 2012; Gallina & Apesteguía, 2005; Apesteguía, 2007; Carballido et al., 2010, 2012; Haluza et al., 2012; Paulina Carabajal et al., 2016; Canudo et al., 2018). Some of the geologically youngest rebbachisaurids in South America ( Ibiricu et al., 2013, 2015) are the highest-latitude (palaeolatitude of ~ 52°S) representatives of the clade ( Ibiricu et al., 2012), but their remains have never been recovered from southern Patagonia. By contrast, titanosaurs are known from the southern extent of Patagonia in the latest Cretaceous (e.g. Lacovara et al., 2014; Novas et al., 2019). Taken at face value, we might therefore infer that rebbachisaurids never ventured into high palaeolatitudes (approaching 60°), as is also supported by their absence from southern Africa ( Mannion & Barrett, 2013). Thus, following this line of reasoning, rebbachisaurids never dispersed as far south as Antarctica and therefore would not have been able to expand into Australia. However, no dinosaur fossils are currently known from southern Patagonia during the mid-Cretaceous, including titanosaurs, and the African record is also extremely patchy (e.g. Benson et al., 2013; Mannion & Barrett, 2013). As such, we cannot be sure that rebbachisaurids were genuinely absent from these high-latitude ranges at this time, with our earliest high-latitude sampling window (Campanian) occurring long after the demise of the group.

If diamantinasaurians could traverse Antarctica to occupy both South America and Australia during the mid-Cretaceous, why could (or did) rebbachisaurids not? Perhaps the barrier to their dispersal was palaeoenvironmental. During theAlbian–Cenomanian, the palaeofloras of the Gondwanan continents (by that time restricted to South America, Antarctica, Australia and Zealandia) were under sufficiently strong palaeolatitudinal control that several distinct floristic provinces have been recognized ( Herngreen et al., 1996; Cantrill & Poole, 2012; Mays, 2014). Lower-latitude Gondwanan floral provinces were characterized by abundant ferns and angiosperms, whereas the highest-latitude (i.e. southernmost) regions were dominated by conifers ( Mays, 2014). A recent comparison of several Gondwanan mid- Cretaceous palynofloras demonstrated that those of the Mata Amarilla Formation in Patagonia are more similar to penecontemporaneous palynofloras of New Zealand and Antarctica than those of the coeval, and geographically more proximal, Cañadón Seco (an equivalent of the lower Bajo Barreal Formation) and Huincul formations from further north in Argentina ( Santamarina et al., 2020). Instead, the latter palynofloras were aligned with those of the Eromanga Basin: the Winton Formation and the underlying Mackunda Formation, Allaru Mudstone and Toolebuc Formation ( Santamarina et al., 2020). Both the lower Bajo Barreal and Huincul formations preserve rebbachisaurids, whereas the Winton Formation does not. Rising temperatures during the late Albian reduced the latitudinal thermal gradient ( Huber et al., 2018) and facilitated the dispersal of angiosperms into the polar regions ( Korasidis et al., 2016; Korasidis & Wagstaff, 2020), along with diamantinasaurian titanosaurs. However, the same warming and floral change does not appear to have facilitated the spread of rebbachisaurids; indeed, it might even have been detrimental for this group (see ‘ Diamantinasaurian palaeoecology ’ section below).

The Australian sauropod record also lacks evidence for titanosaurs with procoelous caudal centra. The geologically oldest titanosaur from South America with strongly procoelous caudal centra is the late Albian Patagotitan mayorum Carballido et al., 2017 . By the Cenomanian–Turonian, titanosaurs with strongly procoelous caudal centra were widespread in Patagonia [e.g. Epachthosaurus sciuttoi ( Martínez et al., 2004) , Drusilasaura deseadensis ( Navarrete et al., 2011) , and Quetecsaurus rusconii ( González Riga & Ortíz David, 2014) ]. Despite their absence in the Cenomanian-aged Winton Formation (and in the Australian record generally), procoelous caudal vertebrae pertaining to titanosaurs have been identified in Campanian- to Maastrichtian-aged deposits in both New Zealand ( Molnar & Wiffen, 2007) and Antarctica ( Cerda et al., 2012). These titanosaurs were almost certainly South American emigrants; if so, their arrival in New Zealand presumably pre-dated the Santonian (~85 Mya) onset of seafloor spreading in the Tasman Sea ( Bache et al., 2014). Before 95 Mya, the same palaeoenvironmental barrier that excluded rebbachisaurids from polar latitudes during the mid-Cretaceous might likewise have impeded the southward dispersal of titanosaurs with procoelous caudal centra, thereby precluding their entry into Australia. However, when that same barrier was lifted between 95 and 85 Mya, derived titanosaurs evidently dispersed, whereas rebbachisaurids were already in terminal decline and soon went extinct.

DIAMANTINASAURIAN PALAEOECOLOGY

The endocranial morphology of Diamantinasaurus is similar to that of Sarmientosaurus . Thus, some of the ecological interpretations inferred for the latter taxon by Martínez et al. (2016), particularly related to feeding height, might be transferable to Diamantinasaurus or to diamantinasaurians generally. However, the two specimens of Diamantinasaurus and the sole specimen of Savannasaurus preserve elements unknown in Sarmientosaurus , thereby facilitating a more complete assessment of the feeding envelope of diamantinasaurians.

Martínez et al. (2016) interpreted Sarmientosaurus to have been a low-level browser, based on the downward tilt of the snout relative to the neck when the skull was oriented in its ‘alert posture’ (determined from the orientation of the semicircular canals relative to the skull overall), and the extensive pneumatization of the cervical vertebrae. Although this interpretation is plausible, both the snout shape and the distribution and morphology of the teeth of Sarmientosaurus are more closely aligned with those of sauropods that are thought to have been higher-level browsers, such as brachiosaurids ( Upchurch & Barrett, 2000). The neck of Diamantinasaurus , based on the vertebrae available, does not appear to have been particularly elongate, potentially supporting a low- to mid-level browsing niche for this taxon. However, the cervical ribs of Savannasaurus are elongate ( Poropat et al., 2020a), a feature more commonly seen in higherbrowsing sauropods ( Upchurch & Barrett, 2000). Furthermore, based on the preserved portions of the limbs of the Diamantinasaurus type specimen ( Poropat et al., 2015b), there would have been little discrepancy between the forelimb, which was ~2.4 m tall (humerus, 1068 mm; ulna, 700 mm; longest metacarpal, 412 mm; plus ~10% cartilage at each joint), and the hindlimb, which was ≥ 2.35 m tall without the pes (femur, 1345 mm; tibia, 795 mm; plus ~10% cartilage at each joint). Forelimb length to hindlimb length ratios of close to 1.0 (rather than close to 0.7) are again characteristic of sauropods often interpreted as medium- or high-level browsers ( Upchurch & Barrett, 2000).

Rebbachisaurids appear to have been the low-level feeders par excellence among sauropods. The highly derived rebbachisaurid, Nigersaurus Sereno et al., 1999 , from the Aptian–Albian Elrhaz Formation of Niger has been identified as a highly specialized lowlevel feeder ( Sereno & Wilson, 2005) that potentially subsisted mainly on horsetails and ferns ( Sereno et al., 2007), based on its anteriorly flattened and expanded Π- shaped jaws, its ‘battery’ of extremely narrowcrowned teeth and the presence of labial wear facets on those teeth (implying abrasion against a flat substrate, i.e. the ground). The few cranial remains known for other rebbachisaurid taxa [notably, Lavocatisaurus agrioensis Canudo et al., 2018 , Limaysaurus tessonei ( Calvo & Salgado, 1995) ( Paulina Carabajal & Calvo, 2015) and an indeterminate rebbachisaurid from the Candeleros Formation of Argentina, MMCh-PV 71 ( Paulina Carabajal et al., 2016)] appear to be broadly similar to those of Nigersaurus , implying that the entire clade was specialized for low-level browsing. Such behaviour might have been suited only to savannahtype biomes ( Whitlock, 2011a); thus, if such habitats were small in extent, non-contiguous or absent at high latitudes during the mid-Cretaceous (as suggested by the heightened diversity and abundance of conifers in palynofloras; see above), the palaeoenvironmental barrier that prevented the dispersal of rebbachisaurids into Australia via Antarctica might have been the conifer-dominated high-latitude floral province.

Assuming, for the sake of argument, that Sarmientosaurus and other diamantinasaurians were low-level browsers, they appear to have been nowhere near as specialized for that way of life as rebbachisaurids or, indeed, some later-branching titanosaurs [e.g. Antarctosaurus wichmannianus ( Huene, 1929; Powell, 2003), Baalsaurus mansillai ( Calvo & González Riga, 2019), Bonitasaura salgadoi Apesteguía, 2004 ( Gallina & Apesteguía, 2011), and Brasilotitan nemophagus Machado et al., 2013 ]. Given that the jaws of Sarmientosaurus are not anteriorly flattened, that the teeth are not entirely (or even mostly) restricted to the front of the mouth and that the downward inclination of the head in ‘alert posture’ was less extreme than in rebbachisaurids ( Sereno et al., 2007), the case for Sarmientosaurus as a specialist low-level browser appears far less robust than that for dicraeosaurids, diplodocids and rebbachisaurids ( Whitlock, 2011a). The microwear patterns observed on the teeth of Sarmientosaurus , comprising grooves that are mostly parallel but that sometimes intersect, in addition to rarer pits of varying size and distributional density ( Martínez et al., 2016), also argue against predominantly low-level feeding in this taxon, and instead are more in line with mid-height (1–10 m) browsing ( Whitlock, 2011a).

If diamantinasaurians were mid-level feeders, they presumably would not have been in direct competition with rebbachisaurids for food. Moreover, as generalist mid-level feeders rather than specialist low-level feeders, they might have been better able to adapt to new environments and food sources than rebbachisaurids. Verification of these tentative hypotheses will have to await the discovery of more complete diamantinasaurian material; however, they could hold the key to why these titanosaurs, but not rebbachisaurids, reached Australia.

CONCLUSION

The complete description of the referred specimen of D. matildae (AODF 836) presented herein, coupled with an expanded phylogenetic analysis, highlights the similarities between it and the type specimen, while concomitantly revealing that Diamantinasaurus forms a clade with the sympatric species Savannasaurus elliottorum and with the contemporaneous South American titanosaur Sarmientosaurus musacchioi . This clade, herein named Diamantinasauria, is presently known only from the Cenomanian–earliest Turonian, but crucially spans both southern South America and north-east Australia. This supports the hypothesis that titanosaurs were able to traverse between these continents across Antarctica during the early Late Cretaceous, while simultaneously suggesting that rebbachisaurids and titanosaurians with procoelous caudal centra (at least initially) were not capable of the same.

Kingdom

Animalia

Phylum

Chordata

Class

Reptilia

Order

Saurischia

Family

Titanosauridae

Loc

Sarmientosaurus

Poropat, Stephen F, Kundrát, Martin, Mannion, Philip D, Upchurch, Paul, Tischler, Travis R & Elliott, David A 2021
2021
Loc

Australovenator wintonensis

Hocknull 2009
2009
Loc

Australovenator

Hocknull 2009
2009
Loc

Santanaraptor

Kellner 1999
1999
Loc

Megaraptor

Novas 1998
1998
Loc

Timimus hermani

Rich & Vickers-Rich 1994
1994
Darwin Core Archive (for parent article) View in SIBiLS Plain XML RDF