The Titanosaurus (in?)validity project Part 4 - Bonatitan

Today's taxon to be compared with Titanosaurus in our continuing series is Bonatitan reigi, described by Martinelli and Forasiepi (2004) as based on two partially preserved individuals, but redescribed by Salgado et al. (2015) as five fragmentary individuals, of which a braincase among the material labeled MACN-PV RN 821 is now the holotype. Other elements under that number and MACN-PV RN 1061 are paratypes, including a mid caudal neural arch probably coming from a smaller individual than the holotype. It derives from the Allen Formation of Argentina.

Titanosaurus indicus lectotype GSI 2191 in ventral (top), anterior (left), left lateral (center) (all after Lydekker, 1879) and left lateral (right) views (after Mohabey et al., 2013; scale = 50 mm).

This case is a bit different from the previous three because we have no preserved mid caudal centra for Bonatitan, and the only somewhat comparable caudal neural arch is from a more proximal position than Titanosaurus. As established before, Titanosaurus shows a more distal-esque morphology than saltasaurines in lacking transverse processes and being narrower compared to its length. But I'm still not sure if this makes Titanosaurus the paravian of titanosaurs or saltasaurines the caenagnathoids of titanosaurs (or both?). Regardless, Bonatitan does have well-developed transverse processes based on the lateral articulations on the neural arch, and the broader morphology expected of a saltasaurine, but its features (tall neural peduncle, prezygapophyses with dorsal projection distally) suggest it's around position 4-5 compared to 6-11 in Titanosaurus. And if there were six vertebrae between these, maybe they could transition from a Bonatitan type to a Titanosaurus type. That being said, the width difference still seems implausible to me and the posterior position of the neural peduncle compared to the anterior centrum edge looks to be less than half as much as Titanosaurus, which has also been a consistent trend in saltasaurines.

Bonatitan reigi paratype MACN-PV RN 1061 in anterior (left) and right lateral (reversed; right) views (after Salgado et al., 2015).

Neuquensaurus' caudal sequence has shown that other neural arch characters like the neural peduncle width and posterior shape between the neural arch and centrum are highly variable within an individual, so we can't use those to distinguish Bonatitan and Titanosaurus. For completeness' sake, there is a ~second caudal vertebra preserved with a deep ventral keel over at least the posterior half of the ventral midline (considered diagnostic by both sets of authors), but no taxon so far has described or illustrated if the first two caudals differ from more distal elements in ventral topography (they often do in theropods).

Is Bonatitan Titanosaurus? Probably not, but we're hindered because due to position and incompleteness they aren't that comparable. In particular, the centra seem to be the most positionally consistent in titanosaurs so far within this portion of the tail, and we just don't have one for Bonatitan. There is a diagnostic character listed by both Martinelli and Forasiepi and Salgado et al. involving the vertebra which could be checked in Titanosaurus- "dorsal to middle caudal vertebrae with deep oval to circular pits on both sides of the prespinal lamina", but the published figures and photo of Titanosaurus don't expose the area well enough. This is also a difficulty in comparing Titanosaurus' neural arches- the lack of a dorsal or posterior view, since the short acute triangle with possible prespinal fossa in anterior view in Lydekker's drawing isn't that useful.

Next time- Ibirania.

References- Lydekker, 1879. Indian pre-Tertiary Vertebrata. Part 3. Fossil Reptilia and Batrachia. Palaeontologica Indica (series 4). 1, 20-33.

Martinelli and Forasiepi, 2004. Late Cretaceous vertebrates from Bajo de Santa Rosa (Allen Formation), Río Negro province, Argentina, with the description of a new sauropod dinosaur (Titanosauridae). Revista del Museo Argentino de Ciencias Naturales 6(2), 257-305.

Mohabey, Sen and Wilson, 2013. India’s first dinosaur, rediscovered. Current Science. 104(1), 34-37.

Salgado, Gallina and Paulina Carabajal, 2015 (online 2014). Redescription of Bonatitan reigi (Sauropoda: Titanosauria), from the Campanian-Maastrichtian of the Río Negro Province (Argentina). Historical Biology. 27(5), 525-548.

The Titanosaurus (in?)validity project Part 3 - Rocasaurus

Next up in my series of comparisons with Titanosaurus, to determine its validity, is Rocasaurus muniozi. Named by Salgado and Azpilicueta (2000), it's from the Allen Formation of Argentina. The holotype is a partial skeleton (MPCA-Pv 46) that includes three mid caudals, with paratype vertebrae including two mid caudals (MPCA-Pv 49 and 58). The type material was later described in more detail by Garcia and Salgado (2013), while Fernandez et al. (2025 online) recently published on its caudal histology, using some paratype specimens as well as MPCA-Pv 161 that they call an anterior caudal but is of comparable position to Titanosaurus, and three more fragmentary mid caudals (MPCA-Pv 311-313).

Titanosaurus indicus lectotype GSI 2191 in ventral (top), anterior (left), left lateral (center) (all after Lydekker, 1879) and left lateral (right) views (after Mohabey et al., 2013; scale = 50 mm).

So again, as apparently typical for saltasaurines, we have much wider vertebrae with transverse processes (also apparent in dorsal view of MPCA-Pv 161 not shown below). The first three vertebrae figured below seem to be distal to Titanosaurus' type based on central proportions and neural spine development. Another obvious difference is that most have dual pleurocoels arranged vertically, with Garcia and Salgado noting they are truly lacking in the holotype mid caudal, but that may be positional as it only has slight bumps for transverse processes (as seen in posterior view; Garcia and Salgado call it a posterior caudal). As with the previous two genera, the morphology of the ventral face is especially distinctive, with Rocasaurus having deep and sharply defined oval fossae with a thin median ridge subdividing it in at least its anterior portion. A subtle difference that seems constant in Rocasaurus (also in MPCA-Pv 57) is the double concavity of the dorsal centrum edge posteriorly, unlike Titanosaurus. As in the previous two saltasaurines, the neural peduncle is placed more anteriorly (0-8% of central length vs. 16%) than in Titanosaurus.

Rocasaurus muniozi holotype MPCA-Pv 46 (left; after Salgado and Azpilicueta, 2000), paratype MPCA-Pv 58 (center left; after Garcia and Salgado, 2013), paratype MPCA-Pv 49 (center right; after Fernandez et al., 2025 online) and referred MPCA-Pv 161 (right; after Fernandez et al., 2025 online) in ventral (top) and left lateral (bottom) views.

Salgado and Azpilicueta propose the short ventral concavity (in lateral view) in both preserved caudals of the holotype may be a species-level difference with paratypes MPCA-Pv 57 and 58. Yet the former looks closer to the holotype in my opinion, and all of the specimens figured by Fernandez et al. have the paratype condition where determinable. I would just ascribe this to individual variation. Similarly, while they say the neural arch is placed more anteriorly in these two paratypes, the ratio of centrum posterior to the neural peduncle is <41% and 57% respectively, versus 49% in the holotype. Paratype MPCA-Pv 49's ratio is 49% and referred MPCA-Pv 161 is 50%, so I don't see a correlation with the ventral concavity length. 

The original diagnosis of Rocasaurus muniozi includes "caudal vertebrae with a deep ventral cavity divided by a longitudinal partition", noted above as different than Titanosaurus, and "posterior articulation notably depressed and extended ventrally forward" which is partly the wide shape noted above but also the proposed interspecific difference with paratypes that I think could be individual variation.

So Rocasaurus isn't Titanosaurus. Next up, Bonatitan.

References- Lydekker, 1879. Indian pre-Tertiary Vertebrata. Part 3. Fossil Reptilia and Batrachia. Palaeontologica Indica (series 4). 1, 20-33.

Salgado and Azpilicueta, 2000. Un nuevo saltasaurino (Sauropoda, Titanosauridae) de la provincia de Río Negro (Formacíon Allen, Cretácico Superior), Patagonia, Argentina. Ameghiniana. 37(3):259-264.

Garcia and Salgado, 2013 (online 2011). The titanosaur sauropods from the Late Campanian-Early Maastrichtian Allen Formation of Salitral Moreno, Río Negro, Argentina. Acta Palaeontologica Polonica. 58(2), 269-284.

Mohabey, Sen and Wilson, 2013. India’s first dinosaur, rediscovered. Current Science. 104(1), 34-37.

Fernández, Windholz and Zurriaguz, 2025 onine. Palaeohistological characterisation of the caudal pneumaticity of Rocasaurus muniozi (Sauropoda: Titanosauria). Historical Biology. DOI: 10.1080/08912963.2025.2481526

How not to illustrate your new skeletal reconstruction in its descriptive paper

A brief one today while I finish the Rocasaurus post. So we have a new basal sauropodomorph described from the Late Triassic Santo Domingo Formation of Argentina- Huayracursor (Hechenleitner et al., 2025). All well and good, except when you look at the only anatomical figure in the main paper (it's Nature- you only get one), this is the skeletal reconstruction-

So the post-cervical skeleton is in an oblique anterolateral view. Thus you can't really get a good feel of its proportions, or see elements like the femur, tibia, fibula, etc. in strict lateral view. And they're never figured that way elsewhere either (Extended Data Figure 3 claims to be proximal femora in lateral view, but their Huayracursor drawing is literally a tracing of Figure 2k correctly labeled as cranial view). There's a reason you don't just take a photo of a mounted skeleton at some random angle and use that in your descriptive paper for scientific purposes. For one, most scored characters depend on a standardized perspective if they involve shape, and anterolateral is never involved in my experience. 

But at least a photo saves time on illustrative work, whereas someone went through great effort to create this 3D model of Huayracursor's skeleton. But that makes it so much worse because the model is not actually representative of the anatomy. Most basically, they just made any element represented by some material completely brown, but e.g. the distal scapula, distal metacarpal III, distal manual unguals, dorsal ilium, and distal pubes are all unpreserved. And even worse, in one of the rare cases you can actually check the model against the real anatomy because it happens to be on the same plane, the scapula clearly has a more prominent acromion, a greater distal expansion prior to midshaft, an obtuse angle at the glenoid, etc.. So the anatomy isn't even correct. And Figure 2q shows metatarsals II and III tightly articulated, but the reconstruction has the right foot with their distal ends splayed which I think would be a pretty tragic injury in the living animal. You might as well represent Archaeopteryx with one of those sculpted models of its skeleton, done by somebody who didn't put in the work, and at an oblique angle to boot. It's just scientifically sorta worthless. 

And this one image isn't the end of it. If you check out the Extended and Supplementary Information, we get these-
 

So the upper left is Huayracursor again, but basically useless because now we have an oblique dorsoposterolateral view. But at least it's the actual effing elements as preserved, so here's our equivalent of taking a picture of a museum mount. In the upper right you have the therapsid Exaeretodon sp. "3D rendering with preserved bones in orange, based on preserved specimens CRILAR-Pv 156-160." So besides cranial material, the authors list "CRILAR-Pv 156, ... and right femur. ... CRILAR-Pv 160, ... left femur, left tibia and fibula, metatarsal (?), and other fragmentary remains", and I can't help but notice we only got the CRILAR-Pv 160 femur but also not the tibia, fibula and possible metatarsal, but instead the radius and ulna that don't seem to be preserved. It and the lower right Hyperodapedon take up space in Supplementary Figures S5 and S7, but if I were a rhynchosaur or cynodont worker I would much rather you just photographed another bone or view of an already figured bone there. 

And at bottom left is Extended Data Figure 1a-e, er... the second time. Because Extended Data Figure 1 actually has ten images labeled a, c, e, b, d, a, b, c, d, e.  Good thing Nature is so discerning. The first letters (elements) match the second letters (3D models) taxonomically, but that's not really how figure captions work. In any case, this is the image you put on the magazine cover, or the first and last slide of your conference presentation, but not in your technical publication. It's not even useful in a vague ecosystem illustration sort of way because due to foreshortening and a vague Z axis, their relative size is imprecise.

To sum up- represent your skeletal reconstructions in strict lateral view, and if you use computer models use the elements as preserved instead of idealized creations. Unless your publication literally involves computer models like a study of vertebral articulation or center of gravity, etc.. And if you want to insert a flashy 'cool' 3D model as a figure, 9 out of 10 serious researchers would rather just have another photo of a bone. Check out the description of fellow Late Triassic Argentinian basal saurischian Anteavis (Martinez et al., 2025) for how it's done in a tabloid.

References-  Hechenleitner, Martinelli, Rocher, Fiorelli, Juarez, Taborda and Desojo, 2025. A long-necked early dinosaur from a newly discovered Upper Triassic basin in the Andes. Nature. DOI: 10.1038/s41586-025-09634-3

Martínez, Colombi, Ezcurra, Abelín, Cerda and Alcober, 2025. A Carnian theropod with unexpectedly derived features during the first dinosaur radiation. Nature Ecology & Evolution. DOI: 10.1038/s41559-025-02868-4

The Titanosaurus (in?)validity project Part 2 - Neuquensaurus

Continuing our project, the next taxon is Neuquensaurus australis, first named Titanosaurus australis by Lydekker (1893). It was then separated as "Neuquensaurus" in Powell's (1986) thesis, becoming official in Powel''s (1992) osteology of Saltasaurus*. Neuquensaurus is based on six holotype caudals (MLP Ly 2-6), to which a sacrum (MLP Ly 1, 7) and two more caudals (MLP Ly 48, 66; now lost) were later added (Powell, 2003; D'Emic and Wilson, 2011). Huene (1929) and Powell referred a lot of material to the species, but of that only three unfigured mid caudals and three distal ones were accepted by D'Emic and Wilson. Importantly however, those authors also accept the referral of a partial skeleton described by (Salgado et al., 2005) that includes a "probably" continuous series of proximal to mid caudals (MCS-5/2-MCS-5/15). Both the type and MCS-5 come from the Anacleto Formation of Argentina.

* Or was it? That's how Powell (2003) and D'Emic and Wilson (2011) cite it, but ICZN Article 13.3 states "To be available, every new genus-group name published after 1930 (except those proposed for collective groups or ichnotaxa) must, in addition to satisfying the provisions of Article 13.1, be accompanied by the fixation of a type species in the original publication", and Powell only ever says "Neuquensaurus australis (see Powell, 1986)", "Neuquensaurus australis (Huene, 1929)", "Neuquensaurus australis (Huene, 1929: plate 9)", "Neuquensaurus (= “Titanosaurus”) australis", AND importantly "Neuquensaurus robustus as described by Huene (1929)" and "Neuquensaurus (= "Titanosaurus" robustus, Huene 1929; plate 19: 1)" (all translated). Powell makes australis the type species in his 1986 thesis, but as a thesis that's not recognized by the ICZN. In 2003 he gives australis as the type species, but also credits his 1992 paper for the genus, which is a problem because it's from after 1999, so Article 16.1 kicks in ("Every new name published after 1999, including new replacement names (nomina nova), must be explicitly indicated as intentionally new").

Titanosaurus indicus lectotype GSI 2191 in ventral (top), anterior (left), left lateral (center) (all after Lydekker, 1879) and left lateral (right) views (after Mohabey et al., 2013; scale = 50 mm).

In any case, MCS-5 is convenient because it lets us determine more precisely what caudal the Titanosaurus type belongs to. It seems closest to caudals 6-11 in Neuquensaurus in elongation, prezygapophyseal shape and neural spine development. As in Saltasaurus, all these caudals in Neuquensaurus have transverse processes unlike Titanosaurus, with even the last preserved in MCS-5 (fourteenth) showing some development of one. The centra are so wide I don't think taphonomy can explain the difference this time and further differ in having a concave dorsal cotyle edge. The ventral edge of all Neuquensaurus centra are more concave than Titanosaurus. Ventrally the centra show a broad fossa with no lateral ridges, unlike Titanosaurus that has anterior and posterior grooves delimited laterally with ridges, and a broad and deep median keel in the anteroposterior middle. Like Saltasaurus, the neural peduncle ends more anteriorly in Neuquensaurus (10-11% vs. 16% in Titanosaurus), which is consistent in all comparable vertebrae. The prezygapophyses are low in anterior view as in Saltasaurus.

Neuquensaurus australis holotype MLP Ly 3 (upper left) in ventral (top), anterior (left) and left lateral (right) views (after D'Emic and Wilson, 2011); holotype MLP Ly 5 (upper right)  in ventral (top), anterior (left) and left lateral (right) views (after D'Emic and Wilson, 2011); referred MCS-5 (bottom) caudals 6-11 in left lateral view (after Salgado et al., 2005).

MLP Ly 5b has comparable round and pronounced posterior chevron facets to Titanosaurus, unlike MLP Ly 3, so that difference noted for Saltasaurus could be positional variation. MLP Ly 5b also has a deep concavity to the dorsal centrum posteriorly like Titanosaurus, but unlike other vertebrae of either the type or MCS-5, so this could be individual variation. The shape of the concavity between the neural arch and centrum posteriorly is highly variable, with caudals 8 and 11 of MCS-5 pretty close to Titanosaurus, and caudals 9 and 10 having completely different shapes from one another (acutely pointed vs. rounded), showing this is positional/random variation. While the neural spine is angled more highly compared to the prezygapophysis in Titanosaurus than MLP Ly 5 and the ninth-eleventh caudals of MCS-5, the fifth caudal is comparable, so perhaps the dorsally broken spines of the sixth, seventh and/or eighth were too.

The only caudal character in Powell's (2003) diagnosis is "caudal centra rather short and with a very concave ventral face bounded by rounded edges", which was also mentioned in his 1992 paper that said in Saltasaurus mid caudals "The ventral side has a longitudinal groove much narrower than that seen in Neuquensaurus". As noted above, this is quite unlike Titanosaurus. Salgado et al. add a character from the 1986 thesis "lateral walls of caudal vertebral centra little exposed in ventral view", but this seems the same as in Titanosaurus and was said by D'Emic and Wilson to be present in several other titanosaurs as well. The latter also list "longitudinal ridge below transverse processes" of the mid caudals as diagnostic, which is absent in Titanosaurus. Finally, they list "[postzygodiapophyseal lamina] present and elongate" which is just a gentle convexity in Titanosaurus (seemingly similar to the eighth caudal of MCS-5, but maybe it's just illustrated inaccurately).

Note no caudal material was ever assigned to Neuquensaurus? robustus, so we won't be dealing with it.

So Neuquensaurus isn't Titanosaurus, which even Lydekker recognized was quite possible- "the reference of the latter to Titanosaurus must be regarded as a more or less provisional measure, rendered necessary by our very incomplete knowledge of the type species."  Next up, Rocasaurus.

References- Lydekker, 1879. Indian pre-Tertiary Vertebrata. Part 3. Fossil Reptilia and Batrachia. Palaeontologica Indica (series 4). 1, 20-33.

Lydekker, 1893. The dinosaurs of Patagonia. Anales del Museo de La Plata. 2, 1-14.

Huene, 1929. Los saurisquios y ornitisquios del Cretáceo Argentino. Anales del Museo de la Plata (series 3). 3, 1-196.

Powell, 1986. Revisión de los titanosaurios de América del Sur. PhD Thesis. Universidad Nacional de Tucumán. 340 pp.

Powell, 1992. Osteologia de Saltasaurus loricatus (Sauropoda - Titanosauridae) del Cretácico Superior del noroeste Argentino. In Sanz and Buscalioni (eds.). Los Dinosaurios y Su Entorno Biotico: Actas del Segundo Curso de Paleontologia in Cuenca. Institutio "Juan de Valdes", Cuenca, Argentina. 165-230.

Powell, 2003. Revision of South American titanosaurid dinosaurs: Palaeobiological, palaeobiogeographical and phylogenetic aspects. Records of the Queen Victoria Museum. 111, 173 pp.

Salgado, Apesteguía and Heredia, 2005. A new specimen of Neuquensaurus australis, a Late Cretaceous saltasaurine titanosaur from North Patagonia. Journal of Vertebrate Paleontology. 25(3), 623-634.

D'Emic and Wilson, 2011 (online 2010). New remains attributable to the holotype of the sauropod dinosaur Neuquensaurus australis, with implications for saltasaurine systematics. Acta Palaeontologica Polonica. 56(1), 61-73.

Mohabey, Sen and Wilson, 2013. India’s first dinosaur, rediscovered. Current Science. 104(1), 34-37.

The Titanosaurus (in?)validity project Part 1 - Saltasaurus

The whole Alamosaurus vs. Utetitan discussion on the Dinosaur Mailing Group made me curious about Titanosaurus' validity, since it's not been tested since Wilson and Upchurch (2003) declared it a nomen dubium in 2003. Their justification was "In summary, all six features forwarded by Lydekker in his diagnosis of T. indicus are now broadly distributed within Titanosauria. ... Because no diagnostic characters could be identified, T. indicus must be regarded as a nomen dubium." But you can't just take an original diagnosis and say "well, those characters are all shared with other taxa, time to sink it" and "we don't see any differences", you have to put in the work yourself to evaluate the existing differences (since no two bones are identical) and show they're at least plausibly individual/positional/ontogenetic/etc. variation. Not only has that never been done for Titanosaurus, there's never even been any suggested taxa it is identical to, of which we would need two to make it indeterminate (because only one identical taxon would make them synonyms). 

So as a little project I decided I'll start from the tip of my sauropodomorph cladogram and go stem-ward, comparing all procoelous titanosaurs with known caudals to see which, if any, could make Titanosaurus a nomen dubium. I'm not a sauropod worker, but at least titanosaur mid caudals have a pretty simple morphology with e.g. no complicated laminae, so the anatomy should be within my grasp. Here's a picture we'll be seeing a lot- the lectotype of Titanosaurus indicus.  Note it is complete except for the fragmented anterior centrum cotyle edge and incomplete neural spine, which would have included the postzygapophyses. Thought to be lost for decades until Mohabey et al. (2013), nobody has figured a posterior or dorsal view.

Titanosaurus indicus lectotype GSI 2191 in ventral (top), anterior (left), left lateral (center) (all after Lydekker, 1879) and left lateral (right) views (after Mohabey et al., 2013; scale = 50 mm).

And our first contender is Saltasaurus loricatus, from the Lecho Formation of Argentina. Described by Bonaparte and Powell (1980) and then in more detail by Powell (1992; 2003), the holotype is PVL 4017-92, a synsacrum fused to both ilia, but mid caudals are represented by up to ten paratype elements including PVL 4017-28, 4017-32 and 4017-33, figured by Powell (1992- Figs. 22-23; 2003- Plate 34, Plate 52 Fig. 3, Plate 53 Fig. 1, 5-6). 

Saltasaurus loricatus paratype PVL 4017-33 in ventral (top), anterior (left) and left lateral (right) views (after Powell, 1992; scale = 100 mm).

Immediately noticeable is despite PVL 4017-33 having a more elongate centrum (length vs. posterior height 2.16 vs. 1.73; a proxy for proximodistal position), Saltasaurus still has prominent transverse processes. Titanosaurus' vertebra is much narrower transversely, but may be taphonomically compressed. Ventrally, Titanosaurus has a median groove anteriorly and posteriorly, but the middle fourth of the ventral midline is so convex it actually protrudes ventrally as a broad keel in both drawing and photo. Both figured Saltasaurus mid caudals instead have a continuous median groove that contains a narrow keel in its middle half. The anterior chevron facets are longer than Saltasaurus', and the posterior chevron facets are distinct circles but not obvious in Saltasaurus. These are apparent in lateral view as a posteroventrally facing surface anterior to the condyle, while Saltasaurus' condyle extends to the posteroventral corner of the centrum. The prezygapophyses are more elongate (especially noticeable in ventral view), and in anterior view form tall triangular processes while Saltasaurus' are very low. The neural peduncle is placed further posteriorly in Titanosaurus (16% of centrum length past the anterior edge vs. 4% in Saltasaurus), which is consistent in all figured Saltasaurus caudals so is not positional variation. The neural peduncle also ends much more anteriorly than Saltasaurus, with 53% of the centrum behind the peduncle compared to 38-39% in Saltasaurus, which combined with the previous ratio makes a narrower peduncle base for Titanosaurus. This posterior peduncular concavity is angled in Titanosaurus (note Lydekker's drawing gets this wrong), but rounded in Saltasaurus caudals until you get to much more distal elements like PVL 4017-38 that have a ridge for a neural spine and an elongation index of 2.62. The centrum also has a deeply concave dorsal border posteriorly unlike Saltasaurus (slightly concave in distal caudals).

Note while PVL 4017-33 has a more basally positioned postzygapophysis than Titanosaurus, PVL 2017-28 does not, so this could be positional variation. The fragmented cotyle edge is why I think we can see the prezygapophyses meet in ventral view, so I'm not counting that as a difference.

So in conclusion, it's pretty obvious Saltasaurus is not Titanosaurus, which is why Powell erected Saltasaurinae in the first place to contrast with Titanosaurinae. His (1992) Saltasaurinae diagnosis includes "The centra of the caudal vertebrae are broader than tall ... The spines of the caudal vertebrae are posteriorly inclined", though the former could be taphonomic for Titanosaurus and the latter is not different between the genera. 

Next up, Neuquensaurus australis.

References- Lydekker, 1879. Indian pre-Tertiary Vertebrata. Part 3. Fossil Reptilia and Batrachia. Palaeontologica Indica (series 4). 1, 20-33.

Bonaparte and Powell, 1980. A continental assemblage of tetrapods from the upper Cretaceous beds of El Brete, northwestern Argentina (Sauropoda-Coelurosauria-Carnosauria-Aves).  Mémoires de la Société Géologique de France. Nouvelle Série. 19, 19-28.

Powell, 1992. Osteologia de Saltasaurus loricatus (Sauropoda - Titanosauridae) del Cretácico Superior del noroeste Argentino. In Sanz and Buscalioni (eds.). Los Dinosaurios y Su Entorno Biotico: Actas del Segundo Curso de Paleontologia in Cuenca. Institutio "Juan de Valdes", Cuenca, Argentina. 165-230.

Powell, 2003. Revision of South American titanosaurid dinosaurs: Palaeobiological, palaeobiogeographical and phylogenetic aspects. Records of the Queen Victoria Museum. 111, 173 pp.

Wilson and Upchurch, 2003. A revision of Titanosaurus Lydekker (Dinosauria-Sauropoda), the first dinosaur genus with a Gondwanan distribution. Journal of Systematic Paleontology. 1(3), 125-160.

Mohabey, Sen and Wilson, 2013. India’s first dinosaur, rediscovered. Current Science. 104(1), 34-37.

No pygostyle for Baminornis

This week's big news was Baminornis (Chen et al., 2025), a bird with a supposed large pygostyle from a well dated Tithonian locality (Nanyuan Formation- in the same layer Fujianvenator is from, which emerges sister to Serikornis in the Lori matrix btw). This supposed pygostyle is at least 74% of ilial length, so is not like the comparatively undeveloped fused structures in Deinocheirus, Beipiaosaurus and some oviraptorosaurs. Instead, it's out of place in the Jurassic and out of place phylogenetically given the authors recover Baminornis as just crownward of Archaeopteryx in both the TWiG and O'Connor matrices, stemward of Jeholornis in both. Chen et al. call this "The observed ‘chaos’ of bird tail evolution" and note the Late Jurassic Solnhofen and Yanliao Biotas lack short-tailed birds, thus concluding "It would be intriguing to explore why no short-tailed avialans have been found from these two localities."

The obvious answer is- that structure in Baminornis isn't a pygostyle. It's pretty obviously a synsacrum. The first thing to note is that despite the clean, rounded end drawn in their Extended Data Fig. 1b (see below), the posterior end is clearly broken in the photo Extended Data Fig. 1a above it, so it may have included six or more vertebrae in life. This isn't an argument for either identification, as Archaeopteryx can have five (most specimens) or six (Thermopolis specimen), while Jeholornis has six. In any case, the arguments for identification as a synsacrum are-

1. Size. As noted above, even with its broken posterior end, the structure in Baminornis is 74% of ilial length despite only including five vertebrae. As Chen et al. say "As in confuciusornithids, the pygostyle is longer than metacarpal II, but the opposite is true in most other early avialans such as Sapeornis and Cratonavis" yet confuciusornithids fuse eight-ten vertebrae to make their longer pygostyle and enentiornithines usually fuse more as well (e.g. Iberomesornis, IVPP V15664) (Rashid et al., 2018). Excluding the very different elongated distal caudals of basal paravians, bird caudals are short, so fusing so few together wouldn't result in such a large structure. Using Wang and O'Connor's (2017) Table 1, the pygostyle/femoral length ratio of Sapeornis with 4 fused caudals is 19-29%, Jehol euornithines' with 3-5 fused caudals is 19-35%, confuciusornithids' is 49-71%, and non-pengornithid Jehol enantiornithines' is 39-80%. Baminornis' ratio is >44%.

2. Curvature. Chen et al. state the bone "curves dorsally—reminiscent of, but to a lesser extent than, the derived condition in ornithuromorphs in which the pygostyle is characteristically plough-shaped. By contrast, the bone is straight in other early avialans such as confuciusornithids and enantiornithines." It actually looks more curved than their figured 'ornithuromorphs' (= euornithines) Bellulornis and 'Abitusavis' (= Yanornis), but in any case I reinterpret the concave side to be ventral, which matches many paravians from Saurornitholestes to Ornithodesmus to Zhyraornis. It's usually not obvious in Lagerstatten birds where the synsacrum is often preserved in dorsoventral view or below the ilium in lateral view. It's of course possible that a basal avialan pygostyle could be curved either dorsally or ventrally, but it definitely matches the listed paravian synsacra more than any pygostyle I can think of.

Top row- upside down synsacra in right lateral view of (left) Zhyraornis kashkarovi holotype TsNIGRI 42/11915 (after Nessov, 1992); (center) Ornithodesmus cluniculus holotype NHMUK R187 (after Howse and Milner, 1993); (right) Saurornitholestes langstoni (?) NHMUK R4463 (after Howse and Milner, 1993). Bottom rows- (a, b) photo and drawing of Baminornis zhenghensis holotype IVPP V33259, note the posterior end in the photo is broken unlike the drawing; tails of- (c) Confuciusornis sp. IVPP V16066; (d) Jinguofortis perplexus holotype IVPP V24194; (e) Jeholornis prima IVPP V13353; (f) Parabohaiornis martini paratype IVPP V18690; (g) Pterygornis dapingfangensis holotype IVPP V20729; (h) Yanornis martini (Abitusavis lii holotype) IVPP V14606); (i) Bellulornis rectusunguis holotype IVPP V17970 (after Chen et al., 2025).

3. Supposed proximoventral processes. The authors say "The pygostyle has a pair of proximally distributed ventrolateral processes which terminate posterior to the proximal articular facet", but they are asymmetrical with the supposed left one being a small prong while the supposed right one is a more posteriorly placed and ventrally projected, longer blade. Taphonomy maybe, but you know what these perfectly match if the structure is flipped dorsoventrally? A prezygapophysis and a neural spine lamina, respectively. Chen et al. note Sapeornis lacks these processes and obviously Jeholornis doesn't have them, so it would be less parsimonious if Baminornis converged with pygostylians in developing them (or they were lost in Sapeornis) anyway.

4. Anteroposteriorly concave vertebral edges. At least the first, third and fourth vertebrae have concave edges along the concave side. This matches the ventral edges of each centrum in most archosaurian vertebrae, while the dorsal edge of pygostyles is either smooth or convex at each vertebra to reflect the neural spines (see Chen et al.'s euornithine examples, or Fukuipteryx).

5. Taphonomic position. Under my interpretation, you only have to rotate the synsacrum 90 degrees instead of also flipping it upside down, plus the only well-preserved free caudal is next to its posterior end. These aren't in themselves good arguments, since the right pectoral girdle, ischium etc. obviously suffered more displacement than just rotation, but they do match this hypothesis better. Similarly, the synsacrum would be missing otherwise, and as a centrally placed element we would expect it to stay close to the skeleton more than the distal tail.

6. Parsimony, as mentioned above. A pygostyle in an early-diverging avialan like this introduces homoplasy, especially such an elongate one with proximoventral processes like confuciusornithids and most enantiornithines, unlike the shorter and simpler form in Sapeornis and basal euornithines. Whereas a synsacrum is exactly expected.

Once you see the synsacrum in Baminornis, it's pretty hard to un-see. The details all make sense, like the simple and smooth centra and the more complicated topology where the sacral ribs are. Indeed, I'd say if it were a pygostyle my orientation would still make more sense with the centra, prezygapophysis and neural spine lamina, but the vertebrae would still be too big, and why wouldn't it be a synsacrum at that point since it looks just like one and a pygostyle in this taxon is unexpected. Another case of supposed giant Nature falling for sensationalism *cough Oculudentavis cough*.

References- Nessov, 1992. Mesozoic and Paleogene birds of the USSR and their paleoenvironments. In Campbell (ed.). Papers in Avian Paleontology Honoring Pierce Brodkorb. Natural History Museum of Los Angeles County Science Series. 36, 465-478.

Howse and Milner, 1993. Ornithodesmus - a maniraptoran theropod dinosaur from the Lower Cretaceous of the Isle of Wight, England. Palaeontology. 36, 425-437.

Wang and O'Connor, 2017. Morphological coevolution of the pygostyle and tail feathers in Early Cretaceous birds. Vertebrata PalAsiatica. 55(4), 289-314.

Rashid, Surya, Chiappe, Carroll, Garrett, Varghese, Bailleul, O’Connor, Chapman and Horner, 2018. Avian tail ontogeny, pygostyle formation, and interpretation of juvenile Mesozoic specimens. Scientific Reports. 8:9014.

Chen, Wang, Dong, Zhou, Xu, Deng, Xu, Zhang, Wang, Du, Lin, Lin and Zhou, 2025. Earliest short-tailed bird from the Late Jurassic of China. Nature. 638(8050), 441-448.