The re-description of Liaoningotitan sinensis Zhou et al., 2018

Description

Skull: Only the left side of the skull of the Liaoningotitan sinensis holotype is visible (Fig. 3) and is approximately 30 cm in length (the preserved most anterior part of the premaxilla to the preserved most posterior part of the quadratojugal) and 20 cm in height (the preserved apex of the narial to the preserved bottom of the quadratojugal), with no developed premaxillary fenestra. The antorbital fenestra is well-defined and triangular, similar to those in Euhelopus zdanskyi (Poropat & Benjamin, 2013), Mamenchisaurus youngi (Ouyang, 2003), and Omeisaurus maoianus (Tang et al., 2001). The orbit is broken. The narial fenestra opens laterally. It does not exhibit a conspicuous anteroposterior expansion, similar to those of the early-diverging Titanosauriformes such as Euhelopus zdanskyi, but differing from those of the late-diverging Titanosauriformes such as Rapetosaurus krausei (Rogers & Forster, 2004). The length of the premaxilla constitutes 15% of the skull’s overall length (the preserved most anterior part of the premaxilla to the preserved most posterior part of the preserved quadratojugal). The premaxilla and maxilla are divided by a suture. The length of the maxilla is three times the length of the dentigerous portion of the maxilla. The anterior side of the jugal is aligned with the anterior side of the antorbital fenestra, representing an autapomorphy. The maxilla is connected with quadratojugal bone, and the posterior region of the maxilla arch towards the dorsal side, similar to that in the late-diverging Titanosauriformes such as Rapetosaurus krausei (Rogers & Forster, 2004). The right lacrimal is preserved and isolated. The form of the lacrimal is similar to those seen in Euhelopus zdanskyi. A fossa presents in the dorsal portion of the lacrimal, speculation that is connected to the maxilla, similar to that seen in Euhelopus zdanskyi and Tapuiasaurus macedoi (Wilson & Upchurch, 2009; Wilson et al., 2016). However, the lacrimal shaft’s dorsal portion is thinner than the ventral portion, dissimilar to that seen in Euhelopus zdanskyi and other Somphospondyli whose lacrimal bones are thick, but similar to that seen in Brachiosauridae (Wilson & Upchurch, 2009; Wilson et al., 2016; Martinez et al., 2016; Poropat et al., 2023; Rogers & Forster, 2004; Wilson & Sereno, 1998; Daniel et al., 2010). The palatine is a laminar and triradiate bone with a maxillary branch. The pterygoid is also triradiate, with a fan-shaped anterior process. The angle between the horizontal branch and the ascending branch of the quadratojugal is an obtuse angle, similar to that found in some late-diverging Titanosauriformes such as Rapetosaurus krausei (Rogers & Forster, 2004) and Tapuiasaurus macedoi (Wilson et al., 2016). This differs from early-diverging Titanosauriformes such as Euhelopus zdanskyi and Giraffatitan brancai, indicating that mosaic evolution occurred in the skull of Liaoningotitan sinensis. The maxilla and quadratojugal cover the dorsal edge of the dentary, surangular, and angular. Because the surface of the mandibula is full of cracks, the borderline of these three bones is not obvious, as shown in the dotted line in Fig. 3B. The dentary is U-shaped and robust, with a circular rostral side. An obvious foramen is between the surangular and angular, the form is similar to the posterior surangular foramen of the non-Titanosauriformes Sauropod dinosaurs such as in the Mamenchisauridae (Ouyang, 2003; Yang, 2014), but has not been seen yet in the Titanosauriformes (Martinez et al., 2016; Poropat & Benjamin, 2013), therefore, our research is cautious about whether the foramen is the posterior surangular foramen or not. The teeth are spoon-shaped with narrow crowns and are only distributed in the anterior region of the premaxilla, and maxilla. The rostral is convex laterally and lingual concave to medial. The slender index of the teeth (the ratio of the tooth crown length to tooth crown width) is nearly 3.76. The cross-section of the crown is elliptical. The teeth had no serrations. The angle between the long axis of the tooth crown and the abrasive surface is approximately 30°. The ratio of the tooth crown and tooth root length is approximately 1.1. The skull is shown in Fig. 3.

Cervical vertebrae: Only one cervical vertebra is well preserved and 30 cm long. However, the anterior and posterior surfaces, prezygapophysis, and postzygapophysis of the vertebra are invisible (Fig. 4). Only the left surface of the vertebra is visible. The neural spine is bifurcated, A shallow pleurocoel is divided by a lamina on the lateral surface of the cervical vertebra, similar to that in Bellusaurus sui (Mo, 2013). The posterior centrodiapophyseal lamina connects the diapophysis. The cervical rib is a double-head type and short. The cervical vertebra preserves the caput costae. The cervical rib develops a rib ridge (Fig. 4B).

Dorsal vertebrae: There are only two preserved dorsal vertebrae (Fig. 4A), referred to as ‘a’ and ‘b’ for distinguishing (a: left dorsal vertebra in Fig. 4A, shown in left side view; b: right dorsal vertebra in Fig. 4A, shown in posterior side view). Both dorsal vertebrae are flat, also presumed to have been flattened by the rock bed.

Dorsal vertebra ‘a’ is a middle dorsal vertebra in speculation. The left surface of the dorsal vertebra ‘a’ is 17 cm long. The angle between the diapophysis and posterior centrodiapophyseal lamina is acute. The posterior centrodiapophyseal lamina and posterior centroparapophyseal lamina are intersected but do not form a K shape, dissimilar to that seen in the middle dorsal vertebrae of Euhelopus zdanskyi and Gannansaurus sinensis (Wilson & Upchurch, 2009; Lü et al., 2013b). The diapophysis is extends dorsolaterally (Fig. 4A), similar to that in the middle dorsal vertebra of Liubangosaurus hei, Daxiatitan binglingiYongjinglong datangi and Euhelopus zdanskyi (Mo, Xu & Buffetaut, 2010; You et al., 2008; Li et al., 2014; Wilson & Upchurch, 2009), therefore the article speculates ‘a’ is a middle dorsal vertebra. The angular surface of the diapophysis and parapophysis is elliptical. The lateral surface’s pleurocoel (pneumatic foramen) is shallow, similar to that seen in Euhelopodidae (Mannion et al., 2013). The dorsal edge of the pleurocoel is broken.

Dorsal vertebra ‘b’ is a posterior dorsal vertebra in speculation. The posterior angular surface of the centrum of vertebra ‘b’ is opisthocoelous and wider than its height. The ratio of lateromedially width to dorsoventral height of the centrum is 1.2, which is greater than that seen in Daxiatitan binglingi (You et al., 2008) and less than that seen in Mamenchisaurus youngi (Ouyang, 2003). The length of the neural spine is shorter than the centrum. A dorsal rib covers the left part of the neural spine. Only the left parapophysis is preserved. The diapophysis is vertical to the hyposphene and extends in a transverse orientation (extends laterally), therefore the article speculates ‘b’ is a posterior dorsal vertebra, similar to that seen in Euhelopus zdanskyi, Patagotitan mayorum, Yongjinglong datangi and Daxiatitan binglingi (Wilson & Upchurch, 2009; Carballido et al., 2017; Li et al., 2014; You et al., 2008). The location of the parapophysis is lower than the hyposphene. The width of the neural canal is greater than its length. The postzygapophysis and hyposphene are Y-shaped. The posterior centrodiapophyseal lamina and centropostzygapophyseal lamina constitute a shallow fossa, similar to that seen in the Ruyangosaurus giganteus and Daxiatitan binglingi (Lü et al., 2014; You et al., 2008). The neural spine is not bifurcated, similar to that seen in Titanosauria (Mannion et al., 2013). The spinopostzygapophyseal laminae are narrow, and their lateral extension is not conspicuous. A shallow fossa is located on the right posterior side of the neural spine, we speculate there are two shallow fossae located on the posterior side of the neural spine but a dorsal rib covers the left one. The ventral profile of the centrum is incomplete, and the ratio of the dorsoventral height of the neural spine to the posterior angular centrum is 0.98 (Fig. 4A).

Sacral vertebrae: Sacral vertebrae I, II, III (s1, s2, s3) are preserved but embedded in the rock. All centra are amphiplatyan. All vertebrae preserve no neural arch and spine (Fig. 4D). s2 is well preserved, while s1 and s3 are fragmented. The left sacral yoke is preserved, and the form is plate-like. The vertebrae are rectangular, with the anterior and posterior regions equal in height. There are no apparent concavities on the lateral surface of the vertebrae. All vertebrae are interrelated but have not fused, suggesting that this specimen was still immature at death.

Caudal vertebrae: All caudal vertebrae are embedded in the rock, so only the left lateral faces are visible. The middle and posterior caudal vertebrae are preserved and interrelated, but only one middle and three posterior caudal vertebrae are completely preserved. They are all procoelous (Figs. 4C and 4E). The left lateral surface is visible with no diapophyses or concavities. The ventral surface has visible concavities, with the concavity of the anterior caudal centrum being shallower than that of the posterior caudal centra. The neural arch is in the front region of the vertebra and extends in an upper posterior orientation. The angle between the arch and vertebra is approximately 25°. In all preserved caudal vertebrae, the prezygapophysis is long, extending to the anterior, beyond the vertebra. The distance the prezygapophysis extends beyond the anterior margin of the centrum in the middle posterior neural arches is 49% of the centrum length, similar to that seen in Somphospondyli. The postzygapophysis of the middle caudal vertebra approximately is aligned with the posterior part of the centrum (Fig. 4E). The angle between the spine and vertebra is approximately 60°. In the posterior caudal vertebrae, postzygapophysis extends to the dorsoposterior and beyond the neural spine. No chevron bones are preserved in all caudal vertebrae. All vertebrae are shown in Fig. 4.

Scapula: The left scapula is preserved. The proximal end extends dorsolaterally. The dorsal side is thick, and the ventral side is thin. The proximal end is medially curved, similar to that in Somphospondyli. The ratio of the maximum dorsoventral height to the minimum dorsoventral height of the scapular blade is 1.2, less than that of Dashanpusaurus dongi (Ren, 2020; Ren et al., 2022) and Jiangshanosaurus lixianensis (Mannion et al., 2019a). The scapula has a lateral ridge in the middle of the shaft and extends to the anterior and posterior, similar to that of Jiangshanosaurus lixianensis (Mannion et al., 2019a), and speculated to be the attachment point of the subcoracoscapularis muscle. The ratio of the overall length of the scapular blade to its narrowest dorsoventral length is 5.06. The acromion process is not preserved. The posterior margin of the acromial plate is concave. The angle between the acromion posterior region and the scapular shaft long axis is 37°. A subtriangular process in the anteroventral corner of the scapular blade is the tuberosity, or called the attachment point of the triceps brachii muscle, similar to that in Yongjinglong datangi and Daxiatitan binglingi (Li et al., 2014; You et al., 2008). The scapular suture and coracoid are lost. The cross-section of the middle region of the scapular shaft is D-shaped. The thickest area of the shaft is located in the 1/3 area of the proximal end. The width of the proximal end is 38% of the overall length of the scapula. The dorsoventral height of the distal end is greater than the dorsoventral height of the proximal end, and the dorsal side of the proximal end extends slightly to the dorsoventral and posterior side. The distal end extends to the dorsoposterior and posterior side, with an attachment point of the teres major muscle located on the lateral distal end of the scapula (Fig. 5A).

Humerus: Both the left and right humerus are well preserved. The humerus is short, with a length around 70% of the femur. The proximal end is fan-shaped; its maximum width is 54% of the length of the humeral shaft, and the minimum width of the proximal end is 31% of the length of the humeral shaft, which is greater than in Brachiosauridae. The proximal end extends lateromedially, indicating a robust forelimb. The lateral and medial surfaces of the humerus are asymmetric, and the medial angle is sharper than the lateral angle. The slender index, or the humerus length ratio to the humerus’s midshaft width, is 4.4, less than that of Fusuisaurus zhaoi (Mo et al., 2020). Proximal humeral robusticity (PHR), or the ratio of the length of the proximal end to the width of the midshaft, is 2.3, which is less than those of both Ruyangosaurus giganteus (Lü et al., 2014) and Notocolossus gonzalezparejasi (González Riga et al., 2016). The humeral head is oval-shaped. The muscle scar located in the anterior middle region of the proximal end is speculated to be the attachment point of the coracobrachialis muscle. This muscle scar is slightly flat rather than concave, dissimilar in the Mamenchisauridae, Ruyangosaurus giganteus, Patagotitan mayorum (Yang, 2014; Lü et al., 2014; Carballido et al., 2017), and other Titanosauriform dinosaurs, indicating that this is an autapomorphy of Liaoningotitan sinensis. The cross-section of the humeral shaft is elliptical. The humerus decreases in size from the proximal end to the distal end. The dorsal side of the deltopectoral crest is broken. The deltopectoral crest is robust, located in the 1/3 shaft region, and extends medially, similar to that in Titanosauria (Mannion et al., 2013). The length of the deltopectoral crest is 26% of the shaft’s length, less than that seen in Omeisaurus tianfuensis and Huangshanlong anhuiensis (Ren, 2020). The deltopectoral crest is the attachment point of the pectoralis muscle. The concave shaft is not conspicuous and is located on the medial side of the deltopectoral crest. The width of the distal end is 41% of the shaft length of the humerus. The narrowest width of the middle of the shaft is 54% of the distal end’s widest measurement. The radial and ulnar condyles are well preserved, extending to the distal end, with almost a 51° angle. The ulnar condyle is slightly larger than the radial condyle. The medial part of the ulnar condyle is greater than the lateral part (Fig. 5B). The robust index of the humerus, or the ratio of the average of the sum of the widest parts of the proximal end, middle, and distal end to the total length of the humeral shaft is 0.38, which is greater than the 0.29 robust index of the humerus in Qingxiusaurus youjiangensis (Mo et al., 2008), and less than the 0.39 robust index of the humerus in Zhuchengtitan zangjiazhuangensis (Mo et al., 2017), the humeral measurements of some Titanosauriformes shown in Table 2.

Radius: Only the left radius is preserved, 42 cm long, robust, and straight. The width of the proximal end is 25% of the total length. The cross-section of the middle of the shaft is elliptical (Fig. 5C).

Ulna: Only the left ulna is preserved, triradiate (Fig. 5D), and is 53 cm long. The width of the proximal end is 26 cm, and the width of the distal end is 12 cm.

Manus: Only the right manus is preserved. Metacarpus I–IV (m1–4) and two phalanxes (p1 and p2) are preserved (Fig. 5E). The distal end of metacarpus I, II, and III extend, similar to a manus from an unnamed sauropod dinosaur excavated from the Tuchengzi Formation, Liaoning Province, China (Dong, 2001). The middle region of the metacarpus is thin. Phalanx III and IV are fused with the metacarpus. The length of the proximal end of metacarpus II is 27% of the overall length of metacarpus II. Metacarpus III is the longest. The length of all metatarsals is greater than their width. The medial surface of metacarpus IV is slightly concave. Phalanx I and phalanx II are covered by gypsum. Phalanx I has a robust claw (Fig. 5), speculated to have been used to defend against carnivorous theropods or attack competitors during courtship. The scapula and all forelimbs are shown in Fig. 5.

Ilium: The ilium is approximately 70 cm long and not fused with the sacrum. The ilium is elliptical and slightly concave in lateral view, and the dorsal side bulges into an arch, similar to that in Analong chuanjieensis (Ren, 2020). The dorsoventral height of the ilium is 42% of the overall length of the ilium. The lateral side of the ventral surface bulges slightly. The preacetabular process extends to the anterior, and the anterior end is acute and triangular, which is different from that in Qiaowanlong kangxii (Li & You, 2009), Dongyangosaurus sinensis (Lü et al., 2008), and Ruyangosaurus giganteus (Lü et al., 2014), but similar to that in Qinlingosaurus luonanensis (Xue et al., 1996). The preacetabular angle is 50°, which is less than that in Qinlingosaurus luonanensis (Xue et al., 1996) and greater than that in Dongyangosaurus sinensis (Lü et al., 2008), but similar to the angle seen in Shunosaurus lii (Zhang, 1988) and Mamenchisaurus youngi (Ouyang, 2003). The lateral side of the postacetabular process is flat (Fig. 6), unlike Ruixinia zhangi, which has a distinct bulge on the lateral side of the postacetabular process (Mo et al., 2022). The acetabulum is semicircular.

Pubis: The pubic bone is approximately 77 cm in length. The proximal end is plate-like and flat, and the shaft is not inflated. The acetabulum is semicircular. The pubic foramen is an elliptical shape (Fig. 6). The pubic apron is located on the ventral aspect of the pubis. The length of the articular surface of the ilium is 19% of the pubis shaft’s length.

Ischium: The ischium is approximately 70 cm in length, Y-shaped, and triradiate, similar to that in Huanghetitan ruyangensis (You et al., 2006). The dorsoventral length of the proximal end is twice that of the distal end and is 57% of the overall length. The middle region of the proximal end is flat. The iliac process is triangular. The acetabulum is semicircular and conspicuously concave (Fig. 6). The ratio of the dorsoventral width of the ischium’s distal shaft to the ischium’s proximodistal length is 0.24. The ratio of the anteroposterior length of the proximal plate to its total length is 0.67. The ratio of the dorsoventral width of the distal end of the ischial shaft to the smallest dorsoventral width of the shaft is 0.9, similar to that in Somphospondylans (Mannion et al., 2013). The pelvic girdle is shown in Fig. 6.

Femur: The right femur is 106 cm long and only the anterior side is visible. The width of the proximal end is 36 cm and the width of the distal end is 40 cm. The width of the proximal end is 34% of the total length of the femur. A distinct lateral bulge is located at the proximal end’s lateral margin, similar to that in Yunmenglong ruyangensis and Patagotitan mayorum (Lü et al., 2013a; Carballido et al., 2017). The femoral head is well developed and is confluent with the proximal end, lacking a distinct neck. The angle between the dorsal margin of the proximal end and the lateral margin is 127°. The shaft is long and robust with an elliptical cross-section (Fig. 7). The greater trochanter is not conspicuously developed. The lateral process is located in the lower part of the greater trochanter and is 1/3 of the shaft’s lateral dimension. In the distal end, the fibular and tibial condyles extend laterally and are concave. The Femoral Robusticity Index (FRI), or the ratio of the sum of the widths of the proximal, middle, and distal ends to the total femur length, is 0.94, greater than that of Ruyangosaurus giganteus and Daxiatitan binglingi (Lü et al., 2014; You et al., 2008). The medial margin of the femur is concave, similar to that in Ruyangosaurus giganteus (Lü et al., 2014). The femoral measurements of some Titanosauriformes are shown in Table 3.

Tibia: The tibial length is 56% of the femoral length. The proximal end extends lateromedially. The tibial ridge is well-developed and extends laterally to the 2/3 position of the shaft. The ridge length is 94% of the width of the tibial proximal end. The fibular articular surface is concave and located behind the ridge. The 1/3 position of the tibia shaft extends. The narrowest position of the tibia is located below the middle point of the shaft and is 1/3 of the width of the proximal end. The distal end extends to both the anterior and posterior. Cnemial crest develops well. The second cnemial crest is absent, similar to that in Euhelopodidae (Mannion et al., 2013). The narrowest region of the tibial shaft is located above the center point of the tibial shaft (Fig. 7).

Fibula: The length of the fibula is slightly shorter than the length of the tibia and is half the length of the femur. The length of the proximal end of the fibula is 66% of the length of the proximal end of the tibia. The shaft of the fibula is straight and narrow with both mediolateral shrinkage evident. There is a tibial ligament muscle scar. The distal end of the fibula extends to lateromedially and is convex medially (Fig. 7).

Pes: The right pes is preserved. Metatarsal I–IV (m1–4) and one phalanx (p) are preserved (Fig. 7). The pes has a developed, robust, and flat claw. The lateral and medial sides of all metatarsals are curved. The claws are regressed in metatarsals II to IV, and the phalanges are fused with the metatarsals. Metatarsal I is robust and short, with a proximal end that extends slightly. The length of metatarsal I is 30% of radial length and 63% of the length of metatarsal II. The phalanx is robust. Metatarsal II is 20 cm long and is distinctly longer and narrower than metatarsal I, similar to those in Opisthocoelicaudia and Notocolossus (Borsuk-Bialynicka, 1977; González Riga et al., 2016). The width of the proximal end of metatarsal II is narrower than the distal end. The ratio of the length of metatarsal II to its radial length is 0.2. Metatarsal III is the longest, but the ratio of the length of its proximal end to its radial length is also 0.2. The lateral profile of metatarsal III is convex. The shaft of metatarsal IV is the thinnest one of the metacarpi. The width of its proximal end is equal to the width of its distal end (Fig. 7). The ratio of the length of metatarsal IV to the radial length is 0.1. The hindlimbs are shown in Fig. 7.

Skull reconstruction

The skull of the Liaoningotitan sinensis holotype has preserved premaxilla, maxilla, dentary, angular, and supraangular bones, partial quadratojugal bones, and some teeth. An apparent slope between the premaxilla and maxilla is present, and the naris opens laterally. These characteristics are similar to those in Mamenchisaurus youngi (Ouyang, 2003), Omeisaurus maoianus (Tang et al., 2001), and early-diverging Titanosauriformes such as Euhelopus zdanskyi (Poropat & Benjamin, 2013). Therefore, in our research, the nasal, maxilla, parietal, and frontal bones of Euhelopus zdanskyi were used as references for reconstructing the same parts of Liaoningotitan sinensis. However, the quadratojugal of Liaoningotitan sinensis differs from that in Mamenchisaurus youngi, Omeisaurus maoianus, and Euhelopus zdanskyi. The angle between the horizontal and the ascending branches of the quadratojugal of Mamenchisaurus youngi, Omeisaurus maoianus, and Euhelopus zdanskyi is close to a right angle, but in Liaoningotitan sinensis, the angle between the horizontal branch and ascending branch of the quadratojugal is obtuse. This characteristic is more similar to that of the late-diverging Titanosauriformes, such as Nemegtosaurus mongoliensis (Wilson, 2005), Tapuiasaurus macedoi (Wilson et al., 2016), and Rapetosaurus krausei (Rogers & Forster, 2004), indicating that the skull of Liaoningotitan sinensis is in a transitional state in the evolution from the early-diverging Titanosauriformes to the late-diverging Titanosauriformes. Therefore, it is speculated that the postorbital bones of Liaoningotitan sinensis are similar to those of the late-diverging Titanosauriformes. Our research used the postorbital, squamosal and quadrate of Rapetosaurus krausei to reconstruct the same parts of the Liaoningotitan sinensis. The result of the Liaoningotitan sinensis skull reconstruction is shown in Fig. 8.

Body type estimation

The Euhelopus zdanskyi was used for estimating the body length of the Liaoningotitan sinensis because the Euhelopus zdanskyi is a well-preserved sauropod dinosaur specimen with a complete skull, appendicular elements, cervical, and dorsal vertebrae. It was unearthed from the Lower Cretaceous Mengyin Formation in Ningjiagou village, Xintai City of Shandong Province, China, and housed in the Uppsala University of Sweden at the moment (Wilson & Upchurch, 2009). It was recovered as a member of the Somphospondyli family (Mannion et al., 2013, 2019a). The ratio of the length of the posterior vertebra to the length of the anterior vertebra (such as dorsal vertebra 2/dorsal vertebra 1) of the Euhelopus zdanskyi was calculated and then applied to the vertebrae of Liaoningotitan sinensis. The method was once used to reconstruct the body type of Ruyangosaurus giganteus (Lü et al., 2014). The result indicated that the total length of the Liaoningotitan sinensis holotype is approximately 10 m. The height of the scapula and the lengths of the humerus, ulna, metacarpus, and pes all indicated that the Liaoningotitan sinensis specimen had a body length of approximately 10 m and a shoulder height of approximately 2 m. However, the sacral vertebrae of the Liaoningotitan sinensis holotype are not fused, indicating the holotype is an immature specimen. Hence, the body of a mature Liaoningotitan sinensis was probably larger.

Phylogenetic analysis

The first phylogenetic analysis of Liaoningotitan sinensis was performed in 2018, using a matrix modified from Wilson & Upchurch (2009). The results showed that Liaoningotitan sinensis was a Somphospondylian clade member (Zhou et al., 2018). To further analyze the phylogenetic location of Liaoningotitan sinensis, this study used the matrix modified from Beeston et al. (2024) in TNT 1.5 software (Goloboff & Catalano, 2016) for the phylogenetic analysis, which the matrix includes 131 taxa (OTUs: Operational Taxonomic Units) and 560 characteristics. Liaoningotitan was added as a genus to the matrix. This study added the Liaoningotitan taxon to the matrix. Extended implied weighting (EIW) analyses were used with the following settings: max.trees was set to 10,000; tree bisection and reconnection (TBR) was used; new technology search was selected; random addition sequences were changed from 1 to 1,000 addseqs; sect search, ratchet, drift, and tree fusing were all used; The larger the weight set (the value of K), the smaller the weight of the isomorphic features, therefore our analysis attempted to set K = 9, similar in the previous analysis (Mannion et al., 2019a; Beeston et al., 2024). All other options were set to default. The results showed that there were three strict consensus trees, a tree length of 2,840, a CI of 0.208, and a RI of 0.584. Standard bootstrap with the number of replicates changed from 100 to 5,000. Therefore, a strict consensus tree analysis was then performed. The final result identified Liaoningotitan within Euhelopodidae and constituted a sister group with Yongjinglong, as shown in Fig. 9 (see matrix in Supplemental Information).

Phylogenetic position of other Cretaceous Titanosauriformes in China

This analysis indicates that all genera of Cretaceous Titanosauriformes in China fit within Somphospondyli excluding Fusuisaurus. Fusuisaurus was reconsidered as a member of Basal-Titanosauriformes, supporting the result of the initial analysis (Mo et al., 2006, 2020). Liubangosaurus was considered a non-Titanosauriformes Eusauropoda dinosaur in initial analysis (Mo, Xu & Buffetaut, 2010), then classified into Lithostrotia (Mannion et al., 2013) or Euhelopodidae (Poropat et al., 2014), and finally classified into Euhelopodidae again in the present analysis. Yongjinglong, Qiaowanlong, Euhelopus, and Gobititan have been classified into Euhelopodidae in many analyses (Mannion et al., 2019b; Beeston et al., 2024), and Qiaowanlong was considered a Brachiosaurid in an initial genus construction article (Li & You, 2009); the results of the present analysis support the classification of these genera into Euhelopodidae. Dongbeititan is classified as a non-Titanosauria Somphospondylan in the present analysis, which aligned with the results of previous analyses (Mannion et al., 2019b; Poropat et al., 2014), Jiangshanosaurus was classified as non-Euhelopodidae and Titanosauria Somphospondylan in the present analysis, but has been classified into Euhelopodidae and Lithostrotia in analyses of past time (Mannion et al., 2013; Beeston et al., 2024). In the present analysis, Huanghetitan liujiaxiaensis Huanghetitan ruyangensis, and Baotianmansaurusare are non-Euhelopodidae and Titanosauria Somphospondylan. Huabeisaurus, Dongyangosaurus, Daxiatitan, Xianshanosaurus, Ruyangosaurus, and Mongolosaurus are Titanosaurian. In past analyses, Mongolosaurus has been classified as Lithostrotian, and Ruyangosaurus, Dongyangosaurus, and Jiangshanosaurus have been identified as non-Titanosauria Somphospondylans (Poropat et al., 2014; Mannion et al., 2019a). The results of the present analysis support that Xianshanosaurus constitutes a sister group of Daxiatitan, and Dongyangosaurus constitutes a sister group of Huabeisaurus.

About Euhelopodidae

The present analysis supports the validity of Euhelopodidae. In this analysis, Euhelopodidae include Euhelopus, Qiaowanlong, Erketu, Yongjinglong, Liubangosaurus, Tangvayosaurus, Gobititan, Phuwiangsaurus, and Liaoningotitan. The result of the analysis supports that Liaoningotitan constituted a sister group with Yongjinglong, Tangvayosaurus constituted a sister group with Phuwiangosaurus. Fossil evidence has shown that the earliest taxon of Euhelopodidae in Asia is Euhelopus, from the Berriasian period in China (Han et al., 2024). Liaoningotitan, Yongjinglong, Euhelopus, Gobititan, Liaoningotitan, Qiaowanlong, Erketu, Tangvayosaurus, Phuwiangosaurus, and Liubangosaurus indicate that Euhelopodidae was a large and diverse taxon in Asia during the Cretaceous Period. In addition, the present analysis identified Australodocus and Astrophocaudia as taxa of Euhelopodidae, indicating that the distribution region of Euhelopodidae might include Africa and North America. This conclusion will have to be verified in the future as fossil evidence continues to accumulate.

The Titanosauriformes skull

The current understanding of the evolution of the skull of Titanosauriformes is poor. As of February 2025, only 10 taxa of Titanosauriformes have been found with complete skulls: Giraffatitan, Abydosaurus, Euhelopus, Liaoningotitan, Malawisaurus, Tapuiasaurus, Sarmientosaurus, Diamantinasaurus, Nemegtosaurus and Rapetosaurus. Except for Giraffatitan, all of these taxa are from the Cretaceous Period, with Abydosaurus, Euhelopus, Liaoningotitan, Malawisaurus, and Tapuiasaurus coming from the Early Cretaceous. No individuals have been found from the Turonian-Campanian interval, making research on Titanosauriformes difficult (Paul, 1988; Wilson & Sereno, 1998; Daniel et al., 2010; Poropat & Benjamin, 2013; Zhou et al., 2018; Gomani, 2005; Wilson et al., 2016; Martinez et al., 2016; Beeston et al., 2024; Wilson, 2005; Rogers & Forster, 2004).

Based on the 10 taxa that have been found, Titanosauriformes skulls can be divided into four types by our research: (Type 1) In the first type, the snout is higher than the back of the skull (the nasal area bulges), all or most of the narial fenestra is visible in lateral view, and the angle between the horizontal and ascending branches of the quadratojugal is a right or acute angle, such as in Giraffatitan, Abydosaurus, Euhelopus, and Malawisaurus, and similar to the skulls of Mamenchisaurus and Camarasaurus (Daniel et al., 2010; Poropat & Benjamin, 2013; Zhou et al., 2018; Gomani, 2005; Ouyang, 2003). (Type 2) This type has a shorter snout compared to the back of the skull (nasal area is low), less of the narial fenestra is visible in lateral view, the skull is elongated in lateral view, and the angle between the horizontal and ascending branches of the quadratojugal is an obtuse angle, such as in Tapuiasaurus, Nemegtosaurus and Rapetosaurus, and similar to the skull of Diplodocus (Wilson et al., 2016; Wilson, 2005; Rogers & Forster, 2004). (Type 3) The skulls of Sarmientosaurus and Diamantinasaurus have a combination of characteristics of Titanosauriforme types 1 and 2. These have an elongated skull, with a nasal area taller than the back of the skull, a low narial fenestra, and an acute angle between the horizontal and ascending branches of the quadratojugal. These characteristics indicated that mosaic evolution occurred in the skull of Sarmientosaurus and Diamantinasaurus, and they are transitional species in the evolution of Titanosauriformes. (Type 4) Liaoningotitan has a bulging nasal area and an obtuse angle between the horizontal and ascending branches of the quadratojugal. Therefore, mosaic evolution also occurred in the skull of Liaoningotitan, and it is also a transitional species but contrary to those in Sarmientosaurus and Diamantinasaurus (Martinez et al., 2016; Poropat et al., 2023; Zhou et al., 2018). With more and more skulls of Titanosauriformes recorded and described in the future, types may increase, but at least one viewpoint will not be changed, is the mosaic evolution may be a common phenomenon in the skulls of Titanosauriformes and presents not only one evolutionary orientation, the evolution of Titanosauriformes may be more complex than we previously considered.