Biomechanics and morphological comparisons of the caudal region of titanosaurs from the Cretaceous of Brazil: Paleobiology and paleoecology inferences

Abstract

Biomechanical studies in sauropod dinosaurs are mainly focused on neck posture and feeding strategy. Few works investigate other aspects such as tail movement and function, especially in the clade Titanosauria, the most diverse within Neosauropoda. This study applied biomechanical concepts of neutral pose (cartilaginous neutral pose) and range of motion to verify the shape and direction of the caudal region of the advanced titanosaurs Adamantisaurus mezzalirai and Baurutitan britoi, in addition to comparing it with other titanosaurs (e.g., Arrudatitan maximus, Lirainosaurus astibiae, and Trigonosaurus pricei). As a result, the tails analyzed have a sigmoidal‐convex shape, probably close to the ground (but not touching it). The sigmoidal‐convex shape could increase the moment arm for the M. caudofemoralis longus, making it possible to use the tail as a fifth stabilizing member. This work expands our knowledge about sauropod dinosaurs by exposing a diversity of shapes for the tails of titanosaur sauropods and brings new possibilities for how these dinosaurs could use their tails.

This work demonstrates that sauropod dinosaurs have a variety of tail shapes that may have influenced their function beyond balancing the tail/neck. Here, we find “sigmoidal‐concave and sigmoidal‐convex shapes” for the tails of South American titanosaurs. This morphology could have provided mechanical advantages for these sauropods, which could have used the tail as a fifth limb in a potential pentapodal posture.

1. INTRODUCTION

In recent decades, the paleobiology of sauropods has gained significance, particularly through biomechanical studies. These studies primarily focus on topics such as neck posture, movement, feeding strategies, and the use of the tail as a weapon (Conti et al., 2022; Myhrvold & Currie, 1997; Stevens & Parrish, 1999; Taylor & Wedel, 2013). Most biomechanical studies are concentrated primarily on occurrences in countries that were part of the supercontinent Laurasia, particularly on the neosauropods clades Diplodocoidea, Brachiosauridae, and some Titanosauria (Díez Díaz et al., 2020; Myhrvold & Currie, 1997; Preuschoft & Klein, 2013; Stevens, 2013; Stevens & Parrish, 1999, 2005b; Taylor, 2014; Taylor & Wedel, 2013; Vidal et al., 2020).

However, another important neosauropod lineage, Titanosauria, remains obscured regarding paleobiological themes in general, but especially biomechanics. Notwithstanding, titanosaurs represent two‐thirds of the sauropod species unknown to date (e.g., González Riga et al., 2016). Despite the importance and diversity of titanosaurs, few studies address the biomechanical capabilities of the tails of these reptiles (Fronimos et al., 2016; Fronimos & Wilson, 2017; Pereira et al., 2024; Vidal et al., 2021, 2024; Vidal & Díez Díaz, 2017). The main arguments are based on the paucity of complete skeletons and the instability of the phylogenetic relationships of this clade, which hampers more complex biomechanical studies.

Brazil currently has 15 formally described species of sauropods, 13 of which are titanosaurs (Bittencourt & Langer, 2011; Pereira et al., 2024; Ribeiro et al., 2023). Although most Brazilian neosauropods are quite incomplete, three biomechanical studies have been conducted, focusing mainly on the axial elements (Pereira et al., 2024; Vidal et al., 2021, 2024). These studies also included the first digital reconstructions of the skeletons of these species.

The first study was the analysis concerning the neutral pose of Arrudatitan maximus Santucci and Arruda‐Campos (2011), observing the range of movement of the anterior portion of the tail (Vidal et al., 2021). The second study focused on the biomechanical properties of the middle caudal region of the titanosaur Tiamat valdecii Pereira et al. (2024) and the last one proposed the posture, mainly of the neck, trunk, and tail postures of Trigonosaurus pricei Campos et al. (2005; Vidal et al., 2024).

However, these previous studies were conducted using replicated reconstructions (digital twins) of the preserved bone elements, and phylogenetically close species to fill in missing elements, which may introduce biomechanical bias. In this scenario, species with complete vertebral sequences, such as Baurutitan britoi Kellner et al. (2005), that bears one of the most complete caudal vertebral sequences (19 vertebrae and 15 haemal arches) known to a titanosaur (e.g., Lacovara et al., 2014), may provide a more reliable model. Furthermore, Vidal et al. (2024) demonstrated that the anterior caudal region can indicate the general “shape” and “direction” of the tail of titanosaurs because it is directly linked to the sacrum. Adamantisaurus mezzalirai Santucci and Bertinni (2006) is another important titanosaur that presents well‐preserved and sequenced anterior caudal vertebrae that may allow testing of mechanical properties of the anterior region of the tail in titanosaurs.

1.1. Biomechanical approach to South American titanosaurs

The caudal region is crucial for understanding sauropod posture, especially since these animals have long necks and tails. In recent years, there has been growing interest in the posture and function of the sauropod caudal vertebrae as a whole (Baron, 2020; Díez Díaz et al., 2020; Ibiricu et al., 2013; Otero & Vizcaino, 2008; Pereira et al., 2024; Vidal et al., 2020, 2021, 2024; Vidal & Díez Díaz, 2017). To date, digitally performed biomechanical analyses that target the interaction between those bones, such as the osteological neutral pose (ONP; Stevens & Parrish, 1999, 2005a) or “cartilaginous neutral pose” (CNP; Taylor, 2014), require well‐preserved and sequenced elements (Mallison, 2010a, 2010b). For these analyses, the articular facets of the vertebral centra and the zygapophyses, when preserved, are the best source of information for the reconstruction of posture, thanks to their articulation and overlapping surfaces (Taylor, 2014; Taylor & Wedel, 2013).

Titanosaurian caudal vertebrae are numerous in the fossil record (e.g., Bandeira et al., 2018; Powell, 2003; Ribeiro et al., 2023), being useful for taxonomic inferences (e.g., Casal et al., 2007; Kellner et al., 2005; Salgado, 1996; Santucci & Bertinni, 2006) but also for biomechanical analysis. Sellers et al. (2013) analyzed the biomechanics of the titanosaur Argentinosaurus huinculensis Bonaparte and Coria (1993), aiming to understand how its musculoskeletal system responds to speed and gait, due to its large proportions. Through a digital reconstruction of the skeleton, and using computer simulations, the authors inferred how this giant organism walked and concluded that an 83‐ton sauropod was mechanically capable of low‐speed locomotion. Fronimos et al. (2016) and Fronimos and Wilson (2017) discussed the concavo‐convex characteristics of procoelous and opisthocoelous vertebrae, through mechanical tests with cantilevers and suggested that this morphology conferred stability between the articulation of the vertebrae centra. However, there are many studies being carried out with the aim of understanding the posture and way of locomotion of titanosaurs, but few that aim to study the tails of these organisms.

Vidal et al. (2021) conducted a study on the tail of the Brazilian Aeolosaurini titanosaur Arrudatitan, through a CNP analysis. The authors concluded that the tails of Arrudantitan and other Aeolosaurini might have been sigmoidal shape (S‐shape) with a subhorizontal direction close to the ground. The authors suggested that the morphology of this part of the tail could have been caused due to the origin of the powerful caudofemoralis longus muscle and that this could result in an increase in the mechanical advantage of the tail. Vidal et al. (2024) performed a biomechanical analysis on the titanosaur Trigonosaurus to revise its posture and concluded that it was a small sauropod with an elevated neck and sigmoidal tail.

Adamantisaurus and Baurutitan can provide valuable insights into the caudal motion of derived titanosaurs (or at least in Lithostrotia), notably as these taxa are not phylogenetically related (Navarro et al., 2022; Pereira et al., 2024). Aspects such as the shape, orientation, and functionality of the caudal vertebrae are crucial to understanding the paleobiology of sauropods, especially since the tail also acts as a biomechanical counterpoint for the sauropod necks (Stevens, 2013). In this sense, understanding the biomechanical aspects of titanosaurs can also aid in paleoecological interpretations, such as their browsing habits and potential competitiveness in their ancient environments (e.g., Baron, 2020; Tütken, 2011). For this work, the CNP of A. mezzalirai and B. britoi is presented to recognize the shape constituted by the anterior and middle caudal vertebrae in titanosaurs from Brazil.

1.2. The M. caudofemoralis in titanosaurs

Several authors utilize shared homologies between living archosaurs (avian theropods and crocodilians) and non‐avian dinosaurs to reconstruct the tail musculature of these extinct reptiles (e.g., Díez Díaz et al., 2020; Hutchinson et al., 2015; Hutchinson & Gatesy, 2000; Ibiricu et al., 2013; Mallison et al., 2015; Witmer, 1995, 1997). However, extant crocodilians serve as the best analogs for reconstructing the architecture and tail musculature of sauropod dinosaurs, since birds have undergone significant structural bone modifications, mainly the tail reduction to the pygostyle as an adaptation for flying (Díez Díaz et al., 2020; Mallison et al., 2015). Nevertheless, most studies on dinosaur tail musculature primarily focus on the caudofemoralis longus muscle (M. CFL) due to its locomotor importance, which primarily functions in the retraction of the hindlimbs (e.g., Mallison, 2011a; Persons et al., 2014; Persons & Currie, 2011, 2012). Previous authors support that the M. CFL played a significant role in titanosaur tails, influencing even the morphology of the anterior caudal vertebrae, particularly on the lateral aspects such as the position and robustness of the diapophysis (Salgado & García, 2002).

The “origin” and “insertion” definition sites of the M. CFL are still in debate. For instance, according to most authors, the “origin” of the M. CFL is on the lateral surface of the vertebral centrum, lateral to the haemal arches, and on the ventral surface of the transverse processes, with insertion on the fourth trochanter of the femur and an adjacent tendon that inserts into the proximal region of the fibula. (e.g., Díez Díaz et al., 2020; Gallina & Otero, 2009; Hair, 1868; Ibiricu et al., 2013; Otero & Vizcaino, 2008; Romer, 1923). Alternatively, other authors sustain that the M. CFL originates at the fourth trochanter, being inserted on the lateral surface of the vertebral centrum and consequently, lateral to the haemal arches. Furthermore, this muscle orientation implies a muscle insertion on the ventral surface of the transverse processes (Mallison et al., 2015; Salgado & García, 2002). However, the surfaces where the muscle is located (regardless of the direction observed, for example, (a) origin to insertion or (b) insertion to origin) are similar among authors.

According to the widely accepted, the origin of a muscle is always the one with the proximal attachment (generally having a broader attachment surface), whereas the insertion has a more distal attachment and typically ends with tendon presence (Hamill et al., 2015). We opted for the most accepted determination to define the “origin” and “insertion” of the M. CFL. Thus, the origin is located on the caudal vertebrae (concerning haemal arches and the transverse processes), and the insertion at the fourth trochanter of the femur. (Díez Díaz et al., 2020; Gallina & Otero, 2009; Hair, 1868; Ibiricu et al., 2013; Otero & Vizcaino, 2008; Romer, 1923).

Indeed, it is generally agreed that the lateral surfaces of the anterior vertebral centra in sauropods represent a significant portion of this muscle origin (Borsuk‐Białynicka, 1977; Díez Díaz et al., 2020; Ibiricu et al., 2013; Mallison et al., 2015; Romer, 1923; Salgado et al., 2005; Salgado & García, 2002). Salgado and García (2002) have observed variations in the lateral surfaces of the anterior vertebral centra in some titanosaurs. The authors identified two types of lateral surfaces: a primary and a secondary one. According to these authors, the primary surface extends from the anterior caudal vertebrae, migrating to the ventral region of the anterior–posterior caudal vertebrae, allowing the development of a new (secondary) surface (Ibiricu et al., 2013: figure 3). As Salgado and García (2002), the primary lateral surface was where the M. CFL originated. As the muscle volume decreased, the origin surface shifted its position, assuming a more ventral development. The authors highlight that the line of contact between the primary and secondary surfaces culminates posteriorly in a lateral protuberance. Here, we opted to use terms such as “distal” or “proximal” to refer to regions closer or farther from the axial skeleton.

Otero and Vizcaino (2008) discuss that the M. CFL origin occurs at the primary lateral surfaces of the anterior caudal vertebrae, as well as the lateral surfaces of the haemal arches. In this regard, Otero et al. (2011) highlight that the morphological variability of sauropod haemal arches is related to caudal musculature, and consequently, to the origin of the M. CFL.

In addition to the lateral surface of the vertebral centrum, the morphology of the haemal arches can indicate the size, shape, and extent of the M. CFL in fossil archosaurs. Thus, a weak oblique roughness appears more dorsally on the lateral surface of the anterior haemal arches for the attachment of the M. CFL (Díez Díaz et al., 2020). Gallina and Otero (2009) suggest that the development of M. caudofemoralis brevis (CFB) and M. CFL is related to the transverse processes variation along the tail, and propose an additional origin insertion for the muscle beyond the lateral surface of the vertebral centra and haemal arches. Furthermore, the transverse processes are also associated with muscle extension, hypothetically marking its distal termination (Ibiricu et al., 2013). In this context, Ibiricu et al. (2013) postulate three evolutionary states for the distal termination of the M. CFL in Titanosauria based on the termination of the transverse processes: at Ca17 or more posteriorly (state 0), related to basal titanosaurs; between Ca9 and Ca12 (state 1), related to non‐saltasaurine Lithostrotia; or at Ca8 (state 2), related to Saltasaurinae dinosaurs (Ibiricu et al., 2013: figure 6).

However, in some cases, the M. CFL could persist beyond the transverse processes (Díez Díaz et al., 2020; Mallison et al., 2015). Also, in some titanosaurs, the transverse processes persist (albeit in a reduced form) after the primary lateral surface (the origin of the M. CFL, as proposed by the authors), indicating that the associated muscle (M. CFL) could have its distal termination before the termination of the transverse processes (Salgado & García, 2002).

Therefore, it is evident that the development and morphology of the transverse processes, the lateral surfaces of the vertebral centra, and the haemal arches are the key indicators for inferring the origin, shape, and extent of the M. CFL in sauropod dinosaurs.

1.3. Geological setting

Adamantisaurus mezzalirai was reported from rocks of the Adamantina Formation (Campanian‐Maastrichtian), Bauru Group, near the municipality of Florida Paulista, São Paulo state, Brazil. According to Santucci and Bertinni (2006). The specimens were collected in a widely oxidized rock package, extensively distributed in fine to very fine feldspathic sandstones, with intercalations of clayey siltstones and gray or greenish mudstones/shales. The depositional environment of the Adamantina Formation was interpreted by Dias‐Brito et al. (2001) as lacustrine or fluvial in nature (Figure 1a).

FIGURE 1.

Map of the occurrence of the sauropod taxa analyzed in this study. (a) Pedreira Caieira where materials from Trigonosaurus pricei and Baurutitan britoi were collected and (b) location near Florida Paulista where materials from Adamantisaurus mezzalirai were collected (according to Campos et al., 2005; Kellner et al., 2005; Santucci & Bertinni, 2006).

Baurutitan britoi was collected in rocks of the Bauru Group, Marília Formation, Serra da Galga Member with Maastrichtian age (Campos et al., 2005; Kellner et al., 2005). However, recently Soares, Basilici, Lorenzoni, et al. (2020) and Soares, Basilici, Marinho, et al. (2020) proposed the elevation of the Serra da Galga Member to the Serra da Galga Formation (adopted in this work), as they are more recent layers and have physical–chemical and paleobiological characteristics different from those of the Marília Formation.

The location where B. britoi was collected was called “Point 1” by Llewellyn Ivor Price, a quarry known as “Caieira” at Fazenda São Luís, Morro do Veadinho, near Peirópolis, Minas Gerais state (Campos et al., 2005; Campos & Kellner, 1999; Kellner et al., 2005) (Figure 1b). The specimen was reported to be fine‐ to medium‐grained white and yellow sandstone rocks, with a silt–clay matrix and disseminated clayey clasts.

2. MATERIALS

The holotype of A. mezzalirai is composed of six anterior caudal vertebrae (Ca2–Ca7) and two hemal arches that are deposited in the collection of the Valdemar Lefèvre Geological Museum (MUGEO), in the city of São Paulo, São Paulo state (Figure 2a). The holotype of B. britoi is composed of one sacral vertebra (S6), 18 caudal vertebrae (Ca1–Ca18), and 15 haemal arches (currently 13 units only) that are deposited in the collection of the Earth Sciences Museum (MCT), located in the municipality of Rio de Janeiro, Rio de Janeiro state (Figure 2b).

FIGURE 2.

(a) Top row, last sacral (S6) and caudal (Ca1–Ca18) vertebrae of Baurutitan britoi (MCT 1490‐R) in left lateral view. (b) Bottom row, anterior caudal vertebrae (Ca2–Ca7) of Adamantisaurus mezzalirai (MUGEO 1282) and two haemal arches (MUGEO 1289 and 1295), in left lateral view. Scale bar: 20 cm.

3. METHODOLOGY

3.1. Digitalization

Structure from motion photogrammetry was used to acquire the 3D elements of Adamantisaurus and Baurutitan following the protocol proposed by Mallison and Wings (2014). The elements were positioned on a surface and photographed in two stages (chunks 1 and 2), where the first (chunk 1) aimed to digitize the dorsal portion of the vertebrae and the second aimed to digitize the ventral portion (chunk 2). Subsequently, the chunks were aligned and merged using Metashap 2.0 software (© Agisoft LLC, www. agisoft.ru). Finally, a three‐dimensional mesh was established based on the alignment of the chunks.

For photogrammetry, the photographs regarding the materials of Adamantisaurus and Baurutitan were taken by the authors. For each element, approximately 70 images were obtained for each face each caudal vertebrae. To generate the 3D models, the photos were processed at the Macrofossil Laboratory (Geology Department, Institute of Geosciences of the Universidade Federal do Rio de Janeiro). It was necessary to reduce the number of faces (Decimate Mesh—Initial faces 500,000) in the 3D elements that, after the procedure, had approximately 100 thousand faces. The resulting meshes were exported in Wavefront .OBJ format. Figure 3 shows the scanning result of the 3D elements.

FIGURE 3.

Digitized 3D elements through photography. (a) Anterior and middle caudal vertebrae of Baurutitan britoi and (b) anterior caudal vertebrae of Adamantisaurus mezzalirai. Not in scale.

The materials of Adamantisaurus and Baurutitan present structures and/or breaks as shown in Tables S1 and S2. The damaged structures of interest in Adamantisaurus and Baurutitan (e.g., zygapophyses and condyle and cotyle) were recovered according to bilateral symmetry or reconstructed from nearby elements following the methodology proposed by Lautenschlager (2016) (see Figure 3). There were no significant deformations in the vertebrae of Baurutitan and Adamantisaurus, therefore, no element underwent retrodeformation. The elements that showed breakage marks (see Tables S1 and S2) were reconstructed according to Lautenschlager (2016) using the law of bilateral symmetry. For example, the anterior caudal vertebrae Ca7 and Ca8 of Baurutitan presented marks of anthropic origin (possible core drill mark during excavation) at the ends of the vertebral centra (cotyle and condyle, respectively: Figure 4a,b). To reconstruct the condyle portion of the Ca7 vertebrae, the equivalent portion of the Ca6 vertebrae was used, which was cut and superimposed on the Ca7 condyle and subsequently modified until it assumed the final shape using free modeling software (Blender 4.1, www.blender.org) (Figure 4a 1 to 4). Likewise, to reconstruct the posterior portion of the vertebral centra of the Ca8 vertebra, the posterior portion of the Ca7 vertebrae was used (closest and most similar: Figure 4b).

FIGURE 4.

The figure represents the process of digital restoration of the caudal vertebrae of the studied titanosaurs. (a) Anterior caudal vertebrae Ca7 of Baurutitan. The circle shows the anthropic marks on the posterior portion over the condyle in the (1) superior and (2) right lateral views. The arrow indicates the result of the restoration of the caudal vertebra Ca7 from Baurutitan represented in numbers (3) and (4) in the superior and right lateral views, respectively. (b) Anterior caudal vertebrae of Baurutitan Ca8. The circle shows the anthropic marks on the anterior portion over the cotyle in the (5) in the right anterolateral view. The arrow indicates the result of the restoration of the caudal vertebra Ca8 of Baurutitan represented in numbers (6) in the right anterolateral view; (c) anterior caudal vertebrae Ca2 of Adamantisaurus with damage of transverse process, neural spine in anterior (7) and right lateral view (8). The arrow indicates the result of the restoration of the caudal vertebra Ca2 of Adamantisaurus represented by numbers (9) in the anterior and (10) right lateral view. Not in scale.

The neural spines of the S6 vertebrae of Baurutitan and Ca2 of Adamantisaurus, which were displaced from their original positioning due to breakage, were repositioned by aligning with the fracture line in the dorsal portion of the vertebral centrum (according to Lautenschlager, 2016). The Ca2 anterior caudal vertebrae of Adamantisaurus has a broken neural spine and transverse process and the absence of the right side of the vertebral centrum. To reconstruct these structures, the left side of the vertebral centrum was cut and superimposed on the right side (bilateral symmetry) and modified until it assumed the final shape using free modeling software (Blender) (Figure 4c 7 to 10).

The breaks and repositioning in the vertebrae of Baurutitan and Adamantisaurus did not require great degrees of speculation during the reconstruction process. For example, in the Ca7 vertebrae of Baurutitan the break mark develops diagonally over the condyle, maintaining most of the features of the edge of the vertebral centrum on the right side which facilitated interpretations of the features on the left side during reconstruction remodeling.

3.2. Cartilaginous neutral pose

To establish the neutral pose of the tail of Adamantisaurus and Baurutitan, the method “CNP” proposed by Taylor (2014), which was an improvement on the method proposed by Stevens and Parrish (1999) (ONP). CNP adds intervertebral soft tissue thickness to neutral pose analyses (Taylor, 2014). For virtual manipulation of the 3D models, 3D MAX software was used (© 2019 Autodesk Inc, www.autodesk.com), as in Molnar et al. (2015).

As proposed by Taylor (2014), we add the thickness of intervertebral cartilage as 2.5%, 5%, 10%, 15%, 20%, and 25% of the length of vertebral centra (Tables S3 and S4). As proposed by Mallison (2010b), only two vertebrae were at a time to minimize preconceived notions. To establish the CNP of the tail of Adamantisaurus and Baurutitan, we consider initially the zygapophyseal joints to be 100% overlapping (Taylor, 2014; Taylor & Wedel, 2013) with the alignment of the canal neural (initial horizontal line: Taylor, 2022). Subsequently, the intervertebral soft‐tissue thickness variables were added at 2.5%, 5%, 10%, 15%, 20%, and 25% of the vertebra and the variation of the angle in the dorsal or ventral direction was plotted in Tables S5 and S6. The alignment of the vertebrae in the CNPs was performed as in Vidal et al. (2024), that is, the vertebrae were positioned from the most posterior caudal vertebrae in relation to the sacrum, that is, in Adamantisaurus from Ca7 to Ca2 and in Baurutitan from Ca18 to Sa6. The results of cartilage thickness variation for Adamantisaurus and Baurutitan are shown in Tables S3 and S4.

The CNPs results were evaluated according to direction and shape, as proposed by Vidal et al. (2024). For the authors, the direction of the vertebral set is established in relation to the sacrum and indicates whether the set is below or above the horizontal support line (horizontal line in relation to the ground). The shape is the result obtained from the alignment of the vertebrae and is the sum of the changes in direction of each segment of the column region in the CNP (In this case, how the caudal region appears: sigmoidal, linear, “U” shaped, and “J” shaped). The sum of these two factors results in the posture.

3.3. Range of motion

To analyze the range of motion (RoM), we considered the concepts of Stevens and Parrish (2005b), Stevens (2013), and Fronimos et al. (2016), for concavo‐convex vertebrae (procoelous or opisthocoelous), where a center of rotation (CoR) was placed within the condyle of the anterior vertebrae in the procoelous caudal vertebrae, where the element rotates around a point outside itself (Fronimos et al., 2016: figure 8A; Stevens, 2013: figure 9; Stevens & Parrish, 2005b: figure 6.14). The vertebrae that were initially positioned in the CNP were moved in the X‐ and Y‐axes corresponding to the maximum dorsal, ventral, and lateral movement and were expressed in degrees (°) in Tables S7–S9.

For maximum movement, the vertebrae were rotated on the CoR so that the zygapophyses preserved substantial overlap (not disarticulation) between the zygapophyseal facets or until there was a strong contact between the edges of the vertebral centrum of the anterior and posterior vertebrae (although in modern birds there may be zygapophyseal disarticulation) (Stevens & Parrish, 1999, 2005b; Vidal et al., 2024). Each vertebra was connected to the previous one using the 3D MAX link tool.

4. RESULTS

4.1. CNP in Adamantisaurus

CNP caudal—in the pose in 2.5% there is strong contact of the Ca3, Ca6, and Ca7 vertebrae in relation to their respective anterior ones and therefore it was not considered (Table S5; Figure 5a). The alignment of the caudal vertebrae at 5% and 10% of the thickness of the intervertebral soft‐tissue appears to be more parsimonious, as there are no strong contact or disarticulation of the zygapophyses (Table S5; Figure 5b,c). In the poses, at 15%, 20%, and 25% there are disarticulations of the zygapophyses (when there is not overlapping of the zygapophysis) (Figure 5d,f).

FIGURE 5.

Cartilaginous neutral pose of the anterior caudal vertebrae Ca2–Ca7 of Adamantisaurus. (a–f) 2.5%, 5%, 10%, 15%, 20%, and 25% of the intervertebral soft‐tissue thickness, respectively. Arrows indicate points of disarticulation in the zygapophyses. Not in scale.

The increase of the intervertebral soft‐tissue thickness decreased the angle formed by the sequence from 13.5° (at 5%) to 5° (at 10%) in the region of the Adamantisaurus tail, making clear the importance of this tissue for the general shape of the tail in sauropods (Figure 5b,c). Furthermore, in 5%, the vertebral units are strongly angled ventrally, giving a curved appearance to this region of the tail in Adamantisaurus. However, this curvature is less pronounced at 10% of the thickness of the intervertebral soft‐tissue, making the sequence almost linear and oriented in a sub‐horizontal direction. There is no strong contact between the haemal arches and the vertebral centra and not even between haemal arches in any of the CNP proposals for Adamantisaurus (Figure 5).

4.2. RoM in the tail of Adamantisaurus

Values for maximum dorsal, ventral, and lateral displacement for the RoM of Adamantisaurus were plotted in Table S7. In Adamantisaurus, the RoM from the Ca2–Ca7 vertebrae in the maximum dorsal and ventral positions in 5% of the intervertebral soft‐tissue thickness is 20° and in 10% it is 25°. There is a considerable difference of 5° between the RoM at 5% and 10% of the thickness of the intervertebral soft‐tissue, which shows an increase in the RoM of the anterior portion of the tail (Figure 6a,b).

FIGURE 6.

Angular variation between the maximum dorsal, ventral, and lateral movement for the tail of Adamantisaurus in (a) 5% and (b) 10% (maximum dorsal and ventral) of the soft‐tissue thickness left lateral view; (c) 5% and (d) 10% (lateral maximum) of the soft‐tissue thickness in dorsal view. Orange color indicates intervertebral soft‐tissue thickness at 5% and blue at 10%. Not in scale.

The increased thickness of the intervertebral soft‐tissue noticeably increases the dorsoventral RoM, enlarging the action space between the vertebrae and, therefore, the movement of the anterior portion of the Adamantisaurus tail. For the laterolateral maximum in 5% the predicted value was 22° and in 10% 12° (Figure 6c,d). Here, contrary to what is assumed (larger thicknesses provide a greater RoM—Taylor, 2014; Taylor & Wedel, 2013), there is a reduction in the RoM of the vertebrae in lateral movements that is more significant than in dorsoventral movements (Figure 6c,d). As in the CNP, there is no strong contact between the haemal arches and the vertebral centra, nor even between the haemal arches in any of the RoM proposals for Adamantisaurus (Figure 6).

4.3. Neutral pose (CNP) in Baurutitan

Caudal CNP: the alignment of the caudal vertebrae with an intervertebral soft‐tissue thickness of 5% and 10% appears to be more parsimonious, as it prevents strong contact or disarticulation of the zygapophyses (Table S6; Figure 7b,c). In the 2.5% pose there is a strong contact of the Ca3, Ca4, Ca6, and Ca9 vertebrae and therefore were not considered (Figure 7a). In poses at 15%, 20%, and 25% there is disarticulation of the zygapophyses and vertebral centra, in addition, the haemal arches are also disarticulated from their respective articular surfaces (ventrolateral protuberances at the anterior and posterior ends in the vertebral centra), specifically after Ca9 (at 20%) and Ca8 (at 25%) (Figure 7e,f).

FIGURE 7.

Cartilaginous neutral pose of the anterior and middle caudal vertebrae of Baurutitan. (a–f) 2.5%, 5%, 10%, 15%, 20%, and 25% of the intervertebral soft‐tissue thickness, respectively. Arrows indicate points of disarticulation in the zygapophyses. Not in scale.

The increase in the thickness of the intervertebral soft‐tissue in the Baurutitan tail does not significantly reduce the angle formed by vertebral sequence from 16.17° (at 5%) to 16.12° (at 10%), a difference of 0.5° (Figure 7b,c). However, for Baurutitan a greater number of vertebrae was considered (17 of Baurutitan vs. six of Adamantisaurus) between the middle and anterior caudal, therefore, it is not possible to make a comparison with Adamantisaurus (in this research) and Arrudatitan (Vidal et al., 2021).

At 5% of the intervertebral soft‐tissue thickness, the anteriormost caudal vertebrae of Baurutitan (Ca1 and Ca2), elevate the direction of the tail to supra‐horizontally while in 10%, it remains linear horizontally (Table S6; Figure 7b). However, from Ca3 to Ca9 at 5% (and Ca10 at 10%), the vertebrae tend to orient ventrally (Table S6; Figure 7b,c). Therefore, it is possible to notice that the vertebral assembly expresses a significant change in direction throughout the sequence assuming a “sigmoidal shape.” There is no strong contact between the haemal arches and the vertebral centra and not even between haemal arches in any of the CNP proposals for Baurutitan. However, the arches are further away from the articular surfaces where there is greater displacement in the dorsal direction (Figure 7d–f).

4.4. RoM in Baurutitan tail

Values for maximum dorsal, ventral, and lateral displacement for the Baurutitan RoM were expressed in Tables S8 and S9. In Baurutitan, the RoM for the maximum dorsal and ventral poses at 5% of the intervertebral soft‐tissue thickness is 83° while at 10% it is 105°. There is a considerable difference between the RoM at 5% and 10% counted in 22°, showing that the thickness of the intervertebral soft‐tissue can expand the RoM of the tail in the dorsoventral direction, as predicted by Taylor and Wedel (2013) and Taylor (2014) (Figure 8a,b).

FIGURE 8.

Angular variation between the dorsal, ventral, and lateral maximums in (a) 5% and (b) 10% (maximum dorsal and ventral); (c) 5% and (d) 10% (maximum lateral) of the soft‐tissue thickness for the Baurutitan tail. Orange color indicates intervertebral soft‐tissue thickness at 5% and blue at 10%. Not in scale.

For the lateral maximum, the difference is 6°, representing an increase in range, but not as significant as for dorsoventral movements (Figure 8b,d). In this case, it was observed that with the increase in the thickness of the intervertebral soft‐tissue, there is also an increase in the RoM, expanding the space of action between the vertebrae. As in the CNP, there is not strong contact between the haemal arches and each other, nor between the haemal arches and the vertebral centra, in a way that could prevent movements in the dorsal and ventral directions, in any of the RoM proposals for Baurutitan (Figure 8).

4.5. Comparison between tail regions

Biomechanical studies concerning caudal movements in titanosaurs are still scarce, and comparisons are very limited. In this case, an example to assist the present study is A. maximus (Vidal et al., 2021). This Brazilian titanosaur is consistently recovered as a sister‐taxa of Adamantisaurus, recovered in the Rinconsauria clade. Nevertheless, Baurutitan is deeply nested as a Saltasaurinae, presenting a series of synapomorphies of this clade since its description (e.g., Bandeira et al., 2024; Navarro et al., 2022; Pereira et al., 2024).

As discussed by Vidal et al. (2021), only CNPs were considered at 5% and 10% of the thickness of the intervertebral soft‐tissue (Figure 9). According to the authors, in Arrudatitan, CNP in 2.5% generates strong contact in the vertebrae (similar to what occurs in Adamantisaurus and Baurutitan) while in 15%, 20%, and 25%, there is disarticulation of the zygapophyses and/or vertebral centra (which also happens in both: see figure 11 in Vidal et al., 2021) therefore, they were not considered for comparisons.

FIGURE 9.

Difference between the angle formed by the vertebral assemblage in Arrudatitan in (a) 5% and (b) 10% of the thickness of the intervertebral soft‐tissue. Not in scale.

Compared with a taxon recovered in the same geological unit, Trigonosaurus stands out with more confidence due to the sacrum presence. Considering this, it was possible to compare the “shape” and the RoM of the tail based on the common element between both: the first caudal vertebra Ca1, which served as a parameter for the RoM analysis, where measurements were taken of the angle formed by the tail in relation to the line horizontal (based on neural canal alignment) for both (Figure 7; Vidal et al., 2024: figure 11).

Adamantisaurus and Arrudatitan, on the other hand, have only part of the sequence of anterior caudal vertebrae (Ca2–Ca7 and Ca4–Ca9, respectively) (Figures 5 and 9) and the “shape” and “direction” were analyzed based on the most anterior vertebra in the sequence, for example, Ca2 in Adamantisaurus and Ca4 in Arrudatitan. However, for a more comprehensive analysis, the respective sequences from Adamantisaurus and Arrudatitan were compared to Baurutitan and Trigonosaurus (Figure 10; Table 10).

FIGURE 10.

Cartilaginous neutral pose (CNP) in Baurutitan: In the Ca2–Ca7 region in (a) 5% and (b)10%; and in the Ca4–Ca9 region; in (c) 5% and (d) 10% of the thickness of the intervertebral soft‐tissue. CNP in Trigonosaurus: In the Ca2–Ca7 region in (e) 5% and (f) 10%; and in the Ca4–Ca9 region in; (g) 5% and (h) 10% of the thickness of the intervertebral soft‐tissue. Orange color indicates intervertebral soft‐tissue thickness at 5% and blue at 10%. Not in scale.

In Adamantisaurus and Baurutitan, the CNPs show a strong tendency of the as of anterior caudal vertebrae towards the ventral direction, forming a convex curve (to a greater or lesser extent) in this part of the tail, similar to what occurs in Arrudatitan (Figures 5 and 7a–d). In Trigonosaurus, on the contrary, the anterior caudal vertebrae assume a dorsal direction and form a concave curve (Figure 10e–h). Although in the most anterior region of the anterior caudal vertebrae Trigonosaurus presents vertebrae directed in the dorsal direction, the anterior–posterior caudal vertebrae change direction and continue ventrally (as do the middle caudals) resulting in the sigmoidal shape for the tail (Vidal et al., 2024: figure 11B,C).

In the CNP with the largest vertebral sequence in Baurutitan (Ca1–Ca18), there is a significant change in the sigmoidal shape of the tail with an increase in intervertebral soft‐tissue by 10%, where the curvature in the ventral direction, mainly in the anterior caudal region, is reduced, making the shape sigmoidal less pronounced than in 5% (Figure 7). In Adamantisaurus there is a reduction in the concave curve with an increase in the thickness of the intervertebral soft‐tissue by 10%, in the same way as occurs in Baurutitan and Arrudatitan in the anterior region of the tail (Figure 5).

In Trigonosaurus, the sigmoidal shape of the tail is above the reference line (horizontal), unlike what occurs in Baurutitan (Figure 7; Vidal et al., 2024: figure 11). With the addition of intervertebral soft‐tissue in 10% for Trigonosaurus, there is also a significant change in the sigmoidal shape that becomes less pronounced than in 5%, however, with a more significant dorsal direction with an increase from −9° to −16° in 10%, different from Baurutitan which presents a reduction in the ventral direction from 21° to 20° (Vidal et al., 2024: figure 11B,C). However, it is possible to observe that the increase in intervertebral soft‐tissue favors displacement in the dorsal direction.

For the dorsoventral RoM in the largest sequences (Baurutitan and Trigonosaurus), it is also possible to observe a significant increase in the RoM of 10% for both (83°–105° in Baurutitan and 78°–83° in Trigonosaurus). In Trigonosaurus the curvature in the ventral direction is more pronounced than in Baurutitan in both 5% and 10% of the intervertebral soft‐tissue. Baurutitan has a greater capacity for movement in the dorsal direction, while Trigonosaurus has the opposite. The movement in the dorsal direction for the tail of Trigonosaurus in 5% still presents the sigmoidal shape (less accentuated) for the vertebral set, showing that even with maximum movement in this direction, the middle caudal vertebrae still allow curvature in the ventral direction in the most posterior region of the vertebral sequence (Ca11–Ca20). In Baurutitan, at dorsal maximum, the anterior vertebral sequence modulates the convex curved shape (Ca2–Ca9) while the others (Ca10–Ca19) form a concave curve (Figure 8a,b).

For the dorsoventral RoMs in the Ca2–Ca7 region, Baurutitan was the only one that showed a reduction in movement capacity from 26° to 23° (≠3°; Table S10; Figure 11e,f). For Adamantisaurus and Trigonosaurus there was an increase in RoM with the addition of intervertebral soft‐tissue by 10%. For the region corresponding to Ca4–Ca9, all tails showed an increase in RoM with increasing intervertebral soft‐tissue thickness. (Table S10).

FIGURE 11.

Range of motion in Baurutitan: Lateral maximum in the Ca2–Ca7 region at (a) 5% and (b)10%; and in the Ca4–Ca9 region at (c) 5% and (d) 10% of the thickness of the intervertebral soft‐tissue; dorsoventral maximum in the Ca2–Ca7 region in (e) 5% and (f) 10%; and in the Ca4–Ca9 region at (g) 5% and (h) 10% of the thickness of the intervertebral soft‐tissue. Orange color indicates intervertebral soft‐tissue thickness at 5% and blue at 10%. Not in scale.

For the lateral RoMs in the Ca2–Ca7 region, Adamantisaurus, Baurutitan, and Trigonosaurus show a reduction in movement capacity with an increase in the thickness of the intervertebral soft‐tissue by 10% (see Table S10; Figures 11 and 12). However, in the region corresponding to Ca4 to Ca9, Arrudatitan, Baurutitan, and Trigonosaurus show an increased capacity for lateral movement. In summary, RoMs from Ca2 to Ca7 present a reduction in movement in this region of the tail with an increase in intervertebral soft‐tissue by 10% for all dinosaurs compared. In all cases where there was a reduction in RoM, they are related to the most anterior portion of the tail between Ca2 and Ca7. However, in most cases, for example, the complete tail or the portion that corresponds to Ca4–Ca9, there is a tendency to increase the range of movement (see Table S3).

FIGURE 12.

Range of motion in Trigonosaurus: Lateral maximum in the region: Ca2–Ca7 in (a) 5% and (b)10%; and in the Ca4–Ca9 region at (c) 5% and (d) 10% of the thickness of the intervertebral soft‐tissue. Dorsoventral maximum in the Ca2–Ca7 region in (e) 5% and (f) 10%; and in the Ca4–Ca9 region at (g) 5% and (h) 10% of the thickness of the intervertebral soft‐tissue. Orange color indicates intervertebral soft‐tissue thickness at 5% and blue at 10%. Not in scale.

5. DISCUSSION

5.1. Baurutitan and Adamantisaurus tail “direction” and “shape,” and comparison with other titanosaurs

Vidal et al. (2021) elaborate the hypothesis that the tail of Arrudatitan and other Aeolosaurini could remain close to the ground, assuming a high sub‐horizontal direction due to the orientation of the anterior–posterior caudal vertebrae (Ca4–Ca9), which are strongly curved ventrally, giving the tail a sigmoidal shape. In this sense, Vidal et al. (2021) tested CNP at 5% and 10% of the thickness of the intervertebral soft‐tissue and obtained values of approximately 17° of displacement in the ventral direction for the vertebral assembly, forming a convex curve in this part of the tail (figure 9: according to Vidal et al., 2021).

This hypothesis was corroborated by the three‐dimensional analyses and the post‐mortem protonic condition of the in situ vertebral assembly, which presented the same shape resulting from the 3D analyses (Vidal et al., 2021). This research confirms the presence of the sigmoidal tail shape in advanced lithostrotians, as observed previously by Vidal et al. (2021). Furthermore, the ventrally arched anterior caudal vertebrae, forming a convex curve, are evident in Adamantisaurus and Baurutitan, taxa that are not closely related. This supports not only Vidal et al. (2021) hypothesis of a “sigmoidal shape” for Aeolosaurini‐like taxa but also for a more widespread tail condition in lithostrotians. Our results also indicate that while Baurutitan exhibits an “S” shaped curvature in the tail, it is less pronounced than that observed in Aeolosaurini‐like taxa. However, due to the absence of the sacrum in these specimens, it is difficult to establish the “direction” of the tail.

In comparisons of the caudal regions (Ca2–Ca7 and Ca4–Ca9) between Adamantisaurus, Baurutitan, and Arrudatitan, these taxa share the ventrally arched shapes with a convex curve to a greater or lesser extent. However, Trigonosaurus (Vidal et al., 2024), although it supports the sigmoidal shape of the tail, differs in the specific CNPs of the anterior region of the tail when compared with the former. In Trigonosaurus the vertebrae are directed above the horizontal line of support, and form a concave curve similar to Lirainosaurus (Vidal & Díez Díaz, 2017).

The variation in direction in the anterior region significantly influences the orientation of the middle and posterior regions of the tail. This variation can elevate or lower the entire vertebral column, in either the dorsal or ventral direction. The CNP in Trigonosaurus shows that the direction of the vertebral to the dorsal direction (low supra‐horizontal) was more favored than in the other titanosaurs studied here, although it is also evident that movements in the ventral direction (RoM) were not restricted in Trigonosaurus.

Although the sacrum is present only in Trigonosaurus, if a direction similar to that of the sacrum of Neuquensaurus robustus Lydekker (1893) and Saltasaurus loricatus Bonaparte and Powell (1980) (two of the phylogenetically closest taxa to Baurutitan) is considered as a resource to infer the direction of the tail in Baurutitan. For example, it would be closer to the ground than in Trigonosaurus. In this sense, this analysis exposes two possible “directions” and “shapes” for the tail of titanosaurs studies here: (1) “sigmoidal‐concave shape,” with an elevate direction (in relation to the ground), as occurs in Trigonosaurus and Lirainosaurus (Figure 13a) and (2) “sigmoidal‐convex shape,” with sub‐horizontal direction (without touching to the ground), as occurs in Arrudatitan, Adamantisaurus, and Baurutitan (Figure 13b).

FIGURE 13.

(a) sigmoidal‐convex shape where the anterior region of the tail forms a convex arc and (b) sigmoidal‐concave shape where the anterior region of the tail forms a concave arc. Arrows indicate concave or convex regions. The silhouettes represent the dinosaurs studied, like Baurutitan and Trigonosaurus. The silhouette in (c) represents a standard sauropod.

In the sigmoidal‐concave tail model of Trigonosaurus, the anterior caudal vertebrae (Ca1–Ca6) are directed above the horizontal line. However, starting from the Ca9 vertebra, the caudal vertebrae are directed ventrally along the sequence to the sub‐horizontal direction, giving a “sigmoidal shape” to the caudal vertebral assembly (Figure 13; Vidal et al., 2024: figure 11). In Baurutitan, as occurs in Trigonosaurus, the first anterior caudal vertebrae (Ca1 and Ca2) are dorsally directed (supra‐horizontal in CNP), however, from Ca3 the vertebrae are strongly directed ventrally toward the distal region of the middle caudal vertebrae, where they return to the dorsal direction giving the tail a sigmoidal‐convex shape (Table S6; Vidal et al., 2024: table S9). The general shapes of the tails of Baurutitan and Trigonosaurus are similar but have an opposite conformation. Baurutitan presents the anterior caudal vertebrae (except Ca1 and Ca2) are directed to the ventral direction (convex curve) and in the middle section return to the dorsal direction, in Trigonosaurus the anterior caudal vertebrae are inclined to the dorsal direction (concave curve), while the in middle caudal vertebrae are directed ventrally (Figure 13a,b; Table S6; Vidal et al., 2024: table S9).

The anterior regions of the tail of Baurutitan and Trigonosaurus have vertebrae that can be compared in the anterior caudal region, and both show a tendency for the first two caudal vertebrae to be dorsally directed, elevating the direction of the tail. This morphological configuration could influence the tail direction in Adamantisaurus and Arrudatitan, where it was not possible to observe directly the first anterior vertebrae (Ca1).

5.2. Thickness intervertebral soft‐tissue variation in the tail of macronarian sauropods

Díez Díaz et al. (2020) concluded that intervertebral soft‐tissue thicknesses between 10% and 15% proved to be more parsimonious for the assessment of CNP in the macronarian sauropod Giraffatitan, where the vertebrae are amphicoelous. For the primitive lithostrotian Tiamat, similar measurements were found for middle caudal vertebrae, which have no defined polarity (i.e., procoelous or opisthocoelous) (Pereira et al., 2024). On the other hand, recent analyses of the caudal region of more closely related Macronaria such as Lirainosaurus (vertebrae deeply inserted into their adjacent ones) and Arrudatitan, which have strongly procoelous or opisthocoelous vertebrae, show measurements ranging between 5% and 10% of the thickness of the intervertebral tissue (Vidal et al., 2021; Vidal & Díez Díaz, 2017, respectively).

There are still few analyses that clearly express the variation in intervertebral soft‐tissue thickness in basal Macronaria. However, previous studies indicate a tendency towards greater intervertebral soft‐tissue thickness in non‐concavo‐convex vertebrae articulation (Díez Díaz et al., 2020; Pereira et al., 2024; Vidal et al., 2024; Wedel et al., 2018). For example, Tiamat, presents two ways: (1) when the vertebrae were aligned in ONP without considering the thickness of intervertebral soft‐tissue, the resulting intervertebral space corresponds to ~15% of the length of the vertebral centrum and (2) when the vertebrae were aligned considering the CNP, the most parsimonious thicknesses for the intervertebral soft‐tissue were between 10% and 15% of the length of the vertebral centrum.

The middle caudal vertebrae of Tiamat do not have convexity at either end, on the contrary, they are amphicoelous and have two slight concavities (Figure S1A). For this, there is a proportionally greater distance between the anterior or posterior surfaces of the vertebral centrum with the adjacent ones. As a consequence, the results presented are greater intervertebral soft‐tissue thicknesses than those that have concave‐convexity (procoelous or opisthocoelous). Still in this sense, shorter intervertebral soft‐tissue thicknesses in more derived macronarians may be related to the polarization of the vertebral centra (the presence of procoelous or opisthocoelous condition) (Figure S1B).

The function of strong procoelous/opisthocoelous articulations in more derived titanosaurs was theorized by Fronimos et al. (2016), and Fronimos and Wilson (2017), as a response to tensional forces caused by the strong traction of the epaxial and hypaxial muscles. Therefore, the vertebrae would be deeply inserted into their adjacent ones, maintaining a reduced space between the articular faces of the vertebral centra proportionally smaller than those presented by amphicoelous vertebrae. More studies to analyze and compare the caudal region in sauropods should be carried out to generate more comparative data that can corroborate or refute these hypotheses.

5.3. The M. caudofemoralis longus in sauropod tails

Baurutitan is the first of the studied Brazilian taxa that presents all morphological features that allow us to infer the origin and insertion of M. CFL due to the completeness of its caudal sequence, while Trigonosaurus, albeit presents a great number of caudal vertebrae (>7), has not a complete sequence, lacking mostly haemal arches. In this way, due to the repetition of caudal elements, this reduces the degree of reliability of the origin reconstruction of the M. CFL. Major anatomical differences were observed that influence the extension of the M. CFL between these taxa (Figure 14a,b).

FIGURE 14.

Reconstruction of the M. caudofemoralis longus in (a) Trigonosaurus; (b) Baurutitan; (c) Adamantisaurus; and (d) Arrudatitan. Light gray represents the entire space occupied by M. CFL (caudofemoralis longus muscle) at its origin. Dark gray represents the point of greatest influence of M. CFL, for example, where there could be greater resistance to forces exerted to move the femur. Not in scale.

As first discussed by Salgado and García (2002), and later in further depth by Ibiricu et al. (2013), the primary lateral surface in Baurutitan is presented solely until Ca9, together with the greater reduction of the transverse process in the caudal vertebrae sequence. The primary lateral are concave, more pronounced in the distal third of the vertebral centrum (more ventrally where there is rugosity, Figure 14b, dark gray) in the most anterior caudal vertebrae. This concavity is interpreted here as being the origin of the M. CFL on the lateral side of the vertebral centra (Figure 14b, dark gray). Between the Ca8 and Ca10 vertebrae, the primary lateral face migrates to the most ventral region of the vertebral centra, marked by the ridge formed between the intersection of the primary and secondary faces (as predicted by Salgado & García, 2002; Figure 14b). Despite the distal termination of the transverse processes and the ventralization of the primary lateral, the protuberance on the lateral surface of the haemal arches (interpreted as the origin of M. CFL) persists until the Ca15, albeit in a reduced form between Ca14 and Ca15. In this case, the persisted presence of the reduced M. CFL, turns this muscle more ventrally situated in the vertebral centrum, or only in the haemal arches up to the 15th, beyond the limits of the transverse processes. This configuration confirms the proposal of Ibiricu et al. (2013), where Baurutitan presents evolutionary state 1, as a non‐Saltasaurini Titanosauria‐Lithostrotia (although the termination is not in Ca12; Ibiricu et al., 2013: figure 6B). Similarly, the M. CFL in Giraffatitan (basal Macronaria) was interpreted as having a distal end beyond the end of the transverse processes, in the most ventralized region of the vertebral centra and lateral haemal arches, up to the 15th (Díez Díaz et al., 2020).

In Trigonosaurus there is no presence of secondary lateral surfaces. In this case, the M. CFL interpretations were limited to the correlates of the transverse and lateral faces of the vertebral centra. This approach may exhibit a lower degree of precision. However, the concavity of the lateral surface, which is more pronounced in the distal third of the vertebral centra (Figure 14a, light gray), with roughness more ventrally (Figure 14a, dark gray: similar to Baurutitan) indicates that the M. CFL originated on the lateral side of the vertebral centra. This concavity continues up to the middle caudal vertebrae (more ventrally than in the previous ones), along with the transverse processes that probably continue beyond Ca20. Therefore, considering these two points, it can be considered that the M. CFL in Trigonosaurus could have been more extensive than in Baurutitan. In the proposal by Ibiricu et al. (2013), Trigonosaurus would have a primitive Titanosauria condition, in evolutionary stage 0 (see Section 1.2. above; Ibiricu et al., 2013: figure 6A).

In Adamantisaurus and Arrudatitan, it is possible to verify the place of origin of M. CFL only in the most anterior portion of the anterior caudal vertebrae (Ca2–Ca7 and Ca4–Ca9, respectively), in addition to the presence of haemal arches (two in Adamantisaurus and five in Arrudatitan) that can support the inferences. For both, the side of the vertebral centra is reasonably concave, being more pronounced in the distal third of the vertebral centrum (Figure 14c,d, light gray), with roughness more ventrally (Figure 14c,d, dark gray: similar to Baurutitan and Trigonosaurus), indicating the surface of origin of M. CFL. In the haemal arches of both, there are lateral protuberances, where the origin of the M. CFL is inferred. In the haemal arches of both, there are lateral protuberances, where the origin of the M. CFL is inferred. Díez Díaz et al. (2020) report the presence of a transverse ridge on the lateral surface of the haemal arches of the titanosauriform Giraffatitan brancai and interpret it as an anchoring point for the M. CFL, however, no conspicuous ridge is observed on the most anterior caudal vertebrae in any of the specimens studied here. Nevertheless, between the 14th and 16th haemal arches in Baurutitan and the 14th(?) and 16th(?) haemal arches (position uncertain, but in the region) in Arrudatitan, there is a transverse ridge on the side of the haemal arches that may indicate distal anchorage points of the M. CFL in these titanosaurs (Figure S2). In these sauropods, the origin of the M. CFL is interpreted as occurring in the concavity of the sides of the vertebral centra, on the sides of the anterior haemal arches (osteological correlates) and possibly being influenced by the presence of the transverse processes.

Díez Díaz et al. (2020) describe the morphology for the caudal sequence of Giraffatitan and discuss the morphological variation at the expense of the tail musculature. According to the authors, the nine anteriormost caudal vertebrae present the ventral half of the centrum as convex, becoming gradually more concave (just below the transverse process). For the authors, the distal lateral surface of the centrum acts as an attachment for the M. CFL. In contrast, in the species studied here there is a concavity that occupies two‐thirds (anteroposteriorly) of the lateral face of the vertebral centra and that is more accentuated in the distal third, which, in turn, is more accentuated in the ventral third. This concavity is more pronounced in the distal third, which, in turn, is more pronounced in the ventral third. Although the concavity of the second third of the lateral side of the vertebral centra is less pronounced, there are foramina in it that may indicate an anchoring site for the M. CFL, most conspicuous in Adamantisaurus but also present in Arrudatitan, Trigonosaurus, and Baurutitan, at least in the most anterior caudals (Figure S2).

In this research, the entire extension of the concavity on the side of the vertebral centra was interpreted as the place of origin of M. CFL. However, we understand that the distal third, where the concavity is deeper (and there is roughness; Figure 14, dark gray), may indicate the location where the M. CFL exerted greater influence. This more pronounced concavity in the anterior caudal vertebrae (e.g., Ca2 and Ca6 of Baurutitan) occupies the largest dorsoventral extension of the distal third of the lateral side of the vertebral centra but migrates to the more ventralized region in the anterior–posterior vertebrae (e.g., Ca7–Ca9 of Baurutitan) and in the middle vertebrae (when they exist). The same thing occurs with the more pronounced concavity in Trigonosaurus (from Ca12 onward it becomes more ventralized), in Adamantisaurus and Arrudatitan (from Ca7 onward for both).

According to Díez Díaz et al. (2020) most of the M. CFL is located in the anterior region, exerting the greatest influence on the tail. As postulated by Salgado and García (2002), the action of the M. CFL on the anterior caudal vertebrae could have significantly changed the anatomical shape of the lateral vertebral centra in some titanosaurs. For example, in the titanosaurs analyzed here, there is a noticeable abrupt change in the anterior–posterior caudal vertebrae proportions, especially in regions where the muscle became less voluminous and migrated to the ventral region of the vertebral centra. This variation was probably due to the need for a very robust lateral or vertebral centra for fixing the M. CFL, different from what occurs in the first caudals where the vertebrae are more robust and with large concavities (sometimes rough and with foramen) on the sides.

5.4. Transition point on the tail of titanosaurs

Russell (1972) defined the transition point as the region at the tail in theropod dinosaurs between the last vertebrae with transverse processes and the first vertebra with distinctly elongated prezygapophyses. However, recently, the determination of the transition point has been modified to recognize the loss of transverse processes, extreme reduction of neural spines, elongation of caudal centra, and gain of inverted T‐shape haemal arches, but not by prezygapophyseal elongation (Senter, 2011). Based on these criteria (although they are adopted for theropod dinosaurs), in Baurutitan the transition point cannot be well defined, since the transverse processes end between Ca9 and Ca10, but the neural spines decrease between Ca13 and Ca16. This point of the tail is also where the haemal arches migrate from a “Y” shape to a “V” shape, this probably also occurs in Arrudatitan (Vidal et al., 2021).

As highlighted by Ibiricu et al. (2013), the transition of the caudal vertebrae in sauropods is complex. For example, unlike Baurutitan, in Trigonosaurus the transverse processes develop beyond Ca9 (beyond Ca20) and this probably also occurs in Arrudatitan (Vidal et al., 2021, 2024). The prezygapophyses in Baurutitan, Trigonosaurus, Adamantisaurus, and Arrudatitan are elongated from Ca3, being even more conspicuous in Aeolosaurini (Kellner et al., 2005; Campos et al., 2005; Santucci & Arruda‐Campos, 2011; Santucci & Bertinni, 2006; Ibiricu et al., 2013; Vidal et al., 2021). With all this variation, it is difficult to establish the caudal regions (e.g., anterior, middle, and posterior).

Despite this, this research shows the variation in proportions between the anterior caudal vertebrae that marks a transition point between the anterior–middle and anterior–posterior caudal vertebrae. In Baurutitan, Adamantisaurus and Arrudatitan, abrupt changes in the shape and proportion between the anterior caudal vertebrae were observed: between the Ca2 and Ca6 vertebrae, for Baurutitan, the side of the vertebral centra is more regular with similar height and length (similar to a square in lateral view), but changes to a rectangular shape (elongation of the caudal centra) and, must be emphasized, with a smaller proportion between the Ca6 and Ca7 vertebrae (Figure 15a,b); this variation is observed most conspicuously in Adamantisaurus and occurs between the Ca5 and Ca6 vertebrae (Figure 15c,d); and is even more striking in Arrudatitan which occurs between vertebrae Ca6 and Ca7 (Figure 15e,f). Even though these transitional vertebrae (with variation in shape and proportion) were not originally articulated (thus sequenced during fossil prospecting), they could be easily digitally reconstructed so that the transition would be smoother than it actually is, that is, in a mistaken way. This is because 3D reconstruction is based on the proportions of the anterior and/or posterior vertebrae (when they exist) and the postulated concept that vertebrae vary little throughout the vertebral sequence (Myhrvold & Currie, 1997; Vidal & Díez Díaz, 2017).

FIGURE 15.

Location of abrupt (or relatively abrupt) changes in the morphology of the caudal vertebrae of (a) Baurutitan between the Ca6 and Ca7 vertebrae; (b) detail of the transition between the Ca6 and Ca7 vertebrae of Baurutitan (less evident discontinuity); (c) Adamantisaurus between vertebrae Ca5 and Ca6 (obvious discontinuity); and (d) Arrudatitan between the Ca6 and Ca7 vertebrae (more evident discontinuity). Yellow and red vertebrae indicate the point of discontinuity (abrupt morphological variation); (e) original Adamantisaurus vertebrae aligned with dashed lines showing the discontinuity and (f) original vertebrae from Arrudatitan aligned with dashed lines showing the discontinuity. Not in scale.

This variation in proportions can also be observed, to a greater or lesser extent, in other titanosaurs: In the colossosaurians, this variation seems to carry an evolutionary signal associated with the anteriormost caudal vertebrae. For example, Rinconsaurus presents a variation occurring between the Ca4 and Ca5 vertebrae (Pérez Moreno et al., 2022; Figure S3A); the unnamed specimen MOZ‐Pv 1221, presents between the Ca9 and Ca10 vertebrae (Otero et al., 2021; Figure S3B); while more advanced forms, such as Overosaurus, this change occurs between the Ca10 and Ca11 vertebrae (Coria et al., 2013; Figure S3C); and in Punatitan occurs between Ca8 and Ca9 (Figure S3D). Nevertheless, in the saltasauroid lithostrotians, the variation can be observed in the posteriormost caudal vertebrae, as in Alamosaurus, which occurs between Ca12 and Ca13 (Gilmore, 1946; Figure S3E); while in Dreadnoughtus, occurs between caudal vertebrae Ca7 and Ca8 (Figure S3F).

In the titanosaurs studied here, Adamantisaurus, Arrudatitan, and Baurutitan, this variation is reflected in the alignment of the vertebrae in CNP, where in this region the curvature in the ventral direction is more accentuated for these dinosaurs (Figure 15a,c,e). In the physical alignment of the caudal vertebrae of Adamantisaurus and Arrudatitan in a linear way and without CNP, this variation is even clearer (Figure 15e,f). This morphology and direction may be associated with the ventralization of the M. CFL in this region until its most distal end, which is where the direction of the tail changes again to the dorsal direction, giving a sigmoidal appearance to the set of caudal vertebrae (Figures 14b and 15a). The lateral concavities of the vertebral centra of the medium and large titanosaurs studied here (e.g., Adamantisaurus and Arrudatitan, respectively), there are foramina (best defined in Adamantisaurus), possibly reflecting a powerful hindlimb retraction muscle.

The neural spines of the anterior caudal vertebrae of Trigonosaurus are directed anteriorly, while in Baurutitan the neural spines are directed posteriorly. This morphology indicates osteological adaptations concomitant with the shapes presented by the vertebral assemblies of these small dinosaurs (sigmoidal‐concave and convex, respectively). In dinosaurs with larger proportions and tails with a sigmoidal‐convex shape, such as Adamantisaurus and Arrudatitan, the neural spines are proportionally larger than in Baurutitan and the proximal third of the neural spines (the base) are directed posteriorly, but become directed more vertically or more anteriorly in the most distal portion of the neural spines, in relation to the vertebral centrum. These variations may be the result of the action of the epaxial muscles that act as stabilizers and extensors of the tail (Díez Díaz et al., 2020).

5.5. Sigmoidal‐shaped tail, its influence on movement and paleobiological interpretations for Brazilian titanosaurs

In other sauropod tail reconstructions, such as Giraffatitan and Spinophorosaurus, the results are sub‐horizontal directions and linear shapes (Díez Díaz et al., 2020; Vidal et al., 2020). Albeit few works concern the shape and direction of the tails in Titanosauria, Vidal and Díez Díaz (2017), observed that Lirainosaurus had a horizontally directed tail, which also had a “sigmoidal shape” due to two variations in the caudal sequence: one in Ca4 that made the set more horizontal (linear) and another abrupt variation of direction in the Ca9 vertebra toward the ventral direction (Figure S4). Since the authors did not use the sacrum to guide the positioning of the anteriormost caudal vertebrae, it is likely that this interpretation may be strongly biased. We can only infer the “shape” of the tail of Lirainosaurus, which would be “sigmoidal‐concave” (Figure 13) due to the concave curvature in the anterior tails, similar to what occurs in Trigonosaurus. However, as in Trigonosaurus, Lirainosaurus also presents an incomplete caudal sequence and, therefore, the reconstruction of missing vertebrae was necessary (Ca3–Ca7) (Vidal & Díez Díaz, 2017), and inferences about “direction” and “shape” of the tail as well as changes in vertebral centra are of low certainty.

The conclusions of Vidal et al. (2021) on the shape of the tail ofArrudatitan, as already mentioned, were based on observations of in situ fossil material that remained in a protonic posture (curved ventrally) indicating that this could be a possible pose. Later, another titanosaur presents the post‐mortem protonic posture (e.g., Punatitan Hechenleitner et al., 2020; Figure S4E), reinforcing the hypothesis of Vidal et al. (2021). The ventral curvature in this region of the tail was also found here (in this research) for the titanosaurs Baurutitan and Adamantisaurus.

Vidal et al. (2021) also reconstructed the sacrum for Arrudatitan based on the sacrum of Trigonosaurus due to the phylogenetic proximity of both taxa and concluded that the distal portion of the tail could have been close to the ground, but without touching it. Furthermore, other sacra used in the reconstruction by Vidal et al. (2021: e.g., Overosaurus and Rinconsaurus) have similar directions to those of Trigonosaurus. The “direction” of the tail of Baurutitan and Adamantisaurus cannot be directly inferred due to the absence of the sacrum, only speculated (similar to Vidal et al., 2021; see Section 5.1), and their shapes would be similar to that proposed by Vidal et al. (2021) with the tail more directed towards the ground to a greater or lesser degree (Figure S5). However, it should be noted that this approach may generate incomplete or misleading data about the direction of the tail.

Arrudatitan and Adamantisaurus were medium to large titanosaurs, whereas Baurutitan was a comparatively smaller sauropod. This could indicate that size does not determine the presence or absence of a sigmoidal‐convex tail closer to the ground, with this morphology probably offering advantages beyond gigantism. Vidal et al. (2021) suggested that the ventrally curved anterior portion of the tail (sigmoidal‐convex shape) could provide a greater moment arm (i.e., mechanical advantage) for the M. CFL, thereby reducing the force required to generate torque for femur movement.

This research agrees with this hypothesis since the relationship between muscular strength (F _mus) and the “moment” developed around the joint (M _joint) can be expressed by the following equation (Nagano & Komura, 2003):

$$M_{\text{joint}}=d\cdot F_{\mathrm{mus}}$$

where “d” represents the length of the “moment arm” of the vector of the force exerted (Figure S6). In this case, as we do not know the exact volume and force exerted by the M. CFL in these sauropods, the moment arm can indicate the variation (more or less) of this force (without necessarily knowing what the force is).

The moment arm can be defined as the shortest distance between the line of action of a muscle‐tendon and the center of rotation of a joint as shown in Figure S6 (Hutchinson et al., 2005; Nagano & Komura, 2003) and can be easily estimated from the calculation of two variables (basic trigonometry): the angle α, formed between the line of muscular action and the length of the line that connects the center of rotation and the point of muscular origin (X) (Figure S6). As a result, a right‐angled triangle is obtained in which one of the acute angles and the hypotenuse are known (Hutchinson et al., 2005; Preuschoft & Klein, 2013) and the moment arm (d) corresponds to the opposite side of the known angle and its value can be estimated using a trigonometric equation (d = X · sen α).

The primary movement of the femur involves flexion‐extension, with the M. CFL serving as the main retractor. The moment arm, therefore, could illustrate potential mechanical advantages in the caudal forms presented here. For instance, in sauropods with sigmoidal‐convex tails that are more ventrally curved in the anterior region, the moment arm is larger in all conditions (neutral pose, flexion, and extension; Figure S6E–G). This, combined with the more ventralized ending of the M. CFL, may indicate a reduced need to apply force during certain movements. Linear and sigmoidal‐concave tails exhibit less mechanical advantage compared to the sigmoidal‐convex tail.

For these dinosaurs, the mechanical advantage gained during femur retraction by the M. CFL could produce/result from a paleobiological adaptation in the tail, where this muscle originates. For example, such an adaptation might support behaviors involving hindlimb retraction, such as assuming a tripodal posture during feeding or engaging in combat. Furthermore, using the tail as a stabilizing support during nesting (digging holes with the hind limbs; Figure 16a) or during combat in a potential pentapodal posture, which may or may not culminate in a tripodal stance (Figure 16b,c).

FIGURE 16.

The illustrations represent the tail closest to the ground in possible situations, as proposed in this work: (a) using the pentapodal posture to stabilize the tail in acts such as digging with the hind limbs; (b) using the five‐legged posture to stabilize the tail during femoral retraction during possible inter‐ or intraspecific disputes; (c) using the tripodal posture in battles; and (d) using the tail as a ventral weapon against a small Theropoda.

In contrast, Trigonosaurus, which possesses a sigmoidal‐concave and horizontally directed tail (more elevated than in Adamantisaurus, Arrudatitan, and Baurutitan), exhibits distinctive anatomical features. These include the absence of a more ventralized primary lateral and the continuation of the transverse processes beyond Ca20, suggesting a relatively long muscle. In this configuration, the M. CFL would likely support greater force or mobility in the lateral direction by contributing to lateral flexor muscles. Such lateral forces could potentially facilitate the use of the tail as a defensive mechanism. These adaptations result in a variety of “tail shapes” that go beyond the “linear horizontalized” tail recognized for basal Macronaria and suggest a diverse range of functions other than counterbalancing the weight of the neck. Tails whose efficiency is found in forms directed at the height or above the hip and with elongated M. CFL (e.g., Trigonosaurus) may indicate that they could serve as a weapon with possible laterolateral and/or dorsoventral movements (supposed lateral or top‐down attack) in disputes over territory or fight for survival against predators (Figure 16d). Lastly, since direct analyses with 3D elements and reconstructions of the M. CFL were not conducted to verify the variation of the moment arm, further studies are necessary to corroborate or refute these hypotheses.

5.6. Biomechanical implications for the concave or convex sigmoidal tail of titanosaurs

5.6.1. Inferences of sauropod habits through biomechanical and ichnological evidence

According to Wilson and Carrano (1999), footprints are the best records of sauropod locomotion, as they represent a moment in time when these animals walked on a surface, making it possible to infer the morphology and movement (gait and posture) of these dinosaurs. Most sauropod footprints depict moments when the dinosaurs were walking or running (alone or in a herd) (González‐Riga, 2011; Lockley et al., 1994; Myers & Fiorillo, 2009; Wright & Breithaupt, 2002), or supposedly fleeing from a predator (Bates et al., 2009; Lockley et al., 1994).

Based on the premise of Wilson and Carrano (1999), the tracks or footprints of sauropod dinosaurs only show some moments of the sauropods' life close to places that allow the preservation of footprints. In this regard, Carvalho and Leonardi (2021) argue that the origin and preservation of a trail are related to many different environmental factors, especially concerning the substrate's cohesion, plasticity, grain size, texture, and water content.

In this sense, far from these appropriate locations, the traces that could represent different habits of these animals would not been preserved and/or recovered so far. For example, there is no ichnological evidence of battles between sauropods and there are few records of possible tail touch marks on the ground (Foster et al., 2000: figure 5). Foster et al. (2000) reported a trail of sauropod footprints associated with tail marks on the ground and Dalla Vecchia et al. (2000: figure 30A–D) demonstrate a possible tail mark touching the ground below a sauropod. Furthermore, several authors demonstrate anatomical and mechanical evidence that indicates the possibility of the tail touching the ground by sauropods and that they could support the tripodal or pentapodal posture (Borsuk‐Białynicka, 1977; Mallison, 2011b; Paul, 1998, 2017; Vidal et al., 2024). Rothschild and Berman (1991) point out that tail drag marks are uncommon in the sauropod footprint record, but argue that sauropods, like probably most dinosaurs, could occasionally drag their tails on the ground, although they kept them elevated to avoid injuries on rough ground or being trampled by other members of the herd.

Biomechanical studies corroborate the hypothesis that sauropods would keep their tails suspended above the ground (Vidal & Díez Díaz, 2017; Vidal et al., 2020, 2021, 2024), but despite the absence of these records, biomechanical analyses pointed to the possibility that some titanosaurs could eventually touch the ground with their tail, for example, in a possible tripodal or pentapodal posture (Borsuk‐Białynicka, 1977; Mallison, 2011b; Paul, 1998, 2017; Vidal et al., 2024; and this work). The absence of tail marks on the ground (ichnological evidence) does not mean the non‐existence of such habits (e.g., tripodal and pentapodal posture) nor does it nullify the anatomical and biomechanical evidence, it only shows (at least moment) that these marks were not preserved due to environmental conditions.

5.6.2. Pneumatized vertebrae in sauropods and the biomechanical implications for tripodal and pentapodal posture

One of the most striking features of sauropod vertebrae is pneumatization (Wedel, 2005, Sereno et al., 2007; Schwarz‐Wings et al., 2009). Wedel (2003) discussed the evolutionary features of pneumatic vertebrae in sauropod dinosaurs and highlighted physical properties based on the architecture of the internal spaces (pneumatic fossae): camellate and camerae (Wedel, 2003: figure 1). For the author, more derived sauropods such as Saltasaurus would have camellate caudal vertebrae (Wedel, 2003: figure 1C), as well as other titanosaurs, as also observed by Wedel and Taylor (2013), Zurriaguz and Cerda (2017) and Navarro et al. (2022: figure 7). In this research, no analysis of internal structures in the vertebral centra of the caudal vertebrae of the titanosaurs studied was performed, therefore, it was not possible to directly observe such structures. However, due to the phylogenetic proximity between these titanosaurs, we assume that they are camellate similar to the vertebrae of Saltasaurini as suggested by Wedel (2003) and Wedel and Taylor (2013) and observed by Navarro et al. (2022) (future studies are needed to verify this hypothesis).

These structures are commonly associated with the respiratory system in sauropods similar to that of birds (in addition to being related to weight reduction for gigantism), with characteristics different from what occurs with trabecular porous bones in mammals (Perry & Sander, 2004; Sander, 2013; Schwarz‐Wings et al., 2009; Wedel, 2003, 2005). However, Sanz et al. (1999) discussed these structures in the vertebrae of sauropod dinosaurs, associating them with the spongy internal structure. In this sense, bones with porosities (internal spaces similar to those of camellates) provide a certain plasticity to the bone tissue, allowing the bone to have a great capacity for energy storage, acting in the absorption of energy and distribution of pressures when loads are applied to the skeletal structure (Hamill et al., 2015). Schwarz‐Wings et al. (2009) analyzed the internal physical characteristics of sauropod vertebrae through finite element analysis (FEA) and agreed that pneumaticity in these vertebrae may play the role of damping large stresses distributed throughout the bone. In addition, Wedel and Taylor (2013) commented that the caudal vertebrae in sauropods have lower levels of pneumaticity than the others (cervical, dorsal, and ilium).

Analyses such as CNP and RoM are attributed to “kinematic biomechanics” which evaluates the movements allowed by the bones without worrying about the elements that produce the movement (e.g., muscles and tendons). On the other hand, analyses that aim to evaluate the resistance capabilities of bones to forces and tensions are in the field of “kinetic biomechanics,” such as FEA (Rayfield, 2007). In this research, concepts of kinematic biomechanics were applied, where CNP and RoM allowed us to infer that sauropod dinosaurs had a sigmoidal‐convex tail that could eventually touch the ground. Although the caudal vertebrae of sauropods pneumatized (but less than in other parts of the axial skeleton), this characteristic could hypothetically confer plasticity to the vertebral unit (and to the set), as occurs in trabecular bones, in response to weight loads.

However, we emphasize that when inferring behaviors in extinct organisms through structural analyses such as FEA, we must consider the complexity involving bones (and not only their mechanical properties), such as soft‐tissues (e.g., muscles, epithelial tissue, fats, etc.). The caudal vertebrae of titanosaurs are inserted into an anatomical context that includes soft‐tissues that, in certain cases, also have mechanical properties that help support large loads. For example, in elephants (living organisms), the feet are formed by a set of small articulated bones that support the great body weight and, for this purpose, they develop thick layers of skin and fat that cushion the loads (Benz et al., 2009: figure 1). For the tail of sauropods, the soft‐tissue (muscles and fats), associated with pneumatized bones with possible plastic structural properties, could have contributed to a posture where the tail touches the ground in possible behaviors (e.g., tripodal or pentapodal posture).

6. CONSIDERATION

Brazil has sauropods mostly belonging to the Titanosauria clade and are represented by small to medium‐sized forms. Biomechanical analyses shows that these titanosaurs could have varied tail shapes and some could be shorter than previously thought. The tail shape of Adamantisaurus and Baurutitan can be assessed as sigmoidal‐convex and would probably be directed closer to the ground, as in Arrudatitan. Trigonosaurus, on the other hand, has a tail with a sigmoidal‐concave shape and being directed dorsally, higher than Adamantisaurus, Arrudatitan, and Baurutitan. Although there is variation in the direction of the tail (higher or lower), the most common shape for these titanosaurs is the sigmoidal (S‐shaped).

The M. CFL probably acted to modify the morphology of the anterior caudal vertebrae in some titanosaurs as it was the main retractor of the femur. The distal termination of the M. CFL could have abruptly modified the shape of the lateral side of the vertebral centra and the proportion of the anterior–posterior caudal vertebrae, conditioning the vertebral assembly toward a ventral direction. In this sense, the sigmoidal‐convex shape (convex anterior caudal region) for the Adamantisaurus, Arrudatitan, and Baurutitan tails could have conferred a greater mechanical advantage for the M. CFL, enabling the use of the tail as a fifth limb in a potential pentapodal posture.

AUTHOR CONTRIBUTIONS

LSV conceived the ideas, designed the methodology, led the writing of the manuscript, analyzed the data, and designed tables and figures. PVLGCP and KLNB contributed to data analysis, methodology, and editing of the manuscript. TBR contributed photographs and reviews. LPB and CRAC are LSV's academic advisors and contributed with the means to carry out the project, as well as writing the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

FUNDING INFORMATION

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) awarded the senior author's grant (88887.492892/2020‐00) that made this study possible.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Appendix S1.

Vidal, L.S. , Bergqvist, L.P. , Candeiro, C.R.A. , Bandeira, K.L.N. , Tavares, S. , Ribeiro, T.B. et al. (2025) Biomechanics and morphological comparisons of the caudal region of titanosaurs from the Cretaceous of Brazil: Paleobiology and paleoecology inferences. Journal of Anatomy, 246, 20–44. Available from: 10.1111/joa.14134

DATA AVAILABILITY STATEMENT

The data underlying this article will be shared on reasonable request to the corresponding author.