Ardetosaurus viator, van der Linden & Tschopp & Sookias & Wallaard & Holwerda & Schulp, 2024

van der Linden, Tom T. P., Tschopp, Emanuel, Sookias, Roland B., Wallaard, Jonathan J. W., Holwerda, Femke M. & Schulp, Anne S., 2024, A new diplodocine sauropod from the Morrison Formation, Wyoming, USA, Palaeontologia Electronica (a 49) 27 (3), pp. 1-79 : 8-54

Ardetosaurus viator gen. et sp. nov.

Holotype. MAB011899 : two cervical vertebrae, 10 dorsal vertebrae, sacrum, five caudal vertebrae, eight dorsal ribs, two chevrons, a left coracoid, a left ilium, both pubes, both ischia, a left femur, a left tibia, and a partial left fibula.

Diagnosis. Ardetosaurus viator is diagnosed by the combination of the following autapomorphies: 1) the presence of distinct, paired accessory laminae in the spinoprezygapophyseal fossae (SPRF) in the posterior cervical and anterior dorsal vertebrae, 2) anteroventrally bifurcating anterior centrodiapophyseal laminae (ACDLs) in the anterior dorsal vertebrae, 3) the presence of centropostzygapophyseal lamina fossae (CPOL-f) in the second dorsal vertebra, 4) a vertebral height/centrum length ratio of <2.5 of the posterior dorsal vertebrae, and 5) reduced or absent centroprezygapophyseal laminae (CPRLs) in the anterior-most caudal vertebrae. Ardetosaurus viator differs from all other diplodocines by having unbifurcated CPRLs in the posterior cervical vertebrae and a highly elliptical femoral cross-section. Ardetosaurus viator differs from Amphicoelias Cope, 1878 , in lacking the rounded, lateral projections of the neural spine tip and the thin neural spine base in the dorsal vertebrae; from Barosaurus Marsh, 1890 , by having tall cervical neural spines, single midline keels, narrower prezygapophyseal rami in the cervical vertebrae, ten dorsal vertebrae, the presence of infradiapophyseal foramina in the dorsal vertebrae, unbifurcated caudal neural spines and a posterodorsally expanded distal end of the ischia; from Diplodocus by having more elongated posterior cervical vertebrae, postzygapophyses that terminate in front of the cotyle edge, the presence of a prespinal lamina (PRSL) and laterally inclined spinoprezygapophyseal laminae (SPRLs) in the posterior cervical vertebrae, the absence of the midline cleft in the dorsal vertebrae and the presence of lateral projections on caudal neural spines tips; from Galeamopus Tschopp et al., 2015a , by having posteriorly projecting interpostzygapophyseal laminae (TPOL) in the posterior cervical vertebrae, strongly opisthocoelous anterior dorsal vertebrae, and the absence of a second cnemial crest in the tibia; from Kaatedocus in lacking the rugose tuberosities and transverse sulci posterior to the prezygapophyses, by having a wider gap between the metapophyses, and postzygapophyses which terminate in front of the rim of the cotyle of the cervical vertebrae; from Leinkupal in having procoelous-distoplatyan caudal vertebrae without distinct pleurocoels and with less developed transverse processes; from Tornieria in lacking strongly procoelous anterior caudal vertebrae, having a mildly concave ischial acetabular surface, having elongate lateral fossae on the ischial shaft and a more transversely expanded distal end of the ischia; and from Supersaurus Jensen, 1985 by being much smaller ( S. vivianae ), the presence of bifurcated neural spines in the cervical vertebrae ( S. lourinhanensis ), the absence of distinct grooves posterolateral to the parapophyses, and paired fossae lateral to the midline keel in the cervical vertebrae, anteriorly inclined neural spines, horizontal transverse processes, and less persistent bifurcation along the series in the dorsal vertebrae, as well as lacking pneumatic foramina and oblique ridges on the thoracic rib heads (both species).

Etymology. ‘Ardeto’ is an inflection of Latin ārdēre, meaning ‘to burn.’ It refers to the history of some of the elements, which were either fully destroyed in a fire, or still show burn scars from the fire. ‘saurus,’ Latinized form of the Greek σαῦρος (saúros), meaning lizard or reptile. ‘viator’ is Latin for traveler, referring to the journey of the specimen from the USA, via Switzerland and Germany, to the Netherlands.

Locality and horizon. Ardetosaurus viator comes from the Howe-Stephens Quarry of northern Wyoming, USA. The quarry is dated, based on magnetostratigraphy and correlation with other sections in the Morrison basin ( Maidment and Muxworthy, 2019; Maidment, personal communication, 2022) at 150.44 to 149.21 million years old, placing it in the Kimmeridgian Stage of the Upper Jurassic.

DESCRIPTION OF MAB011899

Terminology

Terminology from Wilson (1999) and Wilson (2012) is used for the vertebral laminae, Wilson et al. (2011) for the vertebral fossae, and Wilson (2011) for the sacrum. For the definitions of the positional terms for the vertebrae, table 3 from Tschopp et al. (2015a) is followed. Following Tschopp et al., (2015a), Wilson (2002), and Upchurch (1998) and many other authors, ‘anterior’ and ‘posterior’ is preferred over ‘cranial’ and ‘caudal’. As suggested by Tschopp and Mateus (2013), interpre- and interpostzygapophyseal lamina is here preferred over the terms intrapre- and intrapostzygapophyseal lamina of Wilson (1999), as both laminae are positioned in between their respective zygapophyses. Cervical ribs are fused to their respective vertebrae, whereby the tuberculum fuses with the diapophysis and the capitulum fuses with the parapophysis. This creates a structure known as the ansa costotransversaria in birds, but in non-avian dinosaurs, such as sauropods, it is referred to simply as the ‘cervical rib loop’ ( Taylor and Wedel, 2013).

AXIAL SKELETON

The Cervico-Dorsal Transition and Presacral Neural Spine Bifurcation Patterns

Identifying the first dorsal vertebra is important, because the precise location of the cervico-dorsal transition ( Tschopp and Mateus, 2017) or junction ( Taylor, 2022) has implications for the biomechanics of the animal, as well as for phylogenetic analyses and morphological comparisons. Tschopp and Mateus (2017) discuss several characteristics that could be informative regarding the transition in diplodocids, whereas Taylor (2022) broadened the discussion towards all sauropods.

In MAB011899, 16 presacral vertebrae were recorded in the quarry map, which were articulated with the sacrum. Twelve of these vertebrae are still preserved today. The elongation of the anterior-most, preserved presacral vertebrae and the presence of cervical ribs clearly demonstrate that they are cervical vertebrae ( Figure 5 ). Based on photographs taken during preparation ( Figure 5C and 5D ), the two cervical vertebrae and the 10 other, articulated presacral vertebrae, were separated from each other by one additional vertebra, which also bore cervical ribs. The anterior-most element of the 10 articulated presacral vertebrae is identified as the first dorsal vertebra based on a combination of the following characteristics: 1) the assumed general vertebral count for diplodocids (n=25, 15 CV, 10 DV; Hatcher, 1901; Huene, 1929; see Tschopp and Mateus, 2017; Taylor, 2022), thus the position within the axial column as seen in Figure 2; 2 ) the lack of a fused rib, whereas more anterior elements clearly have fused cervical ribs; 3) the location of the parapophysis ventral to the pleurocoel, which follows a similar, gradual transition as in Diplodocus carnegii ( Hatcher, 1901) ; and 4) the distinct shortening from DV2 to DV3 as also occurs in Diplodocus and Barosaurus ( Hatcher, 1901; McIntosh, 2005). Hence, assuming a total of 25 presacral vertebrae, we interpret the preserved presacral vertebrae to represent CV13 and 14, and subsequently DV1 to DV10.

Although CV15 is currently missing, neural spine bifurcation is relatively clear for all vertebrae. Following the terminology of Wedel and Taylor (2013), the neural spines of CV13 through DV4 (including CV15) are all deeply bifid, with bifurcation deeper than half of the neural spine length. Shallowing of this bifurcation occurs rapidly, as DV5 is shallowly bifid, DV6 is notched/unsplit (see description of DV6), and DV7-9 and all sacral vertebrae are unsplit, and do no longer show evidence of dorsal midline indentation (no neural spine tip is preserved in DV10). From the cervical to the anterior dorsal vertebrae, the transverse distance increases between the metapophyses, but is never as short as in Suuwassea ( Harris and Dodson, 2004) .

Cervical Vertebrae ( Figures 5–7 , Table 1 )

Preservation and orientation. The cervico-dorsal transitional vertebra is missing, as well as the three vertebrae anterior to CV13. There are photographs of CV15; this vertebra will be briefly described based on those photographs. Both CV13 and CV14 are well-preserved, although both vertebrae were subject to transverse compression and shearing, especially the centra. Additionally, all cervical rib loops are now missing from both vertebrae, including the accompanying cervical ribs. However, left lateral photographs (see Figure 5 ) exist that show the vertebrae with preserved ribs; these have been used to describe and score these parts of the vertebrae. The vertebrae are described with the long axis of the centra parallel to the horizontal.

Centrum morphology. Both cervical vertebral centra are strongly opisthocoelous. In posterior view, the articular facet has an oval shape, which is higher than wide. Foramina present in the pleurocoels indicate that pneumatic diverticula along the neck invaded the vertebral centra. Because of compression and shear, the condyle and the cotyle of CV13 have a flattened oval shape, but they were probably subcircular in outline originally, as in Diplodocus , Kaatedocus and Galeamopus ( Hatcher, 1901; Tschopp and Mateus, 2013, 2017). The condyle is pronounced, rugose, and bordered by a ridge, which is preserved on the left lateral and dorsal sides of the condyle. The lateral sides of the centra are strongly pneumatized. In CV13, a large pleurocoel, with multiple subfossae, is present on the left lateral surface, extending from the anterior border of the parapophysis to the beginning of the last quarter of the centrum. The pleurocoel is oval, longer anteroposteriorly than dorsoventrally. The anterior-most subfossa is subtriangular and contains two smaller subfossae. The lamina dividing the first and second subfossae is angled anterodorsally-posteroventrally. The second subfossa is sub-oval and located just ventral to the diapophysis. The posterior-most subfossa is larger than the preceding subfossa. The subfossa is subtriangular, is longer than high, and deepens posteriorly. The right lateral surface, on the other hand, is marked by three fossae, which are well separated from each other, and not enclosed within a single, well-delimited pleurocoel. However, the posterior two subfossae possess a joint ventral rim, which fades anteriorly, indicating the original presence of a larger fossa. Transverse compression most likely has caused the dorsal border and part of the ventral border of the pleurocoel to fade, causing the pleurocoel as present on the left lateral side to no longer be present on the right lateral side. The anterior-most of these subfossae is located just dorsal to where the parapophysis would be (if preserved), and is oval in shape, anteroposteriorly longer than dorsoventrally tall. The second fossa is located dorsal to the posterior end of the reconstructed parapophysis and has a subtriangular ‘shark-fin’-like outline. The posterior-most fossa is located just posterior to the second fossa, and is a posteriorly deepening, oval depression, ending just anterior to the anterolateral border of the cotyle. The left parapophysis is located in the first half of the centrum and possesses a mediolaterally elongated fossa on the dorsal side, starting just ventral to the anterior border of the large lateral pleurocoel. The right lateral side does not preserve the cervical rib loop, and most of the parapophysis and the diapophysis are also missing; they were reconstructed symmetrically for mounting purposes. Therefore, it cannot be confidently concluded if a dorsal pneumatization was present on the right parapophysis of CV13.

In left lateral view, CV14 also shows a fossa on the dorsal part of the parapophysis. The fossa is egg-shaped, with the pointed end directed lateroventrally. Dorsal to this fossa, the lateral surface of the centrum is marked by a small ‘half-moon’- shaped fossa, with the convex edge pointing dorsally. Posterior and posterodorsal to this small fossa is a large, sub-oval fossa, which extends until just anterior to the cotyle rim. Within this larger fossa, a smaller, oval subfossa is present in the anterior part. As in CV13, the right parapophysis and diapophysis of CV14 are partly missing, and reconstructed with the cervical rib loop for mounting purposes. Three distinct, smooth, relatively deep fossae are present on the right lateral surface. The first fossa is located dorsal to the proximal part of the parapophysis. The fossa is teardrop-shaped, with the long axis of the fossa oriented nearly parallel to the long axis of the centrum. The pointed end of the fossa faces posteriorly. The second fossa, located posterior to the first, is subtriangular with rounded corners. The posterior-most fossa is similar to that in CV13 on the same side: a posteriorly deepening, roughly teardrop-shaped fossa, with the pointed end oriented posteriorly. Two small, shallow depressions are present dorsal to the first two fossae, in between the anterior centrodiapophyseal lamina (ACDL) and posterior centrodiapophyseal lamina (PCDL), obscured by the diapophyses in lateral view. Both vertebrae contain small, oval fossae on the posteroventral corners of the lateral surfaces of the centra, which are more pronounced on the left sides, with CV14 lacking such a fossa entirely on the right side.

The deformation of the centra of CV13 and CV14 results in a somewhat flattened ventral surface, especially in CV13. Posteroventral flanges are present and are most distinct on the left lateral sides of both centra. Ventral keels are present at the anterior part of the centra, but are difficult to discern due to the deformation and shear of the vertebrae, which is especially true for CV13. Only CV14 preserves a distinct keel, whereas CV13 only preserves severely flattened remnants of the keel. Deformation shifted the posterior end of the ventral keel in CV14 laterally to the left. The surface is transversely concave, although in CV13, the surface is close to flat due to compression. A sulcus is present, most distinct in CV14.

Neural arch morphology. The prezygapophyses of both vertebrae are strongly convex transversely, with the medial part of the rami bending ventrally. Pre-epipophyses are present laterally on the prezygapophyseal rami of both vertebrae, projecting anterodorsally as short, rugose projections. They are more developed on the left lateral side of each vertebra. As the prezygapophyses overhang the condyles in both vertebrae, the CPRLs of both vertebrae extend anterodorsally from the centra. Both CV13 and CV14 contain single CPRLs with no medial or lateral branches, as those present in e.g., Diplodocus carnegii ( Hatcher, 1901, plate V) or Kaatedocus ( Tschopp and Mateus, 2013, p. 873) . In both CV13 and 14, the paired TPRL connects to the posteromedial sides of the prezygapophyseal articular facets. Posteriorly, the paired TPRL meets dorsal to the neural canal in an acute angle. The left and right prezygapophyseal centrodiapophyseal fossae (PRCDFs) of CV13 contain a single, dorsoventrally oriented lamina, which divides both PRCDFs into two cavities. The right lamina is more prominent and less damaged compared to the left lamina. Within the PRCDF of CV14 on the right side, two small bony protrusions are visible which would have formed two dorsoventrally oriented accessory laminae, with the flat side of the laminae facing laterally. The dorsal end of these laminae connects to the ventral side of the PRDL and is positioned more anteriorly compared to the ventral part of the laminae, creating a diagonal outline in lateral view. The more anterior of these two laminae is well exposed, but the posterior lamina is almost fully destroyed, leaving only a small vestige of this lamina. Similarly, on the left lateral side, two small bony protrusions are visible which would have formed these double laminae, subdividing the PRCDF into multiple cavities.

The left diapophysis of CV13, due to the shear, is pushed slightly posteroventrally, whereas the right diapophysis – although largely lost – is seemingly in a more laterally oriented plane. Both diapophyses dip ventrally towards their lateral end. The left diapophysis has a posteriorly oriented, triangular projection, as also seen in Kaatedocus , but it is wider at its base and protrude further posteriorly in CV13 of MAB011899 compared to the holotype of Kaatedocus siberi ( Tschopp and Mateus, 2013) . The diapophyses of CV14 are different, as the right diapophysis is not oriented horizontally, but more obliquely, with the anterior margin located more dorsally than the posterior margin, such that the diapophysis slopes posteroventrally in posterior direction. This is, however, a result of poor preservation of the diapophysis, as well as shear of the vertebra. The diapophysis would have been oriented more horizontally in vivo. The left diapophysis is oriented horizontally, and projects lateroventrally, with a complete cervical rib loop preserved. This diapophysis also has the triangular, posteriorly oriented projection. ACDLs are present on the ventral sides of the diapophyses of CV13 and CV14, and fuse with the centrum at their anteroventral end. On the right side of both vertebrae, however, the ACDLs are more posteriorly restricted; this is not a result of the obvious transverse compression. PCDLs are present in both vertebrae, extending posteriorly onto the centrum. The anterior-most part of the left PCDL of CV13 meets the medial end of the triangular projection of the diapophysis. The right PCDL bifurcates anteriorly, with a short fading branch projecting ventrally close to the diapophysis. This bifurcation is also present in Barosaurus lentus ( Lull, 1919; see Tschopp et al., 2015a), but it is not nearly as prominent in MAB11899 as that seen in, e.g., YPM VP.000429. In CV14, the anterior end of the left PCDL does not meet at the medial base of the triangular projection, but instead proceeds slightly onto the dorsal surface of the diapophysis, meeting the posteroventral side of the postzygodiapophyseal lamina (PODL). The right PCDL is similar in morphology as the right PCDL of CV13. The PRDLs of both CV13 and CV14 are similar. The left PRDLs of CV13 and CV14 are oriented posteroventrally at their anterior end, and subsequently bend lateroventrally to reach the diapophyses. The right PRDLs are slightly different in the two vertebrae. The right PRDL of CV13 is oriented nearly horizontal, only dipping ventrally just anterior to the diapophysis. In CV14, the right prezygapophysis is raised further dorsally, resulting in a diagonal trajectory of the PRDL, as the diapophysis is located further ventrally. However, the dorsal displacement of the prezygapophysis in CV14 is caused by the shear of the vertebra. The anterior half of all PRDLs is roughened laterally but this is most prominent on the left sides of both vertebrae. The PODLs of CV13 are similar in orientation, with their anteroventral ends extending onto the dorsomedial surface of the diapophyses. The edges of the anteroventral halves face laterally, with the posterodorsal halves facing ventrally. This transition occurs earlier in the left PODL of CV13, wherein most of the edge faces ventrally, likely a result of the transverse compression. Both PODLs connect to the anteroventral sides of the postzygapophyses. In CV14, the PODLs are similar in morphology as to those in CV13, except for the posterodorsal half of the left PODL. Although the PODL connects to the postzygapophysis on the anteroventral side of the postzygapophysis, the posterodorsal end of the PODL is overlain by an accessory lamina, which is connected posteriorly to the epipophysis, and anteromedially fades into the SDF.

The metapophyses of both vertebrae are nearly vertical to posteriorly inclined in lateral view. The SPRLs are oriented anteroventrally from the neural spine apices towards the prezygapophyses in lateral view in both vertebrae. Only the anterodorsal tip of the left SPRL in CV14 preserves an anterior projection, which is not as prominent as seen in, e.g., Diplodocus carnegii ( Hatcher, 1901) or Kaatedocus siberi ( Tschopp and Mateus, 2013) . The dorsal tip of the left SPRL in CV13 is deflected medially in anterior view, because of deformation. The SPRLs are inclined laterally proximal to the prezygapophyses, but no fossae are present ventral to the inclined laminae, posterior to the prezygapophyseal facets. Each SPOL has a somewhat different trajectory, due to shear, but both meet their respective neural spine apices at the same height. A median tubercle occurs between the metapophyses, with a faint, rugose PRSL extending along the anterior side of the tubercle. Anterior to the median tubercles of CV13 through DV2 (except for CV15, where it is unknown due to the angle of the photographs, see below), within the spinoprezygapophyseal fossa (SPRF), accessory laminae are present, most prominently in CV13. In CV13, these accessory laminae are positioned anterior to the tubercle and just posterior to the midpoint of the interprezygapophyseal lamina (TPRL). They project anterodorsally, forming a bony protrusion, and extend posteriorly towards the metapophyses, but fade early at the base of the median tubercle. The right accessory lamina is slightly pushed into the left, and both are dorsally rugose. In CV14, these laminae have moved posteriorly, project anterodorsally, are more laminalike, and are located lateral to the PRSL and the tubercle, and posteromedial to both SPRLs. Both laminae fade dorsally just ventral to the dorsal-most edge of the tubercle. Figures 6 and 7 show close-ups of the morphology of these accessory laminae in CV13 and CV14. Similar laminae are present in DV1 and DV2. Therefore, these accessory laminae possibly play a role in laminar capture sensu Wilson (2012, p. 103-105), a process whereby vertebral landmarks ‘capture’ laminae, such as the creation of the spinodiapophyseal lamina (SPDL) through the capture of the SPRL ( Wilson, 2012, figure 10). Posterior to the SPRLs, a dorsoventral coel pierces the lateral side of the metapophyses of both vertebrae, forming a well-delimited, small subfossa within the spinodiapophyseal fossa (SDF). All coels are ventrally open, and the right coels are more pronounced compared to the left, likely a result of compression. An accessory, semi-horizontal lamina is present in the center of the right SDF of CV14, medial to the diapophyses and ventral within the fossa, similar to Barosaurus lentus ( Lull, 1919) . This lamina does not touch the PODL, or the SPRL, but is oriented with the anterior end facing perpendicular to the SPRL, and the posterior end facing roughly perpendicular to the PODL. The left SDF of CV14 contains an unusual sheet of bone, which appears to be an accessory lamina. The dorsal part of this lamina is connected to the ventral-most part of the lateral coel. The lamina is oriented obliquely, with its ventral portion located more anteriorly, roughly parallel to the SPRL. The middle part of the lamina is pushed against the adjacent SPRL. The lamina ends laterally close to the adjacent SPRL, approximately medial to the diapophysis. Although MAB011899 lacks the complex laminae in the SDF seen in Diplodocus carnegii ( Hatcher, 1901, plate III), the accessory lamina is probably best identified as one of the accessory laminae seen in the SDF of D. carnegii , whereby the lamina is partially broken and pushed against the left SPRL.

Within the SPOF of CV14, there are two sub-oval foramina, which pierce the SPOLs and connect to the dorsal part of the postzygapophyseal centrodiapophyseal fossae (POCDFs). Contrary to Galeamopus pabsti ( Tschopp and Mateus, 2017) , these foramina are not visible in lateral view in MAB011899, as the foramina are smaller, and located in the dorsal part of the POCDFs, which is obscured by the PODLs in lateral view. In CV13, these foramina might have been present, but the right POCDF was poorly preserved, and thus infilled with acrylic resin. The left POCDF is better preserved, but at the place where the foramina are located in CV14, acrylic resin has been added. In dorsal view, acrylic resin is visible in the SPOFs, which thus likely obscures the presence of the foramina. CV14 also possesses a ridge on the medial side of the right metapophysis, extending from the middle of the SPOL to the middle of the SPRL, although not directly connected to either lamina. The POCDFs of both vertebrae have different morphologies. The left POCDF of CV13 contains a well-delimited subfossa. Within the subfossa, an accessory lamina projects laterally and is oriented roughly anteroposteriorly, fusing posteriorly with the border of the subfossa. Additionally, a short, thin, dorsoventrally oriented accessory lamina is present ventral to the anteroventral part of the PODL, connecting medially to the anterodorsal part of the PCDL. A similar morphology is found in CV14, although the subfossa is larger, and the accessory lamina fades anteriorly into the fossa, and the additional, dorsoventrally oriented lamina is partially reconstructed (the ventral part). The right fossae of both vertebrae are also similar but have posteriorly projecting accessory laminae. In CV13, a similar, vertical accessory lamina is present in the POCDF on the left side. An additional, dorsoventrally shorter, but similarly oriented accessory lamina is present, just posteromedial to the other lamina in the POCDF. In CV14, within the right POCDF, an accessory lamina is located in a similar position as the first accessory lamina within the right POCDF of CV13, but this lamina bifurcates in CV14, forming two accessory laminae, and is oriented obliquely, not vertical. The bifurcation of this particular lamina occurs roughly halfway of the total length of the lamina, whereby the thicker, dorsal part forms two thin laminae, which constitute the ventral half of the structure. The medial lamina extends further posteroventrally than the lateral branch, with the lateral branch retaining the same trajectory as the initial dorsal part of the lamina. An additional posteriorly-projecting lamina is present posteromedial to the bifurcating laminae in this POCDF.

Due to the shear of both vertebrae, the left postzygapophyses are ventrally displaced compared to the right postzygapophyses. Epipophyses are present on both sides of both vertebrae, but are not as prominent as seen in e.g., Patagosaurus ( Holwerda et al., 2021) or Kaatedocus ( Tschopp and Mateus, 2013) . On the right side of CV13, the epipophysis is located dorsolateral to the postzygapophysis, and is pneumatized, showing a small, oval depression, which is infilled by sediment. The epipophyses are compressed asymmetrically between the left and right sides of both vertebrae. This bilateral asymmetry of the posterodorsal side of cervical vertebrae is also observed for the entire posterodorsal region of the vertebrae in Diplodocus carnegii ( Hatcher, 1901) , although some of the observed variation in D. carnegii may be deformation. In MAB011899, the left epipophysis of CV13, which is pneumatized from the inside via a small infilled hole in the spinopostzygapophyseal fossa (SPOF), and the right epipophysis of CV14, are compressed transversely. The right epipophysis of CV13 is compressed more dorsoventrally, but also slightly transversely, whereas the left epipophysis of CV14 also contains both states, but here, the transverse compression is more prominent. In posterior view, the left TPOL is bent inwards, meeting the right TPOL, which is straighter, above the neural canal. This is more prominent in CV13 compared to CV14, but in both, this is influenced by shear. They both meet in a projection, which protrudes slightly beyond the posterior margin of the neural arch, hanging above the dorsal edge of the cotyle. Lateral to this projection, CV14 shows distinct lateral centropostzygapophyseal laminae (lCPOLs), which are separated by centropostzygapophyseal fossae (CPOFs) from the mCPOLs (or mdCPOL sensu Carballido and Sander, 2014), which run from the projection of the TPOL lateroventrally to the centrum, forming the neural canal margin. In CV13, in contrast to CV14, the lCPOLs are not distinct from the rest of arch, and probably run alongside the TPOL towards the posterior projection. mCPOLs are present in CV13 and similar in morphology to CV14. The neural canals are not perfectly round due to compression, but more triangular in both anterior and posterior view.

CV15 Description ( Figure 5 )

As aforementioned, this vertebra is lost, most likely in the fire at Münchehagen. Although the photographs are insufficient to describe CV 15 in similar detail, several features can be interpreted from a left anterolateral and posterolateral photograph. Only the anterior half of the vertebra was preserved, and the neural spines were missing. The posterior half was likely destroyed when the field jacket containing the cervical vertebrae and the field jacket containing the anterior dorsal vertebrae were detached from each other. This is supported by the fact that the small jacket containing CV15 fitted perfectly with the jacket containing CV13 and CV14. However, posterior to CV15, the jacket ends, which might be an indication that the posterior half of CV15 was already destroyed during the excavation, due to the crumbly nature of the bones (from which the specimen got its nickname).

The centrum of CV15 is strongly opisthocoelous, with a convex condyle, similar to the previous vertebrae. Posterior to the condyle, an extremely deep pleurocoel is seen. Dorsal to the pleurocoel, the diapophysis has a posterior projection, similar as in CV13 and CV14. A cervical rib loop is also present. A glimpse can be seen in Figure 5C of the anterior process of the cervical rib, which appears small and rounded, similar to the process of the rib of CV13 (see below). Anteroventral to the diapophysis, an ACDL is present. A single CPRL can be seen on this side, with no clear indication of dorsal bifurcation into two rami, although recognizing such a division is hampered by the quality and angle of the photographs. The PRDL is preserved, connecting to a near-horizontal left prezygapophysis. The right prezygapophysis is laterally inclined. No transverse sulcus can be seen posterior to the prezygapophyseal facet. From the left metapophysis, part of an anteriorly compressed SPRL is seen. Ventrally, this lamina protrudes strongly dorsally, to subsequently fade anteroventrally posterolateral to the prezygapophyseal facet. Part of a PODL can also be observed on the left side. From the right prezygapophysis, a similar deformed SPRL is visible as the left SPRL. The TPRL is preserved and appears similar in morphology as seen in CV13 and CV14. No accessory laminae are observed in the SDF. The PRCDF is filled with matrix, so it cannot be assessed if multiple laminae are present similar to the previous vertebrae. No other morphological details can be gleaned from the photographs.

Cervical Ribs ( Figure 5 )

Currently, the cervical ribs of both CV13 and CV14 are lost, as well as major parts of the cervical rib loops. However, photographs were taken when the vertebrae were prepared at the SMA, in which CV13 preserved a significant part of the cervical rib loop and the attached rib on the left side. Only the posterior-most end of the rib shaft is missing, and a small part of the anterior process of the rib. A rib was also attached to CV14, but most of the rib is missing in the photographs, preserving only part of the anterior process, a small part of the posterior rib shaft, and the cervical rib loop connecting the diapophysis and parapophysis.

Both ribs are located just slightly ventral to the centrum, unlike those seen apatosaurines (e.g., Gilmore, 1936; Ostrom and McIntosh, 1966; Upchurch et al., 2004 b), whereby the ribs are placed significantly below the ventral margin of the centrum. The posterior end of the rib shaft of CV13 is missing in the photo ( Figure 5A ), but the shaft appears to be tapering towards the posterior end. Both the dorsal and ventral side of the rib shaft is straight, apart from a small indentation caused by deformation/breakage halfway along the visible rib on the posterior side. Anteriorly, the rib is dorsoventrally wider compared to the posterior part of the rib, as the ventral part of the rib shaft curves slowly dorsally towards the anterior process. The anterior process appears rounded in lateral view, although this is difficult to observe in the photograph due to the angle of the photograph, with the dorsal part curving posteriorly against the cervical rib loop. The anterior process appears to be slightly longer anteroposteriorly compared to its height dorsoventrally but is smaller compared to the anterior facets of CV13/14 e.g., Kaatedocus ( Tschopp and Mateus, 2013, p. 873) or Galeamopus ( Tschopp and Mateus, 2017, p. 63) . In lateral view, it is obscured, but a photograph taken approximately anterolaterally ( Figure 5B ) reveals that the anterior process is broken in two, and that the process is quite robust, especially compared to the anterior facets of Kaatedocus and Galeamopus .

Dorsal Vertebrae ( Figures 8–17 , Table 2 )

Preservation and orientation. The dorsal vertebrae were preserved in articulation with the cervical and sacral vertebrae, which were all excavated in four separate blocks. Based on the early preparation photographs provided by the SMA, dorsal vertebrae 1-5 and dorsal vertebrae 6-9 were excavated and prepared in separate blocks. The partially preserved tenth dorsal vertebra was recovered and prepared in a block with the sacrum. Distinction between anterior, middle, and posterior dorsal vertebrae is based on Tschopp et al. (2015a): anterior dorsal vertebrae are defined by having the parapophyses still in contact with the centrum, whereas middle and posterior dorsal vertebrae have a numerical subdivision. Generally, the preservation of the dorsal vertebrae becomes worse along the series, and parts that were missing are reconstructed with acrylic resin for mounting purposes. Especially DV10 is modified and reconstructed, as substantial parts of the neural arch were missing, and the preserved parts were crushed. For most of the reconstructed laminae in the dorsal series, however, small parts were present or are reconstructed based on bilateral symmetry. All dorsal vertebrae are compressed in an analogous way as the two cervical vertebrae, although compression is more pronounced in the anterior and mid-dorsal vertebrae.

General morphology. All dorsal vertebrae are taller than long, and the centra shorten from DV1 to DV5, after which centrum length remains subequal ( Table 2 ). The condyle is distinct in DV1 and DV2, becomes smaller in DV3, and is reduced in DV4-9 (it is not preserved in DV10). The right lateral side of the condyles of DV2 and DV3 are slightly eroded close to the ventral surface, revealing the internal pneumatic structure as polycamerate ( Wedel et al., 2000). All vertebrae have pleurocoels on the lateral sides of the centra, which vary in shape from oval to more irregularly shaped, but this variation is most likely due to deformation. The true shape of the pleurocoels would have likely been similar to that seen in Diplodocus carnegii ( Hatcher, 1901) , although placed more centrally onto the lateral surface as in Supersaurus vivianae ( Jensen, 1985) , and not invading the neural arch pedicles as in D. carnegii or Galeamopus pabsti ( Tschopp and Mateus, 2017) . The size of the pleurocoels increases along the series, with the posterior (DV7- 10) centra having pleurocoels with a length roughly equal to the anteroposterior length of the neural arch pedicles. Due to compression, the transverse processes on the left lateral side are oriented lateroventrally, whereas the right lateral processes are oriented dorsolaterally. In DV6-8, the transverse processes are nearly horizontal, similar to Diplodocus ( Hatcher, 1901) . The SPDLs, where present, follow the curvature of the neural spine and the diapophyses. They first appear in DV3, as the SPRLs seem to transition laterally along the cervicodorsal junction, and are captured by the diapophysis, sensu Wilson (2012). This capture sequence is best preserved on the left lateral side of DV3, wherein the ‘SPDL’ ends ventrally in between the prezygapophyses and the diapophysis. The location of the SPDLs (especially their ventral halves) in more posterior dorsal vertebrae gradually moves more posteriorly onto the lateral surface of the neural spine. SPRLs distinct from the captured SPDLs can first be observed in DV4, but preservation in DV3 is incomplete, so the entire serial transition of these two laminae cannot be clearly identified. Hyposphene-hypantrum articulations are well developed, although most hyposphenes are not preserved, and nearly all hypantra are damaged. The first hyposphene appears in DV3, although poorly preserved. In DV4, the hyposphene is supported by a subvertical lamina ventrally (sTPOL sensu Carballido and Sander, 2014). DV4 is the first dorsal vertebra in which the hyposphene takes a clear rhomboid shape, which is even more prominent in DV6. A supporting lamina is also seen in DV5, but this is a reconstructed lamina. None of the more posterior vertebrae preserve a single, supporting ventral lamina; the hyposphenes in these vertebrae are supported by oblique CPOLs that unite below the hyposphene.

Anterior Dorsal Vertebrae (DV1-3; Figures 8–10 , Table 2 )

Centrum morphology. The centra are all opisthocoelous and their ventral surface is concave anteroposteriorly. Due to breakage and deformation, it is impossible to say whether the ventral surface was concave, flat or slightly convex transversely. The condyle of DV1 and DV2 preserves a relatively distinct bony rim. This may have been present in DV3, but as parts of the condyle are missing, the condyle of DV3 has been reconstructed without a rim. Due to similar compression patterns as in the cervical vertebrae, the condyles and cotyles of DV1-3 are all compressed ellipses. In DV1, the parapophyses are located anteroventrally to the pleurocoels. They are located more dorsally in DV2, anterior to the pleurocoels, and in DV3, the parapophyses are located anterodorsal to the pleurocoels. The pleurocoels in DV1 are roughly oval and located slightly anterior to the middle of the centrum. The right lateral pleurocoel bears the first signs of a vertical bony ridge dividing the coel in two separate chambers, but this ridge is too shallow to truly divide the pleurocoel. In DV2, the pleurocoels are situated slightly posterior to the middle of the centrum and are suboval in outline. DV3 has a poorly preserved centrum, especially the right lateral side. The pleurocoel on the left lateral side is flattened, suboval, and divided from an anteriorly placed coel by an anterodorsally-posteroventrally oriented bony strut, which resembles an earlier state of the rod-like struts dividing the pleurocoels in the mid- and posterior dorsal vertebrae. The pleurocoel on the right lateral side is almost circular and is positioned slightly posterior to midlength. The posterodorsal edge of the pleurocoel is damaged, as part of the neural arch and the complete cotyle rim of the right side is missing, which makes it difficult to assess if the pleurocoel was larger posteriorly, and thus more oval shaped rather than subcircular. No bony strut is visible on this side. As the centrum length decreases from DV1 to DV3, the pleurocoel also shortens in anteroposterior length ( Table 2 ). The ventral surface of DV1 is deformed, and the condyle shows signs of breakage. However, an anteroposteriorly oriented keel is present at the anterior side of the ventral surface of the centrum, albeit faint due to the compressed centrum. DV2 does not preserve a keel, is also deformed, and the ventrolateral side of the condyle is missing, revealing the internal pneumatic structures. The internal structure is polycamerate sensu Wedel et al. (2000), consisting of larger camerae which are separated by branching structures, common for diplodocids ( Wedel, 2003; Tatehata et al., 2023). The structure seen in DV2 is comparable with the pattern seen in Apatosaurus ( Wedel et al., 2000, figure 11C). The ventral side of DV3 is better preserved, although compressed similarly to DV1 and DV2. No ventral keel is present.

Neural arch morphology. The prezygapophyses of DV1-3 are largely preserved. DV1 misses the left prezygapophyses, and DV2 misses small parts of the anteromedial sides of both rami. In all three vertebrae, the rami are inclined laterally and slightly convex dorsally. In both DV1 and DV2, the right prezygapophysis is nearly vertical, which is less pronounced in DV3, caused by the shear of the vertebrae. The left prezygapophysis of DV2 is ventrally displaced. Due to deformation, the left prezygapophysis of DV3 is roughly horizontally oriented, with the entire ramus bending slightly ventrally, whereas the right prezygapophysis is steeply inclined laterally, as well as projecting anterodorsally. In DV1, the right CPRL is well preserved, and bifurcates dorsally to form two CPRLs attaching ventrally to the prezygapophyseal facet. The left CPRL misses the dorsal part. Therefore, it cannot be assessed if this lamina also bifurcates dorsally but based on the width of the lamina compared to the right CPRL, it can be assumed that it did bifurcate. In DV2, the CPRLs are relatively vertical (in anterior view) laminae, which are, in lateral view, inclined anteriorly towards the prezygapophyseal facets. No dorsal bifurcation of the CPRLs appears to be present. The right prezygapophysis is oriented similarly as the right prezygapophysis of DV1, being almost vertical in anterior view due to shear. The CPRLs are both preserved in DV3, but due to the different positions of the prezygapophyses, have slightly different trajectories. In lateral view, the right CPRL is oriented more anterodorsally towards the prezygapophysis. The left CPRL is also oriented anterodorsally, but is bent further anteriorly, therefore appearing closer to the horizontal plane of the centrum. The left half of the TPRL in DV1 is reconstructed. The right half of the TPRL is only posteriorly reconstructed, which in the mount results in a TPRL, which crosses dorsal to the neural canal to the left side. The TPRL of DV2 is largely preserved, missing a small part of the right branch close to the prezygapophysis, and missing half of the left branch, at the same position, but extending further posteriorly. The paired TPRL in DV2 meet dorsal to the neural canal, similar to what is seen in the cervical vertebrae. The TPRL of DV3 is partially preserved, not connecting medially. Most of the left PRCDF of DV1 is preserved, but the fossa does not contain any accessory laminae. The right PRCDF also does not contain any accessory laminae, but it contains a small coel-like depression in the posterodorsal part of the fossa. The neural canal is roughly oval, with the long axis oriented dorsoventrally. The left PRCDF of DV2 is obscured in lateral view by the ventrally displaced PRDL, whereas the right PRCDF is completely visible in lateral view. Neither of the fossae contain any accessory laminae, and only the right fossa contains an additional small depression (similar to DV1). Such a depression, however, might also be present on the other side, but in the left PRCDF, some residual sediment is still present to maintain the stability of the vertebra, which obscures the possible location of this depression. Both PRCDFs of DV3 lack any accessory laminae, but do preserve some additional pneumatic depressions, although the larger depression on the left lateral side is partially reconstructed.

The left transverse process of DV1 is partly reconstructed, but the preserved ACDL and PCDL meet approximately level with the dorsal edge of the cotyle, supporting the diapophysis from below. The right lateral transverse process is located substantially above the dorsal edge of the centrum, halfway between the dorsal edge of the cotyle and the postzygapophysis, but this is probably affected by deformation. In DV2, this shear is more extreme, with the left lateral transverse process projecting even further ventrally, being located in the same plane as the upper third of the centrum, whereas the right lateral process is located just beneath the postzygapophysis at the same side, and is, in posterior view, in the same plane as the left postzygapophysis. In DV3, the shear is less extreme, but this impression is also caused by the reconstruction of the right lateral transverse process, which follows a more horizontal orientation as would be expected in vivo. On the left lateral surface of DV1, the ACDL bifurcates anteriorly, with the dorsal branch extending almost horizontally, and fading anteriorly close to the CPRL, and the ventral branch connecting to the centrum, anterodorsally to the pleurocoel. Similarly, the right ACDL also bifurcates towards the lateral CPRL, however, because the diapophysis is elevated, the ventral branch of the ACDL is oriented almost vertically and connects to the anterodorsal part of the centrum. The dorsal branch of the ACDL extends anteroventrally from the diapophysis and curves anteriorly at about halfway along its length before it disappears close to the CPRL. In DV2, bifurcating ACDLs are also present, but the bifurcating branches are far less prominent than in DV1. On the left side, the ACDL originates at the ventral surface of the diapophysis, approximately at midlength of the transverse process. From here, it extends medially towards the neural arch, and curves anteroventrally, where the ventral branch meets the CPRL at the anterodorsal edge of the centrum. The bifurcation occurs in a similar position as in DV1, whereby a dorsal branch of the ACDL originates close to the CPRL, is oriented anterodorsally, and nearly connects to the CPRL, but disappears just posterior to the CPRL. The right ACDL is oriented nearly vertically, bifurcates ventrally, with a weakly developed branch extending anteromedially, and a more strongly developed branch continuing in the ventral direction of the dorsal portion of the ACDL. In DV3, no bifurcation of the ACDL is present. The left ACDL is oriented dorsoventrally from the transverse process towards the anterodorsal edge of the centrum. However, due to compression, the lamina projects posterolaterally. The right ACDL is only partially preserved, preserving only the part proximal to the centrum, which is also oriented dorsoventrally, with the dorsal part reconstructed in the mount. The ACDL fuses at the dorsal edge of the centrum, just posterodorsally to the flattened parapophysis. The PCDLs of all three vertebrae differ little on the right lateral side, as they are all oriented diagonally dorsoventrally from the diapophysis towards the posterodorsal edge of the centrum. The dorsal-most part of the right PCDL in DV3, however, is reconstructed. The left PCDLs are different, due to the way this side is compressed. In DV1, the left PCDL is oriented almost horizontally. It lacks the proximal part that connects to the diapophysis, and extends to the posterodorsal edge of the centrum. In DV2, the left PCDL is slightly more vertical from the centrum towards the diapophysis, but the lamina curves lateroventrally, and ends posterior to the tip of the transverse process. In DV3, the left PCDL is similar to the right, but the proximal end towards the diapophysis is mediolaterally wider. The left PRDL of DV1 is reconstructed. The right PRDL is preserved, and has a roughened lateral side close to the prezygapophysis. In DV2, both PRDLs are preserved and show a roughened lateral aspect. The left PRDL of DV3 is preserved, but the lateral edge is damaged. Only the anterior-most part of the right PRDL is preserved, with the posterior part fully reconstructed. Neither of the PRDLs show similarly distinct rugose areas on the lamina as DV1 and DV2, but some evidence of rugosities is still present. From the diapophyses in DV1, PODLs project posterodorsally towards the postzygapophyses, which connect to the anterior margins of the postzygapophyseal articular facets. In DV2, the left PODL connects from the posterior margin of the diapophysis to the anterior margin of the postzygapophysis and is less posterodorsally inclined compared to the PODLs of DV1. The entire right PODL is reconstructed and connects incorrectly to the middle of the reconstructed right SPOL. In DV3, the left PODL is oriented mostly posteromedially, but as the postzygapophysis is located above the diapophysis, the PODL is also projecting dorsally, which is caused by the dorsally displaced lateral edge of the postzygapophysis. The right PODL is reconstructed. Infradiapophyseal foramina, as also seen in Giraffatitan brancai ( Janensch, 1950; Taylor, 2009), are present in all anterior dorsal vertebrae, deepening in more posterior elements; these are mainly oval in outline.

The metapophyses in the anterior dorsal vertebrae become less ‘wing-like’ along the series, and instead become more straight, relatively vertical projections, which are V-shaped in anterior view, especially in DV3, although they do not diverge as widely as seen in Apatosaurus ( Gilmore, 1936) . The metapophyses converge throughout the posterior cervical and anterior dorsal vertebrae, eventually to fully fuse in the middle dorsal vertebrae. However, distance between the spine apices increases in DV2 onwards, becoming more V-like. This distance, however, is influenced by the poor preservation of more posterior elements, whereby all dorsal vertebrae with bifurcating neural spines from DV2 onwards lack one or both metapophyses. Therefore, the increasing distance between the spine apices of DV2-4 is an artifact of the reconstructed metapophyses, which exaggerates this distance. There is evidence in DV4, which preserves the bases of the metapophyses, that a V-shape is the correct shape for DV2 and DV3, but it cannot be ruled out that the neural spine tips become nearly parallel to each other. In lateral view, the metapophyses of DV1-3 are all slightly inclined posteriorly compared to the horizontal long axes of the centra. Medially, the metapophyses of DV1 have a rough surface, with small, obliquely oriented ridge-like structures, whereas the medial surface of the metapophyses of DV3 is smoother. The medial surface of the metapophyses of DV2 is relatively smooth, with some smaller ridges and cracks at the posteromedial side of the metapophysis, but this is also caused by the fact that most of the right metapophysis is overlain with acrylic resin, and is mostly reconstructed. However, the ridges in DV1, and on the left metapophysis in DV2, seem to be the result of compression of the metapophyses, and not true ridges as seen in the cervical vertebrae, or as seen in A. ajax ( Marsh, 1877; but see Tschopp et al., 2015a, figure 60C). The SPRLs in all three vertebrae mostly project anteriorly/anterolaterally from the spine apex to the prezygapophyses, terminating posterior to the prezygapophyseal rami. The SPRLs in DV1 are both mostly preserved, except for part of the left SPRL close to the prezygapophysis. In DV2, most of the left SPRL is preserved, only missing the dorsal-most part on the metapophysis, and a small part of the SPRL adjacent to the median tubercle. For the right SPRL, only the dorsal-most part is not preserved. Most of the SPRLs of DV3 are reconstructed. As aforementioned, the ventral half of the left SPRL has moved from the prezygapophysis to attach in between the prezygapophysis and the diapophysis. Because most of the left metapophysis is not preserved, the dorsal half of the left SPRL is also missing. From the right SPRL, only the section on the preserved part of the metapophysis is preserved, fading dorsally near the apex of the neural spine. Most of the SPOLs of all three vertebrae are not preserved. Only DV1 preserves parts of both SPOLs. The left SPOL lacks the dorsal third of the lamina, reconstructed for mounting purposes. The right SPOL lacks most of the ventral half of the lamina, except its connection to the dorsal margin of the postzygapophysis. In DV2, only the ventral-most part of the left SPOL is preserved. In DV3, only the dorsal half on the posterior surface of the neural spine is preserved. The left SDF of DV1 bears no accessory lamina. The dorsal part of the left SDF is reinforced and partially reconstructed, especially the dorsal part towards the neural spine apex. However, based on the preparation pictures provided by the SMA, the left metapophysis including the SDF does have the correct outline, indicating that most of the reconstructive work was aimed at strengthening the metapophysis to prevent further breakage. This becomes more evident based on the right SDF, which bears a more rounded spine apex, or cup of the neural spine sensu Bonaparte and Mateus (1999). Ventral to this dorsally rounded apex, a smaller dorsoventral coel is present on the lateral surface, posteriorly to the right SPRL. Medial to the PODL, an accessory lamina is present on the SDF, likely reaching the SPOL dorsally, and probably the posterior margin of the dorsoventral coel. The main reason for this dubious assignment is the incompleteness of the right SPOL, which is partly covered in acrylic resin, obscuring whether the lamina reaches the SPOL, and whether it is part of the subcircular margin of the coel. In DV2, only the left SDF is preserved, with a short (4 cm) horizontally oriented accessory lamina present on the dorsal surface of the diapophyseal facet. Only the anterior part of the right metapophysis was preserved, which was laterally damaged. Therefore, laterally, the entire metapophysis was reconstructed, whereas medially, some of the bone is still present. Only the ventral part of the postzygapophysis and CPOL are preserved on this side. This is also true for DV3, where the left metapophysis is mostly reconstructed, preserving only the ventral part, anterior to the postzygapophysis. Two accessory laminae are present in the left SDF of DV3. The first is similar to the accessory lamina in DV2, but it connects anteriorly to the displaced SPRL. A second accessory lamina is present more medially, anterior to the postzygapophysis and posterior to the displaced SPRL. This lamina is oriented obliquely. Dorsally, the lamina originates posteromedially to the displaced SPRL. It extends posteroventrally, but also laterally, giving it its oblique orientation, and fuses 4 cm anteriorly to the PODL with the SDF. In contrast to the left side, the right side DV3 only preserves the top half of the metapophysis. Unlike what occurs in Diplodocus carnegii , the cup of the neural spine does not bear a sub-horizontal SPOL ventrally. Instead, the SPDL reaches the spine apex, thus no true cup is seen on the lateral surface of the neural spine apex. The presence of distinct SPDLs thus occurs earlier in the dorsal vertebrae compared to D. carnegii ( Hatcher, 1901, plate VIII). A median tubercle is present in DV1 and DV2 (not preserved in DV3), with a rugose PRSL in DV1 on the anterodorsal surface of the tubercle. This lamina cannot be assessed in DV2 and DV3 due to the preservation of the tubercle, but some weak rugosity marks the ventral base of the median tubercle of DV2, which may represent the PRSL. Laterally adjacent to the median tubercle, as aforementioned, two accessory laminae occur in DV1, as well as remnants of these laminae in DV2. In DV1, these laminae disappear dorsally halfway in between the medial base of the TPRL and the dorsal edge of the median tubercle. In DV2, the exact limits of the laminae are difficult to elucidate, due to the poor preservation of the median tubercle.

The postzygapophyses of DV1 are laterally inclined. The postzygapophyseal facets of DV2 are more horizontally oriented, although the right postzygapophysis is poorly preserved. In DV3, the posterior side is poorly preserved. Only the left lateral postzygapophysis is preserved, missing the dorsal part. However, based on the position of the postzygapophysis, both postzygapophyses are displaced medially compared to the postzygapophyses of DV1 and DV2. Two singular CPOLs originate on the ventral side of the postzygapophyses in DV1, and project ventrally towards the centrum. In DV2, vertically oriented CPOLs are present on each side, which bifurcate ventrally, creating two small fossae (CPOL-f) between the branches. In DV3, the CPOLs are broad and pillar-like, and less laminar in structure than in DV1 and DV2. Medially to the CPOLs, the paired TPOL originate from the postzygapophyses in DV1. The left half of the TPOL is broken ventrally and therefore does not fuse with the right half of the TPOL dorsal to the neural canal. In DV2, both halves of the TPOL are oriented ventromedially. However, the ventral part of both halves is not preserved. Therefore, both halves do not meet dorsal to the neural canal, which causes the neural arch to have an ‘open’ appearance in posterior view. Additionally, the ventral part, where the TPOL would meet, is damaged, opening the dorsal side of the neural canal. In DV3, a TPOL is present medially, which connects to a laminar structure, which runs ventrally onto the neural arch, and terminates dorsal to the neural canal. This laminar structure is dorsally wider, but broken, and very likely represents the first hyposphene in the dorsal series, which is supported by a lamina from below (sTPOL sensu Carballido and Sander, 2014). The neural canals in DV1 and DV2 are subcircular in outline. In DV3, the neural canal is dorsoventrally compressed, but is, as is that of DV1 and DV2, subcircular.

Mid-Dorsal Vertebrae (DV4-6; Figures 11–13 , Table 2 )

Preservation. DV6-9 were all part of a single block, which was partially prepared at the SMA and partially documented photographically. Therefore, DV6, as well as DV7-9 below, are partially described based on these photographs, which were primarily used as a control to elucidate the reconstructed parts from the real bone.

Centrum morphology. The centra of DV4-6 are opisthocoelous, but the anterior condyle and the posterior cotyle become gradually less convex and concave, respectively, along the series. The ventral surfaces are all smooth, apart from cracks caused by the deformation of the vertebrae. The centra are still concave anteroposteriorly and deformed in a similar orientation as the previous vertebrae. Therefore, the posterior articular surfaces are not circular, but more elliptical, with the long axis diagonally oriented, from the right dorsolateral to the left ventrolateral side. This deformation becomes less prominent in DV5 and DV6. Large portions of the anterior surfaces of the condyles are covered in acrylic resin for support, and partly for reconstruction, which is most evident in DV6. As do DV1-3, all mid-dorsal vertebrae bear pleurocoels, however, DV4 is the first dorsal vertebra wherein a true dorsoventral bony strut is present, which separates the pore into chambers. The left and right pleurocoels in DV4 are oval, although the left is close to being circular, and the bony struts are roughly vertical structures in the middle of the pores. In DV5, likely due to compression, the pleurocoels are subtriangular. Bony struts are present, but the right strut is reconstructed. Similar pleurocoels can be found in DV6, where the right pleurocoel is compressed dorsoventrally, flattening the triangular shape of the pore. Bony struts are present, but the left strut is partially reconstructed.

Neural arch morphology. The prezygapophyses of DV4 and DV6 are entirely reconstructed. In DV5, the ventral part of the right prezygapophysis is preserved, but the entire left prezygapophysis is reconstructed. Only the ventral portions of the CPRLs in DV4 and DV6 are preserved. In DV5, both CPRLs no longer directly connect to the prezygapophysis, but connect anteroventrally to the parapophysis (thus becoming an anterior centroparapophyseal lamina - ACPL). This would be similar in DV6.

The parapophyses in DV4 are located in between the centrum and the prezygapophyses at the anterolateral side of the vertebra. The left parapophysis consists of a subcircular bony projection which is heavily deformed. On the right side, the parapophysis is not preserved, but the breakage surface is, which appears to be at the same height as the left parapophysis. DV5 has a relatively well-preserved parapophysis on the right side. The parapophysis is located just ventral to the prezygapophysis on the anterolateral side of the neural arch, and has multiple laminae connected to it. On the left side, the parapophysis is mostly broken off. The lack of a distinct parapophysis results in the appearance of a long ‘lamina’, which extends from the anterodorsal side of the centrum towards the ventral side of the left transverse process. In the middle of this ‘lamina’, the breakage surface of the parapophysis is visible. In DV6, the parapophyses would be expected to be located closer to the prezygapophyses. However, the preservation of the anterior part of this vertebra is poor; there are no signs of parapophyses. In the mount, no parapophyses are reconstructed. PCPLs cannot be identified in DV4, due to the poor preservation of the parapophyses and the surrounding surface. A prezygoparapophyseal lamina (PRPL) connects the left parapophysis of DV4 to the prezygapophysis, but the middle part of this lamina is reconstructed. On the right side, a gap is present between the prezygapophysis and parapophysis, wherein the PRPL is not preserved due to weathering, and wherein the PRCDF is quite damaged. Almost entire paradiapophyseal laminae (PPDLs) are preserved, with only their lateral ends reconstructed. On the posterior side of the right parapophysis of DV5, two laminae are present. One of the laminae is a PPDL, which extends posterodorsally from the parapophysis. The second lamina is oriented anteroposteriorly, posterior to the parapophysis, and ends just anterior to the PCDL, and can be identified as the posterior centroparapophyseal lamina (PCPL), which is also seen in this orientation in e.g., Supersaurus BYU 725-9044 ( Jensen, 1985, p. 698). A PRPL connects to the ventral surface of the prezygapophysis and the anterodorsal surface of the parapophysis. On the left side, ventral to the parapophyseal breakage surface, a short, posteroventrally oriented laminar structure is preserved, which probably represents the PCPL. Two laminae connect the breakage surface of the parapophysis to the ventral side of the prezygapophysis. The dorsal lamina of the two is oriented nearly horizontally, whereas the ventral lamina is oriented ventrolaterally. They are not double PCPLs, as seen in DV9 of Diplodocus carnegii ( Hatcher, 1901, plates VII and VIII) as the parapophysis is still separate from the prezygapophysis, whereas in DV9 of D. carnegii , the parapophysis is laterally adjacent to the prezygapophysis. Instead, the dorsal lamina connecting to the prezygapophysis of DV4 of MAB011899 is short and dorsoventrally thin but does connect to the breakage surface where the parapophysis was located. The ventral lamina is directed more ventrolaterally, and connects to the ACPL, but both laminae connect to the anteromedial part of the parapophysis breakage surface. The dorsal lamina might therefore be an accessory lamina, whereas the ventral lamina can be identified as the PRPL. Interestingly, this PRPL appears to bifurcate ventrolaterally, showing a very small fossa in between the branches. This does not occur on the right side, but this might be due to preservation. Due to the preservation of the anterior side of DV6, all laminae related to the parapophyses (PCPL, PRPL, PPDL) are not preserved.

The transverse processes of all mid-dorsal vertebrae are poorly preserved. In DV4, only the ventromedial part of the left processes is preserved. The remaining parts of the processes are entirely reconstructed. In DV5, the left transverse process has a reconstructed dorsolateral tip. The anterior and anterolateral side of the tip is not reconstructed and shows that the lateral-most part of the transverse process was oriented lateroventrally. In DV6, the left transverse process is partially preserved, missing a posterolateral and the medial section towards the anterolateral surface of the neural spine. A feature reconstructed in DV4, but present in DV5 and DV6, are mediolaterally elongated fossae on the dorsal surface of the transverse processes. The anterior rim of this fossa in DV5 is reconstructed on the left process, which transitions into the SPDL. The posterior rim appears to form a lamina medially as well, but because so little is preserved of this lamina, this might have converged towards the SPDL. In DV6, the posterior rim forms the SPDL. In DV4, distinct PCDLs are preserved, which are almost complete, as only the lateral parts which attach to the transverse processes are missing. The left PCDL of DV5 is complete, extending to the tip of the transverse processes. The right PCDL lacks the lateral part on the ventral surface of the transverse process. Only the left PCDL of DV6 is preserved, forming a right angle with the transverse process. The right PCDL is entirely reconstructed. The PRDLs of DV4 are entirely reconstructed. In DV5, the left PRDL is complete, connecting to the posterolateral margin of the prezygapophyses and the anterior surface of the lateral tip of the transverse process. Only the middle part of the right PRDL is preserved. Only the left PRDL is preserved of DV6, lacking only the end towards the prezygapophysis. Both PODLs of DV4 are reconstructed, as well as the right PODLs of DV5 and DV6. The left PODL of DV4, however, was originally present based on preparation pictures of the Oertijdmuseum, but because of the crumbly nature, most of this lamina is supported and reconstructed with acrylic resin. The left PODLs of DV5 and DV6 connect to the lateral margin of the postzygapophyses, extending to the posterior surface of the transverse processes, fusing at midheight of the process. Well-delimited infradiapophyseal foramina are present in DV6. In DV4 and DV5, they are smaller, and in DV4, filled in with acrylic resin to support the vertebra.

The metapophyses of DV4 are distinctly Vshaped. Part of the left metapophysis is preserved, but the apex is fully reconstructed. Similar parts are preserved of the right metapophysis. Interestingly, photographs made by the Oertijdmuseum during preparations show that at least the left apex was preserved originally. However, the apex appears very brittle and was likely lost during preparation. It was subcircular in lateral view, and distinctly rugose. The medial side, where the median tubercle is located, is original bone, and therefore the metapophyses were correctly reconstructed as separate, as there are no indications yet of a reduced bifurcation of the spine creating a ‘shallowly bifid’ or ‘notched’ appearance sensu Wedel and Taylor (2013). The metapophyses of DV5 are shallowly bifid. When separate, the metapophyses are distinctly V-shaped, although the preserved left apex is anteriorly inclined, unlike those of the previous vertebrae. Because this is the only vertebra that shows this feature, and the apex appears brittle, this inclination is likely due to deformation. In DV6, the metapophyses have fused fully, resulting in a vertical neural spine. In DV4, the SPRL and SPDL are distinct from each other, unlike the morphology seen in DV3 where laminar capture sensu Wilson (2012) in still ongoing. However, both the SPRLs and SPDLs are poorly preserved. Only parts on the base of the metapophyses of the SPRLs of DV4 are preserved. At midheight on the metapophyses parts of the SPDLs are preserved. In DV5, the left SPRL has a similar preservation to SPRLs in DV4. The dorsal part of the right SPRL is preserved. Most of the original left SPRL of DV6 is preserved, missing only parts ventrally, and to a limited extent dorsally. Only the dorsal half of the right SPRL is preserved, but both SPRLs do not join dorsally to form a single PRSL. Instead, the PRSL originates medially to the SPRLs in the dorsal half of the neural spine. The right SPRL is oriented dorsolaterally near the neural spine apex, extending into the prezygapophyseal spinodiapophyseal fossa (PRSDF). The left SPDL of DV6 extends dorsally to approximately the midheight of the neural spine. The dorsal half of the lamina is reconstructed. On the right side, most of the SPDL is reconstructed, except the dorsal-most part of the lamina. The SPOLs in DV4 connect ventrally to the dorsal margin of the postzygapophyses, and dorsolaterally to the SPDLs, although the latter connection is only preserved on the left side. They meet on the posterolateral side of the metapophysis, roughly at midheight of the metapophysis, and form a triangular aliform process. DV5 is the first dorsal vertebra to have a clear distinction between the lateral and medial SPOLs. The bifurcation of the SPOL occurs just dorsal to the postzygapophyses, whereby the mSPOLs converge dorsally to form a rugose POSL. The lSPOLs fuse laterally with the reconstructed SPDLs. On the posterior surface of the spine of DV6, the mSPOLs meet at approximately one-third of the neural spine height, fuse, and form the POSL, which is distinctly rugose. In the SPOFs, lSPOLs are present which meet the SPDLs at approximately midheight, which would form lateral spinal laminae. Dorsal to the lSPOL, in the right SPOF, an accessory lamina is present with the same overall orientation as the lSPOL. Although most of the right SPDL of DV6 is reconstructed, a triangular aliform is preserved, and appears intact. The aliform process, however, is, due to deformation, folded anteriorly. Dorsal to the aliform process, the lateral surface of the neural spine tip bears the neural spine cup, which in DV6, as well as more posterior vertebrae in MAB011899, is a triangular, dorsally widening bony sheet which is slightly rugose. These triangular, rugose sheets are common in diplodocids, as they are present in Diplodocus ( Hatcher, 1901; Herne and Lucas, 2006; Lucas et al., 2006), Barosaurus ( Lull, 1919; McIntosh, 2005), Apatosaurus ( Gilmore, 1936; Upchurch et al., 2004 b), Brontosaurus ( Ostrom and McIntosh, 1966) , and Supersaurus ( Jensen, 1985; Bonaparte and Mateus, 1999). The spine tip is squared dorsally, similar to the posterior-most dorsal vertebra in Diplodocus carnegii ( Hatcher, 1901, plate VIII), but the left side is reconstructed, and the medial ‘notch’ shows some breakage, which might indicate that - similar to more posterior vertebrae - the dorsal tip was rounded, and not ‘notched’ sensu Wedel and Taylor (2013).

Most of the right postzygapophysis of DV4 is preserved, and unlike more posterior vertebrae wherein the articular facets are inclined laterally, the facet is oriented nearly horizontally. The left postzygapophysis is partially reconstructed, including the lateral rim and the section where the PODL attaches to the postzygapophysis. Only the lateral rim of the left postzygapophysis of DV5 is preserved, whereas the remaining bone of the postzygapophyses and the hyposphene are reconstructed using acrylic resin. Only a small part of the ventrally supporting lamina is preserved. In DV6, both postzygapophyses are preserved and are oriented nearly horizontally, inclined laterally similar to DV7 of Diplodocus ( Hatcher, 1901, plate VIII) or DV8 of Barosaurus ( McIntosh, 2005, figure 2.5H). The hyposphene of DV4 is partially broken but appears to be too wide to constitute a laminar hyposphene. This is further supported by more posterior vertebrae, which contain a rhomboid hyposphene. The hyposphene in DV4 is ventrally supported by a single dorsoventral lamina (sTPOL sensu Carballido and Sander, 2014), which projects posteriorly in lateral view. DV6, in contrast to DV4 and DV5, preserves a complete hyposphene. The hyposphene is rhomboid, similar in morphology to the hyposphene of DV 7 in Diplodocus carnegii ( Hatcher, 1901, plate VIII). No ventrally supporting lamina is present, but this part of the neural arch is severely damaged. In DV4 and DV5 the CPOLs are absent. In DV4, lateral to the lamina supporting the hyposphene, two additional laminae extend from the ventral base of the postzygapophyses towards the centrum, which fade ventrally, not extending further ventrally than the supporting lamina of the hyposphene. These supporting laminae could be the presence of mCPOLs in the dorsal series. In DV5, no evidence is present for lateral or mCPOLs, which is also hindered by the poor preservation of this part of the neural arch. The CPOLs are thus more columnar in morphology, indistinguishable from the neural arch. In DV6, the CPOLs are distinct and divided, with lateral branches connecting to the anteroventral side of the postzygapophyses and the medial side of the PCDLs, approximately at the same height as where the PCDLs form a right angle with the transverse processes. The medial branches of the CPOLs connect ventrally to the centrum and dorsolaterally to the neural canal. Dorsally, the mCPOLs connect to the anteroventral margin of the postzygapophyses, medial to the lCPOLs. This is, however, only clear on the left side, as the right side does not preserve the dorsal part of the medial CPOL, which thus appears to connect to the anteroventral part of the hyposphene. The original shape of the neural canal of DV4 is difficult to elucidate, as both anteriorly and posteriorly, the canal is damaged and deformed, and partially infilled with sediment. It was likely subcircular. The neural canal and the surrounding bone of DV5 were damaged, and foam was added internally to stabilize the vertebra, which now obscures the shape and position of the neural canal. The neural canal of DV6 is circular, but the bone surrounding the canal is poorly preserved.

Posterior Dorsal Vertebrae (DV7-10; Figures 14–17 , Table 2 )

The neural arch and spine morphology of DV7-9 can be described in reasonable detail, aided by the comparisons with the original photographs from the SMA. This is not the case for DV10, because the vertebra was part of the block containing the sacrum, and due to the severe anteroposterior compression of the sacral vertebrae, DV10 is very poorly preserved. It will therefore be discussed separately, mainly based on photographs of the SMA and early photographs from the Oertijdmuseum, because in these photographs the vertebra was yet to undergo reconstructions.

Centrum morphology. All posterior vertebral centra are slightly opisthocoelous, with a mildly convex condyle, and a concave posterior articular surface. All centra are ventrally concave in lateral view, but less so than the more anterior dorsal vertebrae. The pleurocoels of DV7 are roughly triangular (left) and oval (right) in outline and include the dorsoventral bony struts in the middle of the coel. The right bony strut is reconstructed. The centrum of DV8 is deformed in a way that the pleurocoel on the left side consists of two fossae; a triangular shaped anterior fossa, a thickened bony strut, and a very small posterior coel, which is roughly pentagonal shaped. On the right surface, the pleurocoel is dorsoventrally compressed, resulting in a flattened oval, in which the thin bony strut is located in the posterior half of the coel, similar to its position in the left pleurocoel. On the right side, dorsal to the pleurocoel, a bulged, irregularly shaped piece of bone projects laterally ( Figure 15 ), which has been attached using acrylic resin. It may potentially represent a deformed, misplaced parapophysis, but this is uncertain. In DV9, the pleurocoels are oval, with the right coel being dorsoventrally compressed. Bony struts are present and are slightly inclined posteriorly. The centrum of DV10 is mostly reconstructed (see below) and is missing the anterior half and parts of the cotyle. The left pleurocoel is oval, but only the posterior half is preserved. The right pleurocoel is a dorsoventrally compressed oval. At first glance both contain dorsoventral bony struts, but both struts are in fact completely reconstructed.

Neural arch morphology. The neural arches and spines of DV7 and DV8 are relatively well preserved, with DV9 less complete. In DV7, the prezygapophyses are preserved, slightly convex dorsally, and are oriented sub-horizontally. They are ventrally supported by CPRLs, but additional lateral branches have been added using acrylic resin, so the laminae wrongly seem to bifurcate dorsally in the reconstructed vertebra. Therefore, these additional branches are not considered here. The right prezygapophysis of DV8 is partially preserved. The articular facet is compressed, facing ventrally, and attached to a deformed PCPL. The left prezygapophysis is partially preserved; the dorsal surface is reconstructed. The left CPRL is entirely preserved, but the anterior margin of the ventral half of the lamina is pushed laterally due to shear. The right CPRL misses the dorsal half, which has broken off. The prezygapophyses of DV9 are entirely reconstructed, but the supporting CPRLs are preserved.

In DV7, the parapophyses appear to have broken off, just ventral to the prezygapophyses. In DV8 and DV9 the parapophyses appear not to have been preserved, supported by the SMA photographs. The laminae observed in Figures 14D and 15D are interpreted as PCPLs due to their orientation, placement, and the fact that parapophyses must have been present in vivo in all vertebrae. On the right side, DV9 only preserves the ventral-most part of the lamina. In DV8, the lamina is oriented nearly horizontally, caused by compression, and not posteroventrally as in DV7. PRPLs are not preserved. DV7, however, does preserve a PPDL on the anteroventral corner of the tip of the left transverse process, which quickly fades medially on the ventral side of the process, just posterior to the PRDL. This short PPDL and PRDL thus forms a small horizontal, tear-dropshaped fossa on the anterior side of the process. This fossa can be identified as a prezygapophyseal paradiapophyseal fossa (PRPADF) sensu Wilson et al. (2011). This fossa occurs in many different sauropods, such as Camarasaurus supremus ( Osborn and Mook, 1921, plate LXXI), ‘Brachiosaurus’ brancai ( Wilson et al., 2011, figure 10), Rapetosaurus krausei ( Rogers, 2009, figure 16), but also in diplodocids such as Apatosaurus louisae ( Gilmore, 1936; but see Harris, 2006, figure 1). The parapophyseal centrodiapophyseal fossae (PACDFs) in all three vertebrae are ventrally open on all sides, similar to those of Galeamopus pabsti ( Tschopp and Mateus, 2017) .

Most of the right transverse process of DV7 is reconstructed, but the left process is mostly preserved. The lateral tip of the left process is directed lateroventrally, but the lateral-most part has likely broken off sometime between the preparation at the SMA and its arrival at the Oertijdmuseum, as the photographs from the SMA show that the lateral-most tip was directed almost completely ventrally. This lateral tip is partially reconstructed on the posterodorsal side, as shown by a small borehole piercing the transverse process in this area. The medial portion of the left transverse process of DV8 is preserved, but the right process is fully reconstructed. In DV9, both transverse processes are fully reconstructed and misaligned, as the right transverse process is almost fused with the prezygapophysis. Only small, medial parts of the left process are preserved. The left PCDL of DV7 is oriented vertically, with only the ventral-most part extending posteroventrally to reach the posterodorsal side of the centrum. On the right, most of the PCDL is missing, preserving only the ventral part, which shows a similar morphology as the left PCDL. The left PCDL of DV8 is only ventrally well preserved, although the SMA photographs do show that more was present, thus it is likely that acrylic resin was put on top of the lamina, possibly for support. Similarly, only the ventral part of the right PCDL is present. PCDLs are present on both sides in DV9, but most of the right PCDL is reconstructed. A very well-delimited right POSDF is present, resembling a large foramen. However, the entire foramen is reconstructed. DV7 preserves the left PRDL. In DV8 and DV9 they are (except for a very small part halfway of the left PRDL in DV9) reconstructed. The left PODL of DV7 is horizontally oriented, fusing with the diapophyses at midlength and the anterolateral margin of the postzygapophysis. Only the medial half of the right PODL is preserved. In DV8, only the left PODL is preserved, which was dorsally damaged; this is covered with acrylic resin. In DV9, both PODLs are reconstructed. Infradiapophyseal foramina are present in DV7-9, but due to compression, the ventral rim is not as distinct in DV7 on both sides. This, however, is prominently present on the right lateral side of DV9, and bears resemblance to the foramina in Giraffatitan MB.R. 3822 ( Janensch, 1950). In DV8 of MAB011899, the left foramen is filled in with support material, and the right side only shows a small foramen deep within the PRCDF.

The neural spines of these vertebrae are all inclined anteriorly and are roughly as long anteroposteriorly as they are wide mediolaterally. The dorsal tips of the neural spines are all dorsally rounded and rugose, unlike those of Apatosaurus ( Gilmore, 1936) or Diplodocus ( Hatcher, 1901) , which are flat, or possess a midline cleft (notched, sensu Wedel and Taylor, 2013). Most of the neural spines of DV7 and DV8 are preserved, whereas DV9 only preserves the dorsal part of the neural spine, missing most of the ventral half. The SPRLs of DV7 and DV8 are partially reconstructed, with reconstruction more extensive in the right SPRLs. Interestingly, at the point of fusion, a lateral branch of the left SPRL extends dorsally in the PRSDF and connects to the anterior side of the SPDL in DV7. These additional laminae can also be seen in the photographs of the SMA of MAB011899, as well as in NSMT-PV 20375 ( Upchurch et al., 2004 b, plate 3) and NMMNH 3690 ( Herne and Lucas, 2006, figure 2). In NMMNH 3690, however, the laminae are more horizontal and appear as accessory laminae instead of continuations of the SPRLs. This lamina also occurs in DV8 of MAB011899, at least in the right PRSDF (the left PRSDF is reconstructed to the point where the PRSL fuses with the SPDL). The SPRLs of DV9 are not preserved, but were likely present, similar in morphology to DV7 and DV8, and thus have been reconstructed as such. These additional laminae are not seen in DV9. The SPRLs in both DV7 and DV8 join dorsally to form a single PRSL. A PRSL is present in DV9 as well, but because most of the spine is missing, it cannot be confirmed if the SPRLs join dorsally to form the PRSL as in preceding vertebrae. The SPDLs of DV7 are largely reconstructed between the ventral- and dorsal-most portions, but the triangular bony sheets on the lateral surface of the neural spine apex are preserved, and also seen in the SMA photographs. The left triangular sheet is fully preserved, whereas the right sheet lacks the anteroventral part. The right triangular bony sheet of DV8 is preserved, as well as large parts of the SPDLs. DV9 only shows the dorsal portions of the triangular sheets, as the remainder of the sheet and the SPDLs are fully constructed. Both lSPOLs and mSPOLs are present in DV7. The mSPOLs converge dorsally to form a POSL, which is partially covered with acrylic resin for support reasons. The ventral halves of the lSPOLs is preserved; the left lSPOL is significantly more prominent compared to the right. Both lSPOLs fuse with the SPDLs, presumably at the same height based on their overall orientation. In DV8, the mSPOLs also converge to form a rugose POSL. Only a remnant of the left lSPOL is preserved. In DV9, only the dorsal part of the POSL is preserved. Triangular aliform processes are partially present in DV7; the ventral halves of the aliform processes are reconstructed. DV8 and DV9 do not possess aliform processes; they are also absent in DV9 and DV 10 in D. carnegii ( Hatcher, 1901, plate VIII) indicating that their absence in the posterior two to three dorsal vertebrae of MAB011899 may be real.

The postzygapophyses of DV7 are oriented roughly similarly to their orientation in DV6. The hyposphene of DV7 is rhomboid, but the lateroventral parts are compressed dorsally, resulting in a subcircular appearance. Only the ventral half of the left postzygapophysis is preserved in DV8. The hyposphene of DV8 is similar in morphology as DV7. DV9 is missing part of the posterior side of the neural arch, with major reconstructions of the laminae. Only the ventral half of the right postzygapophysis is preserved, and inclined laterally, but this is likely an artifact of both deformation and the major reconstructions. The hyposphene is damaged such that only its original position can be elucidated. DV7, like DV6, possesses divided CPOLs, with the mCPOLs attached to the anteroventral margin of the hyposphene, directly ventral to the anterior border of the postzygapophyses. The lCPOLs are oriented similarly but attach dorsally to the anterior margin of the postzygapophyses, and ventrally near where the PCDLs originate. DV8 has remnants of these lCPOLs on the left side, but this lamina is less prominent compared to DV7. A short, dorsoventral accessory lamina is present, ventromedial to the lCPOL and lateral to the mCPOL, on the left side of the vertebra. There is no evidence for this lamina on the right side of the vertebra. The same applies for DV9 on the left side. In DV9, however, the mCPOLs are damaged, as is the bone surrounding the neural canal. lCPOLs are present in DV9, but due to major reconstructions, it is difficult to elucidate the laminae on the lateral and posterior sides of the neural arch. The neural canal shape of DV7 is not exposed, as it is still infilled with matrix. The neural canal of DV8 is rounded anteriorly, but not exposed posteriorly. In DV9, this cannot be assessed, as the arch is filled with supporting material to strengthen the arch.

DV10 neural arch. DV10, although partially preserved, was found in near-perfect articulation with the sacrum. Based on SMA and other early preparation photographs it is missing the anterior half of the vertebra. A left lateral photograph made roughly upon arrival of the sacrum at the Oertijdmuseum elucidates more morphology than any other picture or the current vertebra in the mount, due to extensive reconstruction, see Figure 17 . Because the photograph is taken before full preparation, the right side cannot be assessed. Only parts of the neural arch of DV10 are preserved. Based on the current, mounted and reconstructed state of the vertebra, it is nearly impossible to assess which parts of the arch are reconstructed, and which parts still consist of bone. Therefore, the following description is mainly based on the photograph shown in Figure 17 .

Both prezygapophyses can be recognized, but they are inclined laterally, and displaced taphonomically posteriorly against the base of the neural spine. Ventral to the left prezygapophysis, the left CPRL can be observed. Posterior to the prezygapophyses, a single left SPRL can be recognized, but it is severely damaged. The right SPRL is not preserved. Dorsal to the CPRL, a PCPL has been pushed anterolaterally due to the breakage of the neural arch. Posterodorsal to the damaged PCPL, the partial left transverse process can be recognized, but most of it is not preserved. A PCDL connects dorsally to the remains of the transverse process and posteroventrally extends to the centrum. The postzygapophyses overhang the prezygapophyses of SV1, and appear slightly inclined laterally. From both postzygapophyses, SPOLs extend dorsally. However, because the neural spine is entirely missing, it is not clear if the SPOLs fuse and form a POSL. From the right postzygapophysis, a single lateral lamina is seen disappearing into the matrix. This is likely the right PODL. From the left postzygapophysis, two anterolaterally projecting laminae can be seen. The lamina that is located more dorsally represents an anteriorly bifurcating PODL. The more ventral of the two laminae is the lateral branch of the CPOL, which extends from the lateral edge of the postzygapophysis to the posterior side of the left PCDL. A mCPOL is also partially visible, dorsally obscured by the overhanging prezygapophyses of the first sacral vertebrae. A single SPDL originates medially to the medial branch of the PODL. Anterior to the postzygapophysis, and posterior to the SPDL, a very deep POSDF is present. Other structures of the neural arch are too difficult to discern from the pictures.

Thoracic Ribs ( Figure 18 , Table 3 )

Preservation and orientation. A total of nine ribs were found associated with the articulated cervico-dorsal series of MAB011899 which still preserve the quarry coordinates, so they can be clearly attributed to their correct position in articulation. More ribs were found ( Figure 2 ), and are likely part of the current composite mount, but it cannot be confidently concluded if these are ribs from MAB011899 or from other individuals in the quarry, as the original SMA coordinates or bone numbers given by the Oertijdmuseum have been lost during the reconstructions. We therefore only describe ribs which can unambiguously assigned to MAB011899.

Three left and four relatively complete right ribs can be identified; one of the right ribs consists of a partial head and is not part of the mount. This partial rib head (RR?4, see below) was found in a storage box, which contained small bone fragments, which are all marked ‘G33/90-2’, as well as two larger elements: a partial rib shaft and part of a rib head. The rib shaft bears two sets of coordinates, both attached with tape. One is written in German, stating ‘G33/90-2 Rippe’, and the other appears to be attached later, likely by the Oertijdmuseum, stating ‘G33/90-1 Brösmeli’. Unfortunately, the shaft does not bear an original SMA coordinate on the surface, contrary to nearly all other bones of MAB011899. Because the rib head element bears ‘G33/90-1 Brösmeli’, it seems that the rib shaft element cannot be unambiguously assigned to either the rib head element or RR?4. The quarry sketch of section G33-90 indicates that both ribs numbered ‘G33-90-1’ and ‘G33-90-2’ were relatively complete when found. It is likely that during excavations and due to the crumbly nature of the bones, the ribs were only partially recovered. Therefore, neither the quarry map nor sketches elucidate to which of the two ribs the rib shaft element belongs. This element is therefore described separately from the other elements.

It must be noted that large parts of the ribs are reconstructed. The ribs are described as if they are in an anatomical position, whereby the tuberculum is facing medially and the capitulum is facing ventromedially in the anterior ribs, and medially/dorsomedially in more posterior ribs. The ribs are indicated by their in vivo position, which is discussed after the description of the ribs; the ribs are referred to as RL or RR, abbreviations for ‘rib left’ and ‘rib right,’ followed by their (possible) serial position. None of the ribs bear any pneumatic foramina or oblique ridges on the posterior surface of the rib head.

Left Thoracic Ribs

RL1 ( Figure 18A ). RL1 is a small rib, of which only the tuberculum, the head, and part of the distal shaft is preserved. The midshaft and most of the capitulum are reconstructed in the mount, but the base of the capitulum is preserved. It has a short, subcircular shaft, and an approximate right angle between the tuberculum and the preserved base of the capitulum. The tuberculum bears a very distinct ridge on its anterior surface, which originates from the tubercular facet and extends distally onto the shaft. Unlike Galeamopus pabsti , however, this ridge is not straight ( Tschopp and Mateus, 2017, figure 50), but bulges anteromedially in the middle, thereby slightly folding over the base of the capitulum. The preserved part of the distal shaft tapers distally and is very thin compared to more posterior ribs, see Table 3 .

RL?4 ( Figure 18B ). Only the base of the head is preserved, bearing the ridge which originated from the tuberculum, which connects to the anterior edge of the shaft. The ridge is far less pronounced compared to the ridge in RL1. The midshaft is no longer subcircular, but closer to subtriangular, being wider anteroposteriorly than mediolaterally. The distal part of the shaft is not preserved but reconstructed for mounting purposes.

RL?8 ( Figure 18C ). It is similar in morphology to RL?4, but more of the head is missing, the shaft is significantly shorter, the ridge is more robust, and the overall curvature of the shaft is stronger, although this is possibly influenced by the reconstructions carried out on the rib. Distally, the preserved shaft becomes more oval than subtriangular and is mediolaterally flattened.

Right Thoracic Ribs

RR?3 ( Figure 18D, H ). A long, right rib preserves the entire head of the rib, as well as more than half the shaft. A very distinct ridge is present on the anterior surface of the tuberculum, extending from the tuberculum onto the shaft surface, similar to RL1, but not as medially displaced. On the anterior surface of the capitulum, a similarly oriented, broader, less pronounced ridge is present. The surface between the tuberculum and the capitulum is thin and concave medially. The articular ends of the tuberculum and capitulum are rugose. The posterior side of the rib head between the capitulum and tuberculum is flat. The shaft has a subtriangular outline. As mounted, the shaft is incorrectly connected to the rib head in the reconstructed skeleton, as the anterior face of the head is currently confluent with the lateral face of the shaft. Articulation with the corresponding vertebra would be impossible, as the shaft would be directed posteriorly if the tuberculum and capitulum were articulated with the dorsal vertebra.

RR?4 ( Figure 18I ). This rib is not part of the mount. It consists of a partial head and part of the base of the shaft. The area in between the tuberculum and capitulum is missing, as well as most of the capitulum itself. The tuberculum is well preserved, bearing a ridge on its anterior surface extending onto the shaft. The ridge is straight, similar to the ridge seen in RR?3, but less pronounced.

RR?5 ( Figure 18E ). This consists of a large section of the, lacking the head and the distal shaft. The curvature of the shaft is more significant in RR?5 compared to RR?3. The shaft, however, has been partially reconstructed, which may have altered the overall outline of the rib. Most of the rib shaft has a subtriangular outline, but the distal part of the preserved shaft is more oval.

RR?6 ( Figure 18F, J ). The rib head preserves the tuberculum, which bears a distinct ridge, similar to RL1 or RR?3. The ridge is slightly folded medially in its proximal half, which is likely due to deformation. In life, as seen in RR?3 and RR?4, the ridge would have been straighter. The shaft is rather similar in morphology as RR?3, but less curved.

Rib shaft element. The rib shaft is subtriangular in outline and contains a longitudinal ridge on the posteromedial surface. It is probably part of one of the larger ribs (e.g., RR?4, see above), as the shaft is wide, and no edge of the shaft shows tapering.

Rib head element. This element can be identified as part of a rib head based on the ridge that is preserved, which is present on the anterior or dorsal surface of the tuberculum. This rib head fragment is also part of the right side of the rib cage, as the left side of the fragment is the beginning of the tuberculum, whereas a broken surface to the right of the ridge appears to be the beginning of the capitular facet. It shows no different features from the other ribs.

Serial Position of the Thoracic Ribs

Serial positions are difficult to assign for nearly all ribs, because all ribs miss part of the shafts and part of the head, making comparisons between the ribs and with other taxa nearly impossible. The only rib that is placed unambiguously in the correct position is RL1, due to its short shaft, the approximate right angle between both articular facets, and its similarities with the first rib of G. pabsti ( Tschopp and Mateus, 2017, fig. 50).

For the other ribs, the serial position is ambiguous. Serially, the fourth, fifth and sixth ribs are usually the largest three ribs in diplodocids ( Hatcher, 1901; Gilmore, 1936). Additionally, rib head size can be an indicator of where ribs are placed ( Hatcher, 1901), as in Diplodocus carnegii , the fourth rib bears the largest rib head. However, the ribs of Apatosaurus ( Gilmore, 1936) and Barosaurus , which according to McIntosh (2005), resemble those of Apatosaurus , are different from those of Diplodocus . The ribs of YPM VP.000429 appear to be incomplete and damaged ( McIntosh, 2005), such that comparisons are near impossible. In Apatosaurus , the second rib head is the largest, but based on the shaft length comparisons, the ribs with the largest heads have short shafts. Additionally, the tuberculum becomes less pronounced medially from the fourth rib onwards ( Gilmore, 1936). The latter is also present in Brontosaurus ( Ostrom and McIntosh, 1966) , making comparisons with apatosaurines difficult. The ridge on the tubercular facet can aid in positioning the ribs. In both Apatosaurus louisae ( Gilmore, 1936) and Galeamopus pabsti ( Tschopp and Mateus, 2017) , these ridges become less pronounced in more posterior ribs.

In MAB011899, RR?3 and RL?4 are the largest ribs in terms of shaft length and diameter. RR?4 appears to have an approximately similar head size as RR?3, but the ridge on RR?4 is less pronounced. RR?4 would have been positioned directly after RR?3, or with one rib in between. RL?4 has a slightly smaller midshaft diameter as RR?3, but it falls within the variations seen in G. pabsti , whereby the difference in diameter of left and right ribs at the same serial position can be up to 1.5 cm. Therefore, purely based on the slightly less pronounced preserved part of the ridge, RL?4 is placed directly behind RR?3, but it could be the left equivalent of both RR?3 and RR?4. RR?5 and RR?6 are both significantly more convex ribs, but are still quite large, and would be placed behind RL?4. RR?5 is slightly larger compared to RR?6, although it lacks more of the entire rib, and would thus be placed in front of RR?6. RL?8 is the smallest rib apart from RL1, is more convex than any other rib, and should be placed behind RR?6, likely with one or maybe two ribs in between. In vivo, this could result in the following sequence: RL1 is the first left rib, RR?3 is the third or fourth right rib, RR?4 is the fourth or fifth right rib, RL?4 is the fourth or fifth left rib, RR?5 is the fifth or sixth right rib, RR?6 is the sixth or seventh right rib, and RL?8 is the eighth or ninth left rib. The rib head element preserves a distinct ridge, which could place it in front or behind RR?3. However, because most of this rib head is missing, this is difficult to assess.

Sacrum ( Figure 19 , Table 4 )

Preservation. The sacral vertebrae are only partially preserved and have been heavily reconstructed since the first preparations by the SMA. Several photographs from mostly approximately lateral and dorsal views were made at the time and also subsequently when the sacrum arrived at the Oertijdmuseum, and are used here extensively for further reference, as the reconstructions made during the mounting process are difficult to distinguish from the real bone. The entire sacrum was excavated and prepared as a single block consisting of the remaining parts of DV10, four fused vertebrae that are firmly attached to the left ilium, and SV5, the centrum of which had not fused to SV4 anteriorly. However, the sacral rib of SV5 is fused laterally to the medial surface of the ilium, just as the other sacral ribs. The entire right side of the sacrum, including the right ilium and some of the righthand parts of the neural spines, is missing.

On excavation SV5 was preserved in a near-horizontal orientation, with the neural spine directed anterodorsally and the centrum was directed posteroventrally. This caused the sacrum to have an anteroposteriorly folded appearance in older photographs. Because SV5 was reoriented in this manner, the anterodorsal surface of the neural spine of SV5 was displaced anteriorly into the posterior surface of the neural spine of SV4, which caused the neural spine of SV4 to break at midheight. Subsequently, the forward displacement of SV5 caused the neural spines of SV2 and SV3 to bulge, and become convex anteriorly, as the neural spine of SV4 was pushed into the preceding vertebrae. During preparation, it became clear that both the dorsal halves of the neural spines of SV4 and SV5 were lost. In the mount, the neural spines have been straightened as to approach in vivo position; both the dorsal halves of the neural spines of SV4 and SV5 are reconstructed. During the mounting and preparation processes, an additional sixth sacral neural spine was wrongly added as a reconstruction. To avoid future confusion when studying the specimen, it is important to note which neural spine is reconstructed. Based on the trajectory of the SPDL of the fourth neural spine (as mounted) and the enlarged, both anteroposteriorly and transversely, fifth spine, it appears that this fifth mounted spine is reconstructed and was wrongly added to the sacrum, which is composed of five vertebrae as is ubiquitous in diplodocids and many other eusauropods except derived macronarians ( Filippini et al., 2017).

Description. The sacral neural spines project significantly above the dorsal rim of the ilium, similar to the situation in other diplodocid taxa ( Hatcher, 1901; Gilmore, 1936; McIntosh, 2005). The first sacral neural spine is oriented nearly vertically. The left SPDL of SV1 is located on the anterolateral side of the spine. The lamina is prominent, protruding laterally, and bifurcates ventrally, approximately at midheight. The anterior branch connects to the dorsal surface of the first sacral diapophysis, which is indistinguishable from the tuberculum of the sacral rib, but which are still separate elements sensu Wilson (2011). The posterior branch is attached to the ventral-most part of the spine, close to the posterior edge of the spine and the anteromedial side of the diapophysis of SV2. Dorsally, the SPDL of SV1 becomes a single stout lamina, and expands anteroposteriorly when reaching the spine tip, forming a triangular, rugose sheet of bone. A small part, ventral to the triangular bony sheet, of the SPDL is reconstructed. The POSDF is dorsally bordered by a prominent ventral rim of this bony sheet, and anteriorly by the SPDL. Both SPRLs can be recognized, albeit only partially preserved, which join dorsally to form a single rugose PRSL. The rugose POSL extends dorsally above the dorsal edge of the spine.

The second and third sacral neural spines are fused, similar to e.g., SMF R462 (formerly AMNH FARB 516; Osborn, 1904, figure 3) and UW 15556 (formerly CM 563; Hatcher, 1903, plate IV). A PRSL and POSL can be observed, as well as two SPDLs. The two SPDLs expand dorsally in anteroposterior direction to form a single, expanded bony sheet on the lateral side of the neural spine tip. Ventrally, the SPDL of SV2 is not bifurcated, but single. The SPDL of SV2 is ventrally attached to the dorsal surface of the second diapophysis.

The SPDL of SV3 is ventrally bifurcated, but unlike the SPDL of SV1, the bifurcation occurs near the neural spine tip. The anterior lamina can be confidently identified as the SPDL, as the ventral part of the lamina attaches to the dorsal surface of the third diapophysis. The posterior branch, however, is either a second SPDL as in SV1, or it is a lSPOL which dorsally fuses to the SPDL. The edge of the lamina faces posteriorly, but this appears to be influenced by compression. Ventrally, the lamina either attaches to where the postzygapophysis and prezygapophysis of SV3 and SV4 fuse, thus identifying the lamina as a lSPOL, or, which can’t be elucidated with confidence, the lamina attaches to the anterodorsal surface of the fourth diapophysis, thereby identifying the lamina as a second SPDL. The continuation of this lamina towards the fourth diapophysis is difficult to judge, as the lamina shows some breakage near the postzygapophysis, whereby the lamina could terminate, but also could continue towards the diapophysis of SV4.

The anatomy of the fourth neural spine is difficult to discern due to reconstruction and poor preservation. Only the left ventrolateral half was preserved. Additionally, the neural spine was not fully separated from the third spine, but also not fully fused, due to the prominent POSL of SV3. This appears similar to the sacrum of SMF R462 ( Osborn, 1904, figure 3A), which also, although obscured, seems to have a partially fused fourth neural spine. Only a single SPDL is present. The SPDL attaches ventrally to the dorsal surface of the fourth diapophysis. This SPDL, however, does extend further laterally compared to the SPDLs of more anterior neural spines. Part of the ventral portion of the POSL is exposed, as the dorsal half of the spine is lost.

Because it is separated from the rest, SV5 can be described in more detail. The vertebra resembles an anterior-most caudal vertebra sensu Tschopp et al. (2015a). However, the lateral rim of the left rib folds anteriorly and attaches laterally to the posteromedial margin of the ilium, contacting the fourth rib where both ribs contact the ilium. The centrum is anteroposteriorly short, mildly opisthocoelous, and ventrally concave. The posterior articular surface is subcircular. The neural canal is oval and is taller than wide. The neural canal is bordered by two transversely wide CPOLs. The postzygapophyses are distinctly V-shaped in posterior view and supported by a TPOL from below. A rhomboid-shaped hyposphene is added in the mount, however, this is not found in earlier pictures. This could, however, be an artifact of the age and detail of the pictures from the SMA, which renders it difficult to discern whether a hyposphene was present, as hyposphenes are known to be present in post-dorsal vertebrae ( Apesteguía, 2005). However, it is more likely, based on the SMA photographs, that if the hyposphene was present, it would be more of a laminar structure, instead of the reconstructed rhomboid hyposphene. From the postzygapophyses, two vertical, parallel SPOLs are present, which project posteriorly, and are separate from the POSL. The POSL originates approximately at midheight, medial to the SPOLs, is distinctly rugose, and runs towards the dorsal edge of the remainder of the spine. It is likely, as seen in SV3 and the posterior dorsal vertebrae, that the POSL would have reached the neural spine tip. Just above the dorsal-most point of the postzygapophyses, horizontal PODLs are present which meet where the SPDLs fade into the ‘wing-like’ transverse processes. On the right side, only a small part of the ventral border of the rib is preserved.

The anterior part of the vertebra is poorly preserved, also due to breakage caused by detachment of the vertebra from the fused sacral vertebrae and ilium. A convex, dorsoventrally flattened, small condyle is present. Posterolaterally, the ventral borders of the transverse processes originate. On the left dorsolateral side of the condyle, a half circle of bone is present which likely would have been attached to the fourth sacral rib, bending dorsolaterally towards the neural canal. The neural canal is surrounded by two CPRLs, which like the CPOLs, are stout laminae, with a ‘horseshoe’-like appearance, thus fusing dorsomedially. Dorsal to the CPRLs, two laminar structures are present, oriented dorsolaterally, which likely represented the prezygapophyses. These structures are no longer visible in the mount, as they are overlain by reconstructed prezygapophyses. Medial to the prezygapophyses, a single stout PRSL is present, which is dorsally interrupted as the dorsal half of the neural spine is missing. Posteromedially to the prezygapophyses, the PRDLs originate. On the left side, the edge of the PRDL is dorsally oriented anteriorly, facing laterally towards the lateral side of the rib, and facing anteriorly again ventrally. In posterior view, the margin of the rib curves ventrally approximately halfway along the mediolateral lengths of the rib. This is true for a third of the dorsoventral length of the rib, after which the margin curves lateroventrally. Finally, when the margin meets its lateral-most point, anterior curvature of the rim begins, extending beyond the anterior edge of the neural arch. On the posterior surface of the rib, an enclosed fossa is present lateral to the left CPOL. It is bordered by the neural arch and can be identified as the final transverse foramen sensu Wilson (2011). No intracostal foramen or fossa is observed in this final rib.

The sacral centra are all smooth ventrally, and mildly concave. The first and fourth centra bear pleurocoels, although the first is only partially preserved and infilled with residual sediment, and the fourth is infilled with acrylic resin. The second and third centra cannot be assessed for pleurocoels, due to extensive reconstruction. Intercostal foramina which are anteroposteriorly open are not visible and have likely been infilled with acrylic resin. Intercostal and transverse foramina are present. As the first sacral rib is broken laterally, only a small oval shaped transverse foramen can be observed. This rib would have fused with the anterodorsal-most part of the ilium. Transverse foramina may have been present in the second, third and fourth sacral ribs, but everything has been filled in and covered with acrylic resin for support, resulting in fully enclosed reconstructed ribs. Similarly, describing the intracostal foramina in detail is not possible due to reconstruction, but it appears that at least the third and fourth ribs have them. For the first rib, due to breakage, it is unknown if an intracostal foramen was present. Ventrally, two distinct intercostal foramina are present, laterally in between the second and third, and third and fourth centra. Because only the anterodorsal part of the anterior extension of the fifth rib is attached to the ilium, ventrally, this space is entirely open. For the first rib, because of the breakage, similar to the fifth rib, the ventral space is entirely open. A sacricostal yoke is present, fusing the ventrolateral parts of the ribs in a single anteroposterior elongated block, which is laterally fused to the ventromedial side of the ilium. A small foramen is present in the yoke, lateral to the second intercostal foramen, posterior to the pubic peduncle, but this is probably an artifact of a borehole for supporting the sacrum in earlier stages of the mounting process. The dorsal surface of all ribs is smooth, bearing no ridges or rugose surfaces, with the second and third rib being anteroposteriorly wider compared to the other three ribs.

Caudal Vertebrae ( Figures 20–24 , Table 5 )

Preservation and mounting. Initially, an in-situ plan was made for MAB011899 in 1994, wherein more than 20 caudal vertebrae were assigned to MAB011899. However, in subsequent years, most of these caudal vertebrae were determined to belong to the individuals ‘Twin’ and ‘Triplo’ ( Figure 2 ), as more material was found of these individuals. Figure 2 shows seven caudal vertebrae assigned to MAB011899, three of them articulated with the sacrum. However, the jacket containing most of the material of quarry section G34/93 did only include four caudal vertebrae. The side of the jacket was therefore marked with the numbers “17” (G33/93-1), “18” (G33/93-2), “19” (G34/93-4) and “20” (G34/94-3), indicative of the four vertebrae which were present in this jacket. These numbers were subsequently used by the Oertijdmuseum to indicate each vertebra, as a temporary bone number. Early preparations, however, revealed that the jacket contained five vertebrae. Vertebra ‘17’ turned out to be a very poorly preserved cervical vertebra. This vertebra was not drawn on the quarry map, probably because not enough was exposed to identify this bone as such. It is probable that this vertebra does not belong to MAB011899, as the cervical column was found far from the caudal region, but to either ‘Twin’ or ‘Triplo’ based on flow direction ( Figure 2 ). No anatomical characteristics could be gleaned from the vertebra to further support this.

During the early preparation of this jacket, another caudal vertebra was recognized in front of caudal vertebra (Cd)20. This vertebra was given the number “20.1” (G34/93-2), and was recovered displaced relative to the other vertebrae in the jacket. Further exposure revealed that Cd20 and Cd19 were articulated with each other, and that Cd18 was lying with its anterior articular surface facing towards the neural spine of Cd19, partially on top of the neural arch of Cd19. It thus seems that Cd20.1 very likely articulated posteriorly with Cd20 and Cd19, and that Cd18, and possibly the now lost other vertebrae in quarry sections G33/93 and G34/92, are part of the anterior portion of the tail, but do not necessarily articulate with Cd19. Another caudal vertebra can also be assigned to MAB011899 with confidence. It was numbered “322” and was found next to the lower jaw of SMA 0002/NMZ 100002 in quarry section G33/92-1. In the final maps, Cd322 was assigned to Triplo. However, this assignment is likely incorrect for the following reasons: 1) Cd322 is still quite close to the caudal region of MAB011899, especially compared to most bones assigned to Triplo; 2) the caudal vertebrae that are not drawn as articulated are all displaced in a west-southwest direction, such as Cd18; 3) earlier versions of the map indicate that Cd322 was found much earlier than the bones of Triplo, and when more material of Triplo was found, not directly assigned to Triplo; 4) Cd322 is nearly identical to Cd 18 in morphology, apart from size and spine inclination; and 5) the bone preservation of Cd322 is much closer to the bone preservation of Cd18 compared to the bones of Triplo.

Serial position and orientation. Cd20.1, Cd20, and Cd19 were roughly articulated with no space left in between the vertebrae to fit another caudal vertebra. Cd20.1, although anteroposteriorly compressed, is larger in articular surface diameter, and could articulate with SV5. Therefore, from anterior to posterior, Cd20.1, Cd20, and Cd19 is the correct order for their in vivo placement. Because the jackets were separated, and no pictures are known to capture the direct articulation between SV5 and Cd20.1, the quarry map ( Figure 2 ) and sketches are the only evidence of near direct articulation between these vertebrae. Based on the quarry map and sketches, as well as the minor difference between the articular surface diameters ( Tables 4- 5 ), we can confidently conclude that Cd20.1 did articulate with SV5, so we interpret Cd20.1 as being the first caudal vertebra.

Cd18 and Cd322 are more difficult to place because anterior caudal vertebrae are generally similar in their morphology apart from their size and the development of the transverse processes. Although Cd18 was found on top of Cd19, direct articulation between both vertebrae is unlikely, because of the overall smaller size of Cd18.

Because of the morphological variation among anterior diplodocid caudal vertebrae when the serial position is known, assigning a range of serial positions for each vertebra is a more appropriate approach than definite positions. Measurements of the first five caudal vertebrae of Diplodocus CM 84 ( Hatcher, 1901), Barosaurus YPM VP. 000429 ( Lull, 1919), AMNH FARB 6341 ( McIntosh, 2005), and Apatosaurus CM 3018 ( Gilmore, 1936) indicate a slow decline in articular surface diameter along the sequence. Especially when following the measurements of CM 84, the presence of one or two additional vertebrae between Cd19 and Cd18 cannot be excluded. Cd322 has a slightly taller anterior articular surface, and the neural spine inclination is not as strong as in Cd18, although this might be affected by distortion in the latter. Cd322 likely preceded Cd18. Because the articular surface diameter shows such a slow decline in other diplodocids, there is a possibility that Cd322 and Cd18 did not articulate but were separated from each other along the sequence by a single vertebra. It is possible that Cd322 articulated with Cd19, or that one or two vertebrae were present in between both vertebrae.

Therefore, to account for all variability in serial positions, Cd322 is assigned to a serial position between caudal vertebra four and six, and Cd18 positioned between five and eight. From here, all caudal vertebrae will be referred to by their serial position. Therefore, Cd20.1, Cd20, and Cd19 are referred to as Cd1, Cd2, and Cd3, respectively. Cd18 and Cd322 are referred to as Cd?5-8 and Cd?4-6, to account for the variability in their serial position. The caudal vertebrae are described with the ventral surface of the neural canal parallel to the horizontal.

Descriptions ( Figures 20–24 )

Cd1 (20.1). Only the centrum of Cd1 is preserved ( Figure 20 ), which is the largest in terms of articular diameter of all preserved caudal vertebrae. The centrum is strongly anteroposteriorly compressed, resulting in a flattened centrum in lateral view. The anterior articular surface is slightly concave, whereas the posterior surface is flat. Pleurocoels are present, but weakly developed. Foramina are present on the ventral surface. The remainder of the vertebra is reconstructed.

Cd2 (20). This vertebra is, similarly to Cd1, poorly preserved ( Figure 21 ). It is anteroposteriorly longer than Cd1 ( Table 5 ), but this is most likely influenced by the compression of Cd1. Most of the neural arch is missing, including the neural spine, preand postzygapophyses and transverse processes. The pleurocoels are far less pronounced, lacking a clear ventral rim on the left side. The ventral rim of the posterior articular surface is sheared ventrally. The ventral part of the arch is preserved, but it bears no clear remnants of CPRLs.

Cd3 (19). Parts of the neural arch are missing from Cd3 ( Figure 22 ), as well as part of the posterior articular surface. The vertebra is anteroposteriorly obliquely deformed. Additionally, the left dorsolateral surface of the centrum is displaced medially. The posterior articular surface was flat based on the preserved parts of the articular surface and comparisons with the more complete Cd?4-6 and Cd?5-8. The surface contains numerous cracks. The ventral side of the vertebra is concave anteroposteriorly and does not possess any foramina. This is the only vertebra with well-delimited pleurocoels, albeit deformed. The anterior articular surface is concave, similar to Cd?4-6 and Cd?5-8. The rim is damaged ventrally and dorsally. The neural canal is shaped roughly like an equilateral triangle in anterior view, with the apex placed directly dorsally. Posteriorly, the margins of the canal are broken off, so the original outline of the canal is not preserved. Attached to the centrum are two reconstructed transverse processes. These processes are ventrally reconstructed just dorsal to the pleurocoel, as seen in Diplodocus carnegii ( Hatcher, 1901, plate IX). However, the left process is dorsally connected to the dorsal margin of the centrum, and the right process is dorsally connected to the dorsolateral edge of the PCDL. The left process should connect to the small remnant of the preserved transverse process that is still seen posterolateral to the left prezygapophysis, whereas the right process should connect slightly further dorsally. In vivo, the transverse processes would be distinctly ‘wing’-like, as seen in Diplodocus ( Hatcher, 1901) and Barosaurus ( Lull, 1919) . The neural arch bears a hyposphenal ridge posteriorly; the ridge is partially broken, and because of the overall deformation of the vertebra, it is not straight dorsoventrally. Only the lateroventral parts of the prezygapophyses are preserved, oriented anterodorsally. They are both pointed processes, although the distal ends are not fully preserved, and the end of the right prezygapophysis looks similar to those seen in Cd4-6 of Diplodocus ( Hatcher, 1901, plate IX). This vertebra does preserve CPRLs, but these laminae do not connect to the dorsal surface of the centrum. Instead, they extend approximately level to the dorsal margin of the neural canal and are not nearly as robust as in Diplodocus ( Hatcher, 1901) or Leinkupal ( Gallina et al., 2014) . The neural spine is mostly preserved. However, anatomical detail of the spine is obscured by reconstructions, as the neural spine was separated from the rest of the vertebra during preparations and was ventrally damaged. Therefore, the neural spine is described mostly based on the photographs taken during the preparations. The neural spine was posteriorly inclined. It bears the dorsal parts of the postzygapophyses, which are dorsally rounded. On the lateral sides, the SPRLs and SPOLs can be discerned, which dorsally join to form lateral spinal laminae. Anterior and posterior to these laminae, dorsoventral depressions are present. A PRSL and POSL are present. They are distinct laminae, and both are rugose dorsoventrally.

Cd?4-6 (322). Cd?4-6 ( Figure 23 ) consists of a relatively undeformed centrum and a relatively complete neural arch. Partially preserved pre- and postzygapophyses were reconstructed in the mount. Anteriorly, the articular surface of the centrum is sub-oval, of which the dorsal edge of the rim is compressed ventrally, and the right lateral side of the rim is compressed medially. The posterior articular surface has a distinct, distally projecting rim, but this could be an artifact of deformation. The centrum is concave anteriorly, and flat to mildly convex posteriorly [procoelous-distoplatyan sensu Tidwell et al. (2001), or procoelous-opisthoplatyan sensu González Riga et al. (2009)]. Ventrally, the centrum is concave anteroposteriorly, and bears several elliptical foramina, which are significantly smaller compared to Suuwassea ( Harris, 2006) . Pleurocoels on both sides are significantly reduced compared to the cervical and dorsal vertebrae and consist of rounded depressions ventral to the transverse processes with no distinct borders. Both the left and right ACDL project anteroventrally from the transverse processes. The PCDLs differ, however, as the left PCDL projects posteriorly, with the edge facing laterally. Only the posterior end of the lamina bends ventrally. The right PCDL, however, projects almost fully ventrally from the transverse processes and attaches to the centrum ventral to the ventral margin of the pleurocoel. The dorsal portions of the transverse processes are damaged, and parts of the processes are missing, which would have formed a distinct connection between the processes and the neural arch, creating a ‘wing’-like shape. The processes project laterally to lateroventrally, and terminate laterally as damaged, bulged surfaces. The dorsal edge of the left transverse process contacts the posterolateral side of the prezygapophysis. This cannot be assessed due to breakage for the right side, but a similar structure lateral to the prezygapophysis appears to be present. The neural canal is sub-oval in anterior view, with the rounded edge projecting dorsally, whereas in posterior view, the canal is almost a perfect triangle. Dorsal to the neural canal, the two branches of the TPRL diverge laterally, connecting to the medial edge of the oval prezygapophyses. Interestingly, there is no evidence for the presence of CPRLs in either Cd?4-6 or Cd?5-8, which are notably present in, e.g., Diplodocus ( Hatcher, 1901, plate IX) or Leinkupal ( Gallina et al., 2014, p. 4) . The prezygapophyses appear to have projected anterodorsally. The lateral parts of the prezygapophyses project towards the transverse processes lateroventrally. SPRLs are present, but the ventral part is broken off, so the anterior portion of the SPRLs connecting the prezygapophyses to the neural spine is missing. The postzygapophyses are V-shaped based on the remaining parts. The left postzygapophysis is best preserved, although broken posteriorly, and based on the left SPOL, projects dorsolaterally at the top. The postzygapophyses occupy the ventral third of the neural spine, similar to e.g., Diplodocus carnegii ( Hatcher, 1901) . A single, posteriorly projecting hyposphenal ridge connects the dorsal margin of the neural canal to the ventral margin of the postzygapophyses. The ridge is broken in some parts, but is still recognizable. Lateral to the ridge, the PODLs are present, each of which is connected to the anterolateral edge of the postzygapophyses and to the dorsomedial surface of the transverse processes, just posterior to the prezygapophyses. Medial to both PODLs, dorsolateral to the hyposphenal ridge and lateral to the ventral margin of the postzygapophyses, two distinct triangular fossae are present which are ventrally open. The neural spine bears a broad PRSL and POSL, which both are entirely rugose. The PRSL terminates at the dorsal margin of the spine, whereas the POSL projects beyond the dorsal margin, forming a small projection at the posterodorsal side of the spine tip. The neural spine widens dorsally, strongly in the dorsal third. At midheight, the SPRL and SPOL fuse laterally to form a single, rugose lateral spinal lamina, which projects laterally towards the dorsal tip of the spine, creating a rounded outline in anterior view for the top of the neural spine. Lateral to the POSL, posterior to the lateral spinal laminae, dorsoventral depressions are present, similar to D. carnegii ( Hatcher, 1901) . A similar depression is present anterior to the left lateral spinal lamina. The entire neural spine is inclined posteriorly, including the postzygapophyses.

Cd?5-8 (18). Cd?5-8 ( Figure 24 ) is significantly more deformed compared to Cd?4-6. This deformation is most notable in centrum morphology. Anteriorly, the articular facet is heart-shaped, caused by the mediodorsal compression of the ventrolateral sides of the rim. Like Cd?4-6, the anterior articular surface of the centrum is concave, and the posterior surface is flat to mildly convex. The posterior articular surface, however, is severely obliquely compressed, resulting in a facet that is at least two times smaller compared to the anterior facet. This facet is reconstructed in the mount to similar proportions as the anterior facet. Ventrally, the centrum is anteroposteriorly concave, but lacks the small foramina found in Cd?4-6. A pleurocoel is present on the left side, but on the right side, the pleurocoel is absent, which may be caused by the severe oblique compression. It is likely that both pleurocoels resemble those seen in Cd?4-6. The pleurocoels are oval and located beneath the transverse processes towards the posterior end of the centrum. Transverse processes are partially broken, lacking most of the dorsal parts that connect the processes with the neural arch, as well as the distal extremities. The missing dorsal parts preclude any assessment of the presence of foramina in the transverse processes, but their absence on the left side of Cd?4- 6 leaves their presence in Cd?5-8 unlikely. The transverse processes project laterally, with a subcircular distal end in lateral view. The PCDLs are oriented anteroposteriorly and attach to the rim of the posterior articular facet. The ACDLs are oriented similarly towards the anterior articular facet. The neural canal is ovate, both anteriorly and posteriorly, but this is influenced by the compression of the vertebra. The apex of the neural canal is dorsally with the long axis oriented dorsoventrally. The TPRL is similar in morphology as in Cd?4-6. The prezygapophyses are poorly preserved. However, based on photographs taken during the preparation, the right prezygapophysis was partially present. It projected anterodorsally, had a rounded end, and was quite long compared to the prezygapophyses seen in Diplodocus carnegii ( Hatcher, 1901, plate IX). This prezygapophysis, however, was severely damaged, both dorsally and anteriorly, and it was lost during the mounting process. As currently mounted, the right prezygapophysis is almost entirely broken off, now consisting only of a small bony protrusion. The left prezygapophysis consists of a partially broken facet, which projected anterodorsally, and is roughly elliptical in shape. Like Cd?4-6, the transverse processes fuse medially to the posterolateral side of the prezygapophysis. From the prezygapophyses, two SPRLs extend dorsally onto the lateral side of the neural spine. Posteriorly, partially broken V-shaped, laterally concave postzygapophyses are preserved. They are very large, nearly half the neural spine length, although they do not extend as far dorsally as in Brontosaurus ( Ostrom and McIntosh, 1966) . Only the posterodorsal margin of both postzygapophyses is damaged. Ventral to the postzygapophyses, no clear hyposphenal ridge is present, which is a result of damage. Dorsally, two SPOLs extend onto the posterior, and more dorsally, lateral side of the neural spine. Like Cd?4-6, the distinct, triangular, and ventrally open fossae are present lateral to the postzygapophyses and medial to the PODLs. The neural spine is inclined posteriorly, significantly more so than in Cd?4-6. However, the stronger inclination in Cd?5-8 might be affected by the oblique compression. The remaining neural spine morphology is identical to Cd?4-6, with fusing SPRLs and SPOLs, a rugose PRSL and POSL, of which the latter projects beyond the dorsal margin of the neural spine tip, and the lateral spinal laminae extend to the top of the neural spine.

Chevrons ( Figure 25 , Table 6 )

Two anterior chevrons were found close to the anterior-most caudal centra. Although they were not found in direct articulation with the caudal centra, the first chevron (G34/93-8) was found beneath Cd1. The second chevron (G34/93-5) was found beneath Cd3. They are described as if they are in anatomical connection with the caudal vertebrae.

Both chevrons are ‘Y-shaped’, with a bony crus closing the dorsal side, similar to most flagellicaudatans ( Otero et al., 2012). The haemal canal enclosed by both chevrons is oval, with the dorsoventral axis being longer than the transverse axis. However, there are clear differences between both chevrons. The first chevron ( Figure 25 A-C) has a flattened anterior surface along the blade of the chevron. On this surface, many dorsoventral ridges are present extending parallel to each other; these ridges form a rugose surface for ligament attachment, and fade at the ventral-most part of the anterior surface of the chevron. Based on Alligator , a fascial layer was attached to the anterior surfaces of the chevrons, including the first, which connected the chevron to the cloaca, also serving as an attachment point for the M. caudofemoralis longus (Wilhite, 2023, unpublished data). The ridges are dorsoventrally interrupted by several breaks ( Figure 25A ), as the chevron was broken into multiple parts; these breaks are restored using acrylic resin. The ventral part of the blade tapers anteroposteriorly. In lateral view, the blade is flat. Based on the morphology and the fact that it was found beneath Cd1, as well as comparisons with FMNH P25112, confirms that this chevron is most probably the first chevron in the series. The second chevron ( Figure 25 D-F) is slightly larger, and the haemal canal enclosed is shaped more like a trapezium, with rounded corners. The bony crus is flat and almost horizontal. The blade is morphologically similar to most diplodocid chevrons: thin mediolaterally, and anteroposteriorly wide, tapering ventrally. Unlike e.g., Brontosaurus ( Ostrom and McIntosh, 1966) , the posterior side of the blade does not bear a stepped expansion of the blade. This chevron is, as it was found beneath Cd3, the third chevron in the series.

APPENDICULAR SKELETON

Forelimb Element

Left coracoid ( Figure 26 , Table 7 ). The coracoid is described with the articular surface with the scapula parallel to the vertical. The coracoid was found (coordinate: G35/87-1, not drawn in any of the quarry maps made by the SMA) between MAB011899 and the specimen nicknamed ‘XL’, close to a bone, likely a pubis, which was assigned to XL by the SMA. Several lines of evidence connect this coracoid to MAB011899, rather than XL. First, a single scapula appears to have been assigned to XL, located in the quarry sections G41/ 88 and G41/89, which may or may not have had an associated coracoid. However, based on the quarry map ( Figure 2 ), a coracoid foramen is drawn on the west end of the drawn scapula, indicating that a coracoid was likely fused to the scapula of XL. If this coracoid foramen is not a true foramen, but a possible large crack which was drawn by the SMA, a coracoid is drawn in quarry sections G85/45 and G86/45, which would be more likely to be the coracoid of XL. Second, the coracoid of MAB011899 is smaller than that of G. pabsti ( Tschopp and Mateus, 2017) . XL, however, is considered to be the largest sauropod from the Howe-Stephens Quarry, much larger than both MAB011899 and Galeamopus pabsti . It would therefore be highly unlikely that the coracoid was part of XL. Third, XL was found in a layer above MAB011899, causing a different preservation ( Ayer, 2000), which is supported by the crumbly nature of the bones of MAB011899, including the coracoid. Because the coracoid is closer to MAB011899, is quite small, and has a different preservation, this coracoid can be relatively safely assigned to MAB011899.

The coracoid was not fused with the scapula, as there are no signs of breakage on the coracoid where it articulated with the scapula. Parts of the dorsal surface have broken off but have been restored for mounting purposes ( Figure 26 ). The outline of the coracoid is semi-circular. The outer margin of the coracoid is rough. The anterior margin curves slightly medially. In anterior view, the coracoid is slightly concave medially. In lateral view, the enclosed coracoid foramen is oval, oriented anterodorsally posteroventrally. Medially, the foramen is more circular. The glenoid surface is teardrop-shaped and transversely expanded in the middle. On both ends, the glenoid surface tapers into two sharp points, whereby the posterodorsal end transitions into the articular surface with the scapula. The anteroventral end transitions into the anterior margin of the coracoid.

Hindlimb Elements

Left ilium ( Figure 19 , Table 8 ). The ilium is described as it is oriented in its in vivo position. It is preserved in its entirety. The preacetabular process is pointed and directed anterolaterally in dorsal view with an approximate angle of 30° relative to the body axis. The ventral surface of the anterior lobe of the ilium extends relatively straight towards the pubic peduncle and forms an obtuse angle with the anterior surface of the peduncle. The dorsal margin of the ilium is round anteroposteriorly in lateral view and is highest above the posterolateral margin of the pubic peduncle. The dorsal margin ends, in lateral view, in a pointed, postacetabular process, which has a rough posterior surface. The ischial peduncle, which is present anteroventrally to the postacetabular posterior margin, is distinct and faces posteroventrally. The pubic peduncle is heavily compressed dorsoventrally. The peduncle as preserved is a small extension ventral to the iliac blade. The compression resulted in the ventral-most point of the peduncle forming a flat disc, of which most projects anteriorly and laterally. The base of the peduncle is angled more than 90° from the ventral surface of the preacetabular process. Due to the number of cracks on the surface of the iliac blade, it is difficult to assess if there are any distinct fossae on the surface, however, the triangular fossa, formerly considered to be an autapomorphy of Cetiosaurus oxoniensis ( Upchurch and Martin, 2003) , present in most diplodocines ( Hatcher, 1901; McIntosh, 2005; Lucas et al., 2006; Lovelace et al., 2007; Tschopp and Mateus, 2017), is absent dorsal to the pubic peduncle.

Pubes ( Figure 27 , Table 9 ). Both pubes are nearly complete, with only the right pubis lacking part of the anterodorsal part of the iliac peduncle including the ambiens process. They are described with the long axis of the shaft parallel to the vertical. The obturator foramen is completely enclosed in both pubes, oval in the left pubis, and slightly triangular in the right pubis, and in both elements, it is located above the midline of the ischial articular surface. The ischial articular surface is concave medially. Ventral to the ischial articular surface, the pubic surface sensu Osborn and Mook (1921), which form the pubic apron ( Wilhite, 2003), ends ventrally in a hook-like projection, whereby the pubic surface is not confluent with the rest of the shaft. A projection is also present in e.g., Camarasaurus supremus AMNH FARB 5761/Pb.1 ( Osborn and Mook, 1921) and Supersaurus BYU 725-12424 ( Lovelace et al., 2007), however, in these specimens the projection is not hook-like, but more rounded. This part of the shaft tapers distally. The ambiens process present on the left pubis is evident but shows no sign of a protruding hook-like structure as that seen in, e.g., Diplodocus carnegii ( Hatcher, 1901) . The pubes are elongated, with the posteroventral surface nearly straight towards the distal end of the pubes. The distal end is rounded, rugose, and projects slightly anteriorly. The anterior edge of the shaft is smooth, slightly concave, ending almost horizontally as the ventral surface of the ambiens process.

Ischia ( Figure 28 , Table 10 ). The ischia are nearly complete and distally fused, and therefore figured together ( Figure 28 ). They are described with the long axis of the shaft parallel to the horizontal. Parts of the articular surfaces have suffered damage. The proximal sections of the ischia are concave medially and convex laterally. The articular surfaces, both the iliac peduncle and the pubic articular surface, are rugose. The pubic articular surface of the right ischium is pushed laterally. The acetabular surface is distinctly concave, and not flat as seen in Galeamopus pabsti ( Tschopp and Mateus, 2017) . The acetabular surface and the pubic articular surface are compressed laterally.

Both the ventral and dorsal margins of the shaft are parallel to each other proximodistally. At the base of the shaft, a fossa occurs on the dorsolateral surface. This fossa is elliptical, and rugose; these rugosities extend proximally outside the fossa. Therefore, both fossae and muscle scars are present in the ischia of MAB011899. The ventral margins of the shafts are relatively straight from the pubic articular surface towards the distal end of the shaft. The distal end expands both transversely and dorsoventrally/posterodorsally. In posterior view, the distal shafts are oval shaped, rugose, and fused in an obtuse, V-shaped angle. This morphology is similar to Brontosaurus excelsus ( Ostrom and McIntosh, 1966, plate 67), except for the triangular shape being expanded in all directions and more rounded. The dorsal margin of the distal end of the shaft projects dorsally, and is not confluent with dorsal edge of the shaft, unlike Diplodocus ( Hatcher, 1901; Lucas et al., 2006) or Apatosaurus ( Gilmore, 1936) , in which this distinct dorsal expansion is absent.

Femur ( Figure 29 , Table 11 ). Originally, a complete femur was excavated from the quarry ( Figure 2 ). However, this femur was part of the bones sent to Münchehagen. Subsequent exposure to fire and water damage caused the femur to fall apart, whereby some parts of the femur completely disintegrated. The femoral head, distal condyles and part of the distal shaft are still preserved, with element in the mounted skeleton reconstructed based on other diplodocid femora for mounting purposes. A length for the complete femur is provided in Waskow (2019) as 1344 mm, and the femur is supposedly figured as SMA 0013 (the original specimen number of MAB011899) in Woodruff et al. (2017), but both the measurement and the figure are not of MAB011899, but of SMA 0086 ‘David’, which was found in the northern part of the quarry ( Figure 2 ). A femur length of approximately 1300 mm can be calculated from the quarry map, which would imply, combined with the measurements of Table 11 , a rather gracile femur (robustness index is 0.19, sensu Wilson and Upchurch (2003)). Although calculating femoral length from the quarry map has considerable drawbacks in accuracy, the obtained femoral length is certainly a reasonable estimate given that the femora of SMA 0086 and MAB011899 are approximately the same size. The femur is described with the long axis vertical.

Both the femoral head and the distal condyles have rugose surfaces. The femoral head is separated medially from the shaft. The ventral surface of the head is confluent ventrally with the shaft, and not stepped as seen in e.g., Dicraeosaurus ( Janensch, 1961, p. 208) . The femoral head is transversely wider than anteroposteriorly. From the three pieces, it is clear that the shaft tapers distally from the femoral head and widens again closer to the distal end. The shaft is elliptical, and twice as mediolaterally wide as anteroposteriorly thick. There are no clear indications of severe anteroposterior compression, thus this elliptical outline is more extreme than seen in Diplodocus ( Hatcher, 1901) or Galeamopus ( Tschopp and Mateus, 2017) . The distal condyles are pronounced posteriorly from the posterior side of the shaft, especially the tibial condyle, which is shifted posteriorly, like NSMT-PV 20375 ( Upchurch et al., 2004 b). This projection is more pronounced due to shear but is certainly partly biological.

Tibia ( Figure 30 , Table 11 ). The tibia of MAB011899 was found next to the femur and will be described with the shaft parallel to the vertical, whereby the cnemial crest is placed to the right in anterior view. The tibia is nearly complete and has only suffered minor damage to its posterior surface, both proximally and distally, and is anteroposteriorly compressed as a result of deformation. Additionally, some burn marks are visible from the fire in Münchehagen, mainly at the proximal end. The proximal end is wider compared to the distal end, and has a subtriangular outline, with the cnemial crest strongly expanding. Presence of a second crest sensu ( Bonaparte et al., 2000) cannot be assessed due to the posterolateral damage of the tibia. The distal end is concave, with the posterior part projecting further distally compared to the anterior part of the articular facet, creating the steplike morphology for the insertion of the astragalus. A fibular trochanter is present, posteromedial to the ventral half of the cnemial crest.

Fibula ( Figure 31 , Table 11 ). Only the left fibula is preserved, missing the proximal part. It is described with the long axis vertical. The fibula is slender in anteroposterior view. In lateral view, the fibula tapers towards midlength, and expands again thereafter towards the distal articular surface. In posterior view, the lateral side of the fibula is slightly concave. The insertion of the M. iliofibularis is present just above the narrowest part of the fibula, but its relative position along the bone cannot be accurately determined due to the missing proximal part. The distal articular surface is oval, rugose, and anteroposteriorly wider than mediolaterally.

Figures

FIGURE 2. Quarry map of Ardetosaurus viator MAB011899. Excavation map of the Howe-Stephens Quarry, indicating the major finds from 1992-2000. Individual dinosaurs are color coded, and MAB011899 is coded with dark blue, and named ‘Diplodocus Brösmeli’ herein. The red crosses indicate the missing/lost cervical vertebrae. Note the relatively similar color for ‘Brösmeli’ and ‘David’ (SMA 0086), but their significant separation in the quarry. Figure is courtesy of the SMA. Quarry sections equal 1 by 0.5 m.

FIGURE 5. Cervical vertebrae 13, 14 and 15 of Ardetosaurus viator MAB011899, as photographed by the SMA. Cervical vertebrae 13 and 14 in A) left lateral and B) anterolateral view, with the preserved cervical ribs still present. Cervical vertebra 15 is shown in C) posterolateral and D) anterolateral view. Photographs are courtesy of the SMA. Scale bar is only applicable for figures A and B, as there is no measurable reference for CV15 due to the oblique views. Abbreviations: ap, anterior process; CPRL, centroprezygapophyseal lamina; pl, pleurocoel; PRDL, prezygodiapophyseal lamina; prz, prezygapophysis; SPRL, spinoprezygapophyseal lamina.

FIGURE 6. Cervical vertebra 13 of Ardetosaurus viator MAB011899. CV13 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. A close up of the white box in F is provided of the accessory laminae in the SPRF, shown in anterodorsal view. White shaded areas indicate reconstructed parts. The left cervical rib loop was obscured in ventral view for support and therefore roughly outlined here. White dotted lines in A indicate the remnants of the ventral keel. 1 indicates the triangular projections on the diapophysis. Abbreviations: al, accessory lamina; CPRL, centroprezygapophyseal lamina; epi, epipophysis; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; pre, pre-epipophysis; PRSL, prespinal lamina; pvf, posteroventral flange; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina; TPOL, interpostzygapophyseal lamina.

FIGURE 7. Cervical vertebra 14 of Ardetosaurus viator MAB011899. CV14 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. A close up of the white box in F is provided of the accessory laminae in the SPRF, shown in anterodorsal view. Note the transition of these laminae to a more posterior position in the SPRF. White shaded areas indicate reconstructed parts. White dotted line indicates the remnant of the ventral keel. 1 indicates the triangular projections on the diapophysis. Abbreviations: al, accessory lamina; CPOL, centropostzygapophyseal lamina; CPRL, centroprezygapophyseal lamina; epi, epipophysis; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; pre, pre-epipophysis; PRSL, prespinal lamina; pvf, posteroventral flange; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina; TPOL, interpostzygapophyseal lamina.

FIGURE 8. Dorsal vertebra 1 of Ardetosaurus viator MAB011899. DV1 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. White shaded areas indicate reconstructed parts. Note the bifurcated ACDLs. Abbreviations: ACDL, anterior centrodiapophyseal lamina; al, accessory lamina; CPOL, centropostzygapophyseal lamina; CPRL, centroprezygapophyseal lamina; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; PODL, postzygodiapophyseal lamina; poz, postzygapophysis; prz, prezygapophysis; SPRL, spinoprezygapophyseal lamina; TPOL, interpostzygapophyseal lamina.

FIGURE 9. Dorsal vertebra 2 of Ardetosaurus viator MAB011899. DV2 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. Note the shallow fossae medial to the ventrally bifurcating CPOLs. White shaded areas indicate reconstructed parts. Note the bifurcated ACDLs. Abbreviations: ACDL, anterior centrodiapophyseal lamina; al, accessory lamina; CPOL, centropostzygapophyseal lamina; CPOL-f, centropostzygapophyseal lamina fossa; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; SPRL, spinoprezygapophyseal lamina; TPOL, interpostzygapophyseal lamina; TPRL, interprezygapophyseal lamina.

FIGURE 10. Dorsal vertebra 3 of Ardetosaurus viator MAB011899. DV3 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. Note the displacement of the SPRL/SPDL, as well as the first appearance of the hyposphene, albeit crushed. White shaded areas indicate reconstructed parts. Abbreviations: ACDL, anterior centrodiapophyseal lamina; al, accessory lamina; CPRL, centroprezygapophyseal lamina; hyp, hyposphene; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; prz, prezygapophysis; SPDL, spinodiapophyseal lamina; SPRL, spinoprezygapophyseal lamina.

FIGURE 11. Dorsal vertebra 4 of Ardetosaurus viator MAB011899. DV4 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. White shaded areas indicate reconstructed parts. Abbreviations: hyp, hyposphene; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; PODL, postzygodiapophyseal lamina; SPDL, spinodiapophyseal lamina; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina; tap, triangular aliform process.

FIGURE 12. Dorsal vertebra 5 of Ardetosaurus viator MAB011899. DV5 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. Note the complex morphology of the left parapophysis and the surrounding laminae. White shaded areas indicate reconstructed parts. Abbreviations: ACPL, anterior centroparapophyseal lamina; al, accessory lamina; mSPOL, medial spinopostzygapophyseal lamina; lSPOL, lateral spinopostzygapophyseal lamina; pap, parapophysis; PCDL, posterior centrodiapophyseal lamina; PCPL. posterior centroparapophyseal lamina; PODL, postzygodiapophyseal lamina; PPDL, paradiapophyseal lamina; PRPL, prezgyoparapophyseal lamina; SPDL, spinodiapophyseal lamina; SPRL, spinoprezygapophyseal lamina.

FIGURE 13. Dorsal vertebra 6 of Ardetosaurus viator MAB011899. DV6 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. White shaded areas indicate reconstructed parts. Abbreviations: hyp, hyposphene; mCPOL, medial centropostzygapophyseal lamina; mSPOL, medial spinopostzygapophyseal lamina; lCPOL, lateral centropostzygapophyseal lamina; lSPOL, lateral spinopostzygapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PODL, postzygodiapophyseal lamina; poz, postzygapophysis; PRDL, prezygodiapophyseal lamina; PRSL, prespinal lamina; SPDL, spinodiapophyseal lamina; SPRL, spinoprezygapophyseal lamina; tap, triangular aliform process.

FIGURE 14. Dorsal vertebra 7 of Ardetosaurus viator MAB011899. DV7 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. White shaded areas indicate reconstructed parts. Abbreviations: hyp, hyposphene; mCPOL, medial centropostzygapophyseal lamina; mSPOL, medial spinopostzygapophyseal lamina; lCPOL, lateral centropostzygapophyseal lamina; lSPOL, lateral spinopostzygapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PCPL, posterior centroparapophyseal lamina; PODL, postzygodiapophyseal lamina; poz, postzygapophysis; PRSL, prespinal lamina; SPRL, spinoprezygapophyseal lamina; tap, triangular aliform process.

FIGURE 15. Dorsal vertebra 8 of Ardetosaurus viator MAB011899. DV8 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. Note the attached piece of bone (1) to the lateral surface of the neural arch. White shaded areas indicate reconstructed parts. Abbreviations: hyp, hyposphene; mCPOL, medial centropostzygapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PCPL, posterior centroparapophyseal lamina; PODL, postzygodiapophyseal lamina; POSL, postspinal lamina; poz, postzygapophysis; prz, prezygapophysis; SPDL, spinodiapophyseal lamina; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina.

FIGURE 16. Dorsal vertebra 9 of Ardetosaurus viator MAB011899. DV9 is shown in A) ventral, B) dorsal, C) left lateral, D) right lateral, E) posterior, and F) anterior view. White shaded areas indicate reconstructed parts. Abbreviations: CPRL, centroprezygapophyseal lamina; mCPOL, medial centropostzygapophyseal lamina; lCPOL, lateral centropostzygapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PCPL, posterior centroparapophyseal lamina; POSL, postspinal lamina; poz, postzygapophysis.

FIGURE 17. Dorsal vertebra 10 of Ardetosaurus viator MAB011899. Outline of DV10 is drawn on a left dorsolateral photograph of the sacrum. Photograph is courtesy of René Fraaije. Abbreviations: CPRL, centroprezygapophyseal lamina; lCPOL, lateral centropostzygapophyseal lamina; mCPOL, medial centropostzygapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PCPL, posterior centroparapophyseal lamina; prz, prezygapophysis; SPDL, spinodiapophyseal lamina; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina.

FIGURE 18. Dorsal ribs of Ardetosaurus viator MAB011899. Left ribs: A) RL1 in anterolateral view, B) RL?4 in lateral view, and C) RL?8 in anterolateral view. Right ribs: D) RR?3 in lateral view, E) RR?5 in posterior view, and F) RR?6 in lateral view. Rib heads: H) rib head of RR?3 in anterior view, I) rib head of RR?4 in anterior view, and J) rib head of RR?6 in anterior view. Note the pronounced ridges (1) on the rib heads of RL1, RR?3 and RR?6. White shaded areas indicate reconstructed parts. Abbreviations: cap, capitulum; tub, tuberculum.

FIGURE 19. Sacrum, dorsal vertebra 10 and ilium of Ardetosaurus viator MAB011899. Composite figure of the sacral elements. A) Fifth sacral vertebrae in left anterolateral view. B) Left ilium in lateral view. C) Sacrum in posterodorsal view. D) Sacrum and outline of DV10 in left dorsolateral view. Markings 1, 2, 3, 4, and 5 indicate the sacral ribs in anteroposterior order. Note that in Figure D, the anterior part of the preacetabular process is missing, which had broken off during transport. Figures A, C, and D are courtesy of René Fraaije, made during the preparation and mounting process. Abbreviations: CPRL, centroprezygapophyseal lamina; CPOL, centropostzygapophyseal lamina; PODL, postzygodiapophyseal lamina; POSL, postspinal lamina; poz, postzygapophysis; prap, preacetabular process; PRDL, prezygodiapophyseal lamina; prz, prezygapophysis; pup, pubic peduncle; SPDL, spinodiapophyseal lamina; SPOL, spinopostzygapophyseal lamina.

FIGURE 20. Caudal vertebra Cd1 of Ardetosaurus viator MAB011899. Cd1 is shown in A) left lateral, B) right lateral, C) posterior, and D) anterior view. White shaded areas indicate reconstructed parts.

FIGURE 21. Caudal vertebra Cd2 of Ardetosaurus viator MAB011899. Cd2 is shown in A) left lateral, B) right lateral, C) posterior, and D) anterior view. White shaded areas indicate reconstructed parts.

FIGURE 22. Caudal vertebra Cd3 of Ardetosaurus viator MAB011899. Cd3 is shown in A) left lateral, B) right lateral, C) posterior, and D) anterior view. White shaded areas indicate reconstructed parts. Note the presence of distinct CPRLs ventral to the prezygapophyseal rami. Abbreviations: CPRL, centroprezygapophyseal lamina; POSL, postspinal lamina.

FIGURE 23. Caudal vertebra Cd?4-6 of Ardetosaurus viator MAB011899. Cd?4-6 is shown in A) left lateral, B) right lateral, C) posterior, and D) anterior view. Note the absence of distinct CPRLs ventral to the prezygapophyseal rami. Abbreviations: ACDL, anterior centrodiapophyseal lamina; PCDL, posterior centrodiapophyseal lamina; PODL, postzygodiapophyseal lamina; POSL, postspinal lamina; poz, postzygapophysis; prz, prezygapophysis; SPOL, spinopostzygapophyseal lamina; SPRL, spinoprezygapophyseal lamina.

FIGURE 24. Caudal vertebra Cd?5-8 of Ardetosaurus viator MAB011899. Cd?5-8 is shown in A) left lateral, B) right lateral, C) posterior, and D) anterior view. Note the absence of distinct CPRLs ventral to the prezygapophyseal rami. Abbreviations: PODL, postzygodiapophyseal lamina; POSL, postspinal lamina; poz, postzygapophysis; prz, prezygapophysis; SPRL, spinoprezygapophyseal lamina.

FIGURE 25. Chevrons of Ardetosaurus viator MAB011899. First chevron in A) anterior, B) right lateral and C) posterior view. Third chevron in D) anterior, E) left lateral, and F) posterior view. Note the (1) vertical striations on the anterior surface of the blade in Figure A. The rectangle in Figure A shows the chevron portion shown left of the chevron, wherein the striations are highlighted in white. Figure B includes the outline of the first chevron angled as it would be in vivo. Figure E already shows the third chevron angled as in vivo. Abbreviation: hc, haemal canal.

FIGURE 26. Left coracoid of Ardetosaurus viator MAB011899. Coracoid in shown in A) medial and B) lateral view. White shaded areas indicate reconstructed parts. Abbreviations: cof, coracoid foramen; gl, glenoid.

FIGURE 27. Pubes of Ardetosaurus viator MAB011899. Left pubis is shown in A) lateral and B) medial view. Right pubis is shown in C) lateral and D) medial view. Abbreviations: ac, acetabular surface; isa, ischial articular surface; of, obturator foramen.

FIGURE 28. Ischia of Ardetosaurus viator MAB011899. Ischia are shown in A) right lateral (left ischium cut out), B) left lateral (right ischium cut out), C) ventral, and D) dorsal view. Abbreviations: ac, acetabular surface; pua, pubic articular surface.

FIGURE 29. Left femur of Ardetosaurus viator MAB011899. Femur is shown in A) anterolateral, B) anteromedial, and C) distal views. Femur could be photographed only from the mount during the mounting process, hence the awkward position and angles. White shaded areas indicate reconstructed parts. Abbreviations: fic, fibular condyle; tic, tibial condyle.

FIGURE 30. Left tibia of Ardetosaurus viator MAB011899. Tibia is shown in A) proximal, B) lateral, C) anterior, D) pos- terior, E) medial, and F) distal views. Abbreviations: cc, cnemial crest; fbtr, fibular trochanter.

FIGURE 31. Left fibula of Ardetosaurus viator MAB011899. Fibula is shown in A) lateral, B) anterior, C) medial, D), posterior view. The proximal end is lacking. The number ‘13’ attached with tape to the lateral surface of the fibular is an old catalogue number of the Oertijdmuseum. Abbreviation: tilf, M. iliofibularis trochanter.

Tables

TABLE 4. Measurements of the sacrum of Ardetosaurus viator MAB011899 (in mm).

Dimension
Centrum length SV1	220
Centrum length SV2	140
Centrum length SV3	180
Centrum length SV4	190
Centrum length SV5	115
Cotyle width SV5	280
Cotyle height SV5	258
Rib width SV5	202
Neural arch height SV5	145

TABLE 7. Measurements of the coracoid of Ardetosaurus viator MAB011899 (in mm).

Dimension
Proximodistal length	338
Dorsoventral height	387
Scapula-coracoid articular length	276
Longest diameter glenoid	174
Transverse width glenoid	87

TABLE 8. Measurements of the left ilium of Ardetosaurus viator MAB011899 (in mm).

Dimension
Proximodistal length	930
Anteroposterior length pubic peduncle	70
Mediolateral width pubic peduncle	80
Preacetabular process length	412