A titanosaurian sauropod with Gondwanan affinities in the latest Cretaceous of Europe

The origin of the last sauropod dinosaur communities in Europe and their evolution during the final 15 million years of the Cretaceous have become a complex phylogenetic and palaeobiogeographic puzzle characterized by the controversy on the alleged coexistence of immigrant, Gondwana-related taxa alongside relictual and insular clades. In this context, we describe a new titanosaurian sauropod dinosaur, Abditosaurus kuehnei gen. et sp. nov., from the Late Cretaceous (Maastrichtian) Tremp Group of Catalonia (Spain). Phylogenetic analyses recover Abditosaurus separately from other European titanosaurs, within a clade of otherwise South American and African saltasaurines. The affinity of the new taxon with southern landmasses is reinforced by spatiotemporal co-occurrence with Gondwanan titanosaurian oospecies in southern Europe. The large size and the lack of osteohistological features potentially related to insular dwarfism or size reduction support the idea that Abditosaurus belongs to an immigrant lineage, unequivocally distinct from some of the island dwarfs of the European archipelago. The arrival of the Abditosaurus lineage to the Ibero–Armorican Island is hypothesized to have occurred during the earliest Maastrichtian (70.6 Ma), probably as a result of a global and regional sea-level drop that reactivated ancient dispersal routes between Africa and Europe. The arrival of large-bodied titanosaurs to the European archipelago produced dramatic changes in its insular ecosystems and important evolutionary changes in its dinosaur faunas, especially with respect to the ‘island rule’ effect. The authors report a new genus and species of titanosaurian sauropod, Abditosaurus kuehnei, from the Late Cretaceous (ca. 70.5 Ma) of Spain. A. kuehnei groups phylogenetically with South American and African taxa, suggesting geographical connectivity between Africa and Europe at this time.

A t the very end of the Cretaceous, Europe was an extensive island archipelago in which titanosaurian sauropods were integral components of terrestrial ecosystems. In the large Ibero-Armorican Island (which included the current areas of south-western France, Spain and Portugal), the evolution of sauropod communities was shaped by the 'Maastrichtian Dinosaur Turnover' , a major faunal change in which the late Campanian-early Maastrichtian (LC-EM) pre-turnover dinosaur communities were replaced by early-late Maastrichtian post-turnover assemblages [1][2][3] . The pre-turnover titanosaur assemblage included small-sized species such as Lirainosaurus astibiae, Atsinganosaurus velauciensis and Garrigatitan meridionalis [4][5][6] , and moderate-sized species such as Ampelosaurus atacis and Lohuecotitan pandafilandi 7,8 , some of which have been interpreted to have lowered metabolic rates 6,[9][10][11] . Reduction of body size and shifts in growth rate relative to their contemporaneous sister taxa from continental landmasses may have resulted from selective pressures under insular, limited-resource environmentsan evolutionary process known as insular dwarfism 12,13 . In contrast, the post-turnover forms are apparently larger in size but still taxonomically undescribed 14,15 , and their first appearance on the island co-occurs with a specific group of oospecies of Gondwanan affinity. Recent phylogenetic analyses distinguish two distinct clades: a relictual core of endemic titanosaurs, and a clade of immigrants with African affinities 11 . This partially supports a previously postulated biotic interchange, including sauropod faunas, between the archipelago -mostly and particularly with the Ibero-Armorican Island 16 -and the neighbouring landmasses during the latest Cretaceous [17][18][19][20][21] .
Here we present a new titanosaurian sauropod dinosaur from the Late Cretaceous (Maastrichtian) of Catalonia that represents the most complete, semi-articulated titanosaur skeleton thus far discovered in Europe. Phylogenetic analyses of this informative taxon provide an opportunity to test hypotheses of phylogenetic relationships among European titanosaurs and shed light on palaeobiogeographic events between the southern islands of the European archipelago and the northern regions of the African landmass. Further, the new taxon is substantially larger-bodied than any species within the late Campanian-early Maastrichtian titanosaur assemblage, probably indicating an eventual decrease of the alleged insular dwarfism or the so-called 'island rule' effect in Ibero-Armorican sauropod faunas. meaning lizard. 'kuehnei' honours Professor Walter Georg Kühne , the discoverer of the specimen.
Holotype. An associated, semi-articulated, partial skeleton consisting of isolated teeth, 12 articulated and partial cervical vertebrae, cervical ribs, 7 nearly complete or fragmentary anterior and middle dorsal vertebrae, 6 complete or almost complete dorsal ribs and fragments of others, 3 chevrons, the right scapula, posterior end of the left scapular blade, right coracoid, left sternal plate, a sternal rib, proximal half of the left humerus, distal half of the right humerus, partial right radius, fragment of the left ilium, part of the proximal half of the right femur, right tibia, right fibula, and distal half of the left fibula with the adhered calcaneum (see Supplementary Information 2.1 for a list of materials with repository numbers). We consider these elements to represent a single titanosaurian individual because they formed a partially articulated skeleton within an accumulation area of 6 m × 4 m at the same stratigraphic level. No duplicated elements were found (Fig. 1).
Osteological description and comparisons. The cervical vertebrae (Fig. 2a) have opisthocoelous, internally pneumatized centra with ventral surfaces lacking a ventral keel or ventrolateral ridges, thus differing from all or some of the cervical centra of most titanosaurs such as Lohuecotitan 8 , Rapetosaurus 24 , Saltasaurus 25 and Overosaurus 26 , which bear some of these structures. Unlike most titanosaurs, the posterior cervical centra bear large, oval and deep pleurocoels. The parapophyses extend half the anteroposterior length of the centrum, project laterally and are located on the anterior margin of the centrum, as in most titanosaurs. The neural spine is estimated to be short. Cervical ribs are fused in 9 of the 12 cervical vertebrae and project far laterally from the centrum (Fig. 2a). The anterior process has a projection that curves medially, resembling a characteristic 'batwing' shape, and extends to the level of the anterior margin of the condyle. The posterior process is elongate and thins distally, overlapping 2.5 more posterior centra. The dorsal vertebrae are described with the neural canal oriented horizontally. They are strongly opisthocoelous, have camellate internal tissue and become shorter from the anterior to the middle vertebrae (Fig. 2b,c). All centra lack a ventral keel, and the lateral pleurocoels are oval and undivided. In the anterior and first middle dorsal vertebrae, the transverse processes are low (with the ventral edge approximately level with the dorsal edge of the posterior cotyle) and laterally directed, as in Lohuecotitan 8 and other titanosaurs 23,24,28 . In the anterior dorsal vertebrae, the anterior centrodiapophyseal lamina does not reach the centrum because it intersects the centroprezygapophyseal lamina. The neural spine is large and triangular, moderately tall and angled posterodorsally (Fig. 2b), similar to those of the middle dorsal vertebrae of most European titanosaurs and others. The surfaces of the spinopostzygapophyseal, centropostzygapophyseal and postzygapophyseal centrodiapophyseal fossae are autapomorphically pneumatized by scattered foramina. The intrapostzygapophyseal and postspinal laminae are absent, as in Overosaurus 26 . The long axes of the zygapophyses are at a low angle relative to the horizontal plane, as in Argentinosaurus 23 , and the diapophyses are laterally directed and located at level with the prezygapophyses. Abditosaurus is unique in having the diapophyses ventral to the postzygapophyses in the anterior and first middle dorsal vertebrae (first, second and third dorsals) (Fig. 2b). A secondary prezygodiapophyseal lamina in the first middle dorsal and a parapophyseal centrodiapophyseal fossae internally divided by an oblique, posterodorsally oriented accessory lamina in the middle dorsals are two autapomorphies of Abditosaurus. Unlike in other Ibero-Armorican species and many other titanosaurs, the centroprezygapophyseal lamina is absent. The posterior centrodiapophyseal lamina bears a dorsal and elliptical pneumatic foramen (Fig.  2b,c), a character shared with Lohuecotitan 8 , Rapetosaurus 24 and Saltasaurus 25 . The capitulum and tuberculum of the dorsal ribs ( Fig. 2d) are oriented at a right angle and united by a thin capitulotubercular web that shows camellate texture. The web lacks pneumatic foramina, unlike most titanosauriforms 38 , a feature that we regard as a local autapomorphy of Abditosaurus. The rib head bears a proximodistally elongate and narrow ridge (outstandingly pronounced in the third rib) that extends distally from the tuberculum onto the rib shaft. The rib shaft has a plank-like, asymmetric, D-shaped cross-section. The second through fourth ribs show a distal spoon-like expansion. In the third dorsal rib, this expansion produces a distal peduncle where the rib articulates with the sternal plates. Autapomorphically, the posterior surface of the distal end of the second ribs bears a striated anteroposterior thickening or bulge. Caudal chevrons (Fig. 2e) are dorsally open and 'Y-shaped' , with rounded to oval articular facets. The middle chevron has two dorsal articular subfacets, but unlike Lohuecotitan 8 they lack a distinct groove between them.
to Abditosaurus. The scapula contributes more to the glenoid than does the coracoid, differing from Lirainosaurus (coracoid contributes more than scapula 39 ) and Opisthocoelicaudia (scapula and coracoid contributes subequally 41 ), but similar to Mansourasaurus 16 . The coracoid is described with the glenoid surface oriented posteroventrally (Fig. 2g). It is longer than tall and has a quadrangular outline, a character shared with several titanosaurs, and as in many of these sauropods it lacks a ridge originating near the midpoint of the anterodorsal border of the lateral surface. The coracoid foramen is far from the scapular articulation, as in several titanosaurs, but differing from Atsinganosaurus 11 , Lirainosaurus 39 , Ampelosaurus 40 and some other derived titanosaurs 16,24,42 . The foramen is elliptical and autapomorphically twists its orientation from the lateral through to the medial surface. In articulation with the scapula, the dorsal  margin of the coracoid is situated below the level of that of the scapular acromial plate. The sternal plate (Fig. 2h) is semilunar, with a strongly concave lateral surface, as in other titanosaurs 43,44 . Unlike most titanosaurs 6 , however, it lacks an anteroventral ridge on the ventral surface (as in Mnyamawamtuka 29 ). The ratio of the length of the sternal plate to that of the humerus is 0.68 in Abditosaurus, which falls very close to the value seen in saltasaurids (>0.7) 44 . The sternal rib (Fig. 2i) is elongate and rod-like and challenges the assumption that the loss of ossified sternal ribs might be a synapomorphy of Titanosauriformes 45 . The presence of this element in Abditosaurus is considered a local autapomorphy within the clade. The ossification of at least one sternal rib in the new form might be related to the very advanced ontogenetic age of the holotypic individual.
The humerus (Fig. 2j,k) is robust and remarkably different from the more gracile humeri of Ibero-Armorican and many other titanosaurs. The mediolateral development of the proximal end is similar to those described in Saltasaurus 33 , Neuquensaurus australis 35 and Opisthocoelicaudia 41 . The deltopectoral crest is robust and strongly expanded mediolaterally to reach the central axis of the shaft, as in several titanosaurs, but differs from the unexpanded or moderately expanded crests of all other Ibero-Armorican taxa and many other titanosaurs. It projects unambiguously medially, similar to Gondwanatitan 30 , Nullotitan 31 and Jainosaurus 32 . Unlike other Ibero-Armorican titanosaurs, the Abditosaurus humerus exhibits a distally expanded deltopectoral crest, a synapomorphy of Saltasauridae 44,46 shared with Opisthocoelicaudia 41 , Saltasaurus 47 , N. australis 48 and Alamosaurus 49 . The distal condyles are clearly divided as in saltasaurids 44,46 , and the articular surface of the radial condyle faces anterodistally, as in Paralititan 34 and Saltasaurus 47 . The radius (Fig. 2l) has a mediolaterally expanded distal end that is bevelled posterolaterally more than 20° relative to the long axis of the shaft and a well-defined interosseus ridge, as in other lithostrotian titanosaurs 44 . The interosseous ridge extends proximodistally and is laterally emarginated, as in Opisthocoelicaudia 41 . Unlike the rounded shape of Rapetosaurus 24 or Dreadnoughtus 50 , the distal surface is elliptical and anteroposteriorly asymmetrical, as in N. australis 48 . Autapomorphically, Abditosaurus presents a small, oblique ridge on the posterodistal surface of the radius.
The ilium has a preacetabular process that projects anterolaterally, as in Garrigatitan 6 and Lohuecotitan 8 , but differing from Lirainosaurus 39 . Internally, it is pneumatized, as in most Ibero-Armorican titanosaurs 6,8,11,39 , Alamosaurus 51 , Saltasaurus and Neuquensaurus 52 , among others. The femur exhibits a prominent, proximally located lateral bulge, as well as a posterior accessory ridge. The eccentricity index of the shaft is >185%, as in most titanosaurs 53 . The tibia (Fig. 2m) is gracile, and its proximal end is less mediolaterally compressed than in Atsinganosaurus 11 and Lirainosaurus 39 , but also contrasting with the rounded shape of Lohuecotitan 8 and Ampelosaurus 40 . A small but prominent bulge at the posterior margin of the proximal end is also observed in Neuquensaurus 35 , and the cnemial crest projects anteriorly, as in Gondwanatitan 30 . The distal end is longer mediolaterally than anteroposteriorly, as typical for titanosaurs 43,54 . The anteromedial ridge is more pronounced than in Lirainosaurus 39 , Atsinganosaurus 11 and Lohuecotitan 8 . The sigmoid shape of the fibula (Fig. 2n) is a character shared with most Campanian-Maastrichtian titanosaurs, and the presence of a prominent lateral trochanter is a synapomorphy of Saltasaurinae 48 . The fibula of Abditosaurus is unique in having its distal end bevelled posteriorly 20° with respect to the long axis. The distal articular surface is triangular, as in Rapetosaurus 24 , Lirainosaurus 39 , Alamosaurus 55 and Laplatasaurus 56 . The calcaneum (Fig. 2o) is a small, convex and oval element, as in Euhelopus 57 and the purported calcaneum reported in Elaltitan 37 . The ossification of the calcaneum (and a sternal rib) in the Abditosaurus holotype is exceedingly rare in Titanosauria, and therefore their presence in this senescent individual might indicate that the preservation of these elements is related to its advanced ontogenetic stage.

Discussion
Phylogenetic and palaeobiogeographic analyses. We assessed the phylogenetic affinities of Abditosaurus kuehnei using both parsimony and Bayesian methods with the dataset of Gorscak and O'Connor 29 (Methods and Supplementary Information 1.6). In both analyses, Abditosaurus kuehnei is recovered as a saltasaurid lithostrotian titanosaur on the basis of the following synapomorphies: coracoid with rectangular anteroventral margin; humerus with strong posterolateral bulge around the level of deltopectoral crest apex; humeral deltopectoral crest expanded distally; humeral distal condyles divided; and radius with distal end bevelled ca. 20° proximolaterally relative to the shaft 43 . More specifically, Abditosaurus is positioned within a clade of Late Cretaceous saltasaurines from South America and Africa (Fig. 3). Within this otherwise South American-African clade, Abditosaurus and Paralititan are sister taxa, and the Abditosaurus + Paralititan clade is the sister taxon of Maxakalisaurus. The (Abditosaurus + Paralititan) + Maxakalisaurus clade is the sister taxon of the South American saltasaurins (Saltasaurus + Neuquensaurus). The South American-African clade as a whole (that is, (Paralititan + Abditosaurus) + Maxakalisaurus and Saltasaurus + Neuquensaurus) is supported by two synapomorphies (anterior and middle caudal centra wider than tall; divided humeral distal condyles) and is sister to Opisthocoelicaudiinae 58 , a Pan-American, African, and Eurasian clade that comprises most of the rest of the Ibero-Armorican taxa as well as the North American Alamosaurus, the South American Dreadnoughtus, and the Asian Opisthocoelicaudia, among others. The phylogenetic and palaeobiogeographic analyses (Fig. 3) indicate that Abditosaurus is a derived member of a distinct immigrant clade of otherwise South American and African saltasaurine titanosaurs that reached Europe via a dispersal event from Africa during the post-Cenomanian, very probably during the early Maastrichtian. Further, the direct association of the Abditosaurus skeleton with Fusioolithus baghensis, an oospecies with in-ovo titanosaur embryos 59 and an otherwise Gondwanan (South America, India, Africa) distribution (Supplementary Information 2.4), reinforces the southern origin of the clade. The dispersal event from Africa is hypothesized to have occurred during the KMa2 regressive event (70.6 Ma, early Maastrichtian), a eustatic event 60 that affected the central Tethyan margin and northern Africa. With the subaerial exposure of various carbonate platforms, the Europe-Africa connection was reestablished, and Early Cretaceous migratory routes [61][62][63] were probably reactivated, facilitating the dispersal of titanosaurian taxa ( Fig. 4 and Supplementary Information 2.6).

Implications for titanosaur evolution in the European archipelago.
Abditosaurus is a large titanosaur by Ibero-Armorican standards, representing a very mature, senescent individual estimated to be 17.5 m in length with a body mass of 14,053 kg (Supplementary  Table 1). Overall, it is substantially larger than any titanosaur species described from the region, being more than 70% larger than the largest known adult Atsinganosaurus and Garrigatitan, more than twice the size of the largest individuals attributed to Lirainosaurus or Lohuecotitan, and 20% larger than the largest individual attributed to Ampelosaurus (Supplementary Table 1). The anatomical and osteohistological characters of the senescent Abditosaurus do not indicate a reduction in growth rate nor a reduction in body size ( Supplementary Information 2.3), two common traits found in Lirainosaurus 9 , Ampelosaurus 10 , Atsinganosaurus 11 and Magyarosaurus 64 . Indeed, the Abditosaurus appendicular bones lack modified laminar bone, a bone tissue potentially related to a reduction in metabolic rate and body size 10 and interpreted as evidence for possible insular dwarfism in sauropods [9][10][11]20 . Therefore, there is no evidence for metabolic adaptations to insular settings in Abditosaurus, as are present in other European titanosaurs 13,64 . Our results suggest that Abditosaurus must be regarded as a member of a distinct immigrant clade of large titanosaurs that reached Ibero-Armorica during the early Maastrichtian, and as a representative of the post-turnover titanosaurian fauna of this island.
The scenario during the middle of the early Maastrichtian (lower part of C31r) of large-bodied, post-turnover titanosaurs in the Ibero-Armorican Island replacing small late Campanianearly Maastrichtian forms might be mirrored in other islands of the European archipelago, such as in the eastern Hațeg Island (in modern-day Romania), where large taxa seem to appear later than dwarfed forms (for example, Magyarosaurus 65 ), probably also by the middle of the early Maastrichtian (lower part of C31r 66 ).
Accordingly, these new titanosaur faunas are expected to be associated with fusioolithid oospecies in post-early Maastrichtian deposits. Hence, this proposed titanosaurian faunal replacement might represent the decrease of the 'island rule' effect on sauropod communities throughout Europe.

Methods
Osteohistological analyses. Histological samples were taken from the humerus, femur and tibia, prepared as thin sections, and studied using a petrographic microscope.
Body mass and length estimates. We compiled a dataset of limb bone measurements for the Ibero-Armorican titanosaur species (Supplementary  Table 1) for use in body mass and body length estimations with the formulas of Campione and Evans 67 based on femoral and humeral circumferences, which were estimated following the allometric equation of González Riga et al. 68 when necessary, and Seebacher 69 .
Phylogenetic and palaeobiogeographic analyses. We performed maximum parsimony and Bayesian analyses. Maximum parsimony analyses used the dataset of Gorscak and O'Connor 29 , which included the original 55 taxa plus Abditosaurus scored for 272 independent characters that were treated as unordered in all the analyses. Both equal weights and extended implied weighting analyses with different concavity constants were explored. The Bayesian dataset 29 , which also included the original 55 taxa plus Abditosaurus, was coded for a total of 272 variable characters plus 260 autapomorphic, invariable characters. A non-clock analysis was carried out using MrBayes 3.2.6 73 , running in the CIPRES Science Gateway, using a model of variable rates of character state change, sampled from a lognormal distribution, setting an exponential hyperprior of 1.0 for the rate of variation among characters. Four independent runs of the Markov chain Monte Carlo (MCMC) ran for 10 million generations, with one hot and three cold chains. The chains sampled tree-space every 1,000 generations and the first 25% of the posterior distribution was discarded to eliminate the initial climbing phase. Convergence of independent runs was assessed in Tracer 1.7 using effective sample size for each parameter greater than 200. We also performed a tip-dating Bayesian phylogenetic analysis to estimate divergence dates and branch lengths on the basis of additional data in the form of stratigraphic information. The assumed tree model for this set of analyses was the birth-death-skyline-serial-samplin g 74 . A relaxed clock was assumed under a lognormal distribution of sampled rates. Rates of character change were tested under variable (with an assumed gamma-distribution) assumptions. Four independent runs of the MCMC persisted for 50 million generations, with sampling of tree-space occurring every 1,000 generations, and the first 25% of the sample was discarded. Stratigraphic ranges of each taxon (Supplementary Information 1.6 and Table 2) were sampled under a uniform distribution to account for stratigraphic uncertainty. Again, convergence of independent runs was assessed in Tracer 1.7 using effective sample size for each parameter greater than 200. A palaeobiogeographic analysis following the methodology of Sallam et al. 16 was conducted using the R script BioGeoBEARS 75 over the tip-dating maximum clade credibility tree. Three models (DEC, DIVALIKE and BAYAREALIKE) and alternative models with the additional + J parameter to facilitate long-distance dispersal events alongside the assumptions of each model were evaluated. All figures were created by the authors using Adobe Illustrator CC 2015 2.1.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
This published work and the nomenclatural acts it contains have been registered in ZooBank, the proposed online registration system for the International Code of Zoological Nomenclature (ICZN). The ZooBank LSIDs can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix 'http://zoobank.org/' . All other data supporting the findings of this study are available within the paper and its Supplementary Information.

March 2021
Corresponding author(s): BERNAT VILA Last updated by author(s): Nov 22, 2021 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy Provide your data availability statement here.