Simon Wills

The armoured ankylosaurian dinosaurs are best known from Late Cretaceous Northern Hemisphere ecosystems, but their early evolution in the Early–Middle Jurassic is shrouded in mystery due to a poor fossil record^1,2. Spicomellus afer was suggested to be the world’s oldest ankylosaur and the first from Africa, but was based on only a single partial rib from the Middle Jurassic of Morocco³. Here we describe a new, much more complete specimen that confirms the ankylosaurian affinities of Spicomellus, and demonstrates that it has uniquely elaborate dermal armour unlike that of any other vertebrate, extant or extinct. The presence of ‘handle’ vertebrae in the tail of Spicomellus indicates that it possessed a tail weapon, overturning current understanding of tail club evolution in ankylosaurs, as these structures were previously thought to have evolved only in the Early Cretaceous⁴. This ornate armour may have functioned for display as well as defence, and a later reduction to simpler armour with less extravagant osteoderms in Late Cretaceous taxa might indicate a shift towards a primarily defensive function, perhaps in response to increased predation pressures or a switch to combative courtship displays.

Readily identifiable based on their large, “spatulate” teeth with diagnostic “heart”-shaped crowns, turiasaurians are non-neosauropodan eusauropods known from varied Jurassic and Cretaceous formations across Laurasia and Gondwana. Recently, three teeth with turiasaurian features were collected from the Middle Jurassic El Mers III Formation in the Middle Atlas Mountains of north-central Morocco. Although these teeth are superficially similar to those of the Late Jurassic Turiasaurus riodevensis from Spain, the absence of rounded denticles presence of a prominently peaked apex and a mesially flared margin, differ from other known turiasaurians. Turiasaurians have not previously been described from the El Mers III Formation, and the only named sauropod from the El Mers Group, which lacks preserved teeth, is the dubious taxon “Cetiosaurus mogrebiensis”. Due to lack of overlapping material and its lack of clear diagnostic characters, we refrain from referring these teeth to the latter, and identify them as Turiasauria indeterminate instead. These teeth represent the first definitive turiasaurian remains from Morocco, as well as the geologically oldest occurrence of Turiasauria from mainland Africa.

In a recent Perspective Piece, Cooper et al. (2024) highlight the benefits that large language models (LLM) bring to the coding process and offer some guidance around the use of LLM‐generated code in academic publications. Whilst we broadly support the stance taken by the authors, we offer a slightly different perspective from our experience of using both artificial intelligence (AI) generated code and more traditional resources such as Stack Overflow. We highlight how the integration of generative AI tools such as ChatGPT holds immense promise for advancing academic research, revolutionising coding education and practices, promoting inclusivity, and facilitating knowledge dissemination. However, we also acknowledge that researchers must navigate the ethical and methodological challenges inherent in harnessing AI technologies, prioritising transparency, validation, and critical inquiry to ensure the integrity and credibility of research initiatives.

The Lower Cretaceous Wealden Supergroup of southern England yields a diverse assemblage of theropod dinosaurs, its taxa being represented by fragments in addition to some of the most informative associated skeletons of the European Mesozoic. Spinosaurids, neovenatorid allosauroids, tyrannosauroids and dromaeosaurids are among reported Wealden Supergroup clades. However, the majority of relevant specimens are from the Barremian Upper Weald Clay and Wessex formations, and theropod diversity in the older Berriasian–Valanginian Hastings Group has remained poorly known, the fragmentary specimens reported thus far remaining enigmatic both in terms of phylogenetic affinities and sometimes provenance. A better understanding would be welcome given the paucity of Berriasian–Valanginian dinosaurs worldwide. Here, we describe an assemblage of Hastings Group theropod teeth from the Valanginian Wadhurst Clay Formation, mostly collected from the Ashdown Brickworks locality near Bexhill, East Sussex. These teeth were assessed using phylogenetic, discriminant and machine learning analyses and were found to include members of Spinosauridae, Tyrannosauroidea and Dromaeosauridae, in addition to others that remain of uncertain affinity within Coelurosauria. The taxa appear distinct from those already known from Wealden Supergroup strata: the spinosaurid cannot be referred to Baryonyx or the tyrannosauroid to Eotyrannus, for example, but we have not named new taxa at this time. Combined with other findings in the Wadhurst Clay Formation, our study indicates that Valanginian theropod diversity was comparable to that of younger Wealden Supergroup units, implying that the ‘characteristic’ theropod components of Wealden faunas were established early in the deposition of this famous geological succession.

Gomphodonts form a Triassic radiation of small to medium‐bodied (<0.5–2.5 m in length) quadrupedal cynodonts characterized by labiolingually expanded gomphodont postcanines. They were the dominant cynodont group in Middle and Late Triassic ecosystems from the Southern Hemisphere and the first predominantly herbivorous cynodonts to evolve. Gomphodonts were also the first therapsids to develop hypsodonty and a dentition with complex occlusal patterns, and their highly diagnostic upper and lower postcanines show many different morphologies. Here, we explored dental complexity in gomphodont cynodonts through time using geographic information system analysis and orientation patch count applied on 3D crown surfaces of upper and lower gomphodont postcanines belonging to 32 gomphodont taxa. This study reveals that the peak in postcanine complexity was reached early in the evolution of gomphodonts with the emergence in the Early Triassic of omnivorous or insectivorous forms with postcanines made of well‐separated cusps and cingular cuspules. Traversodontids evolved simpler postcanines via coalescence of cusps into crests and the development of large occlusal basins, and the Middle Triassic radiation of traversodontids led to a sharp decrease in mean postcanine complexity. Simplification of the postcanines in traversodontids is interpreted as being related to a gradual increase in the consumption of plant material. Interestingly, the trend of insectivory/omnivory high postcanine complexity and herbivory low dental complexity in gomphodonts is opposite to the trend of dental complexity reported in some extant mammals, with omnivorous having low dental complexity and herbivorous higher. Postcanine complexity remained relatively stable throughout the evolution of traversodontids and only slightly diminished in the Late Triassic due to the presence of minute forms with particularly simple postcanines in the Rhaetian. The major phylogenetic diversity and taxonomic richness of Gomphodontia are represented in two periods of time: at the end of the Anisian, an age in which the postcanine complexity is simplifying, and at the early Carnian when the postcanine complexity in traversodontids, the only Gomphodontia represented, is stable.

Isolated theropod teeth are abundant in many fossil localities (e.g., Currie et al., 1990; Gerke & Wings, 2016; Hendrickx et al., 2019; Hendrickx & Mateus, 2014; Larson & Currie, 2013; Malafaia et ...

Taxonomic identification of fossils based on morphometric data traditionally relies on the use of standard linear models to classify such data. Machine learning and decision trees offer powerful alternative approaches to this problem but are not widely used in palaeontology. Here, we apply these techniques to published morphometric data of isolated theropod teeth in order to explore their utility in tackling taxonomic problems. We chose two published datasets consisting of 886 teeth from 14 taxa and 3020 teeth from 17 taxa, respectively, each with five morphometric variables per tooth. We also explored the effects that missing data have on the final classification accuracy. Our results suggest that machine learning and decision trees yield superior classification results over a wide range of data permutations, with decision trees achieving accuracies of 96% in classifying test data in some cases. Missing data or attempts to generate synthetic data to overcome missing data seriously degrade all classifiers predictive accuracy. The results of our analyses also indicate that using ensemble classifiers combining different classification techniques and the examination of posterior probabilities is a useful aid in checking final class assignments. The application of such techniques to isolated theropod teeth demonstrate that simple morphometric data can be used to yield statistically robust taxonomic classifications and that lower classification accuracy is more likely to reflect preservational limitations of the data or poor application of the methods.