A pair of newly discovered 70-million-year-old fossils from Mongolia — including material previously lost to poaching — reveals the true nature of one of the most enigmatic dinosaur species, Deinocheirus mirificus. See Letter p.257
Although some species of fossil vertebrate are known from hundreds or even thousands of individual skeletons, most are known from the incomplete remains of a single individual, making it difficult to infer the organism's anatomy, ecology and evolutionary origins. This is especially troubling when the fragmentary remains point to animals of unusual appearance. Thankfully, continued fieldwork can lead to the discovery of additional remains and help to fill the gaps in our knowledge. Such is the case for Deinocheirus mirificus, a dinosaur found in the Nemegt Formation in Mongolia, which dates to around 70 million years ago, during the Late Cretaceous period. Deinocheirus was known only from a gigantic pair of arms and a few isolated bones — until now. On page 257 of this issue, Lee et al.1 describe two new skeletons of Deinocheirus, which add together to give us a nearly complete anatomy of this previously mysterious animal.
In July 1965, Zofia Kielan-Jaworowska of the Polish–Mongolian Palaeontological Expedition discovered the first Deinocheirus remains. All that was found were nearly complete shoulder girdles and forelimbs (the latter an astounding 2.4 metres long, the record-holder for the longest-known arms of a bipedal animal), and a few ribs and vertebrae. These clearly showed the diagnostic traits of a theropod — the group of mostly carnivorous dinosaurs that includes Allosaurus, Tyrannosaurus and Velociraptor, and also, through descent, modern birds. Despite their incompleteness, the remains were sufficiently distinct in size and morphology from other known theropods to allow Kielan-Jaworowska's colleagues Halska Osmólska and Ewa Roniewicz to propose2 that they belonged to a new genus and species, Deinocheirus mirificus, or 'unusual horrible hand'.
The lack of a complete skeleton meant that palaeontologists were uncertain as to the place of Deinocheirus in the theropod family tree. Although Osmólska and Roniewicz observed some similarities with the smaller and superficially ostrich-like ornithomimosaurs (a group of omnivorous beaked theropods), they remained uncommitted as to a particular affinity with any of the then-known groups of dinosaurs. Some early reconstructions for popular audiences portrayed the dinosaur as a generalized Allosaurus-like carnivore with hypertrophied prey-capturing arms; more bizarrely, others speculated that it was a gigantic sloth-like climber that suspended itself underneath the limbs of (unrecorded in the fossil record) enormous trees3. By 1972, however, palaeontologist John Ostrom had agreed that Osmólska and Roniewicz's initial observations were correct and that, bone for bone, Deinocheirus shared several anatomical features uniquely with ornithomimosaurs, and so was most likely an 'ostrich dinosaur' of hitherto unrecognized gigantic proportions4. Although some have been sceptical of this claim5, phylogenetic analyses have supported the hypothesis that Deinocheirus was a primitive ornithomimosaur6, and most modern restorations of this dinosaur have portrayed it as such.
Lee and colleagues' discoveries show that Deinocheirus in fact had a series of unusual features not seen in combination before (Fig. 1). The dinosaur's toothless snout flares out to the sides, similarly to that of its contemporaries the hadrosaurids (duck-billed herbivorous dinosaurs). Its lower jaw is particularly massive for an ornithomimosaur, more reminiscent of other large-bodied theropods. The neural spines of the lower back, hips and base of the tail are extended vertically to form a tall sail, somewhat like that of the distantly related Spinosaurus. And Deinocheirus retains the relatively short and broad foot of a typical theropod, unlike the long, slender feet of the ornithomimid ornithomimosaurs (such as its smaller contemporary Gallimimus) and of the tyrannosaurids (such as the predator of Deinocheirus, Tarbosaurus7). The toes end in unusually squared-off hooves.
Because nearly the entire skeleton is now known, the evolutionary position of Deinocheirus can be estimated much more securely. It is indeed an ornithomimosaur, and is placed by Lee et al. with the earlier Garudimimus and Beishanlong (previously the largest known ornithomimosaur) to form a group to which Osmólska and Roniewicz's name Deinocheiridae has been applied. Both deinocheirids and the smaller ornithomimids lack teeth (more- primitive ornithomimosaurs have simple peg-like teeth), but whereas the ornithomimids evolved numerous adaptations for increased speed, the deinocheirids seem to have specialized for size and weight bearing.
Where did Deinocheirus fit in the ecology of the Nemegt Formation — an environment similar8 to that of today's Okavango Delta, which drains into the Kalahari desert? The anatomy of ornithomimosaurs (including Deinocheirus) is consistent with a diet of plants and small animals9; in addition, Lee et al. document the presence of fish remains in the belly of one of their specimens. Thus, Deinocheirus was seemingly a 6.4-tonne omnivore, inhabiting an ecosystem with several gigantic potential competitors: the comparably large long-necked herbivorous theropod Therizinosaurus; one or more even-larger titanosaurian sauropods (long-necked herbivorous quadrupedal dinosaurs); and the smaller duck-billed Saurolophus. These may have competed for trees as a food resource, and the broader diet of Deinocheirus may have allowed it to access resources that the other tall herbivores of the community could not. The identification of Deinocheirus as a gigantic ornithomimosaur confirms the observation that three separate lineages of non-flesh-eating theropods (ornithomimosaurs, therizinosaurs and the short-beaked bird-like oviraptorosaurs) independently achieved their maximum known size in the later Cretaceous of eastern and central Asia10.
Alongside the anatomical, evolutionary and ecological implications of this discovery, these particular fossils have an unusual collection history. One of the two skeletons was discovered in 2006 and the other in 2009, as part of the Korea–Mongolia International Dinosaur Expedition, but it was realized in the field that parts of both skeletons had already been poached by illegal collectors. Pascal Godefroit at the Royal Belgian Institute of Natural Sciences informed Lee and colleagues in 2011 that he had seen the skull, hand and feet of Deinocheirus in private hands in Europe (see Supplementary Information of the paper1). In May 2014, these specimens were repatriated to Mongolia and the exact match of the joint surfaces of various bones indicated that the poached remains indeed belonged to these skeletons.
A parallel situation played out recently with another 'mystery' theropod dinosaur. Much of the anatomy of the enormous mid-Cretaceous African fish-eating Spinosaurus, known initially from an incomplete skeleton found in 191211, remained obscure until the publication of a new skeleton and assorted isolated elements of this dinosaur earlier this year12. As with the Deinocheirus material, parts of the skeleton had been poached in the field and were later reunited. The anatomy revealed by these more-complete Spinosaurus remains also showed that it was not simply a typical member of its group grown huge, but rather had unsuspected adaptations and perhaps a different mode of life. Both cases reveal that, although inferring the anatomy and ecology of fossil forms from close relatives is often the only reasonable approach, such interpretations can be proved incorrect when primary data are retrieved from the fossil record.