- The Paleontological Society
A remarkable specimen of the small neoceratopsian dinosaur Protoceratops andrewsi (Late Cretaceous, Mongolia) reveals the first nest of this genus, complete with fifteen juveniles. The relatively large size of the individuals and their advanced state of development suggests the possibility that Protoceratops juveniles remained and grew in their nests during at least the early stages of postnatal development. The nest further implies that parental care and sociality are phylogenetically basal behaviors in Ceratopsia. Finally, it reaffirms the conclusion that Protoceratops lived (and died) in the sandy aeolian dune fields of the central Asian craton.
Embryonic, hatchling, and juvenile material are now known for nearly all major dinosaur groups (Ornithopoda, Ceratopsia, Prosauropoda, Sauropoda, and Theropoda). In some cases, based largely on the degree of ossification of the long bones, the young are thought to have left the nests immediately after hatching (Horner, 1984; Weishampel and Horner, 1986; Horner and Weishampel, 1988; Chure et al., 1994). In contrast, some ornithischian taxa are thought to have had nest-bound hatchlings with parental care, based on long-bone ossification and size disparities in concert with the preservation of inferred nests (Horner and Makela, 1979; Horner and Weishampel, 1988; Horner, 2000; Horner et al., 2001).
Here we report on an articulated juvenile assemblage of the dinosaur Protoceratops cf. andrewsi Granger et Gregory 1923 (MPC-D 100/530). The assemblage is interpreted as associated within a nest. Along with aggregations of Psittacosaurus from Early Cretaceous deposits of Liaoning, China (Meng et al., 2004; Erickson et al., 2009), this specimen reaffirms the idea that young dinosaurs remained in the nest for a significant time after hatching, and therefore that parental care was likely involved. This in turn suggests that parental care was a basal behavioral characteristic of ceratopsian dinosaurs.
Locality and Preservation
Mongolian-Japanese expeditions collected MPC-D 100/530 from the Upper Cretaceous Djadokhta Formation of central Asia (Berkey and Morris, 1927; Eberth, 1993; Loope et al., 1998; Dashzeveg et al., 2005; Barsbold et al., 2008; Dingus et al., 2008) exposed at Tugrikin Shire, Ömnögov', Mongolia (Fig. 1). There, the Djadokhta Formation consists almost exclusively of weakly cemented, fine-grained, very well-sorted, very large-scale cross-stratified sandstones. The dip slopes of the huge foresets, which in plan view are arcuate and extend for distances >100 m (Fastovsky et al., 1997; Saneyoshi and Watabe, 2007), are oriented at 52.5°; SD 4.71°; n = 100 (see also Fastovsky et al., 1997, reporting a mean dip orientation of 76.8°; R = 0.929; n = 24). On the basis of textural maturity, the large scale and high angle of the cross-stratification, and mm-scale coarsening upwards sequences observed within the cross-strata, Fastovsky et al. (1997) intepreted the deposits at Tugrikin Shire to represent an aeolian dune field, dominated during Cretaceous time by west-southwest winds. Extensive bioturbation preserved along bounding surfaces, as well as the vertebrate fossil wealth of this locality—mainly articulated specimens of the ceratopsian genus Protoceratops—has suggested to many authors that the Tugrikin Shire dune field was a region of extraordinary ancient biotic activity (Ishii et al., 2000; Gao and Norell, 2000; Dashzeveg et al., 2005; Turner et al., 2007). Descriptions of the petrology and aeolian sedimentary geology of Tugrikin Shire, as well as a description of aspects of the taphonomy of some of the many Protoceratops specimens found there, are found in Fastovsky et al. (1997).
MPC-D 100/530 (Weishampel et al., 2000) consists of a series of fifteen fully articulated, aligned skeletons of juveniles, with the skulls all facing east-southeast. Six complete individuals are closely appressed along the well-defined (eastern) margin, suggesting an originally circular shape about 70 cm in diameter (Fig. 2). All individuals show juvenile characteristics including proportionately large orbits, short snouts, and an absence of adult characteristics such as prominent horns and large frills (Dodson, 1996; Dodson et al., 2004). At least ten are complete (Fig. 2).
Suborder Ceratopsia Marsh, 1890
Infraorder Neoceratopsia Sereno 1986
Family Protoceratopsidae Sereno, 2000
Genus Protoceratops, Granger et Gregory 1923
Protoceratops cf. andrewsi Granger et Gregory 1923
MPC-D 100/530, currently housed at the Paleontological Center of the Mongolian Academy of Sciences, Ulan Baatar, Mongolia.
Locality and horizon
Djadochta Formation, Tugrikin shire, Ömnögov', Mongolia.
Skulls preserve nearly all cranial elements (Fig. 3), including the facial skeleton, braincase, palate, and uncommonly, palpebrals and sclerotic rings. In contrast to adults, the snout is short, blunt and dorsoventrally low. The small rostral bone is situated rostroventrally in the snout region, close to the external nares. Both features are located farther forward in adults. There is no indication of a nasal eminence or horn core; characteristic of adult animals (Makovicky et al., 2007). The elliptical antorbital fossa, characteristic of Protoceratops, has similar relative position on the rostrum to that of adult specimens (Brown and Schlaikjer, 1940; Dodson, 1976); however, it has a rostrodorsal long axis, unlike the adult caudodorsal orientation. The orbits are close to each other and proportionately larger than in adult Protoceratops. Mean skull length is 43 mm (Fig. 3), a number that is signficantly smaller than the skull lengths of adult specimens of Protoceratops that have been recovered from Tugrik and elsewhere, but larger than the smallest specimens known to have been referred to Protoceratops (e.g., IGM100/3003; personal commun., M. Norrell and P. Makovicky).
The width of the supratemporal fenestra is proportionately much greater than in adult individuals. The rostromedial extension of the supratemporal fenestra onto the frontal and parietal is very faintly developed in these juvenile animals, in contrast to adults. The frontoparietal suture is clearly seen at the level of the caudal edge of the orbit. Along this suture are small fronto-parietal depressions. The frill is very short with parallel edges compared to the expanded and caudally flared adult condition. The frill margin is straight, and tapers to a thin edge. Fenestrae are not developed in the frills of these juveniles, although they become well-developed in adult Protoceratops.
Postcranially, these juveniles have 12 dorsal vertebrae and 12 pairs of ribs (Fig. 4.1). Vertebral ossification is incomplete and neural arches and spines are not preserved in any of the skeletons, a condtion that contrasts markedly with the moderately tall neural arches and ossified tendons associated with them in adult Protoceratops; we note, however, neural spines are observed in an embryonic neoceratopsian bearing a braincase resembling that of Protoceratops (Balanoff et al., 2008).
Consistent with their general immaturity, all of the ends of the long bones are composed of poorly-defined, spongy bone. All of the elements of the appendicular skeleton bear poorly-defined articular surfaces. Complete forelimbs, down to the unguals, are preserved in a number of individuals (Fig. 4.2). In all cases, the elements are more gracile than in adults. A low spine crosses from craniodorsal to the supraglenoidal ridge of the narrow scapula, as in adults. The coracoid is not visible in any of the individuals, due to the prone position of the skeletons. The humerus bears a distinct, but low deltopectoral crest lateral on the upper half of the shaft and the long axis of the bone deviates medial from the midshaft onward. The hemispheric head is located more to the caudal surface of the humerus. The ulna is longer than the radius and bears a large olecranon process. The carpus is unossified. The manual digital formula is ?-2-3-4-?-?. Hoof-like unguals terminate each digit of the hand and foot. The length of the forelimb is 67% that of the hindlimb, closer to the condition in basal neoceratopsians than in Psittacosaurus spp.
The ilium is proportionately dorsoventrally low, much more so than in adults, in keeping with incomplete ossification at this early life stage. It is difficult to determine whether the preacetabular process is incipiently everted, as it is in adult Protoceratops andrewsi. The long, slender, and straight ischium lacks an obturator process. The pubis has not yet been exposed (Fig. 4.3).
The hindlimb is also gracile (Fig. 4.3). The femur is bowed and a modest cleft separates that greater from the cranial trochanters. A low, pendant fourth trochanter is located just proximal to the midshaft. The tibia and fibula are 130% the length of the femur, a greater ratio than in adults. The astragalus appears on the medial half of the distal tibial condyle. Beneath the mesotarsal joint are at least three poorly-defined ossified distal tarsals. Overall, the pes is compact and elongate. The metatarsals and phalanges are well-preserved and articulated. Metatarsal V is very small. The pedal digital formula is 2-3-4-5-0 and the unguals are blunt and hoof-like.
Sereno (2000) noted that “generic and specific diagnoses for Protoceratops … do not include any derived features because the skeleton is plesiomorphic … at the level of Neoceratopsia” (p. 489). The issue of diagnosis is further complicated because many of the osteological characters that are used to characterize the adults (e.g., You and Dodson, 2004) are either muted or not developed in these juveniles. In addition to indicating their immature status, several features of the skull, including the relatively flat, straight snout and the short, parallel-sided frill, also constitute the plesiomorphic condition for Ceratopsia (Balanoff et al., 2008). In these features, these ceratopsians resemble the adults and juveniles of more basal members of this clade, like Psittacosaurus spp. (Long and McNamara, 1995; Meng et al., 2004; Zhao et al., 2007), Liaoceratops, and Auroraceratops (You and Dodson, 2004).
For all that, however, several characters suggest Protoceratops affinities including an eliptical antorbital fossa, median keel on the frill, small fronto-parietal depressions (Dong and Currie, 1983; You and Dodson, 2004), and hoof-shaped pedal unguals (Sereno, 2000). Within Protoceratops, two species are recognized: P. andrewsi (Granger and Gregory, 1923) and P. hellinkorhinus (Lambert et al., 2001). Of the characters that distinguish the latter from P. andrewsi—the absence of premaxillary teeth, two nasal horns, well-developed fronto-parietal depressions, and a forwardly curved frill (You and Dodson, 2004)—the presence or absence premaxillary teeth has the potential to diagnose; the other characters are simply not well developed within these juvenile skeletons. In the case of the teeth, however, insufficient preparation precludes positive identification of the character, even presuming that at this juvenile stage they would have erupted. There are, however, no reasons to preclude a specific assignment to P. andrewsi, and given the fact that adult Protoceratops andrewsi are the only ceratopsians known from Tugrikin Shire (Brown and Schlaikjer, 1940; Dodson, 1976; Dong and Currie, 1993; Lambert et al., 2001), we tentatively refer all of the individuals in MPC-D 100/530 to P. cf. andrewsi.
While ideal criteria for identifying dinosaur nests may not all be preserved in MPC-D 100/530 (Chiappe et al., 2004), the shape and size distribution of the assemblage of individuals in MPC-D 100/530 argue for its being a nest. First, the animals in the assemblage are young, the same size, and show the same degree of ontogenic development; they are thus reasonably interpreted to be the same age, such as would be found within a single clutch in a nest. Second, the distinctive curved margin on the eastern side of the specimen (Figs. 2, 4.1) is as might be expected in a circle- or oval-shaped nest (contra Thulborn, 1992). That margin and the distribution of the individuals in MPC-D 100/530 even suggest an approximate diameter for the feature (see above).
An alternative hypothesis is that this specimen is simply some kind of aggregate of individuals, perhaps brought together during a sandstorm (see below). If such were the case, it would be unlikely that the individuals be exclusively juveniles of the same age. Moreover, we view the presence of a well-developed, curved margin in the sandstone at the edge of the assemblage to be inconsistent with this hypothesis. For these reasons, therefore, we prefer the interpretation of this assemblage as a nest of individuals.
Our best inference is that these individuals, although young, are not neonates. First, no eggshell material was found with the specimen, as might be predicted if the animals were freshly hatched. Eggshell is common at this locality so it is unlikely that it would have disappeared diagenetically if present. More significantly, however, the juveniles described here are approximately 60% larger than an a less well-ossified, undescribed specimen from the same locality and referred to the same taxon (IGM100/3003; M. Norrell and P. Makovicky, personal commun.). We infer, therefore, that the animals described here grew after hatching. Growth at the nest implies a life history for Protoceratops andrewsi that involved some degree of of nest-bound parental support.
MPC-D 100/530 provides dramatic evidence of the last minutes of the lives of the juveniles it preserves. The paleoenvironment in which the juveniles were found shows clear evidence of WSW prevailing winds (see above; Fastovsky et al., 1997). The axial skeletons are oriented subparallel to wind direction, with the heads pointing ESE (long axis of the skulls 110.2°; SD 4.45°; n = 12; long axis of the ilia: 103.1°; SD 4.63°; n = 10). The forelimbs of most of the individuals are retracted and manual portions are directed caudally. In contrast with the forelimbs, hindlimbs are fully retracted to lie parallel to the long axis of the body (e.g., Fig. 4). The arrangement of the limbs suggests that the animals tried to move their bodies forward in the then-loose sand, away from the prevailing west-southwest wind, possibly even climbing over one another prior to or during burial.
It has been proposed that the spectacular fossil preservation found at Tugrikin Shire was caused by rapidly migrating dunes and associated sandstorms (Jerzykiewicz et al., 1993; Fastovsky et al., 1997). In this model, the edge along the lee side of the nest would have been the area where the air was the freest of sand and therefore available to the young dinosaurs. Consequently, they may all have struggled to reach this part of the nest. In this case, slipfaces preserved just above the nest itself (see Fig. 1) reinforce this model of episodic storm-driven dune migration in a dynamic, unstabilized sandy environment.
Basing their analysis on the Djadokhta Formation as exposed in the Ukhaa Tolgod locality, Loope et al. (1998) proposed a grain flow model for Djadochta mortality, in which rainstorms caused the catastrophic mobilization of sand down the faces of stabilized alluvial fans. These produced structureless sandy lithofacies such as those associated with the spectacular vertebrate preservation of Ukaa Tolgod and other Djadochta localities (Loope et al., 1998; Dashzeveg et al., 2005). However, the presence of well-developed foresets adjacent to all the fossils preserved at Tugrikin shire, the absence of structureless sandy lithofacies there, as well as the absence of positive evidence for alluvial fan deposits do not support rainstorm-induced grain flows on stablized alluvial fans as the proximate cause of the deaths at Tugrikin Shire generally, or this mortality event in particular (Jerzykiewicz et al., 1993; Fastovsky et al., 1997).
Regardless of which of these two modes dominated in each of these cases, these juvenile dinosaurs lie in situ and all evidence indicates that the inhabitants of the nest were rapidly overwhelmed and entombed while still alive. This then reinforces the conclusion that Protoceratops was not simply passing through the Djadochta dune fields of Tugrikin Shire but rather made its home among the large-scale, sandy, eolian dunes (Fastovsky et al., 1997).
Since the discoveries in the early part of the twentieth century by the expeditions of the American Museum of Natural History (Andrews, 1932, 1933), the Gobi Desert of Mongolia and China has continued to provide a source of extraordinary dinosaur fossils and insights about dinosaurs in the Upper Cretaceous of central Asia. Aside from being among the most spectacular of fossils from a locality known for spectacular fossils (Olsmólska, 1993; Unwin et al., 1995), the inferred Protoceratops nest described here strongly supports the inference that some kind of nest care was involved in the early post-natal developmental stages of these dinosaurs; however, this cannot be directly demonstrated. MPC-D 100/530 represents an in-situ record of Protoceratops nesting in sandy aeolian dune fields, reinforcing the conclusion that it truly lived there. Finally, the animals' nest was overrun by sand, likely during a storm, dramatically recording their last living minutes on Earth.
The late Mr. Lkhagvasuren and Mr. Bayardorj are gratefully acknowledged for early and final preparation, respectively, of MPC-D 100/530. Drs. M. Norell (American Museum of Natural History and P. Makovicky (Field Museum of Natural History) read and edited earlier drafts of this manuscript; their contributions are gratefully acknowledged. We are grateful for thoughtful, constructive reviews by B. Chinnery, F. Jackson, and D. Varricchio. Funding was provided by the Department of Geosciences, University of Rhode Island, and by the Hayashibara Museum of Natural Sciences. This is Hayashibara Museum of Natural Sciences Contribution no. 42.
- Accepted 16 May 2011.