A dinosaur community composition dataset for the Late Cretaceous Nemegt Basin of Mongolia

Dinosaur community composition data for eleven fossil localities in the Late Cretaceous Nemegt Basin of Mongolia are compiled from field observations and records in the literature. Counts were generated from skeletons and represent numbers of individuals preserved in each locality. These data were used in the analyses of Funston et al. [1] “Oviraptorosaur anatomy, diversity, and ecology in the Nemegt Basin” in the Nemegt Ecosystems Special Issue of Palaeogeography, Palaeoclimatology, Palaeoecology, where the results are discussed.


Experimental features
None Data source location Nemegt Basin, Western Gobi Desert, Mongolia Data accessibility Within this article and as Supplementary material

Value of the data
We combined new field observations with an extensive literature survey, compiling an unparalleled community composition dataset for dinosaur palaeontology.
The dataset includes nearly 500 skeletons identifiable to species, which allows for detailed comparison of community composition with other ecosystems around the globe.
Some of the data were collected using GPS, and the map generated from this data allows for examination of finer-scale spatial relations of the skeletons possibly related to taphonomy or palaeoecology.

Data
The data tables presented (Table 1 and Supplementary Table 1) are whole-number counts of the number of skeletons of each type of dinosaur at each locality. There are two tables: the first records occurrence of specimens identifiable to the species-level, the second groups specimens by superfamily. The second table therefore includes some specimens that are not identifiable to species, but are still informative for overall community composition. The columns represent a taxon, and the rows represent localities. The interactive map ( Fig. 1; Supplementary File 1) is a Google Earth (.kmz) file, which has GPS locations of 358 dinosaur skeletons, colour coded by taxon.

Experimental design, materials and methods
Additional details on the methods and materials are available in Funston et al. [1].

Taphonomic considerations
Taphonomy was not directly addressed during data collection, because lithological data were not always recorded with map data, nor with all the specimens reported in the literature. The approach of grouping specimens by locality partly alleviates this issue, because it allowed us to sample from a wider range of taphonomic modes, and provided a time-averaged assemblage that is representative of palaeocommunity structure [20,21]. Future work may find success in correcting species counts using other lines of evidence (eggshell, footprints, or microsites). Table 1 Superfamily-level dinosaur community composition data for 11 Mongolian localities.

Taxonomic considerations
There are four theropod, one hadrosaur, and three protoceratopsian taxa that are each represented by a single specimen, and are interpreted variably. Bagaraatan ostromi [22] has been considered a troodontid [23], tyrannosauroid [24], or indeterminate coelurosaur [25]. Borogovia gracilicrus [26] is a troodontid that may be synonymous with Saurornithoides mongoliensis [27]. Hulsanpes perlei [28] is probably a deinonychosaur [27] but is not distinctive enough to identify. Tochisaurus nemegtensis [29] is a troodontid metatarsus that does not overlap significantly with the material of Borogovia gracilicrus or Zanabazar junior [30]. It is possible that all three of these troodontid taxa are synonymous, a possibility considered by both Osmólska [26] and Norell et al. [30].The hadrosaur Barsboldia sicinskii is known from a partial pelvis and tail. It has been interpreted as a valid taxon [31], a lambeosaurine [32], or as a nomen dubium [33]. Four protoceratopsians have been named from the Baruungoyot deposits in the Nemegt Basin. Bagaceratops and 'Platyceratops' are known from Hermiin Tsav, and 'Breviceratops' and 'Lamaceratops' are from Khulsan. Makovicky and Norell [34] suggest that ontogeny can explain all the variation between these four taxa, and synonymize the other three with Bagaceratops. Determining the true affinities of these taxa is beyond the scope of this study, so they are not included in the species-level analysis, to avoid numerous single-occurrence taxa. These taxa were, however, included in the superfamily-level data. For these analyses, Ba. ostromi was included as a tyrannosauroid. The status of Opisthocoelicaudia skarzynskii is questionable, as indicated by Currie et al. [35]. It is likely that most, if not all, material assigned to Opisthocoelicaudia will eventually be subsumed into N. mongoliensis. Accordingly, all sauropod material from the Nemegt Formation, most of which cannot be identified to species, is treated as N. mongoliensis. Indeterminate material that was identifiable to superfamily but not species was also included in the data set. Where only one species of that superfamily is considered valid (i.e. dromaeosaurids and therizinosaurs), indeterminate material was lumped with that species. Where more than one species is considered valid (i.e. ankylosaurs and oviraptorids), indeterminate material was included together as an indeterminate operational taxonomic unit (i.e. Ankylosauria indet., Oviraptoridae indet.).

Statistical methods
Data were not filtered or modified in any way before analysis. Statistical methods used to analyze community composition and taxon distribution in Funston et al. [1] are described therein, in the main text and the Supplementary information.