Geochemical fingerprinting as a tool for repatriating poached dinosaur fossils in Mongolia: A case study for the Nemegt Locality, Gobi Desert

Over the past two decades, illegal poaching of vertebrate fossils from the well-known Cretaceous exposures in Mongolia's Gobi Desert has reached epidemic proportions. Following recent efforts led by the Mongolian government to stem the flow of illegal fossils on the black-market, numerous important specimens have now been seized from private collectors. A tandem objective has been to identify and document poached quarries in the field, which is now allowing scientists to evaluate which areas and stratigraphic intervals are more intensively targeted and impoverished by poachers. Although illegal specimens are now being returned and becoming available for research, the question of provenance still remains of both scientific and political concern. To address this issue, we investigated the geochemistry of the fossil-rich Baruungoyot and Nemegt formations in the Nemegt Basin, including both sediments and associated fossils, using portable X-ray fluorescence (pXRF). Field data were collected using a stratigraphic criterion and are representative of several localities across the Nemegt Basin, although the most complete datasets come from the Nemegt and nearby Khulsan localities. Secondarily, pXRF data were acquired from specimens housed in Ulaanbaatar, including type, recently described, and poached specimens. The final dataset includes approximately 700 pXRF readings. We used a combination of multivariate analysis and data mining techniques to match known with unknown sample provenances, and to determine the most informative elements from the geochemical array available for doing so. Results indicate that the chemical compositions of the Baruungoyot and Nemegt formations largely overlap with calcium being the most informative element for determining provenance from either bones or sediments. However, data arranged by geographic and stratigraphic occurrence (thus reflecting different facies and palaeoecological settings) resulted in more discriminant clusters. The newly developed methodology represents the first systematic attempts to provide an innovative and rapidly-implementable tool to support legal and scientific assumptions.