DEVELOPMENT OF TILIQUA SPECIES PROVENANCE MODELS FOR USE IN COMBATING THE ILLEGAL WILDLIFE TRADE

Kate Brandis ^1,2* Kyle Zawada ³ Phoebe Meagher ⁴ Daniel Ramp ³ Roxane Francis ²

• ¹ University of New South Wales, Kensington, Australia
• ² Centre for Ecosystem Science, Faculty of Science, University of New South Wales, Kensington, New South Wales, Australia
• ³ Centre for Compassionate Conservation, UTS, Sydney, Australia
• ⁴ Taronga Conservation Society Australia, Mosman, New South Wales, Australia

This study was undertaken to explore the applicability of portable X-ray fluorescence (pXRF) technology in combating the illegal wildlife trade, specifically focusing on Australia's Tiliqua species. The research aimed to develop models that could effectively identify species, distinguish between captive-bred and wild individuals, and predict geographic provenance. The hypothesis was that pXRF could achieve high accuracy in species identification and classifications, thereby providing a useful tool for wildlife enforcement efforts.The study was conducted using pXRF technology to analyse a range of Tiliqua specimens, including shingleback (T. rugosa) and common blue-tongue (T. scinoides) lizards. Specimens were collected and analysed in various states-live, dead, and as animal parts. Species specific XGBoost models were developed and tested for accuracy in identifying species and distinguishing between captive and wild individuals. Geographic provenance models were also created, utilizing predictor variables such as soil nutrient groups and hydrological basins to evaluate model performance.The study found that species specific models could identify shingleback and common bluetongue lizards with an accuracy of 70%. Additionally, the models distinguished captive-bred from wild individuals with up to 81% accuracy for blue-tongue lizards and 83% for shinglebacks. Geographic provenance models demonstrated variable performance, achieving up to 83% accuracy but indicating the need for further refinement and more intensive sampling to improve model resolution.The results imply that pXRF technology has significant potential as a tool for wildlife enforcement, providing valuable information for species identification and the classification of individuals as captive or wild. This finding is consistent with prior research highlighting the utility of elemental profiling in wildlife conservation. The study also identifies a critical knowledge gap regarding the impact of captivity duration on elemental profiles, suggesting that future research should focus on refining geographic models and understanding the dynamics of elemental changes over time in captive versus wild specimens. Overall, the integration of pXRF into wildlife enforcement protocols represents a cost-effective and rapid approach to combatting illegal wildlife trade.