Abstract
Despite being the fourth largest criminal market in the world, no forensic tools have been sufficiently developed to accurately determine the legal status of seized animals and their parts. Although legal trading is permissible for farmed or captive-bred animals, many animals are illegally removed from the wild and laundered by masquerading them as captive bred. Here we present high-resolution x-ray fluorescence (XRF) as a non-invasive and cost-effective tool for forensic classification. We tested the efficacy of this technique by using machine learning on a training set of zoo specimens and wild-caught individuals of short-beaked echidnas (Tachyglossus aculeatus), a small insectivorous monotreme in Australia. XRF outperformed stable isotope analysis (δ(13)C, δ(15)N), reducing overall classification error below 4%. XRF has the added advantage of providing samples every 200 μm on a single quill, enabling 100% classification accuracy by taking the consensus of votes per quill. This accurate and cost-effective forensic technique could provide a much needed in situ solution for combating the illegal laundering of wildlife, and conversely, assist with certification of legally bred animals.