Figure 3
From: Novel detection of provenance in the illegal wildlife trade using elemental data
Predicted classification of samples taken from 12 captive quills (C1-C12) and 11 wild quills (W1-W11) using machine learning on 24 elements quantified with high-resolution x-ray florescence. Shading represents the proportion of classification trees supporting the predicted status of wild-caught every 200 microns (voting confidence), where dark blue confidence values indicate strong support for captive-bred and dark orange confidence values indicate strong support for wild-caught. Position 1 represents the tip of the quill (i.e. the oldest sample), with the number of positions determined by quill length.
