Identification of avian feather product origins: A case study on the tail feather products of Syrmaticus reevesii (Reeves’s Pheasant)

Abstract Since the Yuan Dynasty (1271–1368), the tail feathers of Syrmaticus reevesii (Reeves’s Pheasant) have been used to make the Lingzi, a type of headgear worn in traditional Chinese opera Xiqu. Sourcing these feathers ethically is crucial for promoting the development of Xiqu and the management of wildlife. Through stable isotope analysis, previous studies determined that the Lingzi used by opera troupes were likely sourced from wild individuals. However, these studies did not address the principles of stable isotope selection, the applicability of predictor variables, and forensic reference standards. During the enforcement process, technical barriers and testing costs may reduce the feasibility of enforcement, and the lack of standardized discrimination methods makes it difficult to generalize the identification of feather products to all bird species. Therefore, in this study, we selected δ15N and δ13C as predictor variables to predict the source of S. reevesii tail feathers. We employed random forest classification and linear discriminant analysis to predict the origin of 44 feather samples collected from the troupes. When δ13C and δ15N, and δ13C alone were used as predictor variables, all Lingzi samples were predicted to be of wild origin. Compared to δ13C, δ15N had weaker indicative effect in determining the source of tail feathers and interfered stability in binary model. When the model used only δ13C as a predictor variable, it could still accurately determine the true origin of the tail feathers and had excellent predictive stability. In addition, the partial dependence plot of the random forest model showed that δ13C = –21.5‰ could be used as a discriminant standard for determining whether the tail feathers originated from wild or captive individuals. This refined approach improves source identification of S. reevesii feathers and supports law enforcement efforts.