Abstract: To effectively distinguish natural and cultured populations of Oplegnathus fasciatus, we used both traditional statistical analysis and neural network methods to compare six shape indices and twenty-one truss indices from otoliths of 174 randomly selected specimens (100 from a natural population and 74 from a cultured population). The results showed that among the six shape indices, ellipticity, roundness, and aspect ratio exhibited significant differences between the two populations. Twelve out of the twenty-one truss indices displayed significant differences. Results of discriminant analysis indicated that traditional statistical analysis and neural network methods achieved correct discrimination rates of 57.5% and 81.4% for shape indices, while for truss indices were 69.5% and 85.4%, respectively. These findings indicated that neural network technique is more effectively than traditional statistical method to distinguish natural and cultured populations of O. fasciatus.

Key words: Oplegnathus fasciatus, population discrimination, otolith, stepwise discriminant analysis, neural network