Discriminant analysis of Channa argus , Channa maculata , and their hybrids based on shape and otolith morphology

To compare the external morphology and otolith morphology of four groups, specifically Channa argus , Channa maculata , the hybrid C. argus ♀ × C. maculata ♂, and the reciprocal hybrid C. maculata ♀ × C. argus ♂, we applied three methods, namely traditional morphometrics, truss network morphometrics, and elliptic Fourier analysis of otoliths, with a total of 200 specimens (47–53 per group). From these, a total of 10 traditional measurement traits of fish bodies, 24 truss measurement traits of fish bodies, and 34 elliptic Fourier descriptors (EFDs) were extracted. Differences among species were examined by a one-way analysis of variance (ANOVA), principal component analysis (PCA), and stepwise discriminant analysis (SDA). The one-way ANOVA revealed that 22 of the 24 truss variables were significantly different (P < 0.01). PCA extracted four principal components from the truss data, which explained 68.127% of the cumulative variance, and 10 components from the Fourier coefficients, which explained 76.908% of the variance. Scatter plots of the first three principal components of the otolith EFDs revealed clear separation along PC1 among all four taxa, with slight overlap along PC2, thus indicating that the otolith shape carries a strong group-specific morphological signal, whereas the hybrid otoliths retained diagnostic parental features. The cross-validated SDA correct classification rates were 43.9% for traditional measurements, 91.6% for truss variables, and 81.4% for otolith EFDs. Combining the truss and Fourier data increased the overall success rate to 97.0%. The leave‑one‑out cross‑validated SDA yielded overall correct classification rates of 42.2% for traditional morphometric traits, 87.2% for truss network traits, and 79.4% for otolith elliptic Fourier descriptors. The combined model integrating body truss and otolith shape parameters achieved the highest overall accuracy, 92.9%. This paper demonstrates that a research method based on truss traits and Fourier coefficients can effectively distinguish among four snakehead taxa, thereby further improving the accuracy of discrimination. It also provides reference data for relevant morphological studies within the genus Channa.