Comparison of Two Biometric Methods for Nonlethal Sex Determination of Alligator Gars

Abstract

AbstractMcDonald et al. () proposed a morphometry‐based method for nonlethal sex prediction of the Alligator Gar Atractosteus spatula. Their method correctly predicted the sex of 93% of the males and 72% of the females tested, but failed to predict the sex of 21% of the fish. The authors noted that the method was primarily developed using immature fish from a single Texas population and recommended further validation and refinement using fish of diverse sizes from a broad range of systems. We tested the McDonald et al. () method using 149 fish (standard length, 707–1,920 mm) collected from seven Texas populations. Sex prediction based on the McDonald et al. () method improved for both sexes (98% of males and 86% of females); however, sex still could not be predicted for 29% of the fish. As a result, we explored the use of a cross‐validated, discriminant function analysis (DFA) using the morphometric ratios developed by McDonald et al. () to improve our ability to predict the sex of sampled fish. The DFA facilitated probability‐based predictions of all fish sampled and had an overall accuracy of 84% for males and 89% for females. Logistic regression based on the success of the DFA suggested that fish >1,100 mm SL are easier to identify using the morphometric ratios (>90% accuracy), likely because changes in body shape are associated with sexual maturity. In addition to predicting the sex of all of the fish sampled, the DFA approach facilitated calculation of the probability of correct discrimination for each fish. Due to these advantages, along with its comparable accuracy, we recommend the use of the DFA approach over the method proposed by McDonald et al. () for predicting the sex of Alligator Gars.