A Bayesian Multilevel Ordinal Regression Model for Fish Maturity Data: Difference in Maturity Ogives of Skipjack Tuna (Katsuwonus pelamis) Between Schools in the Western and Central Pacific Ocean

The maturity ogive is vital to defining the fraction of a population capable of reproduction. In this study, we proposed a novel approach, a Bayesian multilevel ordinal regression (i.e., Bayesian continuation ratio model), to model the maturity ogive. The model assumes that the observed maturity stage originates from the categorization of latent continuous variables. We demonstrated this approach by testing whether there are differences in the maturity ogive of skipjack tuna (Katsuonus pelamis) in the western and central Pacific Ocean between two school types, i.e., free-swimming and floating-object-associated schools. The model results show that K. pelamis, given the same fork length, are more likely to have a higher maturity stage in a free-swimming school than those associated with floating objects. The gonadosomatic index revealed the same conclusion. Our results indicate that fish aggregation devices (FADs) could negatively affect the maturity of K. pelamis and consequently reduce the population reproductive potential. This study provides (1) an alternative approach to analyze fisheries ordinal data; (2) important quantitative evidence to evaluate the existing ecological hypotheses; and (3) implications for tuna fisheries management.