Bayesian Estimation of Age and Length At Maturity

Understanding factors that influence age and length at maturity are vital for effective management of exploited fish populations. Traditional frequentist approaches at estimating maturity parameters fail with small sample sizes whereas Bayesian inference of hierarchical models is not always limited to this restriction. Our objective was to evaluate factors influencing maturity of the 1984-2007 yellow perch year classes in southern Lake Michigan using logistic regression models. Several poor year classes preclude frequentist analysis therefore parameters were estimated with Bayesian inference. We tested three hypotheses: 1) Yellow perch exhibit sexually dimorphic maturity, 2) Maturity rate is influenced by year class strength through compensatory mechanisms, 3) Periods of high exploitation influence age and length at maturity. Maturity rate significantly differed between sexes. Year class strength was not a significant factor describing the maturation rate of either sex. Periods of high exploitation significantly influenced age at maturity of both sexes and only length at maturity of females. Both sexes matured at an earlier age and females matured at a shorter length during periods of high exploitation. Our findings suggest commercial fishing harvested individuals with a late maturing phenotype before they could reproduce, thus resulting in an evolutionary response favoring early maturing individuals.