Zero-inflated counting data of juvenile blue crabs (Callinectes sapidus Rathbun, 1896) collected in two estuaries of Rio Grande do Sul are the object of the present study. These data led to the formulation of hierarchical models that quantify the effect of categorical covariates months and location on the probability of occurrence and density of these populations, taking imperfect detection into account. Non-hierarchical models were also developed for comparison. A Bayesian approach was adopted for estimating model parameters using the Markov Chain Monte Carlo (MCMC) simulation technique. Model comparison was performed according to the Deviance Information Criterion (DIC). Hierarchical models resulted in better fits than conventional ones, mitigated the problem of excess zeros and allowed for simultaneous analysis of the probability of occurrence and the density of juvenile blue crab. The probability of occurrence of Class 2 juveniles in the Lagoa dos Patos estuary increases at larger distance from the mouth while in Tramandaí the occurrence is more likely for intermediate distances. In both estuaries the occurrence is more likely in the summer and winter months. Regarding density of Class 2 juveniles, there is marked variation among the months of the year; and is overall higher in the Tramandaí estuary.
Blue crab; abundance; Bayesian inference; hierarchical models; stochastic simulation
