Constructing abundance indices from scientific surveys of different designs for the Torres Strait ornate rock lobster (Panulirus ornatus) fishery, Australia

Fishery-independent surveys of the ornate rock lobster (Panulirus ornatus) population in Torres Strait were carried out annually from 1989 to 2002 with variation in design and implementation due to logistic and funding constraints. Fixed and random station surveys were modeled separately, and their results were contrasted. As all the survey data contain many zero records, a gamma-based generalized linear model was used for non-zero records and a Bernoulli based model for the probability of encountering a lobster. Abundance indices for fished and recruiting year-classes were then constructed by combining the results from both models. To select an appropriate error model, four alternatives-log-normal, log-gamma, Poisson, and negative binomial-were explored. The results show that a log-gamma model best estimated the non-zero encounter rates. Recruiting (age 1+) lobsters exhibit a more variable distribution in space; however, for fished (age 2+) lobsters there is greater temporal variation in the probability of encountering lobsters. The large-scale pattern of lobster distribution among sampling strata remained unchanged over the survey period. In contrast to the results from the annual fixed station surveys (1989-2002), the models for the stratified random surveys (1989 and 2002) showed that the small-scale patterns in lobster distribution over depth did change between years. This may undermine the suitability of fixed station surveys for the construction of relative abundance indices of the Torres Strait lobster population.