Model-based estimates of Length composition time series for Indian Ocean yellowfin tuna fisheries

This report presents model-based estimates of length compositions using a Bayesian Dirichlet- Multinomial (DM) model. The DM model was applied to the commercial catch samples to derive time series of length distributions for the Indian Ocean yellowfin tuna for the Longline and handline fisheries. The model incorporated spatial variability in the population length distributions at 5x5 longitude latitude (the level of reporting) while accounting for sampling errors amongst fleet and seasonal strata. The results suggested the DM model can be a potentially useful method to provide standardized length composition input for the IOTC stock assessments. However further work is required to better disaggregate and quantify various sources of uncertainty in the commercial length samples.