Abstract
Commercial aerial spotting of surface schools of juvenile southern bluefin tuna (SBT), Thunnus maccoyii, is conducted as part of fishing operations in the Great Australian Bight in summer. This provides the opportunity to efficiently collect large amounts of data on sightings of SBT. The data can potentially be used to construct a time-series index of relative abundance by standardising the data for issues such as weather, spotter ability and ocean conditions. Unlike a statistically designed survey, the commercial spotting is governed by business considerations and fishing operations. The SBT dataset is therefore highly unbalanced with regard to spotters operating in each season. This complicates the standardisation of the data, particularly with regard to interactions between covariates. We show how a generalized additive model with random effects can simplify both the fitting of the model and the construction of an index, while also avoiding the need to leave out strata or interaction terms that are important. The approach is applicable to standardisation of more traditional catch and effort data.