Accurate estimates of size-at-age and recruitment variability, as well as fishery specific catch-at-age and gear selectivity are critical to the integrated stock assessments for toothfish in the Heard and McDonald Islands toothfish fishery. Otolith analysis represents a powerful method for improving these estimates.
Currently, a growth model based on fish aged from the trawl fishery between 1997 and 2003 is used to predict catch-at-age for trawl and longline catches and year class strength from trawl surveys. This is done by using the growth model to partition numbers at length into age classes. However if there is variation in this relationship between years or between fisheries then the abundances at age may be falsely estimated resulting in poor estimation of stock status. This is an important potential source of bias in current models and should be addressed by developing age-length keys.
Unbiased age-length keys will require analysing sufficient otoliths, collected from a representative sample of size classes captured by the fishery and survey, such that the age composition of the catch, age-based selectivity of fishing gears and the age structure of the stock can be better estimated.
The otoliths aged in 1997-2003 were sampled primarily to develop a growth model, and are not suitable for evaluating season and gear-specific age-length keys. The majority of age-length estimates currently available result from analysis of otoliths collected before the longline fishery (which catches larger fish than the trawl fishery) commenced, so very few otoliths from larger fish or from longline grounds have been analysed. Furthermore, much of the ageing performed in the past occurred before the latest validation data for toothfish was available.
Hence there is a need to construct age-length keys across gear types and seasons, and conduct sensitivity tests to evaluate the impact of gear and interseasonal variability in assessments.