We present a method to evaluate potential biases and uncertainty in the tagging assumptions of the stock assessment for Antarctic toothfish in the Ross Sea region using spatially explicit operating models. The method allows investigation of potential biases and uncertainty in the assumptions of spatial distribution and fish mixing used in the standard stock assessments for the Ross Sea region (and potentially other CCAMLR areas).

We use the generalised Bayesian population dynamics model, the Spatial Population Model (SPM), to develop spatially explicit movement models of the Antarctic toothfish in the Ross Sea region as operating models in simulation experiments. Simulated observations from these models were then used in a single area stock assessment model derived from the stock assessment model of Antarctic toothfish in the Ross Sea region.

Results from preliminary case studies suggest that the standard single area stock assessment model for the Ross Sea was relatedly unbiased when we simulated from an operating model derived from the best-fitting coarse-scale model that restricted fish to areas inside the historical footprint of the fishery. However, the results when using a similar model that allowed for fish to be present in areas outside the area historically accessed by the fishery suggested the standard stock assessment may be biased low.

While we note that these results are preliminary and further analyses should be carried out, we consider that simulation experiments using spatially explicit models can provide a useful tool to evaluate potential bias and uncertainty in our understanding of the stock assessment in the CCAMLR region. We recommend the further development of this method at future meetings.