In this paper, we address a number of aspects of the model inputs and parameters of the Antarctic toothfish stock assessment for the Ross Sea fishery. In particular we review catch history, length-weight relationships, catch-at-length and catch-at age. In addition, we report some preliminary model runs that investigate the sensitivity of the 2006 stock assessment to changes in these model inputs and parameters.
Tree-regression methods were used to investigate the areal structure of the length distribution of Antarctic toothfish. While tree-regressions suggested strong evidence of a high degree of small-scale areal complexity, we were unable to provide a stratification that resulted in improved or consistent patterns in length frequencies over the duration of the fishery. Including terms for nation, vessel, or vessel type did not provide any additional information as these tended to be highly correlated with the location variables.
The catch and CPUE indices for the Ross Sea Antarctic toothfish fishery were updated, as are some modelling parameters, and methods for calculating age- and length-frequencies. Most of these changes did not have a significant impact on the assessment results.
We also provide an update of the numbers of fish scanned at length by New Zealand vessels, and the numbers of tagged fish recaptured. Inclusion of observations of the 2006 fish recaptured in 2007 had the greatest impact on the assessment model results. Dunn et al. (2007) noted that the locations of the 2007 recaptures were highly aggregated and were mostly located on four key locations in the Ross Sea, and most had moved only short distances. This confirms the concern that the key uncertainty underlying the current model is the impact of movements and spatial structure in the Antarctic toothfish population. In particular, the level and nature of the bias from non-homogeneous mixing assumptions of tagged fish.
Номер документа:
WG-SAM-07/6
Пункт(ы) повестки дня
Резюме