On the relation between ‘true’ numbers of Northeast Arctic cod and VPA- or survey-based abundance estimates

Abstract

When assessing Northeast Arctic cod it is assumed that the estimates of numbers at age generated by a converged VPA type analysis are equal, on average, to the true numbers at age. Based on fishery independent survey data, we show that it is likely that the converged VPA estimates are biased, especially so for small cohorts. The probable cause of this bias is either inaccurate catch data including erroneous age readings and/or using the wrong functional form in the VPA tuning procedure. We provide evidence that it is appropriate to assume that the expected survey-based estimates are proportional to actual abundance. Therefore, we suggest that proportionality should be assumed for all ages when calculating survey catchability (i.e. in the VPA tuning procedure) or using the survey to estimate abundance directly. Since the converged VPA estimates of small cohorts seem to have a relatively high bias compared with those for large cohorts, only the converged VPA estimates for large cohorts should be used in the tuning process. This will not eliminate the bias in the VPA estimates caused by the tuning procedure, but it should reduce it.