A robust method for generating separate catch time-series for each of the hake species caught in the Namibian trawl fishery

Merluccius capensis and M. paradoxus are morphologically very similar and cannot be registered separately by species in the Namibian commercial hake catches. This prevents a biologically plausible single-species stock assessment from being carried out. Here, species-separated data from an observer programme and scientific surveys are used to model the spatio-temporal overlap of the two species, which are then used to predict the catch by species in each commercial trawl. The study presents a method that compensates for both the escapement and codend retention differences in the survey and commercial trawls. The accuracy with which species were identified was found to be considerably higher during scientific surveys compared with that obtained from observers, whereas the seasonal coverage of the observer data was much better than that of the surveys. The estimated spatiotemporal model parameters from each of these two data sources were similar, however, despite these differences. In support of previous findings, M. capensis had a shallower and more northerly distribution than M. paradoxus with depth and latitude together explaining 51% and 85% of the residuals in the models produced from observer and survey data respectively. Model outputs suggest that during the period 1998-2007, M. paradoxus has dominated the annual hake catches. Even though our method is unable to account for abrupt and unexpected changes in the species' geographical distribution, it does open the way for the establishment of a single-species hake assessment in Namibia.