GLM-based standardization of the catch per unit effort series for South African west coast hake, focusing on adjustments for targeting other species

Abstract

Catch per unit effort (cpue) data for the South African west coast Cape hakes Merluccius capensis and M. paradoxus were standardized by applying the Generalized Linear Modelling (GLM) technique. The resulting standardized series indicated that resource abundance declined at a rate of 0.4% per annum over the period 1978-1994. Further investigation revealed that the model applied did not adequately adjust for effort being directed away from hake towards other species. This was because, for low bycatch cpue, there was a positive rather than the expected negative correlation with hake catch rates. A method to correct for this inadequate adjustment, taking account of an assumed underlying reason of positively correlated fluctuations in catchability of hake and bycatch species, was developed and tested by simulation. This resulted in a more optimistic view of resource status, indicating a 0.6% per year increase in abundance over the period considered. This result remained contrary to general perceptions in the industry that there was a substantial improvement in resource abundance over the 17-year period considered, as suggested by a 4% increase in abundance per year in nominal cpue. The reasons that the GLM indicated a lesser increase in resource abundance were that the effective average power of the fleet increased, the fleet moved over time to deeper water, where catch rates were higher, and the increase in fish density in deeper water was more than offset by a simultaneous decrease in density in shallower waters. The inadequate bycatch adjustment arose from the analyses inappropriately including bycatch species that co-occur with hake, and hence are taken incidentally in hake-directed fishing, instead of only species whose capture requires a different targeting of effort.