The surplus production model and ad hoc tuned VPA assessment methods currently used to provide the basis for scientific TAC recommendations for the Cape hake resource off South Africa provide rather different appraisals of the current status and productivity of this resource. The production model approach is based on the Butterworth-Andrew observation error estimator, and takes catch per unit effort (CPUE), as well as biomass survey data into account. The ad hoc tuned VPA is based on the Laurec-Shepherd tuning algorithm and utilizes catch-at-age and effort information. Applications of an age-structured model, which takes both CPUE and catch-at-age data into account, provides similar results to the production model if more weight is given to the CPUE data than the catch-at-age data and similar results to the ad hoc tuned VPA if more weight is given to the catch-at-age data rather than the CPUE data. This led Punt (1993) to conclude that the discrepancies between the various sets of results obtained from surplus production model and ad hoc tuned VPA methods are a consequence of a conflict between the catch-at-age data and the CPUE data and that they are not primarily a result of differences in the two assessment methods. However, the above two approaches are based on certain assumptions regarding recruitment, natural mortality and fishing selectivity. An attempt was made to obtain estimates of fishing selectivity-at-age from an age-structured production model. It is commonly assumed that selectivity-at-age has a slope of zero at older age classes. The estimates obtained all suggest that selectivity-at-age for older age classes (> 2 to 3 years) decreases with age. The results obtained in this study also indicate that the conflict between the observed trends in the catch-at-age data and the CPUE data can be basically resolved by assuming that for older age classes selectivity-at-age decreases.

Bibliography: pages 41-47