SUMMARY

This study aims at examining the potential bias in the VPA estimates by simulating a population, catch and CPUEs over 39 years from 1970 to 2008. In the operational model, two levels of noise were added to catch and CPUEs (50 and 100%), three fishing patterns were explored (F=FMSY, F increases linearly from 0.5 FMSY to 2.5 FMSY, F increases linearly from 0.5 FMSY to 2.5 FMSY over the first 34 years and decreases down to FMSY over the last 5 years). We also tested different hypothesis about the impact of the terminal F-ratios, and the natural mortality used in the current VPA as well as the impact of randomly generated CPUE for long-line and trap CPUE. Results show that the VPA is robust to estimate the SSB whatever the noise level in the catch and the CPUEs. Fs are generally overestimated especially for the last ages and the last year of the run. Randomly generated CPUEs have a strong impact on the estimations for the F at age 8-10+ which is consistent with the fact that the random CPUEs used mainly concern these ages. When fishing mortality decreases during the last years, the SSB estimated by the VPA does not show any sign of recovery while it is the case in the operational model. These results indicate that one of the main bias that could lead to false estimates of SSB and Fs could be the age estimation from the von Bertalanffy curve