Bias in Virtual Population Analysis When the Unit Stock Assessed Consists of Sub-Stocks.

摘要

A simulation model is presented which allows an evaluation of possible bias in the results of standard Virtual Population Analysis in combination with Separable VPA and tuning methods when applied to an heterogeneous unit stock. The results indicate that, if a unit stock defined for assessment purposes is composed in reality by two or more sub-stocks which are exploited differentially, the estimated exploitation curve is dome-shaped, even when the exploitation pattern imposed upon both sub-stocks individually is flat-topped. The bias may cause significant underestimates of fishing effort due to an associated change in apparent catchability. In particular, the apparent recent trend in fishing mortality and spawning stock biomass may deviate considerably from the true trend.