Sources of errors in and limitations of Virtual Population Analysis (Cohort Analysis)

Abstract

Various sources of errors in Virtual Population Analysis (VPA) are discussed. Errors in the assumed value of natural mortality create errors in estimated fishing mortalities of about the same size but with opposite sign if the total mortality varies moderately from year to year and the true natural mortality is constant. Stock size will be over- or under-estimated when natural mortility is over- or under-estimated, but estimated relative changes in stock size from year to year will be approximately correct. Fluctuations in total mortality caused by random fluctuations in natural mortality from year to year will disappear almost completely in VPA if a constant natural mortality is assumed. A trend in the true natural mortality with time or age will be converted to a trend in the VPA estimates of fishing mortalities. Relative strength of weak and strong year classes will be wrongly estimated by VPA if the true natural mortality varies with year class strength, and this may influence various regression lines commonly used for predicting year class strength from young fish survey indices. Errors in VPA caused by uneven distribution of natural or fishing mortality throughout the year are shown to be generally small and negligible. Effects of stock migration on VPA are discussed. For a year class which continuously migrates from an area A to an area B at a constant emigration rate, equations which give the number of fish of the year class at any time in each of the two areas are developed. These equations may be utilized in a technique similar to VPA for retrospective analysis, and the application of this technique is illustrated. If emigration is included in the natural mortality the traditional VPA may successfully be used in area A. There is, however, no simple way to adjust for immigration in area B, and the traditional VPA may give a completely wrong picture of the situation in this area.