contribute significantly to the estim<strong>at</strong>ed TAG. For cod in 4T -4Vn, short term projections are not influenced by this type <strong>of</strong> error (see Table 9.C.1). In fact, a 100% increase in partial recruitment <strong>at</strong> age 3 and a 40% increase <strong>at</strong> age 4 for the years <strong>of</strong> projection will change the estim<strong>at</strong>ed 1981 c<strong>at</strong>ch biomass by less than 2% and the 1981 stock size estim<strong>at</strong>e by less than 0.1%. This type <strong>of</strong> non-random error is thus negligeable in comparison to the rel<strong>at</strong>ive error due to the variance <strong>of</strong> c<strong>at</strong>ch biomass estim<strong>at</strong>es. We also analyzed the effect <strong>of</strong> some other sources <strong>of</strong> system<strong>at</strong>ic errors on projection estim<strong>at</strong>es (see Table 9.C). For instance, a 50% discard <strong>at</strong> sea for age 3, together with a 25% discard <strong>at</strong> sea for age 4 (unreported) does not change signific<strong>at</strong>ively the 1981 stock size estim<strong>at</strong>e and the 1981 c<strong>at</strong>ch biomass (TAG) estim<strong>at</strong>e. Similarly, a 80% discard <strong>at</strong> sea for age 3 (unreported) from 1970 through 1979, does not have a significant effect on projection estim<strong>at</strong>es. In a fourth case, we observe th<strong>at</strong> a 20% under-reporting (or mis-reporting) for all age groups in 1979 will change the 1981 stock size estim<strong>at</strong>e by less than 0.7% and the 1981 c<strong>at</strong>ch biomass estim<strong>at</strong>e by 2.5%. In conclusion, system<strong>at</strong>ic errors <strong>of</strong> th<strong>at</strong> n<strong>at</strong>ure become a minor source <strong>of</strong> concern when using initial stock size estim<strong>at</strong>es derived from Doubleday's method. Ironically, system<strong>at</strong>ic errors appear as the major source <strong>of</strong> concern (and source <strong>of</strong> discussions) when standard analytical methods are applied. Variance estim<strong>at</strong>es are quite important - may be as important as parameter estim<strong>at</strong>es themselves - since they are indic<strong>at</strong>ive <strong>of</strong> the corrective measures to be undertaken for improving the d<strong>at</strong>a base. Analytical assessments do not provide inform<strong>at</strong>ion on such corrective measures. As we know, it is rel<strong>at</strong>ively easy to manipul<strong>at</strong>e analytical solutions, within the range <strong>of</strong> biologically-acceptable parameter values, so as to reflect or to mimic any desired trend. In absence <strong>of</strong> an objective measure <strong>of</strong> the 'quality' <strong>of</strong> the estim<strong>at</strong>es obtained with analytical methods, analytical assessments may become a vehicle for subjective evalu<strong>at</strong>ions. From the management viewpoint, the assessment <strong>of</strong> the limit<strong>at</strong>ions <strong>of</strong> the d<strong>at</strong>a base and the identific<strong>at</strong>ion <strong>of</strong> corrective actions are as important as the provision <strong>of</strong> TAG estim<strong>at</strong>es per se. The preceding analysis is limited to the study <strong>of</strong> one case, i.e. cod in 4T-4Vn; our conclusion th<strong>at</strong> uncertainties <strong>of</strong> mean weight-<strong>at</strong>-age are a major source <strong>of</strong> uncertainties for TAG estim<strong>at</strong>es is valid for this stock only. For other groundfish stocks, the uncertainties associ<strong>at</strong>ed with research survey estim<strong>at</strong>es may become the major source <strong>of</strong> u ncerta inti es for c<strong>at</strong>ch projections. In fact, Table 4 indic<strong>at</strong>es th<strong>at</strong>, for a given precision <strong>of</strong> mean weight-<strong>at</strong>-age estim<strong>at</strong>es, the uncertainties associ<strong>at</strong>ed with initial stock size estim<strong>at</strong>e.s are a key factor for controlling the uncertainties <strong>of</strong> c<strong>at</strong>ch biomass estim<strong>at</strong>es in the years <strong>of</strong> projection. When an assessment is amenable to 109 Doubleday's method, an important benefit is th<strong>at</strong> c<strong>at</strong>ch projections become practically independent <strong>of</strong> system<strong>at</strong>ic errors <strong>of</strong> input parameters (misreporting, discards <strong>at</strong> sea, etc.). REFERENCES Beacham, T.D. 1980. 1980 Assessment <strong>of</strong> Cod in Division 4T and 4Vn (Jan-Apr). CAFSAC Res. Doc. 80/22 61 p Doubleday, W G. 1979 A Sensitivity Analysis <strong>of</strong> Selected C<strong>at</strong>ch Projections. ICES Demersal Fish Committee, C.M 1979/G 53. 7 p. + tables. Doubleday, W.G. 1981. A method <strong>of</strong> estim<strong>at</strong>ing the abundance <strong>of</strong> survivors <strong>of</strong> an exploited fish popul<strong>at</strong>ion using commercial fishing c<strong>at</strong>ch-<strong>at</strong>-age and research vessel abundance indices. In W.G. Doubleday and D. Rivard (Ed.). <strong>Bottom</strong> <strong>Trawl</strong> surveys/Releves au chalut de fond. Can. Spec. Publ. Fish. Aqu<strong>at</strong>. Sci ./Publ. spec can sci halieut. aqu<strong>at</strong>. 58 . (This vo 1 ume l . Gray, D.F. 1979. 1979 Assessment <strong>of</strong> Cod in Divisions 4T and 4Vn (Jan··Apr). CAFSAC Res. Doc. 79/21. 28 p. Lassen, H. 1980. On the quality <strong>of</strong> the TAG's recommended through ICES. ICES C M. 1980/G:39, Demersal Fish committee. 13 p. Lett, P.F. 1978. A multispecies simul<strong>at</strong>ion for the management <strong>of</strong> the southern Gulf <strong>of</strong> St. Lawrence cod stock. CAFSAC Res. Doc. 78/21, 29 p. + tables. Pope, J.G. 1972. An Investig<strong>at</strong>ion <strong>of</strong> the Accuracy <strong>of</strong> Virtual Popul<strong>at</strong>ion Analysis using Cohort Analysis. Int. Comm. Northwest Atl. Fish. Res. Bull. 9:65-74. Rivard, D. 1979. Response <strong>of</strong> Cohort Analysis to Input Parameters. CAFSAC Res. Doc. 79/45. 26 p. Rivard, D. 1980. APL Programs for Stock Assessment. Can. Tech. Rep. Fish. Aqu<strong>at</strong>. Sci. 953: 103 p. Seber, G.A.F. 1973. The Estim<strong>at</strong>ion <strong>of</strong> Animal Abundance and Rel<strong>at</strong>ed Parameters. Charles Griffin & Co. Ltd., London. 506 p.
Survey design Schemas d'echantillolU\age