Improving time series of abundance indices applying new knowledge

Recent information on gear selectivity and fish scattering cross section are used to comet time series of abundance estimates of cod and haddock from surveys in the Barents Sea. Both swept area estimates and acoustic estimates from the period 1981-1993 an corrected for fish size dependent capture efficiency, and the acoustic estimates are also corrected for a change in the scattering cross section-fish length relationship. Ideally, such work ought to be carried out station by station, correcting the length and species distribution from every trawl haul. Because of difficulties in the availability of raw data for the earlier part of the time series, a short cut procedure was established which yields new estimates by using the old total numbers at age as input. The results are compared with those obtained with the correct and comprehensive procedure for some years and the differences were negligible. The general effect of the corrections is an increase in the estimates of the younger age groups, ages 1 and 2 years, in relation to older fish. The corrections did not improve the fit between numbers at age as estimated from virtual population analyses (VPA) and the numbers at age obtained from surveys. Hence, the main advantage seems to be the reduction of a serious bias in the annual length (age) distributions, bringing the survey estimates of fish aged 1 and 2 years more in line with the expectations from population models. Comparison of the various statistics in the linear regression analyses between VPA-numbers and survey numbers may indicate that the natural mortality rate applied by the Arctic Fisheries Working Group for haddock is too low. (C) 1997 Elsevier Science B.V.