Report of the Study Group on the Analysis of Feeding Data

The analysis of feeding habits data and factors influencing their variability is central not only to current multispecies models, but to identifying and testing feeding mechanisms for incorporation in future modeling efforts. At the 1990 meeting of the Multi species Assessment Vorking Group (Anon. 1991), a number of major issues were identified that involved detailed statistical analyses of feeding data. Examination of stomach sampling data for 1981, 1985, 1986, and 1987 in the North Sea revealed, in some cases, large differences in quarterly mean stomach content weights at age. Given the assumption of constant quarterly consumption across years, these partial data suggest that the assumption may not have been completely justified. Confounding this analysis, however, is the fact that precision estimates (CV) of stomach content weight have not been provided for the various years of stomach sampling data [although some estimates of precision have been attempted (Pope and Hunton 1985)). Mean stomach content weights at age are a function not only of the length-stratified sampling program for stomach content, but the application of age-length keys as well. More direct comparisons of feeding data (without the confounding effects of ALKs) would involve