TEST OF A FOOD-CONSUMPTION MODEL FOR THE ATLANTIC COD

A steady-state model of food consumption, based upon rates of gastric turnover in Atlantic cod Gadus morhua L., has been developed and the reliability of the consumption model tested by feeding cod according to different feeding regimes using three types of food. At the end of each series, which lasted for at least 23 days, stomach contents of the cod were recovered and identified. The daily food intake of each cod was then estimated from the stomach contents using the consumption model. In most situations no significant statistical differences were detected between the estimated food intake and the observed steady-state feeding rate, which was set as the arithmetic mean daily consumption in the last 14 days of each series. The statistical power of the estimates to detect a difference of +/-20% in the observed consumption was about 50%, and the model seems suitable for use in the analysis of predation mortality and production of predator populations. Since information was available about the food consumption by individual fish it was possible to analyse the consequences of different sampling procedures on the accuracy of consumption estimates. Pooling of stomach contents data, a practice analogous to that performed in field surveys, resulted in moderately biased estimates of consumption. Although these pooled estimates only rarely differed significantly from the observed values for food consumption, there was a major loss of predictive power. The consequences of this are discussed in relation to the planning of sampling programmes in field studies of fish feeding and consumption.