Bias Associated with Using the Eggers Model for Estimating Fish Daily Ration

Abstract

Estimates of yellow perch (Perca flavescens) daily ration derived from Eggers' model were compared with those based on the Elliott–Persson model and a 3-h within-day sampling interval. Correcting for the bias in Eggers estimates associated with food weight differences at the start and finish of estimation periods significantly reduced discrepancies between the two when Eggers estimates were also based on a 3-h sampling interval. As sampling interval was increased to 6 and 12 h for Eggers estimates, a positive bias related to variability about [Formula: see text] emerged and inflated exponentially with increasing ration levels. A third bias (typically positive) may be introduced when [Formula: see text] in Eggers' model is computed without regard to the skewness that often exits in distribution of food weights among fishes within a single collection. Use of the median or a normalizing transformation will avert this. While each bias alone can have an important effect on daily ration estimates, the possibility of accentuated bias from multiple sources also exists. Two of the potential biases are easily managed, but caution is advised when Eggers' approach is used to estimate daily rations based on expanded sampling intervals.