Sampling channel catfish ponds for pre-harvest off-flavor detection

Earthy-musty off-flavor in channel catfish can adversely affect the catfish industry if strongly off-flavored fish reach the market. Pre-harvest sensory evaluation of fish is commonly performed by processing plants for detecting unacceptable levels of off-flavor. In West Alabama, three to 12 fish are sampled over a period of up to 3 weeks before harvest. Despite this effort, off-flavored catfish regularly reach the market. A study was conducted in summer and fall on ten commercial catfish farms to determine the fraction of off-flavored fish in ponds. Flavor was assessed by sensory evaluation. A larger occurrence of off-flavor was found in summer than in fall. Fraction of off-flavored fish varied from 0 to 54% depending on the pond and the acceptance criterion used. Between-pond variance of the fraction of unacceptable fish was greater than within-pond variance. Therefore, Bayesian sampling should be used instead of conventional sampling based on the binomial function. Different sampling plans are proposed depending on levels chosen for producer and consumer risks. It was recommended to consider a sample size of at least 30 fishes. (C) 2002 Elsevier Science B.V. All rights reserved.