Analytical approaches for addressing the variation in back-calculated age-length relationships for fish

Estimating an age-length relationship is a routine aspect of many fisheries studies and is simplified by the use of commercially available computer programs. These computer programs may be misleading since a result can be produced irrespective of the quality or the extent of the data, and there is some concern that back-calculated age-length relationships are sensitive to the sample size and composition. We investigated this issue by comparing estimates of mean back-calculated lengths at age and growth rates derived from subsets of a large sample of wild channel catfish Ictalurus punctatus (N=788) collected in 2001 and 2002 from 9 rivers in Mississippi, United States. Estimates of growth rate varied among subsets consisting of individual year class (2-6) of channel catfish separated from the overall sample. For nine subsets, comprising randomly-selected and increasing proportions of the overall sample (20%-100% at 10% increments of the overall sample), growth was similar. However, growth differed for a subset representing a random 10% of the overall sample. Lengths at age and growth rates derived from each of the 2001 and 2002 components of the sample both differed. All results were significant at P < 0.05.