The Effects of Otolith Sampling Methods on the Precision of Growth Curves

Abstract

AbstractEstimation of growth curves is a critical component of fish stock assessments. Two widely used otolith sampling methods, the age–length key (ALK) sampling method and the random otolith sampling (ROS) method, have problems that limit their utility for estimating growth curves. First, growth curves based on the ALK method are biased in that otolith samples obtained with the ALK are not selected via simple random sampling. Second, the precision and accuracy of growth curves based on the ROS method are often compromised because random sampling frequently results in a small number of older fish samples. In this study, bias in growth curves based on ALK data that were re‐sampled from a simulated data set describing king mackerelScomberomorus cavallawas corrected with a new reweighting technique. This technique reweighted the growth curves with the length‐frequency distribution of randomly resampled fork length data. The resulting growth curves were compared with growth curves obtained from ROS data to determine which method (reweighted ALK sampling or ROS) was more appropriate for selecting otolith samples for the estimation of growth curves. The results showed that the reweighted growth curves constructed from ALK samples were more precise and accurate than growth curves obtained from ROS data for all sample sizes examined because the reweighted ALK growth curves (1) had less variability in the estimated growth parameters, (2) decreased the probability of drawing wrong conclusions about a fish stock, and (3) provided greater accuracy and precision in predicting mean lengths at age. Results from this study and a previous study support the view that the ALK sampling method is more efficient than the ROS method when otolith samples are used for the determination of king mackerel age composition and growth curves.