Evaluation of procedures to reduce bias in fish growth parameter estimates resulting from size-selective sampling

Abstract

Size-selective fish sampling can strongly bias von Bertalanffy growth parameter estimates. Bias-correction methods have been developed, but they often require previous knowledge of selectivity pattern, capture–recapture data or intensive age-growth sampling over multiple consecutive years. When corrective measures are not feasible, investigators have attempted a number of biologically based procedures to reduce this bias. We evaluated several existing biologically based procedures that could potentially remove bias from growth parameter estimates. We built an age and length structured population model and tested the utility of four procedures to remove bias: 1) fixing t 0 at zero, 2) deleting data associated with ages that are not fully vulnerable to sampling, 3) deleting less than fully vulnerable ages and fixing t 0 at zero, and 4) fixing L ∞ at the maximum value observed in the data. We considered sampling gears that had no size selectivity, asymptotic selectivity, and dome-shaped selectivity patterns for all procedures. Results suggested that none of these procedures would eliminate bias in growth parameters across all three gear selectivity patterns. Investigators should attempt to use methods to correct growth parameters for size selectivity of sampling gears (e.g., mark recapture methods). If such corrections are not feasible, prior information about the size selectivity pattern of sampling gears is necessary for informed selection of biologically based von Bertalanffy fitting procedures.