Aspects of both growth and selectivity affect growth parameter estimation bias

Selectivity curves for sampling size-at-age data are somewhat infamous for biasing growth curve parameter estimates. However, aspects of growth itself can also feed into this phenomenon and further increase bias into parameter estimates. We performed a simulation study by producing size-at-age data under a variety of growth curve parameter values and length-based selectivity curve parameter values for two selectivity functions by altering parameters both independently and in conjunction. In tests where parameters were altered one at a time the greatest amount of bias in growth curve parameter estimates came from sampling the data using dome-shaped selectivity. In addition, certain aspects of growth (namely variance in size-at-age) enhanced the biasing effect of both dome-shaped and asymptotic selectivity. When all parameters were altered together the greatest bias occurred in instances where length distributions overlapped to a greater degree and when asymptotic selectivity curves were more mildly sloped. We conclude that bias in growth curve parameter estimates is a result of the statistical weighting of both the probability of being a certain size at a given age and the probability of being captured at that size.