Incompleteness and statistical uncertainty in competition/stocking experiments

In competition experiments, decisions are made not only about experimental conditions such as initial population densities, of course, but also about population size structure, for instance. Here we use an individual-based simulation model to study the effect of size-grading of mussels. With low individual variability, predicted yield was lower and less variable, there was no density-dependent mortality, and optimal stocking density for aquaculture was lower than with high individual variability, whereby self-thinning occurred and yield was quite variable. Thus, individual variability was a critical factor for estimating survival effects of overstocking, at the expense of precision of growth estimates. Therefore, competition experiments are inherently incomplete. We argue that in practice, incompleteness cannot be overcome by using genetic information as a covariate because evidence from the literature shows that the effect of genetic makeup in competition situations is frequency-dependent. Apparently, the only approach presently available to obtain unbiased estimates is to use a size structure similar to that of the population under study. This contrasts with a literature review of bivalve stocking experiments published in Aquaculture through the last 30 years which clearly shows that the issue of size structure of test populations has been largely overlooked. The same principles hold for competition studies in natural settings.