On the use of tester stocks to predict the competitive ability of genotypes.

It has been recently claimed that the outcome of competition between two phenotypically indistinguishable strains cannot be predicted from comparisons of their respective performances against a mutant tester stock. Our aim in the present paper is to disprove this claim and to show the potential pitfalls of deriving conclusions from a statistical analysis of experimental designs commonly employed for the study of competitive interactions in genetically homogeneous and heterogeneous mixtures. Using our own data, we conclude that evaluating the competitive interactions of phenotypically indistinguishable wild-type strains by competing them against mutant marked stocks still remains a valuable method.