Spurious Associations in Unreplicated Selected Lines

Abstract

Often a single pair of lines that has been selected for high and low expression of a trait is used as an animal model to study new biobehavioral characters thought to be associated with the selected trait. Because of genetic drift at many loci, comparisons of High and Low lines on the new character will frequently produce significant line differences even when there is no association between the selected trait and the new character being studied. In the absence of replicate lines to estimate the degree of genetic drift, effect size can be used to reduce the number of false-positive associations between the original selected trait and the new character. When the heritability of the new character exceeds .40 and the inbreeding coefficient within the selected lines is moderate, High- and Low-line means on the new character will frequently differ by at least one phenotypic SD, but not often differ by more than two SDs, in the absence of any relationship between the selected trait and the new character. If the selected lines are highly inbred, drift effects are greater, resulting in more false-positive associations. Situations posing special difficulty in the absence of replicate lines include the study of characters with low heritability relative to the selected trait and cases in which the lines do not differ greatly on the original selected trait. Studies using selected lines should always report inbreeding coefficients of the generations being studied, relative to the base population from which the lines were derived.