On hidden heterogeneity in directional asymmetry--can systematic bias be avoided?

Directional asymmetry (DA) biases the analysis of fluctuating asymmetry (FA) mainly because among-individual differences in the predisposition for DA are difficult to detect. However, we argue that systematic bias mainly results from predictable associations between signed right-left asymmetry and other factors, i.e. from systematic variation in DA. We here demonstrate methods to test and correct for this, by analysing bilateral asymmetry in size and shape of an irregular sea urchin. Notably, in this model system, DA depended significantly on body length and geographic origin, although mean signed asymmetry (mean DA) was not significant in the sample as a whole. In contrast to the systematic variation in DA, undetectable, random variability in the underlying DA mainly leads to reduced statistical power. Using computer simulations, we show that this loss of power is probably slight in most circumstances. We recommend future studies on FA to routinely test and correct for not only as yet for mean DA, but also for systematic variation in DA.