How repeatable is the estimation of developmental stability by fluctuating asymmetry?

Abstract

The estimation of individual fluctuating asymmetry (FA) is subject to large sampling variabilities. Heritability estimates, as well as correlations between developmental stability and any other individual character and/or between–trait correlations, are consequently biased downward if FA is used as an estimate of an individual's ability to buffer its development against developmental noise. The estimation of the hypothetical repeatability, defined as the ratio of the between–individual component of variation in the unsigned FA divided by the total variance, allows correction for these biases such that patterns observed for FA can be translated to make inferences about the presumed underlying developmental stability. In this paper I show that previous estimates of this repeatability are incorrect. I provide a new method and show by means of simulations that the hypothetical repeatability is in most cases even lower than previously thought. This has important consequences for the analysis of FA with respect to statistical power and the interpretation of patterns in FA.