Palmer, A. R. and C. Strobeck. 1997. Fluctuating asymmetry and developmental stability: Heritability of observable variation vs. heritability of inferred cause. Journal of Evolutionary Biology 10:39-49.

(a critique of the companion paper by Moller & Thornhill (1997) Heritability of developmental stability)


INTRODUCTION: Fluctuating asymmetry -- subtle random deviations from perfect bilateral symmetry -- is an appealing measure of developmental precision because of the apparent ease with which it may be measured and because its developmental origins seem so straightforward (Palmer, 1996). This appeal has led to its wide application as a measure of developmental stability in studies of inbreeding and outbreeding depression (reviewed in Palmer and Strobeck, 1986; Graham, 1992), as a measure of genetic or environmental stress in biomonitoring studies (Leary and Allendorf, 1989; Graham, et al., 1993), and most recently as a measure of fitness or mate quality in studies of sexual selection (e.g., see Moller, 1994; Tomkins and Simmons, 1995). Criticisms of the uses of fluctuating asymmetry (FA) have focused mainly on methodological issues (Palmer and Strobeck, 1992; Fields, et al., 1995), but doubts have also been raised about the developmental origins of subtle bilateral variation (reviewed in Markow, 1994; Palmer, 1996). If proponents of FA are not more reflective about these methodological and conceptual issues, the whole approach may become tarnished. This point was made emphatically by Phil Hedrick in his concluding comments to the symposium on developmental instability at which both Moller and Thornhill were present (Markow, 1994, pg. 434-435).

For bilateral variation to serve as a credible measure of developmental precision, three essential statistical tests or procedures must be applied (Palmer, 1994, 1996; Swaddle et al., 1994). First, because true FA (the result of real differences between sides) is indistinguishable from measurement error (biologically meaningless differences between sides), the between-sides variation must be significantly greater than the variation due to measurement error. Second, because measurement error actually inflates and therefore biases measures of FA upwards, the variation due to measurement error must be factored out to yield credible quantitative measures of bilateral variation. Third, because departures from 'ideal' FA may be caused by factors other than developmental noise, the between-sides variation must meet the statistical criteria for 'ideal' FA. Although other factors may also be important (Palmer, 1994), if these three issues are not addressed, then quantitative measures of bilateral variation and their heritability can not be interpreted with much confidence at all.

Moller and Thornhill provide a valuable service by tabulating the many studies that contain information on the heritability of bilateral variation. For this they are certainly to be commended. Such information is often buried deeply, particularly in the older literature, and is not readily accessible. In addition, their qualitative conclusion may, in fact, be correct: factors that influence the magnitude of subtle bilateral variation may very well have a weakly heritable basis, at least in some traits or species.

So why criticise such an extensive and seemingly worthwhile survey? Unfortunately, it suffers from two serious shortcomings elaborated upon below: I) the quantitative data are tabulated with very little critical evaluation, and II) the interpretation sidesteps or ignores some of the legitimate and widely acknowledged questions about the developmental origins of bilateral variation. Given their visibility, Moller and Thornhill could do the community a great service if they were to lead by example and temper their enthusiasm with greater caution and discrimination. By not subjecting the studies they tabulate to the three key criteria outlined above, and by ignoring or obscuring acknowledged conceptual difficulties, they appear to rely on the size of their table and the magic of meta-analysis to give a false confidence to both their qualitative and quantitative result and therefore an inflated credibility to their interpretation.


A.R. Palmer Publications List, Asymmetry Page, Home Page

Original material on this page copyright A. Richard Palmer. All rights reserved.
(revised Nov. 4, 2002)