Further Applications of EDMA - An Invariant Approach to Statistical Analysis of Shapes - page 262

Fluctuating
asymmetry

a)

<1 1 >1

A = R / L

b)

Directional
asymmetry

<1 1 >1

A = R / L

c)

Antisymmetry

<1 1 >1

A = R / L

Figure 7.1 Distributions of A that are expected for different types of asymmetry. (a)
Fluctuating asymmetry: Because there is no handedness in the sample, the distribu-
tion of A is unimodal and the mean is expected to be zero. (b) Directional asymmetry:
The asymmetry in a typical observation is more conspicuous than in the case of fluctu-
ating asymmetry. The distribution of A is unimodal but is not zero because of the
handedness in the population. (c) Antisymmetry: The asymmetry of a typical observa-
tion is conspicuous; but there is no handedness in the population, so that the
distribution of A is bimodal.

polymorphic for handedness and, therefore, are antisymmetric by def-
inition).
Antisymmetry: Antisymmetry is similar to directional asymmetry
in that each observation in a sample can be considered conspicuously
asymmetric. However, there is no handedness that characterizes the
sample as a whole. Each observation will have a value of A that is
either greater than 1 or less than 1, so that A has a bimodal distribu-
tion ( Figure 7.1c ) . Note that the difference between fluctuating