Biology Reference
In-Depth Information
The comparisons that we just did were intended to be merely exemplary of the methods
used to test a hypothesis of a generalized buffering mechanism. These were not carefully
controlled experiments that could isolate purely microenvironmental, macroenvironmental
or genetic variation. There are, however, several carefully controlled studies that assess
the relationship between FA and variation, both genetic and environmental. For example,
Breuker and colleagues (2006)
compared levels of FA and variation of
Drosophila melanoga-
ster
wing shape across 115 genotypes. To quantify FA, they used both the Procrustes dis-
tance of each individual from the bilaterally symmetric shape and the Mahalanobis
distance. Using the Procrustes distance, they found a correlation of 0.49 between FA and
variation among individuals, which rose to 0.67 based on the Mahalanobis distance. They
also examined the pairwise correlations between covariance matrices of variation and FA
for the 115 genotypes. Including the variances in the analysis produced correlations rang-
ing from 0.54 to 0.91; excluding the diagonals led to correlations ranging from 0.31 to 0.79.
Other studies of insect wings have similarly found moderate to high correlations between
covariance matrices of FA and variation (
Klingenberg and McIntyre, 1998; Klingenberg
and Zaklan, 2000; Klingenberg et al., 2001
). But variance and FA do not always show a
strong relationship to each other. In an exceptionally extensive analysis,
Dworkin (2005b)
found that genetic and environmental perturbations can have a profound impact on vari-
ance, but genetic and environmental canalization appear to be independent of each other.
In contrast to studies of insect wings, those of the mammalian skull usually find very low
or even non-significant matrix correlations between phenotypic variance and FA (
Debat
et al., 2000; Willmore et al., 2005; Breno et al., 2011
). For example, using a quantitative-
genetic analysis, Willmore and colleagues estimated the correlation between FA and the
environmental component of variation at just
0.07; the highest correlation was between
FA and phenotypic variation, which was merely 0.049.
Studies that compare levels of FA to variance, both genetic and environmental, address
one form of the hypothesis of a general buffering mechanism. But the hypothesis could
also be framed in terms of the developmental pathways being buffered. Even if there is no
general relationship between macro- and microenvironmental, and/or between genetic
canalization and developmental stability, there may still be general buffering mechanisms
that indiscriminately buffer a developmental pathway against all sources of perturbations.
But the results obtained by comparing covariance matrices of FA and variance may actu-
ally tell us more about the mechanisms of morphological integration than buffering.
2
PREDICTING THE STRUCTURE OF COVARIATION:
MO
RPHOLOGICAL INTEGRATION AND MODULAR
ITY
Morphological integration and the related property of modularity have long been the
focus of quantitative evolutionary developmental biology. Recent developments in evolu-
tionary theory have stimulated a resurgence of interest in both because of growing interest
in “evolvability”, i.e. the ability to evolve. The fact that organisms are able to evolve had
not prompted much theoretical attention until recently, perhaps because that ability has
been taken for granted (
Wagner and Altenberg, 1996; Hansen, 2003
). After all, it is obvious
that organisms can evolve because they do. What raised questions about the ability to
