Biology Reference
In-Depth Information
linking them all (
Wagner and Altenberg, 1996
). But how they are related biologically has
been a matter of considerable controversy, especially when it comes to the relationships
between the various forms of canalization (i.e. macroenvironmental, microenvironmental,
and genetic canalization and developmental stability, which could be considered canaliza-
tion against developmental noise). Less often debated, perhaps because less often dis-
cussed, is the relationship between properties that regulate the expression of variation and
those that structure covariation. The processes regulating variation include all those listed
above as various forms of canalization and those that structure covariation include
morphological integration and modularity. They are obviously linked in the sense that
covariation requires processes that induce variation, but they are now often linked
methodologically because studies that examine the relationship between canalization and
developmental stability now often compare
co
variance matrices, one representing the (co)
variation among individuals, the other representing the (co)variation due to developmental
noise. Thus, patterns of covariation are analyzed for insight into the processes that suppress
variance and responses to noise. Additionally, the role that developmental modularity plays
in structuring morphological integration is now often analyzed by comparing those same
two covariance matrices.
There are, however, two important distinctions between studying the processes regulat-
ing variation (plasticity, canalization and developmental stability) on the one hand and
those structuring covariation (integration and modularity) on the other. The two distinc-
tions are related to each other because one concerns what needs to be measured, the other
concerns how it is measured. Studies of plasticity, canalization and developmental stability
assess the impact of various factors (either environmental or genetic) on the expression of
phenotype variation. We know how to measure and decompose variation, and the techni-
ques traditionally used to analyze plasticity, canalization and developmental stability are
readily adapted to geometric morphometrics. Conducting the analyses is far from easy
because the logistics of the experiments can be daunting, and the statistical models can be
remarkably complex. But the analyses are all based on sums of squares. Morphological
integration and modularity present more severe technical challenges because the concept
that has been central to them, that of a “trait”, has no obvious analog in geometric mor-
phometrics. An additional problem is that the Procrustes superimposition itself imposes a
pattern of covariances on the data (
Rice, 1989; Rohlf and Slice, 1990; Rohlf, 2003; Adams
et al., 2004
). For both those reasons, adapting geometric methods to the study of integra-
tion and modularity did not prove straightforward. Recent progress has led to several
new methods that replace the idea of “a trait” with “a subset of landmarks” (
Klingenberg
et al., 2003, 2004; Monteiro et al., 2005; Marquez, 2008; Klingenberg, 2009
). But methods for
studying integration and modularity are not as mature and well understood as those for
studying plasticity, canalization and developmental plasticity.
This chapter is organized primarily by subject matter rather than by methods. We
first discuss methods for analyzing plasticity, then canalization and then developmental
stability, and conclude the first section of this chapter by discussing methods for testing
hypotheses about the relationship among these three properties. We then present three
methods for analyzing morphological integration and modularity, all of which are well
grounded in geometric morphometric theory and implemented in freely available
software.
