Biology Reference
In-Depth Information
FIGURE 12.13
A developmental reframing of
the classic concept of modularity, giving a devel-
opmentally explicit version of
phenotype map (after
Klingenberg, 2008
). There
are two developmental modules (M1, M2) com-
prising two groups of
the genotype
T7).
Genes affect developmental pathways rather than
traits; the impact of the gene on a pathway is
shown as a dotted line intersecting a pathway,
shown as a solid line. A direct interaction
between pathways is shown as an intersection
between solid lines.
traits (T1
T4, T5
interactions within and between pathways, such as signaling interactions, partitioning of
tissue, or other mechanisms that can transmit variation along and between pathways. The
distinction is important because, in the case of parallel variation, the source of the varia-
tion is also the cause of the correlation. That is not the case for direct interactions. In the
case of direct interactions, the cause of the correlation is the mechanism that regularly
associates the development of the two traits. If variation arises upstream of a branch in the
pathway, the variation will be transmitted downstream along both branches. Even if the
variation is due to a random developmental perturbation, the variation will be transmitted
downstream or from one pathway to another via a signaling interaction. Modules are thus
highly integrated internally due to many, often strong, direct interactions within them,
exceeding those that take place between modules (
Klingenberg, 2005, 2008
). This develop-
mental view of modules not only reframes the hypothesis of modularity, it also yields a
novel technique for testing hypotheses about modularity. Because direct interactions
within developmental modules can regularly associate traits even when the source of the
variation is a random developmental perturbation, fluctuating asymmetry becomes a use-
ful tool for analyzing the structure of developmental modules. When FA and individual
variation are highly similar in structure, direct interactions within developmental modules
play a large role in integrating the phenotype (
Klingenberg et al., 2001; Klingenberg, 2005,
2008
). Whether the modular organization of development is an intrinsic feature of devel-
opmental systems and a potential constraint on the adaptive evolution of integration
(
Klingenberg, 2004, 2005, 2008
) are open and provocative questions.
To this point, we have not tried to define morphological integration except in terms of
modularity. But the ideas of integration and modularity can be partially separated. They
are not separate when variation actually does have a modular structure, but variation
might not actually be modular. It may be that genetic effects are spatially restricted but
continuous rather than tightly clustered and even partially overlapping, producing what
Roseman and colleagues (
Roseman et al., 2009
) termed “integration without modularity”.
Hallgrimsson has offered a developmental explanation for a non-modular organization of
