Yesterday I ran across this paper by Thomas Flatt in Quarterly Review of Biology, which is a really thorough review of the concept of canalization from its origins to its recent resurgence. Really thorough in this case means that it repeats the same things in several different ways, which is sometimes helpful.
Here's a definition:
Canalization is the reduced sensitivity of a phenotype to changes or perturbations in the underlying genetic and nongenetic factors that determine its expression (see also Meiklejohn and Hartl 2002; De Visser et al. 2003). Canalization is a relative term, and can thus only be defined as a matter of comparison. Thus, a phenotype P is more canalized than another phenotype P* if P remains relatively invariant when the single- or multilocus genotype G, which determines P, is exposed to different environments (environmental canalization) or located in different genetic backgrounds (genetic canalization): P is "resilient," "robust," or "insensitive" to genetic and/or environmental changes or perturbations. Canalization can therefore be recognized by observing that most genetic or environmental changes leave the phenotypic expression of G, and thus the phenotype P, invariant; the expression of G is changed such that specific phenotypic changes (P -> P*) are induced only in some genetic backgrounds or environments (or combinations of genetic backgrounds and environments). Consequently, a canalizing allele or genotype G reduces the phenotypic variation of a trait across a range of genetic backgrounds and environments relative to a noncanalizing allele or genotype G*, and a canalized trait P exhibits a restricted range of phenotypic variation across genetic backgrounds and environments as compared to a noncanalized trait P* (Meiklejohn and Hartl 2002) (Flatt 2005:288).
OK, that's an eyeful. In a nutshell, a more canalized phenotype is one that changes less in response to changes in environment, changes in genetic background, or both.
The definition is complexified by the need to consider behavioral or physiological phenotypes. The paper considers the example of homeotherms, who maintain a single body temperature in a range of environmental ambient temperatures. In this case, the single body temperature is a stable phenotype (upon which much else depends), but it is maintained by diverse physiological mechanisms that are active at different levels and at different times. In this way, canalization is a result of substantial physiological buffering.
Why is canalization important? The intro to the review has this:
Canalization is highly relevant for evolutionary biology. For example, it implies that phenotypes may be stable around their fitness optimum despite genetic and environmental change (e.g., Rendel 1967). By keeping phenotypic variation low, canalization may constrain phenotypic evolution (e.g., Charlesworth et al. 1982; Maynard Smith et al. 1985) and provide a microevolutionary mechanism for character stasis (e.g., Stearns 1994). Canalization also allows genetic variation that is phenotypically not expressed to accumulate. This cryptic variation can lead to the appearance of new phenotypes when development is "decanalized," for instance by environmental stress, thereby allowing evolutionary change (e.g., Rutherford and Lindquist 1998) (Flatt 2005:288).
That is pretty abstract, since it only hints at the adaptive value of canalized phenotypes. What we really would like to know is why some phenotypes would be more canalized than others. The answer could include lots of mechanisms. For one thing, it might be adaptive to have a phenotype that was less vulnerable to environmental modification -- i.e., had less environmental variance. For another as described below, canalization could simply result from the reduction of pleiotropy that results from modularization of genetic or developmental pathways.
Here's one practical application: If you want to find the relationships among a group of species, it is most sensible to choose characters that have a clean distribution of variation --- they should vary relatively little within species, but relatively much between species. Characters that have a great deal of within-species variation are often less useful, because they will often remain polymorphic even within relatively distantly related species.
But a character that has a lot of between-species variation and relatively little within-species variation is exactly the kind of character that may result from canalization. These features may reflect adaptive canalization that differs among a group of species; they may reflect some kinds of developmental constraints on one or more developmental modules. In any event, we should be aware that the kinds of characters that tend to be most useful for phylogenetic reconstruction will have certain evolutionary characteristics.
This section is relevant to me, so I'm including it:
Modularity of development may contribute to canalization (Stearns 1989b; Maynard Smith 1998; Hartwell et al. 1999; Stern 2000). Changes in the organism in one of its parts should not compromise other achievements: independent functions should be coded independently so that the change of one function does not interfere with other optimized functions (G P Wagner and Altenberg 1996). Modularity can be a way to achieve this independence of functions. The significance of modularity for canalization is that perturbing one module does not necessarily perturb the development of the whole organism: the embedding of particular functions into distinct modules allows for phenotypic change by altering connections among the modules while the core function of a given module remains unchanged (G P Wagner and Altenberg 1996; Hartwell 1999; Stern 2000; Schank and Wimsatt 2001). First, if deleterious mutations are highly pleiotropic (but see Stern 2000), then these mutations are likely to have a negative effect on many traits. If gene networks have a modular structure, however, then genetic change in one of the modules does not necessarily influence the others (Bonner 1988). Thus, by restricting pleiotropy, modularity allows some modules to continue to function when others change (Schank and Wimsatt 2001). Second, random mutations of a given phenotypic effect are likely to be more deleterious in complex organisms consisting of many traits as compared to simpler organisms with less traits (Fisher 1930; Orr 2000). Fisher (1930) suggested that mutations of small phenotypic effect are more likely to be favorable than mutations with large effects. In a topological model, he showed that the probability of approaching (or deviating from) the fitness optimum is higher (or smaller) if a mutation has a small phenotypic effect than if it has a large phenotypic effect. In the latter case, a mutation is more likely to go beyond the optimum or to deviate from it more strongly than if the mutation has only a small phenotypic effect. For the same intuitive reason, random mutations of a given phenotypic effect are more likely to disrupt a complex than a simple organism (Orr 2000). It would therefore be evolutionarily advantageous to reduce the number of independent traits by bundling them into modules (Orr 2000) (Flatt 2005:299-300).
The review does not link this topic of modularity and canalization to evo-devo, but it seems like a fertile topic to me. The presence of different canalized pathways that might be alternated by genetic switches of various kinds is also a very interesting topic.
Flatt T. 2005. The evolutionary genetics of canalization. Q Rev Biol 80:287-316. Full text (subscription).