For many developmental biologists there may not be any general principles for the evolution of development (Akam 1998). The claimed opportunistic nature of evolutionary change (Sander and Schmidt-Ott 2004) may be one of the reasons for that. Although for short timescales some genetic changes may seem more likely than others, evolution will, over time, use whatever genetic variation is at hand to produce adaptive variation. As a consequence, previous history about genetic changes may easily fade away over time, precluding predictions about development evolution. Moreover, it has been suggested that, since selection cannot see how the phenotype is produced, mutations that do not alter how development works but do increase its genetic complexity may accumulate over time. This would result in the logic of development becoming more and more baroque over time (True and Haag 2001, Salazar-Ciudad 2006a). Here it will be argued, however, that precisely because of the opportunistic nature of evolution, general predictions about the evolution of development and the effect of development in evolution are possible. These predictions do not involve the exact molecular nature of changes in development but rather involve more general aspects about the logic and topology of genetic networks and how they interact within the developing epigenetic context of the embryo or intermediate phenotype.
In the same way that development can be described as a sequence of pattern transformations, it can also be described as a sequence of action of developmental mechanisms (each mechanism responsible for a specific pattern transformation). Then the evolution of development can be described as changes in the developmental mechanisms used in a lineage over time. These changes can occur because new developmental mechanisms are recruited or because existing developmental mechanisms are replaced by other ones. In this chapter two developmental mechanisms are considered to be different if their gene network topologies are different or if they affect different cell behaviours. This definition is merely a choice of convenience that reflects the fact that gene network topology is among the best-known aspects of developmental mechanisms. Moreover, topological changes have been suggested to have, on average, more dramatic effects on variational properties than other kinds of change (Salazar-Ciudad et al. 2000). From this perspective, the morphological outcome of development can change for two reasons: (1) because mutations (or environment) change which subset of the variational properties of some developmental mechanisms are produced, or
(2) because mutations (or environment) change some developmental mechanisms (meaning topology and then, probably, variational properties).
The chances of a de novo formation of a developmental mechanism depend on its genetic structure. The more genes and gene interactions a developmental mechanism involves, or in other words the more base changes in more genes are required, the more unlikely a mechanism will appear through random mutation. The same applies for the number of genetic changes required to evolve one developmental mechanism from another. The likelihood of an existing developmental mechanism being recruited in a new place and time in the developing embryo depends, instead, on the proportion of genes in the network that can lead to the activation of the whole mechanism.
Developmental mechanisms are not selected by themselves nor by their variational properties but by the specific morphology they produce in specific individuals (which is only a part of the variational properties). If the genetic and variational properties of developmental mechanisms are known, then the evolution of development can be understood by how likely it is that different developmental mechanisms appear or are recruited by mutation, and how likely it is that the morphologies they produce are adaptive in specific environments. Only a limited number of developmental mechanisms are known to that extent. However, it has been suggested on theoretical grounds that there is a limited number of types of gene network topologies that can produce pattern transformations (Salazar-Ciudad et al. 2000, 2003). Pattern formation capacity and opportunism (in the form of likelihood of appearance by mutation) confine the spectrum of developmental mechanisms that can possibly be involved in different situations in development and evolution. This variational approach can be useful because, at least, all developmental mechanisms can be classified into exclusive types that share some aspects of their genetic and variational properties. Predictions about the relative involvement of each type at different stages of development and under different selective pressures are then possible. The next sections offer examples of how developmental mechanisms can be classified and how this approach can give interesting evolutionary insights. This 'variational' approach can be used on different classifications (for example Salazar-Ciudad et al 2001a) and is not dependent on the specific one presented in the next section.
Was this article helpful?