Metodología y procedimientos

Earlier the unit of selection was defined to be a binary string of lengthL. However, there are additional properties that the string must have in order to be considered a unit of selection. There are three states that the unit of selection can occupy – birth, maintenance, death – and it progresses through these states during its lifetime. The first and last, birth and death respectively, occur because evolution is, fundamentally, a birth/death process. Deaths are a way of enacting selection by removing those less adapted to the environment and so, by default, rewarding

those better adapted. A pure birth process, therefore, has no way of selecting

between individuals for reproduction nor any way of removing individuals which have reproduced. As a result, the population will go to infinity having displayed a random walk through trait space and is therefore uninteresting, both from a theoretical and realistic point of view.

Birth-death processes are branching processes, as the process creates tree structures of descent. Phylogenetic trees are used to test evolutionary theory and reconstruct the sequence of evolutionary steps that led to a particular output. How- ever, one criticism of the fossil record is that it is incomplete. Whilst it was argued in Eldredge and Gould [1972] that this incompleteness may be part of the process of evolution, rather than an inconvenience, there are many statistical methods of attempting to construct a “true” phylogenetic tree from data.

However, assume the complete genome sequences of everything that has ever lived are known and accurately dated. Is it possible to construct the “true” phylo- genetic tree from this dataset? This depends on the characteristic time scale that genomes are dated to, since any relatedness measure must assume something about the nature of mutations (namely that there are very few (ideally one) between time steps). The question then becomes whether there are more mutations in a single birth than there are differences between that species and another species. Assuming this to be true results in a set of possible “true” trees, with likelihood contours (as in Billera et al. [2001]) surrounding the various mutational jumps. Therefore, until more is known about the constraints of mutation only approximations of the “true” phylogenetic tree can ever be known. Using only genotypic information and the rate of mutation, the uniqueness of parents cannot be established. This is even more problematic in the case of bacterial evolution, which features lateral gene transfer.

3.4.7 Conclusion

In this section, the theory of evolvable systems was discussed and a comparative

consideration given to different definitions. A simple evolutionary function was

developed, which highlighted the importance of characteristic time scales. This was then used, along with the observations made in 3.2, to establish that the “true” history of phylogeny can never be established, even in the best possible case. As such, assumptions must be made about the nature of the evolutionary process, and such assumptions must be defined as fixed parameters, which are external to the system, rather than internal parameters which may be subject to evolution.

3.5

Conclusion

Although evolution is often reduced to natural selection, there are many processes and intricacies which seem to prevent evolution from proceeding as an optimisation exercise. In the first section of this chapter, three models of evolutionary processes were reduced to their components, to assess what they could and could not explicitly model. This was followed by a theoretical treatment of the main meta-traits of traits – constraint and evolvability. Following the concept of constraint to its natural conclusion, it is found that constraint will always occur in any evolvable system, due to the nature of heritable mutations, meta-regulation, and the dissociative property of the genotype-phenotype map. This suggests that constraint should be explicitly modelled in all evolutionary models which aim to have a predictive effect. It is possible that such an observation will lead to more basic laws of evolution being discovered proposed.

Interesting conclusions also result from the discussion of the paradox of evolv- ability in Section 3.3. Evolvability has the potential to link first tier processes with third tier phenomena, by tying processes at the genotypic and phenotypic level to mass extinctions. As evolvability decreases over time, the ability to adapt to sudden shifts in environment is lowered, and the species become more constrained, until a mass extinction is inevitable. Such a result is due to the nature of characteristic time scales, which are properties of the unit of selection discussed in Section 3.4 and which are explored in Chapter 7. In this section, a formal definition of an evolvable system is discussed, along with the properties of the unit of selection and the environment. This leads to the development of a basic evolutionary function which ties character- istic time scales inherently to the unit of selection. Finally, observations are made about branching processes which suggest that even with complete information the “true” phylogenetic tree cannot be constructed without assuming properties of the

evolutionary process. Among such properties are constraint and evolvability, which are still poorly understood and not explicitly modelled in evolutionary models.