Convergence

We are cluttering up a thread that’s not about convergence with discussions about convergence. Here, we can discuss this specific issue.

Convergence, or homoplasy, occurs when two individuals share a feature that is not inherited from their (assumed) common ancestor. Such a feature may be a phenotype – for example, sharp teeth in many predators, streamlining in fast swimmers, wings in diverse groups – or a genotype. The former is thought to typically arise by natural selection. The same fundamental phenotypic reward – holding on to prey, catching or escaping – leads to similar outward manifestations of the phenotype, although typically the developmental pathway in the different species is different. There is ‘virtually no’ parallel convergence in genotypes. Convergence in gene sequences generally arises as a result of stochastic variation. If one has two diverging lineages, the limited number of changes available to each DNA base pair (3) results in a significant chance that a base pair held in common by two sequences is not there due to homology (common ancestry) but homoplasy.

It’s important to separate out these two senses. Much cross-talk arises because people are talking of one as if it were the other, sometimes with the lame justification that ‘it’s all molecules’, or the convergence directly constrains a biochemical system. These are misunderstandings; it’s the distinction between phenotype and sequence which is important, not whether the phenotype is ‘molecular’ or not.

Now, the reason for this discussion is that some people are arguing that homoplasy should cause one to reject common descent. So here is their opportunity to defend that view.

[eta – whatever one thinks of Biologos, they provide here an accessible introduction to homoplasy and its relation to homology]

103 thoughts on “Convergence

  1. Mung,

    And for some reason they believe it’s easier (more probable) for eyes than for DNA sequences.

    ‘Some reason’ goes by the name of natural selection.

    Selection isn’t involved in the stochastic variation of DNA sequences in diverging lineages, so simple probability will suffice. If either or both of two uncorrelated and non-matching sites undergoes mutation, there is a 1/3 chance that they will end up the same. Take two such sites, it’s 1/3 * 1/3, three and it’s cubed, and so on. But when you add selection, or another source of bias, that changes matters dramatically.

    With typical phenotypic convergence, selection isn’t even acting at the same loci in the two lineages. So the above locus-specific calculation doesn’t even apply, even though one can naively say ‘it’s DNA’. Apples and oranges. Talking of which, is their sphericity convergent?

  2. Allan,

    Apples and oranges. Talking of which, is their sphericity convergent?

    It’s design. Take that, evilutionist!

Leave a Reply