Richard BuggsProfessor, Queen Mary University of London
- Queen Mary University of London
- United Kingdom
One could even go a little further and say that the fact that single genes often produce incompatible phylogenetic trees also refutes Richard Dawkins’ idea of the - selfish - gene as the relevant “unit of selection.” Referring to Douglas Futuyma’s textbook on evolution from 2013 (3rd edition), a passage from Wikipedia aptly boils it down: “One of the resolutions to reduce the implications of incomplete lineage sorting is to use multiple genes for creating species or population phylogenies. The more genes used, the more reliable the phylogeny becomes” (quoted from Wikipedia: incomplete lineage sorting/implications). In other words, entire genomes should be treated as the real units of selection because they produce the most reliable phylogenetic trees. That the molecular reconstruction of phylogeny is not as straightforward as Richard Dawkins tries to make us believe is already shown by the fact that normal sexual reproduction within a given population always involves some amount of horizontal gene transfer or hybridization between a male and a female lineage, which makes it difficult to follow the path of single genes through the population. In contrast, genomes remain relatively stable over time, even if some smaller parts of their genetic content are exchanged from time to time.
Thanks for your comment, Adi. There is a big discussion to be had about levels of selection, but I agree with you that Dawkins' claim about gene tree congruence seems to be driven more by what would be predicted by a simplistic 1970s model for how evolution works than by the data.
Reconciliation of gene and species trees is a popular topic, with plenty of publications, studies and computer programs. However, all of these do not contradict Dawkins' argument. Even if some gene trees may differ, especially with bacteria (HGT), closely related species (ILS) or simply due to reconstruction errors, the gene trees are much closer than expected by chance, with many common splits and subtrees. With a few taxa (e.g. 11 as in David Penny's study), it is still frequent to find the same tree topology (among >600 millions). All the programs you mention assume this closeness between gene trees and species trees, and among gene trees. Thus, explanations and precautions are needed, but the core of the argument is still very strong and convicing. Take care to creationists!
Hi Olivier, thank you for a thoughtful response. However, Richard Dawkins is not just saying that most gene trees are much closer than expected by chance: he is saying that they are all "approximately the same". This is the claim that I am disputing.
Readers may be interested in a discussion of this article, including comments by Joe Felsenstein (University of Washington) and Steve Schaffner (Broad Institute) here: http://theskepticalzone.com/wp...