COI barcode variation within avian species is uniformly low regardless of census population size. This finding directly contradicts a central prediction of neutral theory and is not readily accounted for by commonly proposed ad hoc modifications. As an alternative model consistent with empirical data including the molecular clock, we propose extreme purifying selection, including at synonymous sites, limits variation within species and continuous adaptive evolution drives the molecular clock.
This is the conclusion of a new paper that was published last week in PLoSONE. It is titled: DNA Barcoding Works in Practice but Not in (Neutral) Theory.
The neutral theory of molecular evolution that was proposed by Motoo Kimura suggests that most of the genetic variation in populations is the result of mutation and genetic drift and not selection. As a result sequence differences within and among closely related animal species are synonymous, i.e., do not change the amino acid sequence.
The result of Kimura's theory which was independently formulated by King and Jukes a year later sparked a decade-long debate on to what extent Darwinian selection or neutral processes are driving the molecular evolution of genes and thus the ecological divergence of species. Selectionists argued that a large fraction of those non-synonymous DNA base substitutions in coding genes going to fixation should be driven by positive Darwinian selection. The biological function of these genes should therefore be involved in ecological niche segregation. Under a strict Neutralists view, most fixed amino acid substitutions have no effect on fitness, because purifying selection constantly removes alleles with strongly deleterious effects on fitness and mutations while positive effects were thought to be extremely rare. While the neutral theory was the prevailing model for several decades, e.g. the comparison of whole genome sequences has now produced evidence for a more important role of natural selection (e.g. here and here).
What you can find in almost all textbooks today is the statement that under the assumptions of the Neutral theory intraspecific variation equals 2 Nµ, where N is population size and µ is mutation rate per generation. And that is the prediction the authors Mark Stoeckle (famous author of the other DNA Barcoding blog) and David Thaler pretty much busted as they encountered uniformly low intraspecific variation regardless of census population size when they used COI barcode records from two avian families representing the two major divisions of birds, non-passerine and passerine. Why birds - you may ask. Well, the colleagues have good reasons: Birds are uniquely suited this task: they are well represented in barcode libraries, have the best-known species limits of any large animal group, and, most critically, are the only large group with known census population sizes, a key parameter in neutral theory.
I provided their conclusions already at the beginning of this blog post but I wonder if they would think of looking at their results from the perspective of Tomoko Ohta's Nearly Neutral Theory. The most significant difference between the neutral and the nearly neutral theories is that the latter predicts a negative correlation between evolutionary rate and species population size. The nearly neutral theory states that the interplay of drift and weak selection is important and predicts that evolution is more rapid in small populations than in large populations - what ever a small population in this context means. It might not fully explain the uniformity of the low intraspecific mtDNA variation in the paper but it could partially explain the absence of an increase in intraspecific variation with increasing census population size.
I like this kind of paper as it looks closer at the limits to DNA Barcoding and the evolutionary mechanisms that make it work. It would be great if this paper can spark a new discussion about these questions and motivate further research on the 'Secrets of DNA Barcoding'.