Resetting the molecular clock

A commentary in Nature by molecular biologist David Penny discusses a recent paper in Molecular Biology and Evolution by Simon Ho and colleagues (2005). The subject of the paper is the observation that recent species divergences seem to have higher apparent mutation rates.

If mutations are completely neutral, then the rate of substitutions within species lineages should be the same as the rate of mutations. Of course, not all mutations are neutral: many of them are deleterious, a few are advantageous.

Ho and colleagues focus on purifying selection on deleterious mutations as the most likely reason for the apparent acceleration in rates for recent divergences. The idea is that slightly deleterious mutations remain in the population for a period of time inversely proportional to the strength of selection against them. Very weakly selected mutations can remain for a long time; strongly selected ones will disappear quickly. But none of these are likely to become fixed in a population. Therefore, natural populations harbor a number of slightly deleterious variants that, if observed, tend to inflate the apparent rate of mutations. This inflation is more severe for recent divergences, because these deleterious mutations make up a higher proportion of the total mutational difference between the species considered.

The data were consistent with purifying selection, because the proportion of nonsynonymous mutations (the ones that can have a phenotypic effect) is higher within species than among species, and higher between closely related species than between distantly related ones.

Ho and colleagues (2005:1565-1566) provide new estimates for human and Neandertal divergences based on their correction:

Using an alignment of d-loop sequences from four Neandertals and four humans (African, Caucasian, Chinese, and Indian) (alignment available as Supplementary Material online), we recalculated the ages of the last common ancestors of Neandertals, humans, and of Neandertals and humans. The new date estimates were computed by solving equation (7) numerically with the parameter values estimated from the primate d-loop data (eq. 6) and with HKY85-corrected distances estimated from the alignment using the program baseml (Yang 1997). Ages of last common ancestors were as follows: Neandertals and humans 354 ka (222--705 ka), Neandertals 108 ka (70--156 ka), and humans 76 ka (47--110 ka). These date estimates are intermediate between the values obtained when either the lowest or highest rates of change from figure 1c are used for divergence date estimation.
The three new date estimates were considerably younger than those estimated in previous studies, which gave ranges of 365--853 ka (Ovchinnikov et al. 2000), 550--690 ka (Krings et al. 1997), and 317--741 ka (Krings et al. 1999) for the Neandertal-human divergence; 151--352 ka (Ovchinnikov et al. 2000) for the last common ancestor of Neandertals; and 106--246 ka (Ovchinnikov et al. 2000), 120--150 ka (Krings et al. 1997), and 111--260 ka (Krings et al. 1999) for the last common ancestor of humans (fig. 5). This discrepancy arises because high, short-term rates of change were taken into account by our approach.

Notice here, the largest effect of overestimation occurs for the most recent, within-species comparisons. This observation is consistent with earlier disagreements about the mutation rate of mtDNA -- some studies found that recent humans apparently were mutating at ten times the rate expected from interspecific comparisons.

Notice also, that the TMRCA of human mtDNA on this estimate is between 47,000 and 110,000 years ago. This date is more recent than early modern humans in Africa. The bulk of the confidence interval is later than the first emergence of "modern" humans from Africa. If the TMRCA for humans is really this recent, there is next to no chance that this origin of human mitochondrial variation actually is a signature of the origins of all modern humans. It may be a signature of the movement of some modern populations, but that is a different issue, and one that must ultimately be sorted out from the pattern of selection on mtDNA.

Does the finding have any importance for the way we study variation within humans? Well, there is this:

For some reason, the continuum between population heterozygosity and long-term evolution has not been adequately studied. Although it is a continuum, the techniques required may change as the timescale decreases. For example, some concepts from long-term evolution (binary evolutionary trees with sequences studied only at the tips) have been extended into populations where trees are no longer binary, and ancestral sequences (at internal nodes) are still present in the population. There are hints that a formal multiscale study is necessary, because even though the same underlying process is occurring, different features of trees are observed as the timescale changes (Penny 2005:184).

Just a note that it is a mistake to look at genetic variation within a species and treat it the same way as long-term neutral variation as measured between species over long spans of time. Human variation dates to anywhere between 50,000 years old and over 3 million years old for different genetic loci. The more recent and more ancient parts of this time interval may not be explained by the same pattern of evolution.

And wouldn't it be boring if they were?

References:

Ho SYW, Phillips MJ, Cooper A, Drummond AJ. 2005. Time dependency of molecular rate estimates and systematic overestimation of recent divergence times. Mol Biol Evol 22:1561-1568. Full text online

Penny D. 2005. Evolutionary biology: relativity for molecular clocks. Nature 436:183-184. Full text (by subscription)