A couple days ago, I introduced a new paper by Weaver et al. which continues investigating the effect of genetic drift in modern human vs. Neandertal craniofacial differences. I didn’t have access to the paper then, but now I do and I wanted to share my thoughts and ideas of it with you.
The premise the authors worked upon is through using drift in sequences of microsatellites as a template to estimate the effect of drift in the morphological differences in Neandertal and modern human craniofacial traits. What are microsatellites? I’ve defined this term a million times, so I’ll be brief. Microsatellites are simple repeats of DNA sequences. They aren’t really complex in pattern and when they are found, they indicate an increased rate of mutation compared to other neutral regions of the genome.
To understand how the authors correlated variation microsatellites to variations in morphology, let’s first try to understand how microsatellites are formed. Effectively microsatellites are mistakes, but they aren’t necessarily deleterious. During DNA replication or recombination, slippage can occur, especially in stretches of a simple pattern, like say AAAAA. How? Imagine portion of double stranded DNA, with a sequence like this:
After the two strands are split apart with helicase, and the two ends meet back up, it could slip and mispair. In this case, it would end up like this:
Where the space is where mispairing occurred… on the ends of the AAAAA, since a significant stretch of the DNA, the other AAAA, were perfectly paired. Thus, if this mistake is overlooked by repair mechanisms, a polymerase may come in and extend the space with a complementary A and ultimately extending the original AAAAA into a longer stretch, like AAAAAA or longer. After many generations, this stretch can end up looking like so:
When this sort of phenomenon happens it is called genetic drift. It is random and most often a mistake. Like I mentioned above, rarely are these mutations deleterious. Typically, they are neutral, meaning that they offer no advantageous nor deleterious traits. They persist, and can even get longer over generations. Sometimes they even get shorter.Microsatellites act as good identifiers to figure out relatedness and population structure. How? If one could find out how often changes in microsatellite sequences occur on average, one could begin to begin to make a molecular clock of sorts to calculate temporal rates of neutral evolution.To correlate drift in microsatellites as a framework to understand the impact and temporal mode of drift in the morphological differences in Neandertal and modern human craniofacial traits, the authors had to adapt a mathematical algorithim that I just eluted to called the divergence time estimator (TD). Rather than confusing TD used in molecular evolution, the authors propose a new name for their clock, PTD, standing for phenotypic divergence time estimator.It is critical to understand this equation because the entire study is founded upon applying morphological measurements in modern humans and Neandertals to this clock. I’ll break down all the variable as best as I can, but the final equation the authors give to us for PTD is:
Your mind might be spinning just looking at this equation, I know it immediately sent shivers down my spine of the stuff I had to deal with in multivariable calculus. Wrought with fear, I glanced over this part of the paper at first pass, but under some more scrutiny I realized this really is all algebra… totally doable if we just understand what the variables and constants are.
So from the top, x1 and x2 are the means of population 1 and 2. The authors don’t really define what the mean is, I am assuming it is population size. h2σ21 and h2σ22 are narrow-sense heritability for the measurement and the phenotypic variances for population 1 and 2. Basically the degree of how heritable and variable are these phenotypic traits. V0 represents the additive genetic variance in the ancestral population before Neandertal and modern human lineages went their own ways. Setting V0 to 0 yields the maximum estimate of divergence time.
On the denominator we see m, which represents the mutation parameter. I’m not too sure what this is, but since it is represented twice, I think of it as a constant. Since it is in the denominator and we want to figure out the rate of drift, my best guess is the rate of neutral variation of traits based off of the divergence time estimator (TD) from mircosatellites. σ2P, represents within population phenotypic variance. I’ve already reviewed what h2 is, but to rehash it is the narrow-sense heritability measurement.
If I’m correct with my understanding of the variables, after integrating all them together, I can see that the numerator factors in differences in population size, heritability, and between population variation. The denominator factors in the mutation rate, and within population variation and inheritance.
The archaeological and fossil record tell us that the population sizes of pre-divergence of Neandertal and modern human peoples from several hundred thousand years ago were relatively small. Therefore, initial generations, after the divergence, would also be small and homogeneous both genetically and phenotypically. This assumes that punctuated equilibrium was not a factor in the speciation modern humans and Neandertals. As time progressed and each population grew in size, mutations are expected to increase and introduce variation until another equilibrium is reached.
Personally, I have a lot of questions with this portion of the paper. I understand the equations, but I don’t quite grasp the assumptions one has to make to consider these equations applicable. I can really use some clarification, so feel free to chime in.
My concerns stem off of how the authors corrected for fluctuations in population size? Both Neandertal and modern human populations were not gradually growing. There were waxing and weaning moments in population growth for both species. Futhermore, Neandertals and human populations weren’t both growing and reducing at the same times and rates, which would alter the self adjusting equilibrium the authors operate on.
With a reduction in population size, genetic and morphological variation are also lost. The loss in diversity will take with it neutral diversity as well as negative diversity, like deleterious traits. Some positive diversity will also be lost, too… not all and not as much as losses in negative and neutral diversity. But overall diversity, both from drift and in selection, would be greatly reduced as populations dwindled in size. What I’m trying to get at is, neither selection nor drift can be solely responsible for the differences between Neandertals and humans, because both are filtered out as population sizes grow and reduce.
My confusions aside, the authors took this neutral phenotypic evolution model of theirs and applied it to an recent data set that showed Neandertals and modern humans seem to evolving neutrally in 37 different cranial measurements. The 37 different measurements are illustrated to your right. I’ve plucked this from the article. I must say they are very thorough, much more thorough in measurements than this upcoming paper on Homo floresiensis.
Plugging in the 37 measurements from 2,524 modern human skulls from 30 globally distributed populations and 20 Neandertal specimens into their equation resulted in the divergence time of modern humans and Neandertals to be 311,000 years ago, assuming mutation drift equilibrium. Under this assumption the 95% confidence interval is actually really wide… between 182,000 – 466,000 years ago. This is a huge time span when considering the divergence time of these two Homo species. It raises questions the accuracy of thinking that drift was the only thing at play here. Calculating for the the maximum divergence time yields a narrower 95% confidence interval, of 308,000 – 592,000 years ago divergence time, which is better but still doesn’t resolve how the authors overcame in different fluctuations the different population sizes.
All that aside, these dates extracted from morphology confirm dates that Noonan and colleagues figured out from sequence analysis of ancient Neandertal DNA in 2006, which is really remarkable because it has been hard to make these two lines of evidence speak to one another.
Not that I wanna draw out this post anymore, but I think I should really address why is this important as a prize for you drudging along with me so far. As you may know, there’s quite a lively debate with how ancient Homo species were replaced once modern Homo sapiens started flocking out of Africa. There’s one hypothesis that advocates a total replacement of ancient Homo with modern humans, while another hypothesis that raises the possibility ancient Homo integrated with modern humans. This paper shows that it is possible to look at the phenotypic differences between Homo sapiens and Homo neanderthalensis and see that they really don’t show evidence of admixture, in fact it is possible to trace back the time at which these phenotypic differences began to radiate. Anyways, for a more thorough introduction into the importance of this paper, read Alex’s blog post on it.
In the meantime, I am gonna celebrate Nooruz. Happy New Year everyone!
Weaver, T.D., Roseman, C.C., Stringer, C.B. (2008). Close correspondence between quantitative- and molecular-genetic divergence times for Neandertals and modern humans. Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0709079105