By Matthew Cobb
One of the most important tools in evolutionary biology over the last thirty or so years has been the development of the ‘molecular clock’, which is a technique for measuring how long ago two organisms (or taxa or species) separated on the ‘tree of life’. This approach has been incredibly powerful, and underlies much of our understanding of the rate of evolution, linking fossil and molecular data – all the figures from the excellent Timetree.org are based on the molecular clock. But now it appears that in our lineage at least, the clock may not have been ticking quite so fast as we thought, and some recalibration is going to be necessary.
The basic assumption behind the molecular clock is that mutations – changes in DNA – occur at a constant rate over time, and that the number of differences between two groups can therefore be turned into a figure based on the time since the two diverged. This phenomenon was first noticed in 1962 by Linus Pauling and Emile Zuckerkandl looking at differences in haemoglobin genes, then explicitly turned into a hypothesis the following year by Margoliash, before being fully developed in the 1970s by Allan Wilson. (It is in fact a bit more complicated, as the average generation time of a species has to be taken into account – the shorter the generation time, the higher the mutation rate.)
There are some important provisos to the clock – any stretch of DNA that is subject to selection, for example, is not going to be a very useful source of clock data, as genetic differences will tend to be removed by selection; many genes that are vital to organismal function are therefore highly conserved, showing few differences between groups. For this reason, scientists tend to use either ‘synonymous changes’ in DNA – these are ‘silent’ differences that do not cause any change in gene function (protein structure, gene regulation, or whatever) – or to use stretches of non-coding DNA, which appear to be not subject to natural selection and to evolve ‘neutrally’, just accumulating mutations with time.
The sources of spontaneous mutations are well-known – mistakes in the cellular machinery during copying of DNA, electromagnetic radiation, mutagenic chemicals (most of them completely natural), and so on. These spontaneous mutations are important not only for calibrating the clock, but above all for providing the raw material for evolution by natural selection. The whole infinite variety of life is the consequence of mutations, which have then been filtered through the sieve of natural selection, over vast periods of time.
The molecular clock has been particularly important in helping to calibrate and understand the wealth of fossils relating to human evolution that have been discovered over the past decades. In fact, fossil dates have been used to help calibrate the clock data, and as a result the mutation rate that has generally been used for humans and other apes has been about 1 mutation per year per billion bases of DNA. However, new results from massive programmes of DNA sequencing have revealed that this assumed rate is probably much higher than what has actually been taking place in our gonads.
A review by Aylwun Scally and Richard Durbin, recently published in Nature Reviews Genetics, reveals that over the last decade, nine studies have come out with substantially lower mutation rates, suggesting our clocks have been running far too fast. These studies have looked at mutation rates across the whole genome, and have focused on particular genes, including one study of over 14,000 people. They all suggest that the actual mutation rate is about half that previously estimated. To put this into perspective, a study of 78 families from Iceland (mother, father, child) found that on average, a baby has 36 spontaneous mutations that are not present in either parent. Depending on where you grew up – presence of natural radiation etc – the number of spontaneous mutations in your genes is probably not too far different.
So what happens when this new, lower, figure is plugged into our estimates of divergence times for the various twigs and branches on the tangled bank of recent human evolution? Because the clock is now thought to be ticking more slowly than we originally estimated, the divergence times are being put back. So, for example, the human/Neanderthal split was estimated at between 272,000-435,000 years ago. The new figure would appear to be something more like 400,000-600,000 years.
This removes an odd discrepancy, as previous estimates of human/Neanderthal divergence using mitochondrial DNA (mtDNA is found in the tiny mitochondria of our cells, which are inherited maternally, and which are involved only in the generation of energy) had come up with a figure 0f 500,000-600,000 years ago. So both nuclear DNA and mtDNA now give a similar estimate – we split from our Neanderthal cousins about half a million years ago. Not that that stopped us mating with them and getting some of their genes… In terms of the time when we left Africa (based on genetic differences between non-African and African populations), that figure used to be put at around 70,000 years ago; it now appears to have been substantially earlier, perhaps 90,000-130,000 years ago.
This useful table assembled by Ann Gibbons in her excellent Science magazine piece summarises the changes, and their links with the fossils:
A lot of unknowns remain – in particular the issue of estimating generation time in prehistoric populations, as well as the lack of population-level data for prehistoric groups (e.g. Neanderthals or Denisovans). But the increasing richness of molecular data are producing ever more refined estimates of our past. And that is the power of science – nothing is taken as fixed, knowledge changes and increases, in a uniquely progressive way, enabling us to revise and refine our understanding, and even to reject what we previously thought to be true. Indeed, there is grandeur in this view of life.
References (both hidden behind pay walls, sadly):
Aylwyn Scally and Richard Durbing (2012) Revising the human mutation rate: implications for the understanding human evolution. Nature Reviews Genetics 13:745-753.
Ann Gibbons (2012) Turning back the clock” slowing the pace of prehistory. Science 338:189-191.
You can also learn more by listening to the Science magazine podcast item about this.
Cool.
Although – “we split from our Neanderthal cousins about half a billion years ago.”
You meant “million,” right?
Yes. Sorry and thanks to markkoop, Larry and Bennet. It has now been corrected.
This removes an odd discrepancy, as previous estimates of human/Neanderthal divergence using mitochondrial DNA (mtDNA is found in the tiny mitochondria of our cells, which are inherited maternally, and which are involved only in the generation of energy) had come up with a figure 0f 500,000-600,000 years ago. So both nuclear DNA and mtDNA now give a similar estimate – we split from our Neanderthal cousins about half a billion years ago.
I think you mean half a million years ago.
“we split from our Neanderthal cousins about half a billion years ago.”
Cambrian hominins FTW!
I don’t think the human-chimp split can really be dated from fossils: before 4 Mya certainly, but how much before is rather a guess. However the suggested split of 34-46 Mya is AIUI not only well before when we thought humans and orangutans split, but before old world monkeys split off from the apes. So something is not right there. Does this suggest a significant slow-down in the clock somewhere between 0.6 Mya and 4 Mya?
Ann Gibbons’ table shows the dates one would estimate assuming the mutation rate (either the old or the revised one) was constant. However you are correct that the implied orang speciation is far too ancient, and one of the things we discuss is a slowdown in mutation rate during great ape evolution, to reconcile these estimates. There is some debate about how big the slowdown was and what caused it. We also discussed this in the gorilla genome paper earlier this year (http://www.nature.com/nature/journal/v483/n7388/full/nature10842.html), which is open access.
So, if the DNA clock is variable-speed…is there any hope of further calibrating it in order to figure out what rate it was running at at a particular point in time?
b&
We might be able to get (e.g. from fossils) a better estimate of how long a typical generation was at some point in the past, which is relevant to this discussion. In theory we might also be able to use ancient DNA samples to measure mutation rates within the last hundred thousand years or so – e.g. if we could extract DNA from a family of fossil individuals. But outside that timeframe it is difficult to see how we could ever measure past mutation rates directly. The uncertainty is something we are just going to have to handle in the context of evolutionary models.
Thanks — makes sense.
Sounds like, for the time being, we should accept increasingly-widening error bars on the data with increasing age, and hope for the error bars to tighten in the future…but never for them to get as tight as one might hope for.
Still, it’s an amazing tool!
And I imagine that, even when the error bars are very wide, it’s still useful for comparative analysis. That is, you might not be able to pin down with much certainty when a particular event happened, but you should still be able to be confident that it happened earlier than / later than / contemporaneously with another event.
Cheers,
b&
Thanks for that – interesting reading.
Regarding Out-of-Africa migration, the re-calibrated DNA clock is in much better agreement with both the archaeological evidence and the palaeolinguistic inference accumulated.
For the Neanderthals split, it’s not just archaeology: palaeoanthropology and geology would have to have been wrong about too many things to account for such large a discrepancy by errors or bias on their side only.
Colleagues who specialise in these older periods (I don’t) tended to keep their nerve and wait for geneticists to come around with better data and better dating. As just happened.
Not for the first time, by the way. Early, uncalibrated, radiocarbon dates tended to throw a monkey wrench into the works of consolidated relative sequences by favouring shorter chronologies. Then tree-ring calibration came about, leaving the most vocal proponents of short chronologies with egg on their face.
Whoa, this is very useful for Ardipithecus. They didn’t fit very well with chimps being only at 4.5-5 m.y. from us.
I was always uneasy with the 5 million year divergence time for Human-Chimp given by the molecular clock. I assume Scally and Durbing use theoretical fixation rates for neutral mutations. However, the new estimates seem to be based on DNA segments with few bases, such that a sample size of 14,000 may mean little in terms of statistical precision. A few studies involving unsuspected selected loci could have biased the estimate downwards. Certainly, the question is not closed.
Just an aside: it seems that Nature is an example of the left hand not knowing what the right hand is doing. Nature endorsed the ENCODE assertion that the genome is 80% ‘functional’ and then publishes the present paper whose cornerstone premise seems to be that very little of the genome is functional. One is prompted to believe that biochemists – and Nature editors — are a most heterogeneous group of the very competent and the very incompetent.
How do they find a part of the genome that they think is “selection neutral”? Are there really parts hanging around that have no sequence-dependent function? If so wouldn’t they be lost? And won’t synonymous mutations in coding regions change tRNA usage and translation efficiency and thus not really be neutral?
A few studies involving unsuspected selected loci could have biased the estimate downwards. Certainly, the question is not closed.
Just an aside: it seems that Nature is an example of the left hand not knowing what the right hand is doing. Nature endorsed the ENCODE assertion that the genome is 80% ‘functional’ and then publishes the present paper whose cornerstone premise seems to be that very little of the genome is functional.
Yeah question not closed…and seem to be more open than ever with the ENCODE results. Seems like our idea of gene conservation (and where it needs to take place) has just been lit up. Interestingly molecular clocks have always been fodder for creationists because of how…hmm unscientific it seemed (ie fossil dating/sequence pretty much determined/calibrated molecular clock values since there was little genetic data.) Now we are getting some in-depth genetic data but a long ways to go in reconciling the two disciplines. Have there been any recent studies between phyla to determine breaks from very ancient ancestors?
Yet another example of how science is guaranteed to always be worng…
…but it’s also guaranteed to also always converge on the truth.
Or, in other words, science has phenomenal accuracy, though the precision will always leave something to be desired.
Religion, on the other hand, has unbelievable precision and no accuracy whatsoever.
I’ll have to keep an eye on TimeTree to see how they incorporate these new findings….
Cheers,
b&
great post but the last bit is not accurate.
mtDNA is found in the tiny mitochondria of our cells, which are inherited maternally, and which are involved only in the generation of energy
e.g., metabolism and apoptosis.
Yes. My bad.
And that is the power of science – nothing is taken as fixed, knowledge changes and increases..
But.. so many millions of lies told at so many dinner parties.
That depends. Did your dinner-party conversation include a reasonable indication of the scope of the error bars?
Even if you happen to know the exact standard figures, it’s rhetorically not a bad idea to round everything to one or two significant figures and prepend everything with “about,” unless the precision is the point you’re trying to make.
For example, I’ll state that the universe is about a baker’s dozen billion years old; that the Sun is about a third of the age of the Universe, and that life began sometime fairly early after the Earth had cooled, probably sometime in the first half-billion years after it had coalesced. Dinosaurs died out a few score million years ago, but the last common ancestor of the mammals lived before then, at about a hundred million years ago.
There are a number of advantages. First, it saves you should you misremember or if there are later refinements. It also emphasizes the provisional nature of all such measurements, and it makes the point that it’s the scale that matters more than the precision.
If the age of the universe is later refined to be estimated at 13.62 BYA instead of 13.75 BYA, no big deal. But if it’s later revised to 5,773 years old…well, then we’ve got a problem.
Cheers,
b&
I wonder how this will all affect the story of the Great Bottleneck.
It pushes back the divergence time between African/non-Africans to before the emergence of behaviorally modern humans (without even mentioning the Khoisan!) This is making things warm for those who claim a genetic change was responsible for the behavioral change. hehehe.
You see, this is where I think I can see where the creatnists (did a really good typo there – cretinists)are coming from. I read that (and will read it properly, promise!) but the scale is too huge and immense – it’s just hard for me to understand and my eyes were glazing over about halfway down. However, I am happy to accept that there are people who know more about these things than me and I am very glad they are doing the research, even if I personally don’t quite get it. It’s a bit like my lodger who asked me how to cook a joint of gammon – his eyes were glazing over by the time I’d got to “…and then you stick the cloves into the diamonds you’ve cut into the fat”. He was happy to accept my expertise and knowledge (yes I did cook it in the end) I hope that makes sense? But whilst I think most creatonists would be happy to accept advice on cooking from me, they WON’T accept the evidence that is given to them by people who know more about evolution than they do!
Whoa, you appear to have accumulated a “trackback/pingback” from some spammer/ scammer. Quite what that means, and what if anything can be done about it, I’m not sure. But that link looks more spammy than a serving of “Spam, Spam, sausage, egg and Spam”, with extra Spam.
Sorry about the paywall at NRG – review articles tend to be that way unfortunately. There were also some more details in Ewen Callaway’s piece at http://www.nature.com/news/studies-slow-the-human-dna-clock-1.11431.
Reblogged this on Mark Solock Blog.