I was going to wait to post this until I’d sorted out my issues with primates, evolutionary trees and australopithecines, so I could post the topics in order. But actually I don’t think it matters if you read about Neanderthals first. So here we go.
Over the last decade, techniques for generating and analysing genetic data have sophisticated at an exponential rate. 2001 marked the completion of the Human Genome Project, which used a technique called Sanger sequencing. Since then, the cost of sequencing a single genome (ie. having every one of its ‘A’s, ‘T’s, ‘C’s, and ‘G’s read off by a machine and put into a computer) has plummetted from $100 million to less than $10,000, and continues to fall as next-generation, high-througput sequencing technology improves. As you can see, the process massively outpaced Moore’s law for a time.
The onset of the genomic era we now live in changed the face of evolutionary anthropology by enabling direct comparison of the genomes of humans and other primates via the principle of the molecular clock. The molecular clock relies on the fact that some parts of species’ genomes are non-functional and so not under selective pressure, with mutations to those areas not causing a reduction in reproductive success and therefore being free to accumulate changes willy-nilly. If you know a species’ mutation rate, then the number of nucleotide- (DNA ‘letter’-) differences between two species’ genomes can be used to calculate how long ago they diverged.
Svante Pääbo and ancient DNA
As genomics gathered momentum, it precipitated the development of a swathe of new practical and statistical methods for analysing ancient DNA which, unlike the DNA of extant (living) species, has invariably undergone significant degradation. These methods, as well as techniques used for actually extracting DNA from fossils, were largely devised by Svante Pääbo and his team at Max Planck institute in Leipzig. He gave a talk in Oxford the other night. Afterwards, there was a wine reception. I assumed I’d have no chance to talk to him because he’d be constantly surrounded by swarms of people (all of them taller than me). But perhaps people were intimidated, because when I looked around for someone to stand next to, I found him behind me, by himself! I introduced myself and then bombarded him with questions. Managed to keep him for at least 15 minutes. HE GAVE ME HIS CARD. I think I felt even more fame-struck than I did when I met Richard Dawkins, who has a certain homeliness about him. Have a listen to him speaking – he’s SO cool.
But I digress.
For the rest of this post, I’m going to talk about how genomics, and particularly ancient DNA analysis, has revolutionised our understanding of human evolution. You should bear in mind that sometimes I’ll talk in terms of individual genomes (“a” genome), which across humans (for example) only differ, on average, by 0.1%, and other times I’ll talk in terms of “the” genome of a species. The human genome differs from the chimpanzee genome, on average, by 1%.
Mummies, which look beautifully preserved on the surface, and are often ‘only’ around 10,000 years old, normally no longer contain any useful genetic information because heat has caused the DNA to fragment completely, leaving the molecular equivalent of piles of four-letter alphabet pasta. Over time, that’s what happens to DNA – it gets fragmented. Under cold enough, dry enough conditions, however, this process is slowed down. In well-preserved Nanderthal bones (semi-fossilzed ones), which are around a couple of hundred thousand years old, fragments can be a few hundred letters long. Since all cells contain copies of the genome, and each copy fragments differently (by chance), sequencing fragments from as many cells as possible allows them to be put in order again, by looking for stretches of sequence that are shared in different fragments, and matching them up. I hope the diagram helps you understand this.
Very old fossils don’t provide any DNA at all, because all the biological material has been replaced with minerals via permineralisation.
The making of Homo sapiens
One of the most compelling motivations behind the development of new methods for analysing ancient DNA was settling the debate surrounding admixture (inter-breeding) between different species of human during Homo sapiens‘ colonisation of the globe. The picture that is emerging is one that to some extent bridges the gap between two competing models of human expansion.
The first of these models is popularly known as ‘Out of Africa’. In the text-books it’s called the ‘Recent Single Origin’, or ‘Replacement’ model. In a nutshell, the RSO model says that humans became anatomically modern in Africa by about 200,000 years ago (200 KYA), and then a subset of this African population left the motherland, and wiped out –ie, replaced rather than merged with– all the other hominin species (Homo erectus, Neanderthals, Denisovans, and most probably others) populating the rest of the Old World at the time. The ousted hominins are widely considered to be descendents of Homo erectus, who in turn are widely considered to have descended from Homo ergaster, a hominin species which began to breach Africa’s borders at around 2MYA (ergaster’s extra-Africa fossil record starts in Southern Eurasia at 1.75 MYA).
The alternative is known as the ‘multiregional evolution model’ which, much to the chagrin of its proponents, is often conflated with an older and now entirely untenable model described by the ‘polycentric theory of human origins’. The multiregional model sees recent human evolution as being continuous across the globe from about 2 million years ago (2 MYA). It doesn’t delineate between Homo erectus, (the putative ancestor of ) H. heidelbergensis, (the putative ancestor of both) H. neanderthalensis and Denisovans (more on them later), and instead sees all of these as one species (sapiens). It has this species falling roughly into four fuzzy-edged categories (morphological clades) existing, and evolving, across time, in different areas of the world, with constant but low-level gene-flow (interbreeding) between each clade (see below). This model is proposed to explain persistant regional differences between human populations, which proponents put down to the maintenance of adaptedness to local climes. So, for instance, Chinese Homo erectus has been argued to have the same high cheekbones and flat faces as Oriental populations today, whereas Javanese Homo erectus fossils have more protruding faces and robustly-built cheekbones, argued to correspond with Australian Aborigines. The proposed maintenance of region-specific variation through time is called “regional continuity”.
Below is a schematic diagram of the multiregional model. The dotted lines represent gene-flow – a process which might be described as the key distinction between multiregionalism and the aforementioned ‘polycentric theory of human evolution’, which
proposed that populations in each region evolved in isolation (ie, without interbreeding). The polycentric theory would predict significant genetic divergence between different races because, as I mentioned in relation to the molecular clock, when populations are left to evolve in isolation from one another, they accumulate different mutations (by chance). The prediction doesn’t hold up – different races are strikingly similar genetically.
Mitochondrial DNA (mtDNA)
…is a molecule (genomic DNA comes in two sets of 23 molecules called chromosomes; mtDNA comes in one, circular molecule) that has various interesting properties. First, it mutates at a faster rate than genomic DNA, making it useful for molecular clocking over short time periods. Second, it is passed down the maternal line. It is also much smaller than the nuclear genome, containing ~16,000 base pairs as opposed to ~6 billion.
In 1987, the first large-scale study of human mtDNA in human populations, conducted by Rebecca Cann and Alan Wilson, showed that it contained very little variation – only about a tenth of that found amongst chimps. Moreover, genetic divergence, it was shown, tends to be larger between two individuals taken from a single population than it is, on average, between populations. This finding was curtains for the polycentric theory of evolution (which if you remember proposed completely separate evolutionary histories for each race of human as of about 2 MYA, right up until modern times).
“Lucky Mother” and Human mtDNA
You will have heard of the phrase “mitochondrial Eve”, which is the pop name given (“regrettably”, according to Wilson) to the most recent maternal common ancestor, or “lucky mother” of all living human beings.
Coalscent theory is a mathematical framework for working backwards in time to trace a gene and all its alleles (variants) in a population back to one copy, from which all current copies are descended. This 1987 study placed mitochondrial DNA coalescence (“Eve”) in Africa, around 140-290 KYA, which initially seemed like terrible news for the multiregional hypothesis, not because of the Africa component (both it and the SRO model place human evolution in Africa), but because of the date – remember that the multiregional model has our species evolving 2 MYA (which would be when our common ancestor lived).
Another thing about the mitochondrial genome is that each cell contains thousands of copies of it, widening its odds of successful extraction from fossils. In 1997, a fragment of mitochondrial DNA (mtDNA) from a Neanderthal specimen was successfully extracted and sequenced (by my homie Svante), at great cost, using what was then the absolute state of the art in genetic sequencing technology. Comparison with the corresponding section of human mtDNA supported the idea that Neanderthals were significantly different from us genetically – something which had, until then, to be inferred. Since then, nine more partial Neanderthal mtDNA sequences and six complete Neanderthal mtDNA genomes have corroborated this distinctiveness, which falls well outside the range of human variation in modern populations. This was bad news for the multiregional model, which, remember, sees all recent hominins as one continuous species.
The more work that has been done on evolutionary genetics, the more complicated the picture has started to look. In the 90s, re-analysis of the Eve data showed that it didn’t justify the cut-and-dry conclusion that humanity’s lucky mother was an African. The same data can be used to produce plausible evolutionary trees that start outside of Africa. And, more importantly, whether she was Asian or African, “mitochondrial Eve” is not the same as “last common ancestor”. She would not have been the only female living at the time, or the only female to have contributed nuclear (genomic) DNA to current humans.
All subsequent higher-resolution studies of mitochondria have continued to point towards a single, recent origin in Africa but, crucially, it is now appreciated that different sections of our DNA can have wildly different evolutionary histories. Eve’s provenance and age tell us nothing about where and when the rest of our genomic DNA (which gets constantly cut up into pieces and shuffled about and is orders of magnitude more vast) originates. Our genomes are like mosaics, with each piece having made its own, unique journey to get here. These pieces don’t tell us about population history, they tell us specifically about their own histories. Mitochondrial DNA, since it doesn’t get cut up or shuffled about, and is simply passed down in full, can only tell us as much about the history of our species as any other single mosaic piece, which is really not a lot. We also have Eve-equivalents for every other gene in our genome, and they will all have existed at different points in our evolutionary history. I know – head-spinny.
To summarise this:
MtDNA studies have pointed towards a recent African origin, but they couldn’t conclusively reject multiregional evolution, although they did throw it into some doubt, particularly in how divergent Neanderthal and human mitochondrial DNA turned out to be. If there had been some level of constant gene-flow between them, we’d probably expect them to be more similar.
No Neanderthal mtDNA in humans
Although it seems very likely that humans and Neanderthals crossed paths (particularly since bones of both species have been found in the same cave, in the same sediment layer), this wave of mtDNA studies didn’t find a single Neanderthal mitochondrion in any human population sampled, which suggested that Neanderthals didn’t make any mtDNA contribution to our species.
An important corollary of mtDNA being passed down in full, rather than being chopped up and shuffled, is that it makes them highly susceptible to genetic drift, which I think translates quite well as “getting completely lost by accident”, for the purposes of this post, but can be thought of more generally as random sampling error. Small sections of nuclear (non-mitochondrial) DNA (of an enormous 3 billion bits-worth of initial input) may remain in a population even if most of it is lost over time. Perhaps there were mtDNA contributions, that have just been lost.
Furthermore, if there had been a reproductive bias, such that Neanderthal men were happy to sleep with human women, but Neanderthal women weren’t interested in human men (perhaps too puny), then we shouldn’t expect mtDNA to show up in human populations anyway. For it to have made it into modern humans, we’d have to have had neanderthal women sleeping with sapiens men AND the offspring of those pairings then being raised in their fathers‘ tribes (as humans), AND for it to have been lucky enough to happen to make it into every subsequent generation.
Here are some amazing reconstructions of different hominins, including Neanderthals. Why don’t you see whether you think you would’ve.
Combining coalescence times for multiple regions of DNA
Because each piece of our DNA mosaic has its own evolutionary history, a better estimate of when our most recent common ancestor lived can be made on the basis of combining coalescence (emergence) times for different genes and looking at where these cluster. By 1997, 14 coalescence times from across the entire human gene-pool had been calculated. 9 of these clustered around 200KYA, with the rest scattered at 0.5, 1.2, 1.3, 3.0 and 3.5 MYA.
All evidence was pointing away from a fully multiregional model. However, a simple version of Out of Africa — an exclusively single-origins, outright replacement model — was becoming untenable too.
Breeding the two models: genetic admixture revealed
In 2006, two research teams became the first to sequence portions of the Neanderthal genome. One, Noonan et al, sequenced 60,000 base pairs, and the other, Green et al (Svante’s team), sequenced a million. The latter study suffered from very high levels of contamination from modern human DNA – a problem I won’t go into here in any depth, but that can occur at any stage from discovery of a fossil to manipulation in the lab, and is almost impossible to completely prevent, since all humans are surrounded by a veritable cloud of their own DNA.
Over the next few years, better statistical methods were developed for identifying contamination and eliminating it from analysis and, by 2010, a draft sequence of the Neanderthal genome –from Leipzig– had been published that was estimated to include less than 1% contamination.
Analysis showed that between 1 and 4% of the human genome had been derived from Neanderthals, in non-Africans. This is precisely what would be predicted if admixture had occurred between Neanderthals and humans who had left Africa. In his lecture, Svante said that this was a huge shock to him, and that he had felt almost certain that no evidence of admixture would be found.
The analysis also a couple of large sections of the human genome that contained no Neanderthal DNA (“deserts”), suggesting that inheriting Neanderthal genes at those positions, as a human, had a significantly deleterious effect on survival or reproductive prowess, and was therefore under strong negative selection. Chromosomal crossover (shuffling of genes) during sexual reproduction is a random process – disproportionately large untouched sections are not predicted under random shuffing – unless some shuffle outcomes are systematically eliminated after the event. It turns out that one of these deserts is the home of genes coding for testical function. The implication is that Neanderthal testical genes don’t work properly in the context of human DNA, and those unlucky to have inherited the unfortunate combination died childless.
In case you can’t picture what I mean by ‘shuffling’, here’s a very quick video. Imagine one of the chromosomes comes from a Neanderthal, and the other comes from a human.
Since 2010, two complete Neanderthal genomes have been sequenced to very high quality, and the current best estimate of the extent of human DNA derived from Neanderthals is 1.5-2.5% in non-Africans.
Interestingly, the Northern African genome (but not the sub-Saharan genome) has been found to contain some signatures of Neanderthal DNA as well, though at a lower level than non-African populations. The most plausible interpretation is that, subsequent to Homo sapiens‘ migration out of Africa and into Neanderthal territory, a portion of them (well, their descendents, that is) went back into Africa, taking some Neanderthal genes with them. Back-migration is something Iain Morley (one of the course supervisors) tells us gets disproportionately little attention, probably because it’s less exciting than the idea of going forth and colonising virgin territory.
In 2008, a fragment of hominin finger-bone was found in a cave in Siberia. Just think how skilled the archaeologists doing the excavation must have been to recognise it as human:
The cave was named “Denisova”, because… a guy called Denis lived in there! (LOL.) DNA extracted from the bone had been preserved exceptionally well and, by 2010, a high-quality nuclear and mitochondrial Denisovan genome had been sequenced. By the beginning of this year, the Denisovan genome sequence had an average coverage of 50 – meaning that each base pair had been read on average 50 times, which is exceptionally good – the more redundancy, the higher the confidence in the analyses produced, since the process of reading off individual nucleotide bases (DNA “letters”) has an inherent error-rate.
In that same cave, a Neanderthal toe-bone was also found, in the same sediment layer, indicating that Denisovans and Neanderthals co-existed. This is reflected in comparative analysis of Neanderthal and Denisovan DNA – the Denisovan genome includes about 0.5% Neanderthal-derived material.
Intriguingly, comparison of the Denisovan genome with that of modern humans (by Svante Pääbo’s team again) shows significant contributions to Melanesian populations, with 4-6% of the modern Melanesian genome being identified as derived from Denisovans. A very small amount (about 0.2%) of Denisovan DNA is also found in Asian genomes, but this is probably the result of subsequent gene-flow from modern Melanesians to Asians.
Research on the Denisovan genome also seems to point to genetic input from another archaic hominin – a “ghost species” that has left its footprint on Denisovan DNA. The most likely candidate at the moment seems to be Homo erectus, whose fossil record spans from 1.3 MYA to about 143 KYA across wide geographical range from Africa throughout Asia.
Although it’s far from the final word, the combined, simmered down ingredients of fossils, genetics and inference seem to suggest that around 800 KYA, a sub-sample of the population of hominins ancestral to Neanderthals, Denisovans and human beings (most probably Homo heidelbergensis), branched away, moving up and out of Africa. A couple of hundred thousand years later, this branch then split again, with one sub-branch heading northwest into Europe and the West of Asia, and the other heading eastwards into East Asia, and evolving into Denisovans, with the two branches overlapping in time and space to an unknown degree. Then, some time around 70 KYA, Homo sapiens, having evolved, in the meantime, from Homo heidelbergensis (and after staying put for about 130 KY since then), started to emerge from Africa, whereupon it began to encounter its long lost cousins. Who knows under what circumstances, but these encounters led to Humanderthals and Denisapiens, some of whom managed to get their hybrid DNA into modern human history – a small consolation for the extinction of their own kinds. Here’s a diagram to help you visualise all this. I borrowed it from a 2010 article in Nature called “Human Origins: Shadows of early migrations”, by Bustamente and Henn.
If there’s one thing I hope to have put across here, vaguely at least, it’s that evolutionary genetics is frustratingly complicated but totally fascinating!