Natural Science Forum / Biology / Paleontology / March 2007

I posted this to sci.bio.evolution when I should have posted it here. Sorry
about that. Written by Dan Jones.

Mixing with the ancestors
AFTER the boy died, he was buried in a shallow grave along with some pierced
shells and red ochre, as was customary among his people. There he lay for
24,000 years until his near-complete remains were unearthed by
anthropologist João Zilhão at Lagar Velho in Portugal. He was expecting to
find the remains of an early modern human - Neanderthals were thought to be
long extinct by that time - but the boy's skeleton was different. Realising
that he had something unusual and potentially significant on his hands,
Zilhão called in Erik Trinkaus, an expert on Stone Age humans at Washington
University in St Louis, Missouri.

In 1999, Trinkaus and Zilhão, who is at the University of Bristol in the UK,
published their analysis of the Lagar Velho child. They argued that his
bones provided the answer to a long-standing and delicate question about
human evolution: did our ancestors interbreed with Neanderthals? The child,
the team argued, was clearly a human-Neanderthal hybrid. He had the
prominent chin and facial features of a Cro-Magnon, but also the stocky body
and short legs of a Neanderthal. The only possible explanation was that he
was the product of long and extensive interbreeding between early Europeans
and the Neanderthals.

This interpretation was - and still is - controversial. While the
possibility of interbreeding between our direct ancestors and other human
species has long been recognised, there has never been much evidence to
support it. Since the discovery of the Lagar Velho child, however, new lines
of evidence have started to emerge, largely from genetics but also from new
fossils (see "Wisdom of bones"). As the findings stack up, researchers are
edging towards the conclusion that interbreeding not only happened, but that
it played an important role in our evolution. Like it or not, we may have to
accept that our species is, to some extent, a hybrid. There's a little bit
of Neanderthal in all of us.

For the past 20 years the prevailing view of the origin of modern humans has
been fairly straightforward. About 160,000 years ago a small, isolated
population of archaic humans, most likely in east Africa, evolved the
anatomical characteristics that define modern humans. According to this
"single origin" or "out of Africa" model, their descendants spread across
the globe, completely replacing existing species, such as Neanderthals and
Homo erectus, that were widespread at the time. If there was any
interbreeding, it was insignificant.

That picture replaced an earlier consensus called multiregionalism.
Multiregional theories propose that humans evolved towards modernity in a
more distributed manner, with modern human genes arising in various
sub-populations across Africa and Eurasia and then spreading throughout the
entire human population through regular breeding between these
sub-populations. Until the mid-1980s most palaeo-anthropologists were
multiregionalists, based on fossil evidence hinting at widespread, parallel
evolution towards modern forms.

Then genetic evidence entered the debate. In 1987, a team led by Allan
Wilson of the University of California, Berkeley, published an analysis of
mitochondrial DNA (mtDNA) sequences from 147 people from five geographically
distinct populations. Mitochondria are very useful for tracking evolutionary
history: their DNA passes directly down the maternal line, remaining
unchanged unless a mutation occurs. Measured over thousands of years, these
mutations occur at a regular rate, ticking like a molecular clock. Each new
mutation gives rise to a new lineage of mtDNA, like the branches on a family
tree. By analysing the mtDNA sequences of a large number of people,
geneticists can build a "gene tree", working backwards in time and
eventually converging on a common ancestor. The gene tree can also tell you
where the ancestor probably lived.

Mitochondrial Eve
The one Wilson and colleagues drew up came out strongly in favour of the
single origin model. It pointed to a recent common ancestor for all modern
humans - a single woman, the famous Mitochondrial Eve, who lived in Africa
about 170,000 years ago. Later studies on the Y chromosome, which passes
exclusively down the male line, told pretty much the same story, converging
on a single man - Y-chromosomal Adam - who lived about 100,000 years ago.
"Subsequent genetic data either backed this up or at least didn't refute
it," says Dan Garrigan, an evolutionary geneticist at Harvard University.
"By the mid-1990s the 'out of Africa' view had become the dominant view of
human evolution," adds Chris Stringer, a palaeontologist at the Natural
History Museum, London, and an early proponent of the model.

The story told by mtDNA and the Y chromosome supports the single origin
model, but these are not the only source of genetic information about
patterns of human evolution. In terms of size, the nuclear genome dwarfs
mtDNA and the Y chromosome, making it a potentially richer resource for
reconstructing human history.

Nuclear DNA is harder to work with, though. Unlike mtDNA or the Y
chromosome, which are both passed down intact, the nuclear genome is chopped
u p and recombined into novel combinations every generation. This genetic
shuffling makes it very difficult to build gene family trees: you can't be
sure whether sequence differences arose through shuffling or mutation. For a
long time that made it all but impossible to derive information on
evolutionary history from nuclear DNA.

In recent years those hurdles have been overcome. It turns out that there
are small chunks of nuclear DNA called haplotypes that tend not to be broken
up by recombination, and so, like mtDNA, pass from generation to generation
intact and can be used to build gene trees. In recent years sequencing
technology, and the computational tools for analysing sequence data, have
improved to the point where haplotypes can provide useful evidence about
human history - evidence that is at odds with the single origin model.
"There are patterns of variation in the genome that don't really fit," says
Michael Hammer, an evolutionary geneticist at the University of Arizona in
Tucson.

The first such odd pattern was discovered in the late 1990s, when
anthropologist Eugene Harris and geneticist Jody Hey at Rutgers University
in Piscataway, New Jersey, looked at a haplotype within a gene called PDHA1.
By sequencing DNA samples taken from 35 men across the world, they found
that there were several versions of this haplotype in the modern population.
So far, so unsurprising. But when Harris and Hey constructed a gene tree for
the sequences, something stood out.

They found that the sequences could be clumped into two basic types, or
lineages, which last shared a common ancestor a whopping 1.8 million years
ago. Then 200,000 years ago one of the lineages split again (Proceedings of
the National Academy of Sciences, vol 96, p 3320). But if humans evolved
from a small, reproductively isolated group about 160,000 years ago, how
could the PDHA1 haplotype have diverged 1.8 million years ago, and again
200,000 years ago? "The pattern is completely incompatible with a model in
which modern humans derive from a single population," says Garrigan.

In the parlance of population genetics, PDHA1 shows "deep ancestry". This
poses a big problem for the single origin model. If the model is correct,
all our genes should converge on a single common ancestor who lived fairly
recently - that is, they should show shallow ancestry. On the whole, they
do. But PDHA1 does not, and it isn't alone. "We're repeatedly finding
genetic lineages with deep ancestry that stick out from other areas of the
genome," says Sarah Tishkoff, an evolutionary geneticist at the University
of Maryland in College Park. "The tough part is explaining these patterns."

One solution is to revive the multiregional model, which Harris and Hey
proposed doing. But there is another, more dramatic explanation:
interbreeding. In this model, modern humans did evolve from a single
population in Africa, but occasionally acquired genes from other human
species by having sex with them.

Interbreeding would explain why our genome contains some chunks of DNA with
deep ancestry: they evolved in archaic species and "introgressed" into us.
If that's true then we are, to some extent, a hybrid species - a mosaic of
"our" genes, Neanderthal genes and possibly even Homo erectus genes too.

To some that's a step too far. Surely our direct ancestors would not have
been remotely interested in inter-species sex. And even if they were, what
are the chances of such dalliances producing viable, fertile offspring? Many
experts, however, think human-Neanderthal mixing would have been entirely
possible. "They were very closely related, so there could be interbreeding,"
says Stringer, even though he thinks the biological significance of this is
likely to be low.

"They were very closely related, so there could have been interbreeding"
Until recently, the available evidence suggested that there was no
interbreeding. All Neanderthal mtDNA genomes sequenced so far are distinct
from our mtDNA. But that still left plenty of scope for finding introgressed
genes in the nuclear genome. Last year, dramatic and compelling evidence
emerged for this type of gene flow.

For the past few years Bruce Lahn, a geneticist at the University of
Chicago, has been studying genes potentially involved in human cognition, in
particular one called microcephalin. Mutations in microcephalin cause the
condition microcephaly, characterised by a small head and various
neurological symptoms.

Like many genes involved with brain development, microcephalin has evolved
rapidly in humans. In previous studies, Lahn showed that one variant of
microcephalin appeared about 40,000 years ago and has since swept through
the population, propelled by the power of natural selection. The new variant
is found in 70 per cent of living people. "We don't yet know exactly what
this variant does or why it is being selected for - it could be something to
do with cognition," says Lahn.

The obvious interpretation is that the new version arose 40,000 years ago
via a chance mutation in the microcephalin gene. Lahn thinks otherwise. In a
paper published last year, he looked at a haplotype within microcephalin. On
the basis of sequence differences between the old and new versions of the
gene, he concluded that the two are so different that they must have
diverged at least 1 million years ago (Proceedings of the National Academy
of Sciences, vol 103, p 18178).

This combination of deep ancestry on one level and shallow ancestry on
another suggests that something very unusual might have happened. It is as
if the new version of microcephalin split off from our evolutionary lineage
a million years ago, then jumped back in 40,000 years ago. According to
Lahn, that is exactly what happened. By far the most likely explanation, he
says, is that the newer version of the gene evolved in a separate species of
human - probably Neanderthals - and then entered our lineage through
interbreeding.

"These dates roughly correspond to human-Neanderthal divergence 1 million
years ago, and the time when they coexisted in Europe 40,000 years ago,
which naturally leads to the hypothesis that the new microcephalin gene
introgressed from Neanderthals to humans," says Lahn. "Once in the human
gene pool, the new variant was selectively favoured and now represents about
70 per cent of the worldwide frequency." In this case multiregionalism
cannot explain the pattern: the gene is so strongly favoured by natural
selection that if it arose in a subpopulation of humans that was in regular
sexual contact with others it would have spread throughout our lineage much
earlier.

There's an irony here. If Lahn is right, a gene potentially underpinning the
power of the modern human brain originally arose in Neanderthals, popularly
portrayed as our intellectual inferiors. With the Neanderthal genome
expected within two years we may have confirmation of this introgression.

Microcephalin and PDHA1 are hardly anomalies. "These are just two of a
growing list of regions of the genome that do not fit with a strictly single
origin model," says Hammer, whose lab has found several other cases and is
searching for more.

Hammer is taking a different tack from Lahn. Instead of looking at genes
like microcephalin, Hammer is concentrating on haplotypes in non-coding, or
neutral, regions of the genome - "junk" DNA that can accumulate mutations
without any biological effect.

The reason for taking this approach is to move the introgression story
another step forward. A gene like microcephalin can tell you that
interbreeding probably happened, but it can't tell you how often. Because it
has been strongly selected, microcephalin could have entered the human
population from a single copy that introgressed 40,000 years ago. In other
words, its presence could in theory be the result of the only
human-Neanderthal sexual encounter ever.

Neutral regions, by contrast, are much more informative about how much sex
our ancestors had with archaic Homo species. Natural selection is blind to
these regions, so their frequency in the gene pool drifts up and down by
pure chance. Any introgressed sequences will be few in number, and the vast
majority will at some point drop out of the gene pool altogether. Just a
few, however, will win the genetic lottery and persist in modern humans. The
more interbreeding occurred, the more introgressed neutral regions remain.

Hammer's group has already found several neutral regions that look like they
are introgressions. "It doesn't seem that it was a particularly rare event -
it looks like it's happening enough that neutral regions can introgress into
the genome and persist in modern populations," says Hammer.

One example even tells a possible story of interbreeding between humans and
an even more distant ancestor, Homo erectus. The pseudogene RRM2P4 - a
remnant of a now-defunct gene - shows even deeper deep ancestry than PDHA1.
RRM2P4 comes in two basic types that diverged 2 million years ago - around
the same time that Homo erectus first moved out of Africa into Asia.
Crucially, one type is found almost exclusively in people of east Asian
origin. According to Hammer and Garrigan, the most likely explanation for
this deep ancestry and geographical distribution is that the pseudogene
evolved in the Asian branch of Homo erectus and introgressed into Homo
sapiens (Molecular Biology and Evolution, vol 22, p 189). That event is
certainly not ruled out by the fossil record: recent finds suggest that Homo
sapiens and Homo erectus coexisted in Asia for several thousand years (see
Map).

While most biologists accept that interbreeding was possible, introgression
is not the only way to explain patterns in the genome that don't fit in with
the single-origin view. Multiregionalism is one alternative. Another is that
natural selection has acted on what we wrongly believed are neutral regions
in the genome, distorting the frequency and distribution of genes across the
globe.

"We have a lot of different models," says Tishkoff at the University of
Maryland. "The real challenge is trying to distinguish between them." Hammer
is also wary of jumping to conclusions. "When we find a region of the genome
that shows this pattern of introgression we really have to argue that the
pattern didn't arise by some form of selection, which might also produce
similar patterns."

Even so, a broad consensus seems to be emerging about our ancestry, and it
includes interbreeding as an important element. "There was a great genetic
contribution from one African population, but the genetic material that
existed in other localised archaic populations was not lost forever - it was
integrated into the modern human genome," says Garrigan. Trinkaus, who has
long argued that humans picked up genes from other archaic humans, sees a
similar picture. The extremes of single origin on the one hand and global
multiregionalism on the other are "intellectually passé," he says. "The
basic model is 'out of Africa' - with admixture. The issue is how much,
where, and when."

"Genetic material that existed in archaic populations was not lost forever"
As always in science, the answer to those questions lies in gathering more
data. With the advent of $1000 genome sequencing, predicted to be a reality
within five years, it will be possible to sequence vastly more genomes than
are available today. Researchers can then seek a complete picture of the
puzzling patterns of ancestry locked away in our genomes. Then, at last, we
may know whether the Lagar Velho child was part of a hybrid population
heading down an evolutionary dead end, or an ancient reminder of the
Neanderthal in all of us.

Dan Jones is a science writer based in Brighton, UK
From issue 2593 of New Scientist magazine, 03 March 2007, page 28-32
Wisdom of bones
The Lagar Velho child unearthed in Portugal isn't the only skeleton that has
been identified as a possible human-Neanderthal hybrid. In the past few
years Erik Trinkaus of Washington University in St Louis, Missouri, has
amassed more fossil evidence that he says tells the same story.

In 2002, a team of cavers discovered a human jawbone in a cave called
Pestera cu Oase ("cave with bones") in south-west Romania. Carbon dating put
the remains at about 40,000 years old, which made it the earliest
unambiguous modern human specimen found in Europe.

Even a single bone can contain features characteristic of either
Neanderthals or modern humans. According to Trinkaus, the Oase 1 jawbone has
a mixture of both. Though less dramatically a hybrid than the Lagar Velho
child, the find still suggests interbreeding.

Further exploration of Pestera cu Oase has yielded even greater treasures.
In January, Trinkaus and colleagues described a human skull which they
called Oase 2 (Proceedings of the National Academy of Sciences, vol 104, p
1165). This seems to be the remains of an adolescent who also died about
40,000 years ago and, like the Oase 1 sample, has a mixture of modern and
Neanderthal features (see below).

Trinkaus has also reanalysed some 35,000-year-old human bones discovered in
1952 at another site in Romania, Pestera Muierii, and says that these too
show a mosaic of features.

To Trinkaus, these finds paint a fairly clear picture of the evolution of
humans in Europe. "Early European humans are basically modern - their
anatomy is overwhelmingly like that of the ancestral African population -
but in individual specimens you find features that are absent from or have
already been lost from the ancestral African group," says Trinkaus. "By far
the easiest way to explain this is through interbreeding."

This is not to suggest that the Lagar Velho boy or the Romanian specimens
are the product of occasional, one-off meetings between Neanderthals and
Cro-Magnons. Trinkaus suggests a more radical notion: the hybrids come from
a population of humans that regularly interbred with Neanderthals. In other
words, they are the result of generations of sex between Neanderthals and
Cro-Magnons.

Not everyone agrees with his argument. Human ancestry, like beauty, is in
the eye of the beholder. For instance, Chris Stringer, a palaeontologist at
the Natural History Museum in London, is not convinced that the Lagar Velho
boy is evidence of hybridisation. "In many respects, including face and
teeth, it's a modern human; the only place where it might look archaic is in
the body proportions, but to me they overlap with those of other modern
humans," he says. "I just don't see the Neanderthal influence that Erik
does."

Stringer believes more fossil evidence is required. "When we have a
reasonable sample of early moderns dating from the same time period as the
main sample of Neanderthals in Europe - 40,000 to 70,000 years ago - from
regions such as western Asia or north Africa, then we will be able to see
what their morphology was and will be able to better determine whether
features have come from Neanderthal admixture."