An Ancient, Mysterious Scrap of Human DNA

I can still remember back in 1997 being shocked that a team of scientists had managed to extract a few hundred bases of DNA from a 40,000 year old Neanderthal fossil. Neanderthal DNA! In the years that followed, scientists made huge advances in recovering ancient DNA, with the entire Neanderthal genome published in 2010. But for all that amazement, I had to learn to be resigned that scientists probably wouldn’t get human DNA older than about 100,000 years. Beyond that vintage, the DNA was just too busted up to be recoverable.

Wrong. Very wrong.

In the New York Times, In the New York Times, In the New York Times, I report on how scientists have now extracted 400,000-year-old DNA from a human fossil. That’s ten times older than the first pieces of Neanderthal DNA.

I’ll point you to the newspaper for the straight story. But there is, of course, a lot more to the tale than can fit into 1000 words. Let me lay some of the extra stuff out in Q/A form:

So, how did they manage this now?

One problem with retrieving DNA is that it fragments into tiny pieces in fossils, and the older it gets, the tinier the pieces become. But scientists have made big advances in gathering tiny DNA. Some of this is pretty straightforward, like tweaking the recipe of chemicals used to pull DNA out of bone powder so that the smallest pieces of DNA don’t get washed away with impurities. It also involves finding clever new ways to compare fragments of DNA to sequences of related humans that have already been sequenced.

Did they get the whole genome of a 400,000-year-old human?

Far from it. They only got DNA from the mitochondria of the human in question. Quick mitochondria refresher: these are blobs in our cells that generate fuel for us. They carry their own genes because they were once free-living bacteria. Each cell has hundreds of mitochondria, each with a nearly identical collection of DNA. Sperm don’t pass mitochondria to eggs in fertilization, so we all get mitochondria from our moms.

Mitochondria is an easier place to look for ancient DNA than the nucleus, where we keep the vast majority of our DNA, because every cell has hundreds of copies of mitochondrial DNA and just one of the nuclear genome. The scientists got the entire mitochondrial genome from the human fossil. But that’s just 16,000 base pairs–tiny compared to the 3.2 billion in the nuclear genome.

What’s the family tree they got from comparing mitochondrial DNA from different kinds of humans?


To make sense of this tree, let me point out a few things.

1. The branch lengths show how many mutations have accumulated on each branch. Roughly speaking, the longer a branch, the more time has passed. (Only roughly, of course–all the Africans, Asians, and Europeans are alive today, despite their different branch lengths.)

2. Denisovans are known from a single cave in Siberia, where 80,000 year old fossils have preserved their genome. They have sometimes been called “eastern Neanderthals.” They interbred with ancestors of living humans–people now found in Australia and elsewhere.

3. Sima de los Huesos humans were believed to be Neanderthal forerunners because their fossils have Neanderthal features, like a somewhat beak-shaped nose and upper jaw, as well as peculiar spaces in the back of their lower jaw.

So…what’s going on?

Hard to say at the moment!

The simplest interpretation of this tree on its own would be that Sima de los Huesos people are an early offshoot of Denisovans. The problem with that is that they are older (400,000 years) than the estimated split between Neanderthals and Denisovans (300,000). Plus, if these Denisovans were spread across Spain to Siberia, you’d think we’d have known about them before the past couple years.

So now we get into the trickier ideas–the ideas that are based on the fact that genes can take funny courses through history.

1. Imagine that the ancestors of Neanderthals and Denisovans have many different mitochondrial DNA variants. That’s natural–we humans have many variants, too. As Neanderthals and Denisovans diverge, each population acquires some of those variants. After a while, some variants just disappear. That’s natural, too. If a woman has ten sons and no daughters, her mitochondrial DNA goes extinct. Perhaps at Sima de los Huesos, scientists have found one of those ancient variants that disappeared later in Neanderthal evolution, and which survived in Denisovans. So these people have Neanderthal-like bodies, and a little DNA that is Denisovan-like.

2. Or maybe the people of Sima de los Huesos are very different from what we thought. What if they belongs to another species–the enigmatic Homo erectus that I wrote about last month in the Times? Perhaps Denisovans mated with Homo erectus in East Asia and acquired their mitochondrial DNA.

There are a number of other possibilities.

How can scientists test these possibilities?

With more DNA!

From where?

From the same fossil they got the first batch from, for starters. Their methods may not let them recover a whole nuclear genome, because they’re dealing such tiny fragments. But even if they got a million bases of nuclear DNA, that would be great. They could draw a second family tree by comparing the 400,000 year old DNA to the corresponding segments in other humans and see if the tree looks the same. It might not (we’ve seen this mismatch happen before).

Then scientists could look at the other human skeletons at the same site for more DNA.

Does this usher in a whole new age of ancient human DNA studies?

I sure hope so, but it’s worth remembering that Sima de los Huesos is a unique place. Around 30 people were buried there–perhaps in an underground graveyard. In a deep shaft, their bones were maintained at a cool temperature and damp conditions for 400,000 years. That may be required for preserving ancient DNA for so long. But even if this become our only window into early human genetic evolution, we can still learn a lot.


Harkanwal Hothi asks: How far do you think we can go in terms of explaining the past with the help of DNA fossils? Obviously, this is only going to improve with technology, as it has since 1997. But what could be the absolute limitations for our understanding of human history?

We don’t know what the absolute limitations are going to be. Svante Paabo, the scientist who sequenced that first Neanderthal DNA–and who led the latest project–has actually stated on the record in the past that we can’t go far older than 100,000 years with ancient DNA. If Paabo can be wrong, then I don’t dare set my own limits! Even if we don’t finally reach an outer limit for ancient DNA, we will still learn a lot, especially about how variations in genes flow down through lineages, and between them.

Samantha asks: Will this discovery change our views on Neanderthals in relation to the human evolution? (ie: Neanderthal’s are direct ancestors)

Not that I can see. The combined evidence from fossils and DNA suggests that Neanderthals, Denisovans, and Homo sapiens share an ancestor that lived in Africa about 500,000 years ago. Our ancestors stayed in Africa while the ancestors of Denisovans and Neanderthals moved out to Europe and Asia. Homo sapiens evolved in Africa about 200,000 years ago, and then humans expanded out of Africa 60,000 years ago, after which they interbred with Neanderthals and Denisovans. So, yes, many people on Earth today are have direct ancestors that were Neanderthals. Some have direct ancestors that were Denisovans. But in both cases, most of these people’s ancestors were Homo sapiens.

This new DNA (or, rather, this really old DNA) is from long before Neanderthals and Denisovans made contact with humans–but probably after their ancestors split from ours.

Mike Lewinski writes: I’m unsure about this point and think it would make a fascinating followup story some time:

“Sperm don’t pass mitochondria to eggs in fertilization, so we all get mitochondria from our moms.”

I’ve read in Nick Lane’s book “Oxygen: The Molecule that Made the World” that sperm probably do contribute some mitochondria, but that they’re quickly diluted out of existence because they’re old and worn out and can’t reproduce nearly as fast as maternal mitochondria (which in any case vastly outnumber them).

We do know women with mitochondrial diseases are infertile, suggesting that the paternal contribution is in any event too small and/or too damaged to sustain a growing fetus.

One benefit of this uniparental inheritance mechanism (however it is accomplished) is that the two mitochondrial lineages won’t be in conflict. Natural selection might drive their competition in ways that benefit them but are detrimental to us.

I’ve also read that they’re actively eliminated in the egg. In any case, mitochondrial DNA remains a way to track maternal inheritance.

Richard Jowsey asks: John Hawks: “The difference between Sima and Denisova [mtDNA] sequences is about as large as the difference between Neandertal and living human sequences. It would not be fair to say that Denisova and Sima represent a single population, any more than that Neandertals and living people do.” Splitter or lumper? Are Sima and Denisova different hominid “species”, but could still interbreed?

Scientists who study recent human evolution don’t like to talk firmly about species. Certainly, when I spoke to Svante Paabo and his colleagues about their discovery of the Denisovan girl, I asked them if she belonged to a new species. They wouldn’t settle on a yes or no. There’s just not enough information to really say. If reproductive barriers are how we define species, then they’re not separate species, because clearly we can interbreed. If a distinctive selective pressure on a population is our definition, then we don’t know enough about how Denisovans lived. We’ve just got a genome from a finger bone and some mitochondria DNA from a tooth. As for interbreeding, that’s made even trickier to answer because of time: the Sima people lived 400,000 years ago, and the Denisovans we know of  lived 80,000 years ago.

Steve Ferry asks: Is it fair to say that early humans/middle Pleistocene hominids were far more mobile than we thought? That they did not have ‘ranges’, east or west Eurasia for a particular group or species, but instead a particular region would have had representatives from many groups. For example Denisova had three ‘species’ of human around, perhaps not at exactly the same time.

It would be interesting to see how much genetic diversity there is in the group from Sima de Los Huesos.

No, I don’t think that this argues for more mobility than we thought. There is a 320,000 year gap between the Denisova DNA in Siberia, and the Denisova-like DNA in Spain. It is true that Denisovans, Neanderthals, and humans all used the Denisova cave, but they may have used it thousands of years apart from each other. They weren’t subletting together.

Jay Stern asks: Are we now firm in the belief that mitochondria were once free-living organisms themselves that became symbionts to multi-cellular organisms? But if that is so, what would prevent more than one species of these organisms from being taken up by multicellular organisms? In other words, why couldn’t the mitochondria in a given cell be of multiple origins and possess different DNA themselves?

The evidence is very strong that bacteria became symbionts inside the cells of our ancestors–not in our animal ancestors, but in our single-celled protozoan ancestors a couple billion years ago. This is based on their presence in all eukaryotes, and the strong similarity of their genes to certain kinds of bacteria. It is interesting that it only happened once, no doubt. Maybe it could only happen in certain circumstances, and once it did, the mitochondria-fueled eukaryotes could outcompete any new rivals. On the other hand, many species of eukaryotes have taken up other bacteria as symbionts–most spectacularly, the algae ancestors of plants, which swallowed up photosynthesizing bacteria.

Kirk Maxey asks: I’m curious why you have placed modern Africans on the top branch of your modern humanity tree, closest to Neanderthals, when one of the clear facts emerging from this story is that Africans never interbred with them. On the other hand, Eurasians have a distinct (3-5%) crosslink to the Neanderthal line, and the Denisovans are even more richly interbred with them as well. To be accurate, shouldn’t you pivot your maize bar and place it above your royal blue bar? and then, add some dotted lines between all the branches except the African one?

The tree is from the paper published this week in Nature–it’s not mine. The tree shows the relationship of mitochondrial DNA in different humans only. Africans are not any closer related to Neanderthals in their mitochondrial DNA than Europeans or Asians. All living humans share the same common ancestor when it comes to mitochondrial DNA, a common ancestor that lived after the split of the lineage that led to Homo sapiens and the one leading to Neanderthals. The authors of the paper could have shown the same results by rotating the African+Eurasian branch 180 degrees so that the Africans appear at the bottom rather than the top. When it comes to mitochondrial DNA, Eurasians are NOT closer to Neanderthals than Africans. The few percent of DNA from Neanderthals and Denisovans lie on the nuclear genome.

Richard Jowsey returns to ask: Today I was trying to explain the nuances of ancient hominid “speciation” and “varieties” to a couple of smart 15 yr-olds, while discussing your article. Yikes! Science communication is hard!! Could you please explain what you mean by “distinctive selective pressure on a population” with respect to Homo erectus evolution, and our labelling of branches on the tree?

One feature of many species is that they can’t interbreed easily with other species. But that’s not an absolute rule. Scientists often find two sets of animals that look distinct, that have different ecological niches, have been reproductively isolated for the most part, and yet sometimes interbreed and produce viable hybrids. These may be distinct species, if you look at species as groups of organisms that belong a single lineage that are adapted to the same ecological niche. See, for example, this research on fish that share a lake. Unfortunately, paleoanthropologists can’t study ancient humans like living fish in a lake.

 Lisa Strohlein asks:  Dear Carl, I read your article with great interest. Maybe my questions are a bit naive but I wonder about the following to aspects: 

1) How can scientists reconstruct the DNA sequence from all those tiny pieces being sure, that it matches the original sequence? Is it sufficient to compare them to related species?

2) Does the mitochondrial sequence say anything about the nuclear sequence? Would an analysis of nuclear sequences possibly lead to completely different results?

Not naive at all! These are the basic questions any science writer should ask scientists. They matter.

1) There are many steps to reconstructing DNA sequences from fragments. Here are a couple key ones. First is the sequencing of the fragments. Because there are so many copies of mitochondrial DNA in a single cell, scientists can get lots of fragments of mitochondrial DNA from a fossil (if they use the right chemistry). Some of those fragments will overlap. For example, you might find one fragment that runs from position #1 to position #50 in the fossil DNA sequence. Then you find another segment that runs from #25 to #75. A computer can line up those fragments based on the overlap. Then you can compare the sequence you build up to similar sequences from relatives–in this case, humans and Neanderthals and Denisovans. If the fragments have been assembled correctly, you’ll expect them to share long stretches of identical DNA.

2) The mitochondrial sequence does not say anything about the nuclear sequence–except that the two sequences had to end up in the same human. The nuclear sequence might very well produce a different tree. That’s what happened when scientists discovered the Denisovans. They started with the mitochondrial tree, which indicated Denisovans split off from the common ancestor of Neanderthals and humans (i.e., Neanderthals and humans are siblings, and Denisovans are cousins, as it were). But the nuclear tree showed Denisovans were “siblings” of Neanderthals. Both findings can be true, because each tree only shows the history of a particular chunk of DNA.

Bigfoot Party asks: What was the climate like in the area the fossils were recovered from, 400,000 years ago? Was there a land bridge at Gibraltar during this time? or was it closer to Africa? Did the Neandertal migrate through Suez, and these people cross at Gibraltar?

The climate was in the midst of the Ice Age cycles that have dominated the planet for over 2 million years. There definitely were humans in North Africa at the time (see this pdf for details). But there wasn’t a land bridge at Gilbraltar at the time. Unless humans were sailing at the time (unlikely), the evidence indicates that the common ancestors of Neanderthals and Denisovans expanded out through Sinai, as other human lineages did before and after. (See this pdf for more on this.)

David Bump asks: No matter how much biological material gets recovered, no matter how much older than anyone thought any trace at all could possibly exist, nobody is going to start wondering if maybe these samples aren’t quite so old after all, are they?

The scientists themselves are always asking whether DNA recovered from fossils is really ancient or not. Contamination with DNA is always a threat to contend with, especially when the fossils are of ancient humans, who had DNA a lot like our own. If you look at the original papers, you’ll find a mind-numbing catalog of ways in which scientists have been able to reject the possibility that ancient DNA is not so old.

 Serge D. asks: Is there evidence of continuous human migration out of africa from Homo erectus onward or is the evidence suggesting periodic “out of africa” homo species expansions? In other words, how long were the homo populations living in Eurasia isolated from homo populations in Africa (perhaps allowing for occasional gene flow)? Was there enough time for speciation by allopatry? Evidence of interbreeding seems to suggest that this was not the case. I guess these questions could be answered by looking at the nuclear DNA from many individuals of various homo species. Anyway, a very interesting article. Thanks.

It looks as if there were a series of waves of hominids coming out of Africa since the first ones 1.8 million years ago. When the ancestors of Neanderthals and Denisovans moved out about 500,000 years ago, it doesn’t seem as if there was much gene flow back into Africa. If there had been, then Neanderthals, Denisovans, and humans would share lots of young genes in common. Instead, it looks as if the only gene flow between the three came when Homo sapiens expanded out of Africa 60,000 years ago. Is 500,000 years enough for two populations to evolve into separate species? In some cases yes, in some no. It depends a lot on the biology of the animals in question.

Matt R. asks: Are there any proposed classifications for the Sima de lost Huesos remains? I’m assuming that these might be Heidelbergensis, though of course there is always disagreement for Homo classification. Would you mind explaining what species these could be? (Love your writing, and thanks for taking the time to answer so many questions!)

Are there ever! Juan-Luis Arsuaga, a co-author on the new study, has been working on the fossils for 30 years and proposes they are Homo heidelbergensis. Chris Stringer of the Natural History Museum in London has proposed they are Homo neanderthalensis.

zackoz asks: How firm is the date of 60,000 years ago for modern humans moving out of Africa?

I have a fairly vague memory of reading a few years back that humans were already in India at the time of the Toba eruption in Sumatra, which I think was about 80,000 years ago.

And I understand that some dates for human settlement in Australia are pushing up into the 50,000-60,000 ya area, too. If so, that would make a 60,000 ya date for movement out of Africa look a bit late.

As always I really appreciate all your writing, and the fascinating information about modern science that we get here.

In my article, I wrote “about 60,000 years.” It’s always a challenge to accommodate the genuine fuzziness in our understanding of human evolution without ending up leaving readers thinking that scientists don’t know a single thing about human evolution (a ploy commonly used by creationists). To find a meaningful age for the expansion out of Africa, I looked at recent reviews and found one by Chris Stringer that dates it to 60,000 years. There is some evidence of  modern humans outside of Africa as far back as about 120,000 years ago, but there’s also some debate about whether those people represented just a failed settlement or were really part of the lineage that gave rise to today’s Asians, Europeans, and people of the New World.

–I can update this post with more Q & A’s if you leave them in the comments.

Go Further

Subscriber Exclusive Content

Why are people so dang obsessed with Mars?

How viruses shape our world

The era of greyhound racing in the U.S. is coming to an end

See how people have imagined life on Mars through history

See how NASA’s new Mars rover will explore the red planet

Why are people so dang obsessed with Mars?

How viruses shape our world

The era of greyhound racing in the U.S. is coming to an end

See how people have imagined life on Mars through history

See how NASA’s new Mars rover will explore the red planet

Why are people so dang obsessed with Mars?

How viruses shape our world

The era of greyhound racing in the U.S. is coming to an end

See how people have imagined life on Mars through history

See how NASA’s new Mars rover will explore the red planet