What evidential role should (ancient) DNA play in paleoanthropology?

Guest blogger Helen De Cruz writes...

Of all the paleontological subdisciplines, paleoanthropology is both the most controversial and the most prestigious. It taps into our deep desire to know about our origins. Paleoanthropological studies frequently find their way in the most prominent journals such as Science and Nature. As Tim White observed, “No suid skulls grace the covers of Nature or garner headlines like “new pig skull completely overturns all previous theories of pig evolution.”

In spite of its prestige, paleoanthropology is fraught with long-standing debates that, until recently, seemed to be unresolvable. Did the first hominins (human ancestors after the split with chimpanzees) originate in Africa or Asia? Did modern humans originate from a large population of European, Asian and African Homo erectus mixing their genes, or is a more isolated, recent African origin more likely?  Did anatomically modern humans and Neanderthals interbreed? When did humans first arrive in Australia or the Americas? Eyeballing the fossils does not seem to settle these questions. The bones tell different stories depending the background beliefs, whether you’re a lumperor a splitter, a Multiregionalist or an Out-of-Africa proponent.

DNA extracted from living humans and fossils offers, on the face of it, a decisive way out of these impasses. In this blogpost I will examine what sort of evidence DNA, especially ancient DNA might be, and whether it deserves the privileged evidential status it currently enjoys.

Prior to 1987, both the chronology (millions of years, or just a few hundred thousand) and the origins (Africa or Asia) of our species were a continued matter of debate. That year, Cann et al’s study of mtDNA in Nature came as a dramatic intervention. Doctoral student Rebecca Cann had been collecting mtDNA samples from women of different ethnic backgrounds. Given the limited methods of DNA extraction at the time, a lot of material was needed: the mtDNA was extracted from placentas across US hospitals. Cann was unable to obtain mtDNA from African women, so she relied on African American women. The study made big waves: showing a relatively recent origin of mtDNA in Africa, with a most recent common ancestor of humans in the maternal line at around 200,000 years ago. H. sapiens arose from a relatively isolated population in Africa: multiregionalism was out.

This demonstrates the power of DNA collected from living humans in reconstructing our past. More recently, labs have succeeded in extracting mtDNA and nuclear DNA from fossils, so-called ancient DNA. Climatological circumstances need to be right (cold and dry), and DNA degrades over time (sorry, no dinosaur DNA). Even in samples that are useful, ancient DNA tends to be highly fragmented. In spite of these limitations, extractions of ancient DNA have led to many spectacular results, such as a draft sequence of the Neanderthal genome. It has also brought to light ancient features paleoanthropologists never hoped to uncover such as hair color and blood types. The debate about Neanderthal-human interbreeding seems to have been settled by the discovery of significant (but relatively small) introgressions of Neanderthal DNA in the human nuclear genome.

Further, ancient DNA has helped us identify an entirely unknown, entirely new hominin clade on the basis of genes sequenced from a pinky bone of a girl found in Denisova cave (Siberia), dated to about 40,000 years ago. These hominins, provisionally called “Denisovans” were more closely related to Neanderthals than to Homo sapiens and co-existed with both species. It turns out people in Oceania, and to a lesser extent in South Asia, have some Denisovan DNA, while in south East Asia and Melanesia (Oceania) small amounts of Denisovan DNA are found. Without ancient DNA, it is hard to imagine how paleoanthropologists could have realized there was another hominin species in the late Pleistocene, since all the physical remains we have of the Denisovans are a couple of teeth and the pinky bone. 

Resurrecting the genome of an entire species from a few scraps of bone would have seemed like pure science fiction just a few decades ago, but consider this: ancient DNA can potentially uncover species for which there is no direct fossil evidence. In the genome of Denisovans we find tiny traces, about  0.5-8% of their DNA, that could be of yet another hominin species, which interbred with the Denisovans about 1.1 and 4 million years ago. Given the limited successes in recovering ancient DNA from older samples (the oldest successful recovery is from a horse-like animal that lived 700,000 years ago, and the oldest hominin that yielded ancient DNA is a 400,000 year old individual from Sima de los Huesos, see below), we may never find out what that other hominin species was. This may point to a worrying disconnect between the ancient DNA data and the data paleoanthropologists traditionally rely upon, such as morphological analysis of fossils, analysis of archaeological data, or archaeo-botanical reconstructions of the environment.

Ann Horsburgh writes, “there has been a reversal in the place of genetic data in that it is now privileged over other sources of data. This kind of molecular chauvinism leads to overreach in interpretation and is no less likely to hamper our progress. Moving forward we would do best be judicial in the use of genetic data alongside other independent archaeological evidence in reconstructing the past.” (here).

One illustration of this molecular chauvinism is the interpretation of mitochondrial ancient DNA of a hominin femur from Sima de los Huesos, Spain, dated to approximately 400,000 years ago. The Sima de los Huesos is a rich source of hominin fossils. These fossils are traditionally regarded as a link between an older human population that came into Europe, Homo heidelbergensis (somewhere between 500 and 800,000 years ago) and the more recent Homo neanderthalensis. Initially, the mtDNA suggested that the Sima de los Huesos hominins were more closely related to Denisovans than to Neanderthals, whereas fossil evidence would lead us to expect a close affinity to Neanderthals.  This discovery overturned previous ideas about human evolution in Europe, ”baffling experts“, as the Nature News commentary piece puts it. The archaeologist Clive Finlayson, renowned Neanderthal expert, called the finding “sobering and refreshing”, and added, “The genetics, to me, don’t lie”. 

Or do they? A more recent sequencing of nuclear DNA reveals a different picture. As Meyer et al observe, “mitochondrial DNA does not reveal the full picture of relationships among populations”, thus nuclear DNA can provide more clues. They found that Sima de los Huesos hominins are more closely related to Neanderthals than to Denisovans, showing in fact that they are “early Neanderthals” which is what traditional paleoanthropological methods support (morphological similarity, continuities in material culture, etc.)  The anomalous mtDNA is explained by Meyer et al as either an introgression of some other hominin interbreeding, or mtDNA in Late Pleistocene Neanderthals as the result of gene flow from Africa (for which there is now also ancient DNA support).

This story suggests that caution is necessary with ancient DNA. Ancient DNA is well-suited for drawing up family trees. Gene flow histories are messy. There are lots of different ways you could draw the same cladogram, depending on background assumptions. While ancient DNA could provide a piece of evidence that could prompt us to change some elements in the cladogram, they may also show that some background assumptions need to be re-evaluated. Thus,  we need not redraw family trees each time there is an anomalous result. Ancient DNA may even be less suitable for uncovering other traits, such as language or social relationships. Does a mutation in the FOXP2 gene of Neanderthals similar to that in humans mean they spoke? Can we claim that Neanderthals were patrilocal on the basis of just six genomes? In questions like these, where the relationship between genes and behaviors or other traits seems tenuous at best, it would seem greater caution is required.

Philosopher of science Carol Cleland claimed that historical scientists were always on the lookout for what she terms “smoking guns”, “a trace(s) that unambiguously discriminates one hypothesis from among a set of currently available hypotheses as providing “the best explanation” of the traces thus far observed.” While a smoking gun does not disprove alternative hypotheses on offer, it “(so-to-speak) cinches the case for a particular causal story”.  In some cases it seems that ancient (and modern) DNA can fulfill the role of smoking gun: ongoing debates where the fossil and other traditional evidence did not seem to tilt in any direction. The question of whether modern humans and Neanderthals interbred, for instance, was traditionally approached by looking at fossils and looking for signs of Neanderthal/human interbreeding.

But for many other questions, we do have compelling traditional data available, such as for how the Sima de los Huesos fossils fit in the picture of hominins populating Pleistocene Europe. In such cases, ancient DNA can play a role in furthering a consilience of inductions, and any story that radically deviates from the accepted view (such as the earlier mtDNA evidence) should be scrutinized carefully. 

Is DNA evidentially special then? The answer, it appears, turns on context. We should recognize its power in deciding certain kinds of questions – the origin and nature of our species’ radiation, for instance. But we should be wary of getting carried away: DNA, particularly ancient DNA, has its limitations. It often should play a supplementary rather than central role in reconstructing our past.


Helen De Cruz is a senior lecturer in philosophy at Oxford Brookes University. She has published in philosophy of cognitive science, philosophy of religion, and philosophy of the historical sciences.