The Human Evolution Blog

One of the most interesting times in the history of the human species is the period when fully modern humans (Homo sapiens sapiens) lived alongside and interacted with the Neanderthals (H. neanderthalensis or H sapiens neanderthalensis, depending on your perspective). It is now widely accepted that there was some interbreeding and that all non-African populations have some contribution of Neanderthal DNA in their genomes.

However, preceding the period when humans and Neanderthals reunited, the two groups must have been separated for quite a long time indeed. We know this because partitioning is absolutely necessary for speciation to occur. If the populations that would give rise to Neanderthals and modern humans were in close contact, they would have merged, rather than remaining distinct. Even a small trickle of gene flow would have kept the two species together, especially considering how similar their lifestyle and survival pressures were.

The same, but different

While we are still discovering how and where the many branches of the hominin family tree explored the world, most evidence points to Africa as the place where the H. sapiens lineage evolved into our current form. Meanwhile, Neanderthals were evolving from their ancestors in Europe and Western Asia. During that time, we acquired some differences in our morphology (physical structure). Neanderthals were shorter, more thickly built, with bowed legs and strong upper bodies. They also had a very prominent brow ridge and a protruding lower face.

We also evolved some things in common with Neanderthals. This is called parallel or convergent evolution and can sometimes confuse our normal process of studying how and when features evolved. In the case of humans and Neanderthals, we both evolved increasing cranium size and fine hand-finger motor skills, probably as a result of tool use.

Usually, when two closely related species share a feature, we assume that it was inherited from a common ancestor that also had that feature. Every so often, that’s not the case. Similar selective pressures can lead to similar adaptations and so physical likeness is not always the result of shared ancestry. For example, both bats and birds have wings, but that’s not because they evolved from a common ancestor that had wings. The wings of bats and birds are examples of convergent evolution.

Sorting out which features common to humans and Neanderthals are the result of shared ancestry and which are the result of parallel evolution can be very tricky. Sometimes, we simply don’t have all the fossils we need in order to make the determinations definitively.

The common ancestor

Paleoanthropologists estimate that the human and Neanderthal lineages diverged from their common ancestor between 350,000 and 400,000 years ago. This estimate comes from a variety of techniques, most especially the “molecular clock” method which utilizes the known and stable rate of mutation in certain noncoding regions of DNA.

But what was the common ancestor of humans and Neanderthals? The most likely candidate is the species we call Homo heidelbergensis (for having been first discovered near present-day Heidelberg, Germany). This species existed at the right time: first appearing in Africa at least 700,000 years ago and persisting in Eurasia until as recently as 60,000 years ago. H. heidelbergenesis also existed in the right places: besides those that remained in Africa (where H. sapiens would evolve), a population migrated to Europe and Asia between 300,000 and 400,000 years ago, around the time that the Neanderthal lineage began to diverge. Of course, the morphology of this species fits with what a common ancestor would have to be as well, possessing all the necessary shared features.

The claim that sapiens, neanderthalensis, and the mysterious Denisovans all evolved from heidelbergenesis is not universally accepted partly because of deep disagreements about the organization of the Homo genus and partly because some view the hominin lineage as a mostly linear progression, rather than a bushy evolutionary tree. Notwithstanding the objections, the fossils that most anthropologists call H. heidelbergensis likely come from a population, whatever their species name, that gave rise to Neanderthals and modern humans.

Digitizing morphological change

In order to address the question of what the ancestors of humans and Neanderthals might have looked liked, a research group at the University of Cambridge took a digital approach. They catalogued a large set of morphometric variables in the hominin fossil record, focusing on skull features. (Morphometrics are the measurements of variation in some feature.) They selected 797 specific skull measurements and plotted how these features changed over the course of 1.6 million years of evolution, going back to Homo erectus. The analysis included not just human and Neanderthal skulls, but representatives from many other populations along the way.

The power of this technique is that even fragmentary fossils can be used (and they’re pretty much all fragmentary when we’re talking about the human fossil record). Any fossil that can provide data for even just a single measure can add something to the analysis, as long as its approximate age is reliably known and it can be placed within a species with other fossils. Their analysis reveals how skull features changed over time in different lineages, with the idea that gaps in the progression can be filled in by inference.

Using three-dimensional image analysis, the researchers analyzed the plots of all 797 features simultaneously and then projected how they changed in the various Homo lineages. From here, they aligned the 3D morphometrics of the human and Neanderthal lineage to see where they most overlap. This overlap point represents the morphometric state at which the human ancestor was most similar to the Neanderthal ancestor.

Because the fossil record is always an imperfect scatter of information, there are gaps in our knowledge, and the researchers were faced with at least three points in the historical record at which the common ancestor might have existed. In other words, their were three “models” that could have explained when the divergence of sapiens and Neanderthals might have occurred. They used the morphometric models to calculate which was the most likely.

Ultimately, the researchers were able to generate a hypothetical composite image of what the skull of the common human/Neanderthal ancestor might have looked like. Here it is:

The value of a virtual ancestor “fossil”

The generation of this composite skull represents an emerging phenomenon: virtual ancestor reconstruction. This new technique has the mind-blowing power to generate and 3D-print images of predicted fossils that we can all see and handle. The value of this in evolutionary research and education is unprecedented as it “creates” fossils that scientists have only speculated about until now.

However, there is also a rather obvious danger: the fossils aren’t real; they are conjecture. They are the product of inferences and are thus hypotheses, not evidence. Inferences are integral to science, but they are always put to the test through the acquisition of actual primary data. A virtual ancestor is the equivalent of a proposed phylogenetic tree or a tentative “best guess” about what a common ancestor would look like. These “fossils,” in and of themselves, don’t prove or even add weight to any hypothesis. They are merely the manifestation of an hypothesis.

The reason I call this a danger is because any time we can see, touch, and feel something, it seems real. Indeed, that’s the whole point. Because of that, scientists, museum curators, and the general public must be mindful to resist the undue influence that imaginary fossils can have on our thinking. They can lead us to self-fulfilling prophecies and circular reasoning, in which composite projections are treated as verifications and then influence the crafting of further hypotheses, creating a paleontological house of cards that could all come tumbling down. Hypotheses must be built on evidence, not on other hypotheses.

What this virtual ancestor “fossil” tells us

In this case, the authors present their work clearly and humbly. It is without question a spectacular artifact to behold. It would be tempting to compare it to real hominin fossils and attempt to draw conclusions about evolutionary relatedness, but that would be misleading. Data scientists might call that “over-fitting” the data.

For example, the composite common ancestor skull looks a great deal like the recovered skulls of Homo heidelbergensis. A careless approach would be to use this as supportive verification of the hypothesis that heidelbergensis is indeed the common ancestor of humans and Neanderthals. But that would be circular logic. After all, similarities in skull features are why we think heidelbergensis is the common ancestor in the first place. Furthermore, data from heidelbergensis skulls were used in the morphetric analysis and played a key role in how the projections were made. Of course the composite skull looks like heidelbergensis!

On the other hand, if the hypothetical ancestor skull shows key differences with the recovered heidelbergensis skulls, this could potentially be an important matter to explore. While it wouldn’t prove or disprove anything, it would help scientists visualize where the projections and the actual data differ so that they can figure out why. Either the predictive model needs adjustment or we are wrong about heidelbergensis being the common ancestor of humans and Neanderthals.

To that end, the authors of this paper report that the model that their data supports is that the most recent common ancestor of humans and Neanderthals is much older than previously thought. While it is doubtful that the paleontology community will up and abandon the currently dominant hypothesis because of this one predictive model (based only on skulls, no less), it may get people debating the various lines of evidence. And that is always a good thing.

Therein is the value of predictive models like these: they could help expose weaknesses or contradictions in our current models. Ultimately, those controversies cannot be resolved by digital projections or predictive modeling anyway. Only additional fossils could provide the necessary data.

And that captures it perfectly: virtual fossils are nice, but what we really want are more real fossils.

-NHL