Monday, October 21, 2013

How many 'human' species are there? Is it even a real question? Why does anybody care? The Dmanisi skulls

I was a graduate student in Anthropology at the University of Michigan in the 1970s.  It was one of the leading departments at the time, and the home of an ecological and evolutionary genetic framework for viewing human evolution. The view was based on what is known as the 'modern evolutionary synthesis' from the 1930s, in which Mendelian inheritance of genetic variants, and Darwinian evolutionary dynamics were united.  This union allowed one to interpret the fossil record in terms of species diversity, adaptation, and evolution within a single genetic framework--at least in theory.  The evolutionary population genetic approach differed from the more stereotypically categorical approaches of morphological schools of thought that, to oversimplify, took type specimens to represent species.

Michigan's leading lights at the time were the late (and much missed) Frank Livingstone who modeled genetic population dynamics, and Loring Brace who analyzed fossils.  Milford Wolpoff showed up towards the end of my time there.  Many distinguished future paleontologists were there as students.

Genetic and morphological variation and species formation: a problem!
The modern synthesis allowed one to relate intra- and inter-population variation, rather than 'type', to evolutionary principles and dealt among other things with how much variation is found within a species and how much between species.  Variation represented genetic diversity (even if the actual genes were not known), and the idea was that after a certain amount of time had elapsed and diversity had accumulated, new species formed.  Darwin himself just hand-waved about this (Chapter 4 of Origin of Species), but the implication was that when you had enough variation you could infer new species.

This idea has had widespread acceptance but if you think about it, it doesn't really hold water because there is no definition of what 'enough' is--and only a somewhat artificial definition even of 'species'.  As human beings make very clear, morphological diversity is not the same as species diversity: people from the ends of the world can mate successfully even though we differ in adaptive morphological ways (e.g., skin color) and humans from the ends of the world have been separated by around 100,000 years.

A challenging example of morphological species assessment is sexual dimorphism.  Males and females from the same species can be quite different and while we can try to explain sexual dimorphism in terms of the behavior of living species, there's no simple rule.  Humans, chimps, and gorillas are all very closely related, but hugely different in their sexual dimorphism.  This means we're on shakier ground when it comes to fossils. 

Behavioral variation and species formation:  also a problem!
In the atmosphere of the 1970s, we had to use more abstract genetic and ecological theory to try to make sense of the scattered fossil record known at the time and when we had little genetic data.  One of the major theoretical arguments in the Michigan school was what is called the Competitive Exclusion Principle (CEP).  This asserted that two species could not occupy the same ecological niche: they must differ in important ways in the places, foods, times, and so on that they lived.  If they overlapped too much, they would in a true Darwinian sense compete too severely, and one would die out.  Or, perhaps, the two could never have evolved in the first place to be so similar.

This idea has since been revised here and there, but if you think about it, it's rather vague if not tautological just as is the problem of defining species.  One can always find some difference in diet, location, habits, or behavior, and proclaim that the CEP was being confirmed--each species has its own 'niche'!  That's rather vacuous, no matter how neatly Darwinian it sounds.  After all, not even two members of the same population in the same species eat or behave exactly alike.

In any case, how would the CEP work in the case of humans or hominids (our ancestral species since separating from common ancestry with chimpanzees)?  As Loring Brace used to argue in papers and texts, culture is the human ecological niche.  Culture includes technology, language, ritual, symbolism, and so on, and Brace's argument was that a species with culture would simply drive out of existence any other species that also had culture.  The idea was that only one species could have culture.  There's only one species of humans, right?  As with the CEP in general, this may sound OK unless you think too carefully about it.  It's a kind of argument by definition. We have different cultures around the world, or even nearby, and they tolerate each other.  Or, if it is culture per se, rather than cultural difference, how can that be defined, and how can one in any serious way 'prove' that no two hominid species could have culture in the same place (not to mention different places)?  Does using stone tools count as culture, or not?  What level of verbal activity is culture, and what is just, well, grunts?  Single Species is about as non-definitive or even tautological as CEP.

The Single Species hypothesis
Regardless of these epistemological problems, which generally weren't raised at the time, the bottom line at Michigan was the Single Species assertion.  This was a hyper 'lumper' view of human paleontology, as opposed to the 'splitter' view in which many different contemporary species were asserted based on the fossil record (investigators often naturally wanting to be able to give species names, and hence confer special importance, to the bones they found--something that continues to this day).

Most anthropologists  at the time accepted the existence of a diversity of contemporary species often if not typically during the time since we and chimps shared a common ancestral species. This hyper-splitter view was held even to the point (as Brace often satirically pointed out) that many leading anthropologists seemed unwilling to accept that any of the known fossils, morphologically crude relative to us moderns as they seemed, were actually our noble species' ancestors:  all the fossil specimens were from side branches that became extinct!

One  aspect of this had to do with whether modern-looking fossils represented humans who expanded out of Africa around 100,000 years ago and exterminated the other existing hominids for whom we have fossils dating back even well before a million years ago in Africa and Eurasia.  The replacement hypothesis was opposed by the hypothesis, in some ways based on Single Species thinking, that humans had evolved globally always as a single species, continuous one-species evolving as new genetic abilities arose here or there and spread by diffusion through mating between adjacent populations.

In the last several decades, a steady march of new finds has led to a diminution of the Single Species hypothesis, especially in the earliest hominid finds around 2 million years ago.  Even some of the Michigan types (not to be named) claimed they'd never really argued for a single species.  There is at present at least little or no argument against the existence of multiple species at least at our hominid origins (Australopithecines and other forms).

Single vs multiple species issues recently have pertained mainly to the more recent period.  However, a new find in Dmanisi, Georgia, that has just been described (see below) is viewed as answering that question.  But does it?

Modern genetics to the rescue?
Earlier, I referred to the viewpoint back in my graduate student days as a genetically based one.  It was that, but only in principle.  Other than a few blood group markers, like Rh and ABO, we had very little actual data on genetic diversity even among humans and even less between species (that is, few variants--'alleles'--in humans even were known also to exist in chimpanzees).  So genetic arguments were largely theoretical.

But from the 70s onward, genetic data (on protein variants) from global samples of modern humans began to accumulate.  This led to controversy because the date of separation of all of us moderns from our common ancestor that was estimated from this genetic variation corresponded to the finding of modern-looking fossils first in Africa and then expanding elsewhere at around 100,000 years ago.  But the existence of million year old fossils from Eurasia tended to undermine the Single Species hypothesis and support a the view that two contemporary hominid species, one old and one modern  competed, and our ancestors won out completely.

People were arguing about this, and in one of my very first papers, in 1976 (Am J Phys Anth 44:31-49), the population geneticist Takeo Maruyama and I looked at the allele-frequency data, sampled from different continental populations (Europe, Asia, and Africa), from which the 100,000 year global human separation time had been estimated.   We did computer simulations to show that gene flow between neighboring populations--the kind of mate exchanges that human hunter-gatherers as well as our primate relatives engage in and hence likely applied at the time, would be able to spread genetic variation fast enough, to reduce inter-group differences enough, that an actual 1,000,000 year separation could look like a 100,000 separation, that is, by making populations look too closely related if you assumed no gene flow among them as the typical genetic tree-making models of the time did.  Indeed, we argued that
"It seems futile to employ population genetic models and present protein data to attempt to reveal about the past what the actual, ancestral specimens themselves cannot."
This was in essence a single-species argument that suggested that no literal replacement of one hominid species by another need have taken place.  Where does this argument stand now?  Since that time, we have vastly more direct genetic data from fossils themselves, and not just from frequencies of various alleles but from whole DNA sequences.  This enables one to compare sequences among Neandertals and other fossil groups as well as chimpanzees and modern humans and look at their differences and similarities.

What has been seen suggests that there was in Eurasia some inter-group admixture.  For example, one can see close sequence similarity across a given chromosome among the fossils, but this is interrupted here and there by runs of rather different, and seemingly more distant sequence that seem to come not from the same lineage but from more distantly related ones.  Such 'introgression' as well as sequence elements in these same genomes that may have quite different apparent times of common ancestry.  The general interpretation of such evidence is that after some thousands of years of separation, individuals came together (so to speak) and produced offspring. 

That of course implies that, by the classical species definition of reproductive incompatibility, not only did these guys not exterminate each other or manifest the CEP (at least, not immediately), but they were reproductively compatible with each other.  Nor do these data preclude the gene flow time-diminution issue.  Hello, single species hypothesis again?

The new Dmanisi find
Now we see the news, published in the Oct 18, 2013 issue of Science, and reported all over the media, of a remarkably well-preserved new fossil from Dmanisi in Georgia (the European one).  Here is one view of this skull:

Dmninisi specimen, Georgian National Museum

This is about 1.8 million years old, and deposited remarkably close in time to this one, archeologists say, were several other skulls that look 'very' different from it.  The variation is said to be similar to that of "patterns and ranges of variation in chimpanzee and bonobo demes....and in a global sample of H. sapiens," that is, of worldwide human variation.  Based on these interpretations, the conclusion is that these are of the same species.  Hence the splitters have been wrong.

But to compare this very local variation to our species and its 'racial' variation, is rather curious.  How, given their very local means of support and movement, could ancestral groups as diverse as, say, Amerindians, Koreans, Nigerians, and Finns be found in one local site?   This perspective would argue strongly against the inferences from the site as to times of deposition, or their population assertions.  Local demes that a single carnivore could prey on would intermix, if basically all other primates are a model.  They couldn't have stayed so diverse--unless they were, in fact, different species.

But if the interpretation of the coming together of human-like variation to a single site stretches the imagination, so does the alternative: how could such different groups be so locally different in the first place?  We're not talking about Columbus and some African slaves suddenly arriving in the Americas from long-distance sailing voyages!  How could they have ever become so diverse--unless they were, in fact, different species?  How much variation is too much variation for a single species in a single location?  The point here is to ask, rather than answer these questions.

 The instant and breathless pronouncements about the new Dmanisi specimen seem to replay the old Single vs Multiple Species hypotheses. But, perhaps as usual in the raucously public arena of human paleontology, there seems to be more noise than signal here, in the sense that all elements of the population ecology, as I've tried to suggest above, are vague and verge on tautological.  For example, the authors argue for single species using the arguments summarized above, invoking the idea that this is the most 'parsimonious' hypothesis--that is, the simplest explanation consistent with the facts.

But in the context of the above considerations among others, how is one to determine what parsimony means here---and, more importantly, what does one imagine justifies the 'parsimony' principle in such a situation?  Is the explanation 'consistent' with the facts?  Evolution is usually polyphyletic (branches of contemporary related species), so is 'single species' really simpler?  One might say that Loring Brace had to essentially coin the cultural-niche hypothesis to assert theoretical support for a single-species view.  From a process point of view, does evolution necessarily follow the simplest path, even if we could define what that is or might have been?  By what reasoning would one assert that?

Parsimony is a nice organizing idea, and usefully tends to reduce explanatory clutter and ad hoc or post hoc invention of reasons to fit an author's predispositions.  But in fact it has no authority in science (one could argue not even in physics), and in cases as vague as paleontological reconstruction is entirely too flexible, mainly serving as an assertion.  Single-species may be right or it may be wrong, but conclusions are being jumped to with rather minimal theoretical underpinning.  At 1.8 million years old, there will be no DNA to ride to the rescue.

Does the question that Takeo and I asked, nearly 40 years ago, still stand, or has it been answered?  If the admixture interpretation of the Neanderthal data is correct, then at least to some extent, admixture did occur.  By the usual species definition, these creatures at that time were a single species, no matter how genetically variable.  But the separation times of modern humans, based now on huge amounts of global DNA sequence data still are about 100,000 years, matching what some, at least, argue on the basis of known fossil morpologies.

Could gene flow have artificially reduced this apparent time?  Evidence for introgression--admixture--between what had been long-separated groups seems convincing at present, and in a technical sense settles the argument in favor of a single species.  But how far back?  My thinking is that while direct sequence rather than indirect allele-frequency data are better, the issues are not closed--neither by much better genetic data, nor by many more and better fossils. Perhaps new simulation studies are in order (I've developed a program that could do that, but haven't done it yet).

None of this makes the Dmanisi specimen unimportant, though whether it's the Find of The Century (for today) or just a very interesting find, is as debatable as one could ever want an issue to be if one wants to spill ink over it.  The species question is at least fascinating to many people.  What practical evolutionary difference it would make in the case of such very similar peoples, the question will keep many anthropologists gainfully employed for years to come, whether or not it settles any arguments!

The Dmanisi find is being discussed by many others, including very knowledgeably in blog posts by John Hawks and Adam van Arsdale.


  1. Is it possible that a single group at the same time showed a high degree of variation due to gene flow from different regions? Maybe because it was on the boundary of different subspecies? That is, a single population represented the diversity of the whole species.

    1. I think it's unlikely in the way you put it. Where, for example, in Europe or Asia or the America's would you find regional populations that were as different as all humans but living close enough that the populations' edges met at some central place?

      Gene flow erases differences, more slowly at great distances but rapidly locally. Humans (and our closest primates) have very high rates of inter-demic mate exchange each generation.

      If differences as great as, say, between Asians, Africans, and Europeans arose (that long ago) how would they come together, living in small bands as they must have been?

      So lots of things are possible, but that much difference essentially at the same place and time, if things are being interpreted correctly (and I'm not a paleontologist!), raise interesting questions of interpretation and (to me) no obvious answer at present.

    2. I wonder. Is there more variation within pops than between pops (a la lewontin) when we're just looking at cranial metrics? Seems to me maybe the answer's no... because we can use crania to infer ancestry so well.

    3. I'd say a few things. First, Lewontin was referring to genetic varianCE--and importantly, not genetic varianTS. I can't comment on cranial metrics, but the issue (to me) is that if there is really human global-like variation locally, how did it get that way? Were there distinct local populations surrounding Dmanisi who came together there? Was this just a very variable species, that is, with lots of variation within a single population, for some reason?

      I think these are interesting questions but don't really address the species question. There could be various reasons why this species had relatively ample shape variation in their heads, that could have to do with selection, or population size and structure, or whatever.

      Even if there was only a slight amount of admixture the species question in its strictest form is answered 'yes' (as in the case of the Denisovan/Neandertal/Modern story that is much more recent).

      But, personally, I don't see that just the amount of variation in scatterplots of a few samples can really answer the question.

      And I think Lewontin's point isn't really relevant here since we only have one population, if that's what Dmanisi is.

    4. I didn't anticipate how my comment would be taken so far from where I intended or else I'd have written more. Too late now. Will wait til skype or real face time to continue this as I had imagined :)

  2. Could you please clarify one vague place?
    You wrote,
    --Local demes that a single carnivore could prey on would intermix, if basically all other primates are a model.
    What does "single carnivore could pray on" said for? Does this indicates deme size or imply some influence of carnivores on demes? Sorry, for my knowledge of English isn't too good and I'm not experienced in population ecology, so it maybe stupid question.

    1. We just meant that if there were local populations of such closely related populations that a single species of predator could prey on at the same time (in the geological sense) it is unclear how those local populations could have evolved to be separate species. It's just the question of how such different species, but basically so closely related (the same variation in humans evolved globally, not locally, so there was isolation) could have evolved without always inter-mating. We just raised the question as something to be explained if the hypothesis of separate species were true.

    2. So single predator species is simply a sign of similarity of it's prey. Ok. Thank you.

    3. If the same predator individual, or predator species, was dropping its prey in the same place, and they seemed so different and yet living so nearby, it seems to require explanation.

  3. Great article. Your points in the paragraph two above where the "Parsimony" section begins echo my main quarms with the single species hypothesis. How can such a level of intraspecies/intrapopulation variation have occurred at one point in time?

    Having not read the paper from front to back I'm shocked to learn that parsimony was the explanation cited for support of the "single species" theory! Parsimony is flawed. Evolution is a blind process not one that is pre-determined to take the simplest route every time. You denounce it well here.

    What are your views on the new article (Bermudez de Castro et al. 2014) to emerge on the finds?