The Tzores of Taxonomy

How can one ever justify creating a new species name based on scraps of fossil bones or teeth?

Go to the profile of Henry Gee
Sep 27, 2017
Upvote 4 Comment

What are Human Beings, that we are Mindful of Them? The question is as pertinent now as it was back in the days of the psalms. Such is the question on my mind on returning from the 7th annual meeting of the European Society for the Study of Human Evolution (ESHE) in the charming town of Leiden, in the Netherlands. 

The fossil record seems awash with fossil hominin species. In the past few years we’ve seen Homo naledi, Australopithecus sediba, Australopithecus deyiremeda, Australopithecus garhiKenyanthropus platyops, Sahelanthropus tchadensis, Ardipithecus kadabba, Orrorin tugenensis, Homo antecessor, Homo floresiensis and no doubt more I’ve missed, each one described from one or a few often fragmentary cranial, dental and skeletal remains. The question – the big question – is how you finesse this small box of rocks into biological species. Species in the sense of the late evolutionary biologist Ernst Mayr: that is, populations of living, breathing creatures, which could actually or potentially interbreed with one another to produce fertile offspring, and were reproductively isolated from other such populations.

I am not alone in such ponderings. Another ESHE delegate, John Hawks, reflects on how the division and subdivision of these various relics is a constant niggle, and how some people lump a whole load of them together, whereas others work to split them into more exclusive groups. 

Any field biologist knows how hard it can be to tell members of one species from another, even with animals or plants that are still alive and which one can observe. Some creatures that look very similar do not, in fact, interbreed. Others look rather different and yet are perfectly interfertile. Some show pronounced sexual or regional dimorphism, or have juvenile or larval forms that look entirely different from the adults. Some species appear to be somewhat variable, with a phenotypic plasticity you’d never guess from their karyotype. Others look very similar but have different numbers of chromosomes. Each species is prone to a degree of variability.

Now, fossils.

Fossils represent an unknowably small sample of what were once living creatures. Worse – an unknowably small sample of parts of what were once living creatures.  With rare exceptions we know nothing of their soft-tissue anatomy or their genetics, still less of the variation, life history and behaviour they might have exhibited when they were alive. These things are hard enough to establish even with organisms that live, thrive and survive today. How, then, can one ever be justified in naming, as a new species, some poorly preserved relic of a species that is extinct, and for which any comparison with living species is at best uncertain? Is it evidence, or – as Hawks says – ego?

It was problems like these that confronted me every day as I worked for my Ph.D. thesis, many years ago when the world was young (OK, it was the mid-1980s). My task, back then, was to learn how to tell the difference between cattle and bison. Sure, you’ll say, these are extant animals, and any fool can tell the difference between one and the other. Ah, well, you can, at least with cattle and bison that live today. And you can do it pretty well with the skulls of the various extinct kinds of bison (in the genus Bison), and the aurochs (Bos primigenius), the wild progenitor of today’s cattle, which only became extinct in 1627 when the last one was shot, in what is now Poland. The skulls look entirely different, and are easy to tell apart. Anyone can do it. But whisper it soft – bison and cattle can interbreed, and the European bison (Bison bonasus) shows signs of introgression from cattle, sometime in the past. And if they can interbreed, does Bison bison, or Bison bonasus, or any Bison species, really merit its own species, as distinct from Bos primigenius, or Bos taurus, or whatever – let alone its own genus?

I wasn’t looking at skulls, however. I was looking at limb bones, which are extremely difficult to tell apart, Bos from Bison. The problem of telling aurochs limb bones from bison limb bones was (and perhaps is) a major problem in Pleistocene palaeontology. For, in contrast to hominin remains, bovid bones fair litter the landscape, and museum collections are crammed with enormous boxes full of bones labelled ‘Bos or Bison?’ Well, they are in Britain. I was so busy with all these I had no time to look anywhere else. The upshot of this tale is yes, it is possible to tell Bos limb bones from Bison limb bones, but it’s not easy, and sometimes you have to resort to equations. I wrote this all up in a paper, which, if you want, you may unearth. This paper has attracted a princely 26 citations since its publication in 1993. So perhaps the problem isn’t that pressing, after all.

The point is, it’s very, very difficult indeed to tell the difference between the fossil bones of two closely related species, even when you have such large sample sizes that you can afford to discount anything incomplete or scrappy, and when the species concerned have very close extant relatives whose biology is well known. How, then, can one ever justify naming a new species of extinct hominin, or any other extinct species, based on fragments?

Most palaeontologists silently ignore the biological species concept. When describing an extinct species from one or a few fragments, the best one can achieve is something one might call the morphospecies – a pragmatic concept which allows one to distinguish one form from another on the basis of a prescribed list of characters.

The most important thing when describing a species in the literature is to establish a differential diagnosis. That is, to show how the new form differs in some recognisable and measurable way from every other species with which it might be confused. Formal taxonomy is not an optional extra. It is data, and meant to be as useful as possible to anyone who discovers a specimen sometime in the future and wants to establish that it doesn’t belong to a species already named, before erecting a new one. I suspect that poor diagnosis might have contributed to an inflation in the number of hominin taxa. That, combined with the understandable urge that people want to have the honour (and publicity) of naming a new thing, which is much more exciting, in the popular media, at least, than finding more of the same. Maybe I am an old cynic, but to me the days of palaeoanthropology as trophy-hunting are over. Time was when hominin species were described – in Nature, even – with poor or non-existent diagnoses (one thinks of Dart’s description of Australopithecus africanus). Such times are over, or should be.

In practice, people who study extant creatures use the morphospecies concept, too, as it is almost always impractical to establish reproductive continuity or isolation. The issue with palaeoanthropology, as distinct from palaeontology in general, is that because the fossils are vanishingly scarce; and at the same time have a bearing on a subject as interesting as our own origins (which in turn fuels an understandable if uncritical public enthusiasm); people will want to extract every tiny crumb of information they can from the meagre evidence available. However, it is important – vitally so – not to confuse biological species with morphospecies. Still less is one justified in drawing lines between them and presupposing ancestor-descendant relationships (a bugbear of mine I address in a recent book.)

The advent of ancient DNA (aDNA) and the ability to extract entire genomes from fossils has transformed the field. For the first time, we have prima facie evidence for interbreeding in ancient hominins. We now know, for example, that modern humans (Homo sapiens) interbred with the now-extinct Neanderthals (Homo neanderthalensis); that both interbred with the Denisovans, another extinct species known mainly from its genome, the actual fossil remains being too scanty to allow any kind of morphology-based diagnosis (which is why they have not been given a formal name); and that there aresigns of introgression from yet other, otherwise known extinct hominins in the genomes of people alive today.

This poses a nice problem – given we know that this introgression has taken place, can we invoke the biological species concept and say that modern humans, Neanderthals and Denisovans belong to the same biological species? Well, one could – except that for modern humans and Neanderthals, at least, it is possible to distinguish between the two morphologically. Humans and Neanderthals had different bodies, brains and life histories. So different, that had we met them back in Europe 40- or 50,000 years ago, they would have seemed entirely distinct. As distinct as – say – Bos and Bison. This raises an obvious question of whether the biological species concept is of any use. Sure, species are leaky. There is gene flow between one species and another, but the species themselves remain distinct in all the ways that really matter: their habits, habitats, life history, behaviour and relations with other species.

I wonder if anyone has tried the following exercise in experimental palaeontology? Take a hundred or so skeletons of laboratory mice, tie them in a cloth bag, and bash them around a bit with a cricket bat (readers in the Thirteen Colonies may substitute a baseball bat, if they wish). From this bag, randomly extract lots, each no more than one per cent by weight, and give each lot to one member of a panel of volunteers unaware of the nature of what went into the bag. Ask the members of the panel to pretend they are palaeontologists, and ask them to describe what they see. How many species are represented by the sample? How would one quantify the variation? Get them to measure things – bone lengths (assuming any bones survive the processing), tooth shape and so on.

Now the fun part. Ask the members of the panel to compare their samples and see if they can agree that they are the same or different.

I should add that DNA sequencing equipment may not be used in this experiment. Neither should one be allowed to use ZooMS (zooarchaeological mass spectrometry, which can extract an amazing amount of taxonomic information from otherwise uninformative scraps of bone), or compare proteomes extracted from ancient bone (another rising source of information in the identification of fossil material.)

If anyone has already done anything like this, please do let me know. If they haven’t, this could be the basis of someone’s next grant application. You read it here first, folks.

Go to the profile of Henry Gee

Henry Gee

Senior Editor Biological Sciences, Springer Nature

Ph.D. Zoology, University of Cambridge. B.Sc. Zoology & Genetics, University of Leeds. Joined Nature in 1987. Handles manuscripts for Nature in integrative and comparative biology, aliens from outer space, and other stuff. Recovering palaeontologist, contrarian, author, Rock God and wit.

No comments yet.