Cladistic analysis is certainly a powerful, logically rigorous tool for the determination of relatedness among species. In the simplest case you have three species and want to know which two are most closely related. You look at features shared by two of the three. But since evolutionary rates can vary widely in different lineages, two of the three might look similar because the third one changed a lot, not because those two are actually closest relatives. So you look for a subset of shared features, the shared derived features, (or synapomorphies, from the Early Hittite), which you determine by finding a species slightly more distantly related from the three that you are really interested in. Then you tally up the traits shared by two of the three, and not by the outgroup. Those are your synapomorphies and, in principle, they should tell you which two of any three species are closest relatives.
It’s quite straightforward, actually. Trust me.
[In fact, aside from the Hittite vocabulary, much of the philosophy of phylogenetic reconstruction is articulated in William King Gregory’s 1910 monograph on The Orders of Mammals, which was thrust into my hand back in 1983, when I attempted to raise the subject with Dr. Simpson. Some history-of-science graduate student really ought to “do” Will Gregory, BTW.]
But a cladistic analysis contains a number of important assumptions. First, it presumes that the species only get their features vertically, through a process of what Darwin called “descent with modification”. The goal of the method is to establish relatedness by descent. If traits are acquired some other way, such as horizontally, by airborne viruses, by Lamarckian inheritance, or by dirty toilet seats, then the system breaks down, for then there is no way to establish synapomorphy, which is what the method is predicated on. Shared derived features might be there either by virtue of proximity of descent, or by virtue of infection, or perhaps even by sheer force of will in different animals.
Second, it only works on species. After all, if the taxa in question could receive their similarities via gene flow, the system would likewise break down.
There are also other constraints. The outgroup can’t be too distant, and the rate of convergent evolution can’t be very high. One simply assumes that a single character distributed across three taxa in two states, is the result of a single evolutionary change from one form to another, inherited in the two species that share it, from their common ancestor. Not from parallel mutations, or miasmas.
So I just don’t get how people can build research programs based on the application of cladistics to cultural traits, where the method logically fails. Worse yet, that’s why paleoanthropological systematics will never get cleared up. To do cladistics, you need species, and the surest way to get species in paleoanthropology is to “split” fossil samples, and pretend that the groups you name correspond to some biological reality. Yet a cladistic analysis is the most straightforward paper in the field. One measures a bunch of things and throws the mess into PAUP (for a phylogenetic analysis using parsimony).
But suppose the taxa you are making nested phylogenetic clusters out of, aren’t phylogenetically nested in the first place? Suppose, for example, they were linked bio-historically like a rhizome, or a trellis, or a capillary system (as indeed, generations of physical anthropologists have thought)?
Hooton, E. A. (1946) Up From the Ape, 2d edition, New York: Macmillan.
Holliday, T. (2003) 'Species Concepts, Reticulation, and Human Evolution ', Current Anthropology, 44: 653-673.
Possibly because they were actually members of rather few real biological lineages.
The point is, that if you impose four species on Homo (as the authors of The Human Lineage seem to, although they do some mighty fancy tap dancing around that issue) or if you impose 12 species on Homo (as the authors of The Last Human do – ridiculously, but at least clearly), you are not just having a sterile disagreement about a sterile issue of classification. You are actually disagreeing over what you can fundamentally do methodologically.
Cartmill, M. and Smith, F. H. (2009) The Human Lineage, New York: Wiley-Blackwell
Sawyer, G. and Deak, V. (2007) The Last Human: A Guide to Twenty-Two Species of Extinct Humans, New Haven, CT: Yale University Press.
A cladistic analysis is done on species, not specimens, and is meaningless below the level of the species. The best you can do is a “numerical taxonomy” analysis – the kind of thing that Chris Stringer was doing back in the 1970s – and that will (if you’re lucky) tell you what is most similar to what else. But it won’t make false assumptions about how they got that similar, which the cladistic analysis probably does.
I confess, I’m an just ol’ country anthropologist, but you know what? I think human evolution is strongly rhizotic, just like the post-modernists might-or-might-not have said.
Deleuze, G. and Guattari, F. (1987) A Thousand Plateaus: Capitalism and Schizophrenia, St. Paul, MN: University of Minnesota Press.
[For the love of God, please don’t give me a quiz on this one, but I know it says something about the rhizome as a powerful metaphor.]
But here’s the scary (i.e., challenging) thought: What if it’s rhizomes all the way down?
Arnold, M. (2009) Reticulate Evolution and Humans: Origins and Ecology, New York: Oxford University Press.