So there are a couple of interesting developments about fossil apes. One is the retraction by the author of the claim 14 years ago to have found a jaw bone that was evidence of Homo habilis, a precursor species (arguably) of H erectus, in a recent Nature. Previously he and his coauthor had claimed that erectus may have evolved in Asia and then returned back to Africa. On a re-examination of the evidence, Russell Ciochon now thinks that there is a “mystery ape”, around chimpanzee size, in the Asian forests, a possible precursor to orangutans. In the light of Homo floresiensis in the Indonesian Archipelago, however, it is clear that there had to be another hominid in South East Asia apart from erectus, as it is not likely to have evolved from an erectus precursor. If I recall a talk given by Colin Groves, of the ANU, he thinks that it is likely to have been a descendent of H heidelbergensis, which many paleoprimatologists think was also the ancestor of H neanderthalensis habilis.
So what is going on around 4-1 million years ago in Asia? There has long been a debate over two competing hypotheses: the Out of Africa hypothesis, and the Multiregional Hypothesis, for H sapiens, but this is independent of sapiens‘ evolution. More recently, Alan Templeton argued on genetic grounds that there were several Out of Africa events over a period of nearly two million years, with gene flow back to Africa as well. This, ironically, leads back into the other paper [pdf], by Jeffrey Schwartz and John Grehan, in the Journal of Biogeography, which argues that humans are more closely related to orangutans than to chimps and gorillas [razib has a good summary], a view Schwartz has argued before.
Schwartz and Greham argue a dual case: one is that molecular data is not to be trusted without some serious caveats. This is a line that has been run for some years as an attack upon “phenetics” by certain critics of molecular systematics. For those who don’t know, “phenetics”, AKA “numerical taxonomy”, was an attempt to group taxa (indeed, organisms at all levels) by “overall similarity”. It basically failed because one got, as we now would expect, different groups from different principal components used to measure the similarities. Pattern cladists have attacked molecular systematics as a kind of phenetics, only one that uses only one principal component – molecular sequence.
Schwartz in particular needs to argue this, because molecular metrics place humans and chimps in a clade, then those two in a clade with gorillas, and orangs as an outgroup to these taxa. He wants to place humans and orangs together, and chimps and gorillas in another. He does so on the basis of morphological characters, in the “old fashioned way” that has been used for a long time, well before molecular characters. If molecular systematics is unreliable, then he can make a priority for morphology, and thus support his preferred phylogeny.
I am not so sure about this, myself. I think that molecular data, suitably and carefully used, forms a line of evidence that need not be phenetic – the caveat being that if you can restrict the characters (i.e., sequences) used to actual homologies, then it is as informative as any other set of characters, maybe more. But there’s the rub, and it is that which leads pattern cladists to challenge sequence data – what is a homology, what is a pseudogene and homologous but possibly uninformative, and what is a homoplasy? The sheer volume of the data makes manual tweaking all but impossible except for short sequences, and in that case you may as well be using morphology as molecules. If you use manual tweaking to align sequences (and I gather most do, although there are numerous automated algorithms for doing so), then you are back in the territory of subjective interpretation (no matter how good the taxonomist – if it isn’t repeatable, then the methodology is suspect, as the history of taxonomy shows). I suspect that each domain of data can illuminate each other; I am more sanguine about it than the critics, but I am a lot less sanguine than the proponents of molecular systematics, most of whom I suspect don’t even see there’s an issue.
But, believe it or not, that’s not the interesting point for me. Rather it is the revival or implementation by our authors of Leon Croizat’s “panbiogeography” technique. Let me quote their words:
These complicated and convoluted dispersal models [of conventional accounts of Asian ape origins] are not grounded in empirical evidence, but rather on preconceived notions about the existence and locations of centres of origin. Following Darwin (1859), the underlying assumption is that an ancestor’s distribution is geographically narrow with respect to the distributions of presumed descendants located in different (vicariant) geographical areas. The problem with this preconception, however, is that the very ability of individual descendants to move between different locations obviates the very criterion – geographical isolation – that is also invoked as essential for their differentiation and speciation. An alternative model of vicariant differentiation (Croizat, 1958) suggests that ancestral dispersal occurs before the differentiation of descendant taxa, with the ancestor establishing a widespread geographical distribution that encompasses the combined ranges of all vicariant descendants (Craw et al., 1999).
We take the position that integrating theories of relatedness between fossil and living taxa with those concerning biogeography can be accomplished only if shared–derived morphological similarities are first delineated and then used to generate and subsequently test theories of relatedness (Schwartz, 1987, 2005).
The term “centres of origin” is a key term that denotes an old and largely discredited nineteenth century obsession with finding, on no real evidence at all, the point of origin of some group. Instead, the panbiogeographic approach for living species presumes that the ancestral group (not necessarily a species) for some existing group of taxa had already dispersed, and each new daughter species evolved in situ (“vicariantly”), more or less. This is called vicariance biogeography sometimes.
Now, like many techniques and issues in systematics, this has become an object of quasi-religious fervor, and derogation of heresy, at various times. But here Schwartz is applying it to the paleontological record. Instead of trying to find the point of origin of a group in Africa or Asia, evidence for which is largely if not totally lacking, he and Grehan are taking the fossil record at its “word”. The ancestral group actually was distributed at least in the regions where the various fossils were found. Taking from Croizat the “tracks” that the fossil distribution suggest, using a method called “minimum spanning tree”, they reconstruct the dispersal of these groups with the following figure:
Caption: Geological and spatial correlations for dental hominoids. (a) Tectonic correlation of the dental hominoid track (yellow) with the East African Rift System (approximate boundary as red dotted lines) and Tethyan Orogenic Collage (approximate boundary as blue dashed lines). (b) East Asian Node (as open black circle) encompassing geographical proximity or overlap of tracks between four largely or entirely vicariant dental hominoid genera: Lufengpithecus, Gigantopithecus, Khoratpithecus and Pongo (with overlapping ranges of Gigantopithecus and Pongo indicated by blue circles with a yellow border).
Compare this with the figure in the Ciochin note:
It looks remarkably like Schwartz’ and Grehan’s tracks. Maybe the mystery ape is related to orangs too. Almost certainly it is.
One of the nice things about the dispersal track map is that it explains why there are hominid and pongid fossils as far away as Spain, Russia, China, Indonesia, and Africa. But does it imply that humans are related more closely to pongids than to Pan and Gorilla? I don’t think so, myself. Correlation and causation and all that. But it is clear that the evolution of all the Great Apes is way more complex than we now know, and maybe we have been wrong all this time. The joy of science is finding out that we have been wrong, and learning something new.
Late note: Having been grumpy at Graham Lawton before for poor reporting in New Scientist, it is nice to report that he got this one right. Kudos where deserved.