Human-orangutan phylogeny

[John Grehan]

Some TAXACOM respondents earlier comments on the plethora of morphological
'evidence' for the human-chimp clade in terms of number of characters or
number of papers. I am continuing to examine such papers and I am beginning
to wonder whether the numbers will really add up to anything.

Right now I am reading through a paper by Collar dand Wood (2000) How
reliable are human phylogenetic hypotheses? (PNAS 97: 5003-5006). This
might be cited as another paper providing additional support of the
human-chimp connection, but the majority of characters are extracted from
Shoshani et al who extract many if not most of theres from Groves (1986 and
modified 1995). So one gets a layering of character sources, apparently
without any direct individual observation of said characters.

Further Collard and Wood extract out only 96 craniodental characters for
Gorilla, Homo, Hylobates, Pan, Pongo and an outgroup (Colobus). This seems
to result in some strange characters. For example their character 8 is
presented as autapomorphic for Pongo so its totally uninformative!  Same
for character 24! (which in Shoshani et al is also recorded in several
monkeys so its effectively a plesiomorphic character and probably should
not have been included at all). There are other autapomorphies (which are
probably plesiomorphies in Shoshani et al) as well.

They even include characters that are declared to be in the out group. For
example, character 26 is in Gorilla and the outgroup! Perhaps I have missed
something about cladistic characters, but my understanding was that a
character present in the outgroup suggests it is a plesiomorphy within the
group analyzed.

Character 28 has two states - not developed (0) and developed (1). My
understanding of a binary state synapomorphy is for (1) to represent the
derived state (which is how the distribution goes most of the time in this
paper, but for 28 their distribution is as follows:

Homo 0, Pan 1, Gorilla 1, Pongo 1, Hylobates 0, Colobus 1.

But the most interesting part of this paper was the tallies for particular
clades. There were 11 for Human/chimp/gorilla, 5 for human/chimp, and 10
(!!!) for human/orang (despite not taking all the potential human-orang
synapomorphies.

So on the face of it, 35 characters have been proposed for human-orangs
(for which so far I have not seen any clear-cut refutation of any)
outnumbers the five proposed by Collard and Wood for chimps, and even far
outnumbers their human-chimp-gorilla clade.

Collard and wood prejudice their analysis by stating (p. 5004) to support
the hypothesis that craniodental characters are reliable for reconstructing
the phylogenetic relationships of fossil primate species and genera, the
resulting cladogram matched the molecular cladogram, or a partially
resolved cladogram comprised only molecular clades or the strict consensus
comprised only clades compatible with the molecular cladogram. So here it
is clear that the authors do not view morphology as independent evidence of
phylogeny!

Their parsimony analysis of two quantitative data sets resulted in Homo
being the sister taxon of Gorilla-Pan-Pongo (although I found only three
characters giving this direct support. There are a lot of mutistate
characters and quite a few are confusion in their polarity (eg absent or
not known in outgroup, derived state in Hylobates).

Their parsimony analysis of a qualitative matrix resulted in a Homo-Pongo
in one clade (based on straight numbers for their data this is of no
surprise) and a Pan-Gorilla clade.

Of course nether of these hypotheses are 'supported' because they do not
agree with the TRUE molecular tree.

I could add more, but unless I am missing something fundamental (and I may
have made some mistakes in what is admittedly a quick reading) it seems
that this paper is problematic, not only for its choices and assertion of
putative synapomorphies, but also for its prejudice in favor of molecular
trees as the truth and arbiter of phylogeny

Since there were some very strong orangutan critics posting on TAXACOM I am
sure they will be able to demonstrate how Collard and Wood actually do have
a robust morphological character set, and also demonstrate that any
morphological character not in harmony with phenetic molecular characters
should be discarded - and while we are at it throw out all systematists
working on morphology since genetics is at the truth of the matter anyway
(has anyone yet made this sort of proposal to NSF (or other funding
organizations in other countries) to eliminate funding for morphological
systematics?).

John Grehan




Dr. John Grehan
Director of Science and Collections
Buffalo Museum of Science
1020 Humboldt Parkway
Buffalo, New York 14211-1293
Voice 716-896-5200 x372
Fax 716-897-6723
jgrehan at sciencebuff.org