Human and ape phylogeny

[pierre deleporte]

Hi John,

the current way to deal with cladistic analysis of morphological data is as 
follows :

built a data matrix of terminal taxa X characters  -  you have to define 
the characters and delineate their character states (generally a headache)

set one or several "outgroup" taxa in the series of taxa contained in this 
data matrix (in other words, introduce some putative "outgroup" taxa in the 
data matrix)

the analysis will find the shortest tree(s), and root the tree(s) according 
to your a priori choice of outgroup(s), i.e. manage to make them appear 
"outside" the clade constituted by the remaining taxa, i.e. place the root 
between outgroup and ingroup

this is always possible with only one outgroup taxon, this is not always 
possible if you have set several putative outgroup taxa (if they don't 
appear contiguous on the tree, you will have no unique solution to root 
your tree, and this will indicate that you have made a mistake in your a 
priori ingroup / outgroup delineation).

as you describe the method in the Gibbs et al. paper, they have fixed 
Hylobates as outgroup, so they are in fact analysing the relationships 
between the remaining anthropoids, implicitly assuming that Hylobates is 
effectively outside the clade constituted by the remaining species

this "arbitrariness" in the choice of outgroup(s) is inherent to all 
cladistic analyses
so the choice should be argued (explain what external evidence support the 
a priori out/in group decision)

an obvious limit of the method is the difficulty to find any outgroup in 
order to root the whole tree of life in the first place... you should 
introduce a non-living thing in the data matrix (!)...

thus this way of rooting cladograms is obligatorily based on some previous 
decision about a larger scale phylogeny (who is in, who is out), but this 
does not constrain the internal structure of the cladogram, which will 
result from the analysis (most parsimonious tree)

in the present case, setting Hylobates as outgroup does not influence in 
itself the relative connections between Gorilla, Pans, Pongo and Homo

the "old fashion" procedure consisted in trying to a priori polarize at 
least some of the characters, which could be done either by an implicit 
"outgroup rooting" applied to these characters, or assuming a particular 
polarity of the evolutionary process for these characters

Gibbs et al. cannot be said "pheneticists" just because they use the 
outgroup criterion for tree rooting. They would be pheneticicts if they 
computed an overall similarity index to perform a cluster analysis instead 
of minimising the evolutionary events (character changes or "steps") on the 
tree (parsimony criterion applied to choose the optimal tree under a 
"minimum evolution" model = "minimising homoplasy" = "maximising homology").

maybe a difficulty comes from a possible "presence/absence" coding of some 
characters. The two states of a character may be "present" and "absent", 
but this will not make the result of the cladistic analysis similar to that 
of the phenetic analysis of the same data matrix. In case of regular 
divergence on all branches, the result will be the same ("clocklike" 
evolutionary process); but in case of differences in evolutionary rates on 
the branches ("acceleration" and "slowing down" of evolution), the 
cladistic (parsimony) analysis will tend to group by synapomorphies only, 
when the phenetic analysis will happen to group on symplesiomorphies when 
they match overall similarity between two taxa.

this remains true in the general case, wherever your outgroup will happen 
to be situated and "root" the tree... and there stands also a difficulty I 
guess, for people not familiar with the approach: identifying the shortest 
tree and rooting it by trhe oyut rgroup criterioj provides exactly the same 
result as polarizing every character on an outgroup basis and performing a 
traditional phylogenetic analysis by hand "Hennig's fashion".

hope this helps,
best,
Pierre

---------------------------------
A 09:55 04/04/2003 -0500, John Grehan wrote :
Here's a question for all the cladistic experts on this list. I admit to
having a relatively superficial expertise with cladistics so I want to
measure my current perceptions against any responses on this list. My
question concerns the 'cladistic' quality of an article cited on this list
as increasing the support for a human-chimp/african ape clade vs human
orangutan by S. Gibbs, M. Collard, and B. Wood (2002) Soft-tissue anatomy
of the extant hominids: a review and phylogenetic analysis. Journal of
Anatomy 200: 3-49.

At first glance the paper seemed very impressive. The authors utilized 171
characters and a cladistic analysis. The number of taxa was limited to
humans and extant apes only which did make me wonder if some of their
characters might not qualify as synapomorphies with further comparison
including at least the old world monkeys. Then I read that "No a priori
judgements were made as to the primitive or derived condition of
characters". My reading of that statement is that the characters were not
'cladistic' in the sense of each standing as proposed synapomorphies.
Instead they were phenetic.

The authors then state that "Hylobates was assumed to be the basal hominoid
genus and the cladograms were rooted accordingly". Since the characters
were not defined as synapomorphies with respect to an outgroup, making one
of the taxa the basal lineage seems to be an externally imposed criterion
rather than being generated from the characters themselves. With this root
the characters were subject to 'cladistic' analysis.

Am I correct to view this paper as a 'cladistic' analysis of phenetic
characters with an arbitrary rooting of one of the taxa being analyzed. If
I am then this paper hardly seems to me to stand up as a reliable support
for the human-chimp clade. I am getting old and perhaps my understanding of
cladistics based on what I read by Hennig, Nelson where cladistics was all
about the analysis of proposed synaomorphies is now out of date and I
missed the boat where cladistics is now all about the analysis of phenetic
characters.

John

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

Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 66
Télécopie : 02 99 61 81 88