Linked or associated characters

[Curtis Clark]

On 2005-07-08 04:15, Robert Mesibov wrote:
> One thing that's always puzzled me about character analysis, say in
> cladistics, is how you tell whether two characters are independent, or
> instead are linked or associated.

There is a general problem with null models in cladistics, in that
perfectly informative homologies can never co-occur randomly (for
example, we would never expect to find an organism simultaeously
possessing mammalian hair and an avian syrinx). So there are (at least?)
three non-homoplastic sources of congruence:

1. "Independent" characters that co-occur as synapomorphies (e.g. hair
and temporomandibular joint).
2. "Independent" characters that are linked by development or selection
(your "association"?) so that the occurrence of one without the other is
impossible or improbable (e.g., trivially, presence of petals and petal
color).
3. Characters that are manifestations of the same genetic system, hence
are not independent (e.g. increased pollen size and increased guard cell
size in a polyploid).

Parsimony models cannot distinguish the three.

This would not be an issue in the absence of homoplasy, since you'd get
the same tree anyway, but homoplasy is a fourth source of congruence.
Incorrectly counting linked or associated characters as synapomorphies
will skew the results, and perhaps lead one to draw conclusions that are
not supported by the data.

--
Curtis Clark                  http://www.csupomona.edu/~jcclark/
Web Coordinator, Cal Poly Pomona                 +1 909 979 6371
Professor, Biological Sciences                   +1 909 869 4062