Autapomorphy in multistate characters

Byron J. Adams bjadams at UFL.EDU
Thu Nov 8 10:17:19 CST 2001

on 11/8/01 2:47 AM, David Orlovich at David.Orlovich at BOTANY.OTAGO.AC.NZ

> Hi Taxacom people.
> This question might be obvious but I'm not able to find a firm
> definition anywhere:
> Is a multistate character that has a different state in each taxon
> still called autapomorphic?
> e.g:
> Taxon   Character
> -------------------
> 1       G
> 2       A
> 3       C
> 4       T
Hi David -

By definition, I would agree with you - a character state fixed within its
lineage that is not shared in any other taxon defines an autapomorphic
character state.  However, because it is multistate and not binary, it
cannot be used (unambiguously) to reconstruct the character transformation
or delimit independent lineages.

Using your example, let's assume that taxon 1 is the outgroup taxon.  Let's
also assume that an independent phylogeny yields the relationship among the
taxa as (1(2(3,4))).  There is ambiguity involved in reconstructing the
transformation from the assumed ancestral state (G) to the present states at
the tips of the terms.  For example, we might guess that (A) is the
ancestral condition in taxa 3 and 4 (which subsequently evolved to T & C
respectively).  But it is equally reasonable that A -> T prior to the
divergence of taxa 3 and 4, such that the current presence of (T) in taxon 3
is plesiopmorphic, not apomorphic; a similar argument could be made for the
presence of (C) in taxon 4: A -> C prior to the divergence of taxa 3 and 4).

You don't run into this problem with a binary character because the
direction of the transformation is made unambiguous by character
polarization (via outgroup comparison).

Your example is an important consideration when looking for evidence of
independence among evolutionary lineages.  Detecting lineage independence
requires character independence, and it is difficult to successfully argue
that multiple states of a single character are independent estimates of
evolutionary history.  Thus, for delimiting independent evolutionary
lineages, a single multistate character will not suffice.  The different
character states may be diagnostic (i.e. can diagnose "metaspecies") but are
insufficient to unambiguously delimit independent evolutionary lineages.


Byron J. Adams
University of Florida
Dept. of Entomology and Nematology
Natural Area Drive
PO Box 110620
Gainesville, FL 32611-0260

More information about the Taxacom mailing list