[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

cladistics (was giant birds)



I don't do cladistics myself, but...

Rob Meyerson <rjmeyer@ix.netcom.com> wrote:

> Let's say that a researcher is attempting to build a cladigram for
> the dromaeosaurs.  That person goes through the rigors of a cladistic
> analysis.  Now, lets say that, by some horrible bookeeping error, one
> finds that the data for a sauropod and (HORRORS) a ceratopian has
> been included with the data for dromaeosaurs.  Now, according to
> cladistic theory, the resulting cladigram should look something like
> this:
> 
>         DM        SP     CT
>         \         /      /
>          \       /      /
>           \     /      /
>            \   /      /
>              V       /
>               \     /
>                \   /
>                 \ /
>                  V
> 
> Where CT = Ceratopian; SP = Sauropod; DM = Dromaeosaur Mess
> 
> The question I have is: Has this been done?  Will we get the result
> above or something else?  If we get a cladigram that includes CT
> *within* DM, we could be in trouble.

The situation you describe is in fact standard practice.  When
constructing a cladogram for a group, representatives of a few
'outgroups' are included in the analysis.  A ceratopian and a
sauropod would do.  (It would be even better to use theropods that are
outside Dromaeosauridae.)

The cladistic analysis will show that the dromaeosaurs are more
closely related to each other than to the outgroups.  But it will also
find which dromaeosaurs are more closely related to each other.

So what are the outgroups for?  The analysis will look at all the
characters and produce an 'unrooted tree'.  Of course the results are
statistical and only say what is more or less likely, but I'll ignore
that aspect.  For dromaeosaurs, the tree would look something like
this:

A-----------B
      |          E
      |          |
C------------------F 
            |
            |
            D

This shows, for example, that A has more character states in common
with B than with anything else.  But it doesn't tell you whether these
are derived states (because A and B are descended from a more recent,
specialised dromaeosaur), or primitive ones (because A and B both
retain lots of primitive characters from the common ancestor of all
dromaeosaurs).  We are only interested in shared derived characters.*

By including representatives of outgroups, it is possible to root the
tree.  If we find:

A-----------B
      |          E
      |          |
C------------------F
            |
   SP-------|
      |     D
     CT

This shows that D shares some character states with SP and CT.  D
cannot share a more recent ancestor with SP and CT than it does with
the other dromaeosaurs.  So these character states are probably
primitive states.

Using this kind of logic, we produce the cladogram:

A  B  C  E  F  D  SP CT
\  /  /  \  /  /  /  /
 \/  /    \/  /  /  /
  \ /     /  /  /  /
   \     /  /  /  /
    \   /  /  /  /
     \ /  /  /  /
      \  /  /  /
       \/  /  /
        \ /  /
         \  /
          \/

(((((A B) C)(E F)D)SP)CT)

Of course, this depends on the assumptions.  The outgroups must really
be outside the group you're analysing, a safe assumption in this case. 
If we assume SP is closer to dromies than either is to CT, that's less
certain.

> Along these lines, since the researcher is the one who decides what
> characters are used in a cladistic analysis, does this make the
> cladigram subective?

By picking and choosing characters, you can generate a cladogram to
support almost any hypothesis.  But you'll have trouble convincing
other people (except those who already agree).  The best policy is to
include as many informative characters as possible.

Bill Adlam

*Only shared derived characters are useful for reconstructing phylogeny
(common ancestry).  Whether classification should be based on phylogeny
is another matter.  I think it should.
__________________________________________________
Do You Yahoo!?
Bid and sell for free at http://auctions.yahoo.com