[Taxacom] sloppy cladistic analyses

[Kenneth Kinman]

Dear All,
     Stephen is correct in my opinion.  The narrative which attempts to
explore intermediate steps in the evolutionary process can be largely
lost in the labyrinths of large datasets (which may not be all that
accurate anyway if workers are taking "shortcuts" and/or don't fully
understand or attempt to fully evalutate the datasets they have
inherited from others).  It's even worse for fossil taxa which are
constrained by very limited data from the get-go.  No molecular data at
all to challenge them.
      The new Haplocheirus fossil is an excellent example.  Even Mickey
Mortimer (who still believes a broad "Alvarezsauridae" is a holophyletic
group) complained that the 2010 phylogenetic analysis left out
characters which could potentially unite "derived alvarezsaurs" with
birds.  I have long believed that these "derived alvarezsaurs" (my Order
Mononykiformes) are indeed more closely related to pygostylian birds,
and that the more "primitive alvarezsaurids" are only convergent with
Mononykiformes.  Mortimer has so far disagreed with me, but is equally
frustrated by sloppy cladistic analyses which don't really clearly
address the issue.   I strongly suspect that it is somewhat similar to
the evidence that has long pulled Collembola into a grouping with
insects and other hexapods.  Although I am almost totally convinced that
Hexapoda sensu lato is polyphyletic, I am not quite as certain about
Alvarezsauridae sensu lato also being polyphyletic (although it would
not surprise me).  
       If we can't even resolve hexapod relationships with the relevant
groups being extant, it is not at all surprising that fossil groups like
primtive Alvarezsaurines and derived "Mononykiformes" are so
problematic.  Probably best to classify the primitive forms in Order
Alvarezsauriformes and the derived  forms in Order Mononykiformes.  The
best Mortimer can hope for is that Order Alvarezsauriformes is
paraphyletic with respect to Mononykiformes, but if I am right, they are
separate clades which would be polyphyletic if united as so many others
have advocated due to various (perhaps faulty) cladistic analyses in the
last 15 years.
         --------Ken Kinman
     
--------------------------------------------------------
Stephen wrote:
No, I didn't mean that! I mean that giving a huge numerical dataset and
saying "this is what comes out of the analysis" leaves us with little
hope of spotting errors, and a high probability of coding errors
happening, as lots of numbers don't mix well with scientists in a hurry.
A "narrative" doesn't mean a preconceived wish for a desired conclusion,
it means using words to explain each step of the process so that we can
understand and follow the steps that lead to the conclusion... 
Back to my dungeon! 
Stephen 
________________________________________ 
From: taxacom-bounces at mailman.nhm.ku.edu [taxacom-bounces at
mailman.nhm.ku.edu] On Behalf Of Jason Mate [jfmate at hotmail.com] 
Sent: Thursday, 4 February 2010 11:57 a.m. 
To: Taxacom 
Subject: Re: [Taxacom] sloppy cladistic analyses 
I mistakenly sent my previous email to Ken, but I have changed it in
light of recent postings. 
Ken: 
-Define "information" versus information. 
-Linked characters matter a lot more than coding mistakes (as long as
these mistakes are random and not somehow biased due to the reasearchers
preconceived ideas) 
-I have never taken information from the internet seriously. 
John: "...but in the absence of any demonstrated evidence to the
contrary I won't lose sleep worrying about that with respect to the
current phylogeny produced by a particular analysis." I believe they
call it congruence (or lack of in some cases). 
"...'law of large numbers' that overrules the 'law of shared derived
characters'..." They are not laws, they are more like guidelines...
Seriously, the old quality versus quantity? 
Stephen: "systematics which depends heavily on "number crunching". They
have a tendency to lose the "narrative" - a huge data matrix and
associated tree are not a narrative!" I am guessing that you mean a
prior hypothesis of relationships somehow guiding (= massaging) the data
til the "congruent" answer emerges? The "fleshing-out" of each step
speaks of a posteriori interpretation so I am unsure. 
Back to my hole. 
Jason