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

Re: 19 Questions About Mickey's Analysis For Mickey



David Marjanovic wrote-


> You must. When it'll be finished, it'll be better than any 2 published
> analyses combined.

Thanks.

> > Perhaps.  I'm confused by the attraction Microraptor zhaoianus and
> > Utahraptor have for each other.  Not always sister groups, but often
> closer
> > to each other than M. zhaoianus is to M. gui and Cryptovolans.
>
> Could there be a typo involved...? ~:-|

I doubt it.  M. zhaoianus and M. gui only differ in their coding of
interclavicular angle (which M. zhaoianus is polymorphic for), sternal
fusion and sternal width/length ratio.  Besides that, the differences are in
what can be coded for each.

Jaime Headden wrote-

>   Well, with the possible exception of those analyses that have examined
the
> material first hand. Note that Mickey himself has tossed out characters
because
> he did not understand them, though it is likely ther original authors the
> characters are largely pruned from did have a concept in mind. An analysis
is
> only as good as its data and the origin of it. This is not a stab at
Mickey, but
> an argument at data and testability.

This is an interesting theory, but one I think is less clear cut than you
make it.  Examining specimens firsthand is certainly preferrable (Carrano et
al.'s ceratosaur analysis should be excellent), but is it always better than
having more characters and taxa if one must choose?  I would say no.  It's
been shown many times adding taxa to cladistic analyses has a huge effect on
the topology, as does adding characters (to a lesser degree per character, I
would say).  If I were to go examine firsthand the twelve relevent taxa used
by Makovicky and Sues (1998), for instance, and code my 200 characters based
on that, I don't think the result would be as accurate as my current ~110
taxon analysis.  The different combination of codings in various OTU's just
change trees so much.  In my current analysis for instance, the taxa
included greatly affect the position of Avimimus.  Sometimes it's an
alvarezsaurian, sometimes an enigmosaur, sometimes an avialan.  Sometimes
alvarezsaurids are arctometatarsalians, sometimes basal maniraptorans.  The
craziness that ocurs when I constrain seemingly obvious topologies should
show what affect a taxon in a certain position has on character
distribution, and thus topology itself.  I could say similar things about
character inclusion.  For all its faults, molecular analysis does have a
certain objectivity that morphological analyses are clearly lacking.  If one
were to add Sereno's 10-13 valid alvarezsaurid + ornithomimosaur characters
to (almost?) every published coelurosaur analysis, I bet alvarezsaurids
would come out as arctometatarsalians.  Even with my 200 characters, none
were designed to support Enantiornithes because they were a single OTU in
the original analysis my characters came from.  So naturally I get
paraphyletic enantiorithines.  They may stay that way, but I'm not even
giving the clade a chance yet.  Everything in my trees is being arranged by
characters meant to see if tyrannosaurids are arctometatarsalians, or if
troodontids are deinonychosaurs, etc..  I'm not going to back my results
until I've scoured the literature for every attempt to support all relevent
clades, and have done a massive detailed comparison between taxa myself.
As for rejected characters, all 42 characters I've rejected have been for
reasons other than a failure to understand them.  Rejection is usually done
because the supposedly apomorphic taxa display the same morphology as
supposedly primitive taxa, or because only one OTU actually shows the
morphology.

Mickey Mortimer