[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Princeton Field Guide
Denver Fowler wrote:
Without wanting to get too drawn into species vs genus etc, one of
the key points I presented at SVP, and that Greg has touched on in
his post, is that by using analysis of stratigraphy and ontogeny
(amongst other factors), species become testable hypotheses. Defining
taxa is NOT just opinion: this is the old paradigm, little changed
form the days of Marsh and Cope, where every slight variation in
morphology was defined as a new taxon. Coding "taxa" for cladistic
analyses cannot test the validity of species in the same way ...
although we can use this approach to show that juvenile specimens not
only tend to plot more basally than the adult forms, but also that
they are often the very reason for polytomy disasters, erosion of
node strength and CI/RI.
Having watched your talk (and John Scannella's) with great interest, I
think the taxa you talk about (anagenetic lineages bounded by
cladogenesis and extinction) are LITUs. They are species under some
concepts but not, or not necessarily, under others; you should not call
them "species" unless you make this explicit.
(Obviously, calling them "species" is probably the best way to deal with
them under rank-based nomenclature which requires that you call
_something_ "species" anyway. Still, whether two organisms belong to the
same species only becomes a testable hypothesis once you've picked a
Defining all variation as 1's and 0's makes morphology black and
white, but with a good fossil record (and the Late Cretaceous of
North America is exceptional) you get so many shades of grey that
this approach is becoming less useful.
This is simply an argument for treating (potentially) continuous
characters as such in phylogenetic analyses. The best way I know is
stepmatrix gap-weighting (Wiens 2001, Syst. Biol.): each measurement is
treated as a separate state of an ordered character (which usually means
that every taxon gets its own state), and the costs of transition
between the states are made proportional to the actual morphological
distances between them (rather than being set to 1 or to 1 / [number of
states in the character] or suchlike).
(Complications can arise because PAUP* only allows 32 states per
character. Wiens  suggested averaging in such cases.)
I have used this approach for two characters in the paper accessible
if you scroll down to issue 3. Was very time-consuming, but worked nicely.