Joe, you are still trying to defend the BSC by attacking the=20
PSC, and we are not going to get anywhere arguing about the=20
PSC. The BSC'=92s weaknesses are independent of the PSC or any=20
other species concept. The inadequacy of the BSC to people=20
who desire genealogically coherent taxa is the issue=20
here. Old time BSCers were no=92t concerned with genealogic=20
coherency. Many modern systematists are concerned, though. =20
Whether we settle on something that looks like Cracraft'=92s=20
version of the PSC or something that is more explicitly=20
phylogenetic remains to be seen. =20

Consider the following real-world example involving orioles:

---------------Bullock'=92s *
|
|
---------------Streak-backed
-----|
| ---------Altamira
| |
------|
|
---------Baltimore *

The diagram above is based on Freeman'=92s 1991 (U Washington PhD=20
diss.) analysis of mtDNA restriction sites. Baltimore and=20
Bullock'=92s orioles hybridize with one another in the Great=20
Plains and were lumped together as a single species on that=20
basis -- the Northern Oriole. But Baltimore is genealogically=20
closer to Altamira, which is regarded as a distinct species,=20
that it is to Bullock'=92s. In other words, one BSC subspecies=20
(Baltimore) is more closely related to a different species=20
(Altamira) than it is to a conspecific subspecies (Bullock'=92s). =20
This simply makes no sense. Similar examples of the=20
genealogic flaws of the BSC'=92s hybridization criterion exist in=20
chickadees (Gill et al. 1993) and Towhees (cited by Zink in a=20
talk here; ref?). =20

You still have no=92t told us why we should accept the BSC's
lumping of phenotypically similar allopatric populations on=20
the basis of "potential interbreeding," when we know that=20
gross phenotypic similarity is a notoriously unreliable=20
indicator of genealogic relationships. This flaw and=20
the previous one exist in the BSC simply because its architects=20
were not very concerned with genealogy. Your claim that the=20
BSC does no=92t make phylogenetic statements is simply wrong -- it=20
is precisely because the BSC makes *misleading* phylogenetic=20
statements (e.g., relationship of Bullock'=92s, Altamira and=20
Baltimore orioles in above example) that many systematists are=20
looking for alternatives. =20

Moving away from the classic BSC does no=92t mean we have to=20
ignore gene flow among populations, etc., and rely on a starck=20
version of the PSC. A reasonable approach is to actually make=20
phylogenetic trees that relate populations in question to one=20
another and use these trees as a context for evaluating=20
significance of gene flow among populations, etc. You could probably=20
even do this and hang onto subspecies, which is what I mean by=20
"revamping the BSC." =20

David Wright
dwright at u.washington.edu




The Zink paper (Evolution, 1994) cited by Dennis in an earlier=20
message in this discussion shows than Zink has retreated from=20
the "one barbule" position of McKittrick & Zink 1988. =20
Basically he is looking for agreement in disparate sets of=20
characters (for example, plumage and mtDNA) in diagnosing=20
different populations as species). This "retreat" simply puts=20
Zink more in line with Donoghue=92s version of the PSC. =20

Mitochondrial DNA is not a derived character. Derived=20
characters can be recognized in mtDNA, in the form of=20
nucleotide substitutions, loss or gain of restriction site,=20
etc. But if these are distance data, which is how you=20
identified the human data (I haven=92t seen the paper), they are=20
not derived characters, characters are not pratitoned into=20
"primitive" and "derived." At any rate, I suspect even the=20
most hard-boiled PSCers would decline to split humans into=20
separate species on the basis of any evidence, for the same=20
reason you are reluctant to divide us into subspecies. =20