Exist sub-species?

[Frederick W. Schueler]

Panza, Robin wrote:
>
> The only bird literature I know of on this topic is anti-subspecies.
> However, I don't agree with this.  My background in population genetics
> gives me a good but somewhat impractical definition
>   When gene flow is near 100%, there is a monotypic species.
>   When gene flow is near 0%, there are two species.  When there is
> an intermediate level of gene flow, "significantly" below 100% but
> "significantly" >0%, there are two races of one species. When I say
> gene flow, I do not mean hybridization. There must be introgression.
>   I cannot define "significantly",
>  which is part of the impacticality of my definition.
>
> just my 2 cents.
> * well here's my two cents, which I tried to send earlier in the thread,
but which seems not to have come through:

Historically, subspecies started out as separately described species that
were later found to intergrade.  They became a way of partitioning
species to describe geographic variation, and were delimited (Mayr, 1942)
on the basis of the percentage (often shockingly low) of specimens that
could be correctly identified. No distinction was made between subspecies
that were 'taxon-like entities' which happen to interbreed when they are
in contact, and ones that were simply conspicuous cases of geographic
variation, or extremes of a gradual cline.

In the case of widespread continental species, to name subspecies on the
basis of one's ability to distinguish them may divide the range of a
species into a reasonable number of sub-units if only a few characters
are considered. As more characters are considered multivariately, it
becomes possible to distinguish very many local populations, and
subspecies become less useful as a way of dividing a species.

Multivariate geographic variation can be described statistically either
by partitioning the range into units that are considered to be
homogeneous (subspecies; statistically equivalent to an analysis of
variance), or by a spatial model that describes the variation along one
or more geographic or environmental axes (clines; statistically a
regression of some kind). As a simplest case, if a species varies mostly
with latitude, if partioning populations into Xus s. septentrionalis and
Xus septentrionalis australis accounts for 75% of the variation, while a
regression on latitude only explains 50%, you've got a case for using a
partition (subspecies) rather than a cline as your way of discussing the
variation. But regressions on other, non-geographic, axes are also
possible: maybe the break in the variation occurs at the southern limit
of the Boreal Forest, so that percent-Spruce-in-the-habitat explains as
much variation as the subspecies classification.

It seems to me that if subspecies are 'somethings' (='taxa') the
intergrades between them should be hybrids and show both the intermediacy
and increased variability of multilocus Mendelian hybrids. This increased
variability 'shows' that the subspecies are, in some sense, 'as different
as species,' but they just don't happen, for whatever reason, to be
reproductively isolated.  The theory of this is expounded in: Schueler,
F.W., & J.D. Rising. 1976. Phenetic evidence of natural hybridization.
Systematic Zoology 25:283-289, but the taxonomic application isn't, as
one of the authors, was, at that time, a radical binominalist. It's
applied to hybridization between subspecies in Cook, F.R., 1983. An
analysis of toads of the Bufo americanus group in a contact zone in
central North America. Publications in Natural Sciences, National Museums
of Canada 3:1-89.

This depends on the species being widespread enough, and with enough
potential for gene flow, for a surface to be a plausible model of the
variation, and it depends on the acceptance of the biological species
concept, and of the acceptance of the possiblility that similarity
delimited by interbreeding may not be congruent with a non-reticulate
cladistic scheme of evolutionary divergence (since interbreeding
subspecies may be phylogenetically more different than other 'species'
pairs that don't interbeed). But I think it's a reasonable middle ground
between the use of subspecies to describe clinal variation (as decried by
Wilson & Brown, 1952), and using a binomen for every population which may
be, or may have ever been, hypothetically 'evolutionarily independent.'

doubtless I make myself perfectly clear?

fred schueler.
------------------------------------------------------------
         Eastern    Ontario    Biodiversity    Museum
                Grenville Co, Ontario, Canada
(RR#2 Oxford Station, K0G 1T0) (613)258-3107   bckcdb at istar.ca
------------------------------------------------------------