Archaeopterygid bird from China

[Richard Pyle]

> It's the spatial aspect that's slippery, just as it is for "endemism".
> Everyone knows what "endemic" means: distributed in this place and nowhere
> else. Clear, yes? Actually, no. A taxon endemic to central Florida is also
> endemic to Florida, to the southeastern USA, to North America, to the
> Western Hemisphere and to planet Earth. Which level are you interested in?

The only way I've been able to define "endemism" in a meaningful way is as:
the smallest polygon that can be drawn to circumbscribe all known concurrent
natural occurrences of a specified taxon.  To me, the hairy parts are the
words "known", "concurrent", "natural", and "occurrences".  The word "known"
acknowledges that our current undersatnding of distributions may ne be
complete.  The word "concurrent" assumes one is interested in distributions
at a particular point or defined scope of time.  "Natural" is meant to
except cases of human-mitigated introductions of taxa to places they would
not otherwise occur (ranging from organisms in zoos to species that have
been "naturalized"). The word "occurrences" needs to be defined in the
context of vagrants & waifs, vs. reproducing populations.

> Try zooming in on the map and it's equally hairy, because every taxon is
> disjunct in its distribution at finer spatial scales, and the disjunctions
> get more numerous and more annoying the finer the map scale, until at the
> limit you have point localities whose limits are the physical
> boundaries of the individual.

Yes and no. Certainly you are right, but the "smallest circumscribing
polygon" approach works reasonably well in most cases.  Where it gets dicey
is when you have a *really* disjunct distribution -- such as anti-tropical
distributions (many examples in reef fishes, and I assume other groups as
well).  But that is really an issue of "concurrency" and "occurrences".  If
the taxon is considered to be the same in (for example) both populations of
an anti-tropical species, then "concurrency" is sufficiently broad to assume
that gene flow has recently or does intermittently occur in geographic
regions between the two populations.  Also, the issue of "occurrences" needs
to be defined as to whether it applies only to established/reproducing
populations or (e.g., in fishes) the presence of larvae at intervening
regions between the otherwise disparate populations.

> For practical purposes, many biogeographers delimit the area to which a
> taxon is endemic with a line drawn around all known localities.

This is where we ae of like mind:

> Well, no, again, because (a) for most taxa all peripheral
> localities aren't
> known,

"known", as above

> (b) you get different lines when delmiting by hand, by
> minimum convex polygon and by convex alpha-hull,

....not to mention the inclusion/exclusion of non-relevant habitat (e.g.,
having the contour trace the shoreline of an embayment for the distribution
of a terrestrial species occuring along the entire coastline, or excluding
mountain peaks for a lowlands species) -- something I failed to address
above.

> and (c) for many taxa the time-window for
> which the localities are known is inappropriate

"concurrent", as above

> All of which raises the questions, why think about "endemism" at all? and
> what difference does it make? Now try those questions out on "centre of
> origin".

I believe both are legitimate considerations in science, provided
appropriate context is identified.

Aloha,
Rich