More on Biogeographic memory

Pierre Deleporte Pierre.Deleporte at UNIV-RENNES1.FR
Mon Oct 15 17:52:44 CDT 2001

At 13:11 12/10/2001 -0400, John Grehan wrote:

>As long as the criteria are explicit the choice is open to critique(...)
>There are many possibilities for track construction and even
>though our book emphasized the strict minimum spanning tree this is not to
>preclude those alternatives.

All right for me, provided the criteria are not only explicit (clearly
stated), but justified in historical biogeographical terms (why these
criteria, what does the ensuing pattern is supposed to mean).

>What may be interesting is the spatial congruence of groups that do not
>have comparable ecological 'skills.' This was a major finding in Croizat's

Indeed this may be interesting, if the result appears paradoxical regarding
what our models predicted. Provided that we had models in the first place,
i.e. an expected pattern given some ecological skills and geology, to be
constrasted with an "observed" pattern (in fact a constructed, or
"reconstructed" historical pattern).
Here the methodological analogy is quite clear with "testing predicted
evolutionary scenarios on a reconstructed phylogeny".

>So is this to say that comparing a distribution pattern of butterflies (for
>example) with river spatial networks could not make sense?

It depends on what you pretend to know of the ecological skills of your
butterflies. If they have aquatic larvae and very limited flight dispersal
abilities, treating them as aquatic taxa may make sense. On the contrary,
if they are terrestrial and cannot easily cross a river, treating them with
the terrestrial fauna of inter-river blocks makes sense. If you have no
idea, what can I say? Don't be surprised of any pattern you will find, if
you can't figure out what can constrain a butterfly's distribution.

>The problem in biogeography is that the link between a distribution and
>such geological features has been arbitrary. Often the choice appears to be
>minimal spanning (baseline) links as the implicit operational procedure.

This is likely the case, hence my insistance for explicitation.

>As far as I am aware, panbiogeography is the only historical biogeography
>method that provides an explicit formulation of spatial relationships.
>Form systematics is about biological homologies. Panbiogoeography is about
>spatial homologies and spatial correlation

The puzzling question remains: what is the historical signification of this
"pattern of spatial relationships"? How do we interpret it (historically),
and why?
What are "spatial homology" and "spatial synapomorphy" or "spatial
correlation", and do they what mean? I guess I know what they are (shortest
spatial distances some way). But what do you consider they mean, and why?

>So far it [panbiogeography] has been very successful indeed in
>mapping the global structure of biogeography with its successful
>(corroborated) geological predictions and even some biological predictions.

Specifying these "predictions" could perhaps help answering the questions
above (meaning of a spatial "homology"?).

>Further developments in spatial analysis for historical biogeography would
>be very welcome. I have always been frustrated by my lack of
>methodological knowledge to be able to investigate graph theory and
>develop algorithms etc. So I must sit on the sidelines.

As already underlined by Hovenkamp, I don't think the problem is with graph
theory first. It is with clarifying our goals and assumptions in
biogeography, before we go and ask the "graph theory people" exactly what
we need.

>>The presentation of a method in biogeography should ideally begin with
>>statements of this kind (exactly as for methods for phylogeny
>>reconstruction: which process is required for the particular kind of
>>pattern to make sense?).
>This is just one opinion of what is ideal. There will no doubt be others.

No doubt, but we can argue: I suggest this is "ideal", I could say "highly
recommended", because a logic is thus provided from the beginning for
interpreting the patterns we will draw this or that way. Otherwise we are
at risk of arbitrarily drawing a pattern and trying afterwards to charge it
with some meaning... which could never succeed.

>One may also draw tracks within a taxon solely on minimum distance
>criteria irrespective of sister group relationships.

Of course "one may" (and a lot of other possible things)... but how would
you interpret this historically, and why?
Maybe one could contrast the spatial connections with the phylogenetic
connections in order to test some model (e.g. that shortest distances
should stand between phylogenetically closest groups for some reason, like
a process of generalized speciation by vicariance and limited
dispersal...). But to my knowledge this is not part of Panbiogeography
presently, because Panb. is "just maps"...

By the way, you are considering a taxon, and sister groups relationships
within a taxon. Seems clear that the taxon stands possibly above the
species level. Do you have a criterion for fixing at what "level" you
delineate a "taxon" you use in the analysis (because maybe the whole tree
of life is a taxon...).
This question is not specifically addressed to panbiogeography anyway.
Rarely do biogeographers argue about this problem of delineation (should
involve absolute datations in relation with geological features at stake?).

>If one defines systematics broadly enough it could include biogeography.

Are you suggesting to make localities or spatial distributions a taxonomic,
or even phylogenetic, character? Combined with biological characters of
organisms? Or rather characters for a separate classification ("Pacific /
Atlantic organisms" for instance...) to be contrasted with systematics /

>Morrone also showed how apparent incongruence of form relationships may
>represent the presence of more than one track relationship that can be
>separated out using panbiogeography.

This could perhaps help ( understanding what a track (and a
track  relationship) means.

>>Otherwise, Panbiogeography subsumes to systematics some way. Guess this
>>is logically unavoidable, given that we deal with geography of TAXA.
>Or we deal with the taxa of GEOGRAPHY!

Seems that you are trying to deal with some "geographical taxa"
(geographical relationships, tracks...) of GEOGRAPHY OF TAXA (spatial
minimum spanning trees for localities of TAXA). Thus, "taxa" of geography
of taxa.


More information about the Taxacom mailing list