More on Biogeographic memory

John Grehan jrg13 at PSU.EDU
Wed Oct 17 08:07:36 CDT 2001


In reference to Pierre Deleporte

>At 15:53 16/10/2001 -0400, John Grehan wrote:
>
>>A baseline is a hypothesis of historical relationship between the origin
>>and evolution of a track (individual or standard) and the baseline feature
>>crossed by the track.
>
>Many thanks for this statement.
>If I understand well, in Panbiogeography when straight lines (tracks)
>connecting neighbouring localities harbouring identical taxa cross a
>geological feature, and the more if they do this redundantly for different
>taxa, this means that the history of the distribution of taxa (origin and
>evolution of this neighbouring distribution = track) has something to do
>with the considered geological feature?

In my own view the more tracks involved does not necessarily suggest with
some kind of increased statistical confidence that they have any more to do
with the feature than they do individually. That is, at least at present,
there is not a statistical basis for saying if one gets 10, 100, or 1000
tracks with the same baseline that one has reached some sort of threshold
for being 90%, 95%, or 99% confident of that relationship. This is not to
say that this sort of approach may not be possible. My statistical
abilities are minimal in this respect. Henderson ran some Monte Carlo
simulations to demonstate statistical support of global track and node
patterns being non-random.

>If we keep at this level of generality (I translate "hypothesis of
>historical relationships" by: "have something to do with, historically"),
>and if I cross this with your specification further in the message that:
>>I don' think ecological screening is an assumption in panbiogeography.
>
>then I am reinforced in my view that Panb. as it stands is largely
>conceived as an "exploratory" approach (rather than "descriptive",
>finally), trying to identify "patterns of spatial distribution" some
>specific way (minimal spatial spanning networks) without assumptions (or
>restrictions) concerning ecological requirements of the organisms involved
>in the analysis (be they marine, freshwater, flying or subterraneous
>organisms, everything would go without discrimination).

Minimal spanning approaches do not require such discrimination. This is not
to say that they cannot be applied should one wish to do so as you have
suggested. The minimal spanning method is applied in the absence of other
information -  phylogenetic, ecological, etc. that one may wish to apply.

>May I just suggest that you recently stated another way, and quite
>explicitly, treating organisms with coastal habits not in straight line
>connections but along the coasts? (see quotation further below for
>memory). Hence ecological screening would, not surprisingly fo me, also be
>a common (even if generally implicit) assumption in Panbiogeography.

Perhaps I misunderstood your description of ecological screening. I took
the interpretation that ecological screening involved different track
analysis for groups in different ecological categories. However, it seems
that this was not what you were referring to. In terms of your suggestions
about, for example, drawing tracks via river systems - this is a technique
that one is free to apply, such that absolute shortest distance is not
applied. Panbiogeography as I understand it certainly does not preclude
such an approach.

>In a large-scale approach supposed to deal with continental drift (be it
>"just exploratory" or attempting to test specific models) I would admit
>that you treat freshwater organisms like terrestrial ones (all being
>"continental" organisms some way), but even in this case I really don't
>think you would treat marine organisms (or interpret their tracks) just
>like continental ones. Would you?

Heads (1983) pointed out how tracks of coastal marine fishes in the Pacific
showed tracks paralleling terrestrial organisms on various Pacific islands
(eg. the classic Hawaii-Rapa connection). Further, Ocean baselines can be
applied to both marine and terrestrial taxa (applied by Croizat, and for
example, recently for marine mammals in the book I posted earlier).

more later

John Grehan




More information about the Taxacom mailing list