dispersal fantasy / panbiogeography

[pierre deleporte]

A 14:45 27/06/2002 -0400, John Grehan wrote:

>There is a parallel concept in panbiogeography for the apparent
>incongruence between biogeographic patterns suggesting in situ survival of
>old lineages and the young stratigraphy beneath (eg marine sediments
>supporting terrestrial life). A micro-dipping model suggests (for example)
>terrestrial life could persist in an area as it was sufficiently mobile to
>colonize new the terrestrial landscapes that were in a state of flux
>between terrestrial and marine topography. It is possible that the
>terrestrial landscapes were so ephemeral and localized that there may be
>no longer any geological evidence of their existence within the
>stratigraphic record. In a sense the only 'geological' evidence available
>are the animals and plants above - literally the uppermost geological layer.

John, at least your recent posts had a possibly interesting effect (from
your point of view): I just re-read the introductory chapter of
Craw-Grehan-Heads 1999 !

My questions still concern interpretative models.
(Fellows not interested in the panbiogeographic debate, stop here...)

In the above example, you mean old terrestrial lineages (so you have some
molecular clock or else to date the clades) living on young sea deposits
(like wandering mud-banks in mangroves I presume) presently isolated from
solid land masses.
I guess that you introduce an assumption about dispersal abilities of the
organisms in your explanatory model ? In order to discard their possible
recent dispersal onto he spot ?
But Craw et al. 1999 insists on the general irrelevance of dispersal habits...
Maybe I'm over-simplificating Craw et al...

More about "What is panbiogeography" according to this book:
- panbiogeography does not privilegiate vicariance over dispersal for
species formation. The given example is a same distribution both sides of a
barrier resulting from either dispersal or vicariance. OK, but this is not
peculiar to panbiogeography, this very point was made by Nelson and
Platnick for vicariance biogeography: in such a case, when dispersal leads
to the same pattern as vicariance both sides of a barrier (and not anything
else at random), you have "oriented dispersal", effectively mimicking
vicariance in the resulting pattern. Both patterns are indistinguishable a
poseriori, while dispersal at random would give a pattern different from
vicariance.
So, panbiogeography implements the interpretative model: "vicariance or
dispersal", but this should be specified under the restricted acception:
"vicariance or oriented dispersal", which is not dispersal in the general
case as the book suggests.
If I am correct, this would be just an example of "hidden implicit models"
I appeal to dig out of (or from behind, or from under...) everyone's
current practice.

Track: spatial minimal spanning tree connecting localities of a taxon
(population limits ? their geometric centers ?). Phylogeographers routinely
confront minimum spatial spanning trees with minimal genetic distance
spanning trees of the populations they study. It is not, again, a
pecularity of panbiogeography to deal with such spatial data. But
phylogeographers implement models of genetic drift with absolute distance,
of infer bottle-necks.
A track is used to identify "the spatiotemporal coordinates most closely
associated with the origin of [a] distribution"; question: what are the
explanatory models linking tracks and origins of distributions ?
Also, the geometry of a track may be compared to that of "other patterns"
in order to propose "hypotheses or predictions about their origin";
question: which way and why this way ?

Main massing (main area of endemicity): it is "the greatest concentration
of biological diversity within the geographic range of a taxon". I guess we
could say it is the "larger part" of a track (in terms of number of
connected "elementary units" of the taxon involved in this part of the
track). Looking at Fig. 1-6, it seems that some notion of distance is
involved (relatively long length of an edge = "branch" or "link" of the
graph) because several spots are excluded from main massing of B on a
continent. Now taxon "A" shows two main massings: the largest stands on the
central continent (connecting 9 spots), the smaller on the left continent
(7 spots). Note that according to the definition, "the" greatest
concentration would be on central continent (9), not on the left one (7).
Seems that main massing would not be "the" largest concentration, after
all, and rather be a relative notion with nested local definitions
("locally" larger versus smaller adjacent "massings").
There is still a little problem: applying this definition, the two main
massings of A should first include all spots on central continent (10), and
all spots of left continent plus the island close by (10).

Main massings and tracks: main massings in panbiogeography are clearly
equalled with classic areas of endemicity by Craw et al. They do not imply
"centers of origin", they are used to identify "track polarity" (orienting
an edge of the graph from larger toward smaller adjacent massing if I
understand well). Question: if the larger massing is not considered a
center of origin, how should oriented tracks be interpreted? What do they
connect and why? (Model again...).

Node: point of intersection between two or more tracks. May be a boundary
between different tracks (= different distributions): this corresponds
exactly to a distribution limit between sister-groups in Hovenkamp's
approach. But a node may also be an intersection between different tracks,
may exhibit or not local endemics, or local absence of widespread taxa, or
diverse phylogenetic and geographic affinities, or the geographic or
phylogenetic limits of taxa.
I must question the utility of graphical feature covering such a diversity
of aspects and possible interpretations. Why not jump to these aspects and
interpretation directly and dispense with "nodes" ? Just try it...?
(Not to mention the notion of "prominent" nodes being "gates", because
"they are points connecting past history with present-days distributions"...?).

Baseline: spatial correlation between a track and a specific geographic or
landscape feature.
Just one question: on fig. 1-6, we see two baselines. They cross oceans. If
there is no notion of distance in the definition of a baseline, why not
identify a baseline between continent on the left and the nearby island ?
Because it crosses a sea channel. Conversely, if distance matters, why not
identify several baselines inside the continents, between main massings of
A or B and relatively isolated spots, and also two baselines on left
continent before accepting a baseline connecting to the island ? I remember
of a discussion on this list questioning why panbiogreographers focus a
priori on oceans as relevant "geographic and landscape" features and not
other features. What is the model behind ?

Finally, "The baseline does not "prove" any causal association between a
track and a geographic character, but it does provide an explicit
hypothesis of relationship that may be tested by comparaisons with further
distributions". Here appear the notions of prediction and test. But as
stated here, a baseline is a graphic feature without specific
interpretation, but anyway "predicting" that other taxa will have the same
(non interpreted) feature... Correct ? But what would be the interest of
this for biology and geology ? What is the interest of identifying a
baseline, even a striking, huge and beautiful one, if we have no
interpretation ? And if the main massing is not a center of origin (but
what is it) ?
Alternatively, if a link in a track (maybe a baseline) connects spots
resulting from "vicariance or oriented dispersal", then we are facing the
classic program of vicariance biogeography (identifying redundant patterns
of same localities harbouring sister taxa), so why claim superiority of
panbiogeography ?

When panbiogeography has no explicit explanatory model, I suggest it should
have. When it has explicit models, it implements logics already present in
other biogeographic approaches.
A sound defense of panbiogeography should involve the explicitation of all
the explanatory models supporting it and a justification of the
track-nodes-baselines technique as superior or at least equivalent to
competing approaches for historical interpretation of the data under these
explicit models.

cheers,
Pierre