Deleporte and Kinman on the Pacific

[John Grehan]

At 04:53 AM 7/3/02 +0000, you wrote:
>And my point was that the Pacific has been an effective barrier for
>long distance dispersal "over the water".

Thanks for the clarification. My point is that the evident trans-Pacific
patterns of distribution for many taxa means that the Pacific has not been
a 'barrier' to their establishment on either side.

>This would not apply in the far
>north (Bering Strait) or far south (Antarctica) where the Pacific has been
>bridged (or nearly bridged) in the past.

This might be the crux of Ken's position. Since it is not explicit so far I
may be wrong in my inference. Ken seems to be basing his Pacific
biogeography on the prior acceptance of a particular geological history for
the basin. This history is that the Pacific has always had the status of
being oceanic (a water filled body with volcanic islands without any
geological formations supporting a terrestrial life that have had any
contact with the surrounding regions). From this geological position Ken is
forced to 'explain' trans-Pacific patterns as either 'long-distance'
dispersal over water, or circumglobal travels via landbridges to the north
or south. Thus Oreobolus is on both side of the Pacific because it must
have dispersed from one side to the other via a landbridge - that being
Antarctica in Ken's opinion. If this is not Ken's position then he will
correct me. It is certainly the kind of model postulated by Darwinian
biogeographers in both pre and post-plate tectonic times.

>Why panbiogeography would
>downplay these disperal routes makes no sense to me at all.

I presume Ken has no familiarity with panbiogeography and the analyses
carried out from the time of Croizat. Croizat (1952) embarked on a test of
theorized dispersal routes by examining the concordance or lack thereof,
between patterns of geographic distribution and both dispersal ability and
current geography. He found no correlation with either, but rather with
tectonics, and from that correlation he made geological predictions that
have since been corroborated.

>I find extremely dubious and only
>supported by an arbitrary and unproven "minimal" spanning convention and a
>cladistic analysis which the author (Seberg) himself admits is rampant with
>homoplasy.

Again, the 'unproven' assertion is a bit misleading given the literature
published on the subject. Even the Pacific patterns have been shown to have
a non-random structure using Monte Carlo techniques. Agreed the cladistics
of Oreobolus may be dubious, but given the current arrangement there is a
distinct difference in the findings of vicariance and panbiogeographic
techniques applied to the same data. It will be interesting to see how the
tracks and nodes identified with the current data stand up against future
revision.

>      You have already abandoned your long trans-Pacific track for primitive
>frogs, and I predict you will have to do so for a lot of other groups as
>well.  If your Oreobolus paper doesn't convince me (someone who thinks
>panbiogeography does have great potential, although in a less strict form),
>then how can you expect to convince those who don't think panbiogeography
>has any potential at all?

Being convinced is the right of the individual. The purpose of my papers is
not to 'convince' them of anything. Thinking people convince themselves or
they are simply sheep. I present a case just as a lawyer might do. The Jury
will make its own decision right or wrong. There are people who are
'convinced' of the efficacy of panbiogeography. I think the writings of
myself and my colleagues have had something to do with this, but not
everything.

Pierre Deleporte wrote:
Not that biogeographers, even using phylogenetic information, are
neglecting distributional data. Thay merely integrate this in their
reasoning a non-formal, generally implicit and eclectic way. The recent
post by Ken Kinman in reply to John Grehan is exemplary: Ken, pressed by
questions, explained the geological arguments standing behind his
acceptation of minimal spatial spanning tree as meaningful across Atlantic
rather than across Pacific. Thus, he certainly does not ignore spatial
information in his reasoning.

What Ken was doing was 'explaining' spatial patterns according to a
particular model of geological history and a belief in active dispersal. He
was not using spatial information as an analytical criterion.

  It depends on the dispersal abilities
of taxa combined with geological information about barriers.

If that were so, then panbiogeography could not have discovered the
existence of correlation between tracks and tectonics, or been able to
predict novel geological facts. I think that whatever position one takes
for or against panbiogeography on philosophical grounds, the critical
question is why, if the method is so suspect, did Croizat get it (the
geological predictions) right?

For me, it depends on the landscape, geological information, and inferred
dispersal abilities of the taxa... precisely not "just-so distance", Ken !

This is the foundation of Darwinian dispersalism so it is not surprising
that Pierre does not like panbiogeography.

Isn't the composite tectonic structure of the Americas appearing from a
vicariance biogeography analysis too ? If so, no superiority of
Panbiogeography could be claimed, only identity of result in the particular
case, with no guaranty for other cases.

The composite tectonic structure is BACKGROUND KNOWLEDGE for the vicariance
studies. It was not background knowledge for panbiogeography. At the time
Croizat predicted the composite tectonic structure of the Americas it was
not a known fact. This is a critical difference. Panbiogeographic theory
and methodology make predictions about geology that went ahead of the known
facts, and these predictions were later corroborated. This has not happened
yet, to my knowledge, with vicariance cladistics although it is conceivable
that such novel predictions could be generated.

The empirical argument "in worked in this case" is of no use for defending
the logical relevance of a method. the argument should be why it should
"work" in most cases (if "working" makes general sense at all for
historical inference: we have no time machine to go back and check...).

I'm not aware than any of Croizat's tectonic predictions did not work.
Perhaps from a philosophical perspective working in particular cases may
prove nothing in general. I take the view that the fact that
panbiogeography was right in these particular cases is something of general
interest.

And anyway, isn't any tentative historical
reconstruction more or less convincingly argued "speculation"? So where is
the point?

Panbiogeography offers a method for constructing geological hypotheses
based on the principle of spatial correlation between biological patterns
of distribution and tectonic patterns of distribution. This approach is
less about speculations as it is about analytical procedures connecting
biology and geology without simply relying on historical narratives from
the latter.

John Grehan