More on Biogeographic memory

[Pierre Deleporte]

At 10:33 16/10/2001 +0200, Peter Hovenkamp wrote:


>The basic point of disagreement remains. I would like to let the patterns 
>we observe speak for themselves. Opposed to that, we find John Grehan, who 
>claims that knowledge about tectonic features should be used to decide 
>exactly which patterns can be compared and which had better not; and 
>Pierre Deleporte (bien ?tonn?s de se trouver ensemble...), who claims that 
>knowledge of ecology should be used in the same way.

I'm saying that ecology CAN be used if convincingly argued. The same vails 
for tectonics.
I would possibly object to John Grehan a certain lack of justification of 
the "ocean basin" tectonic constraint for the analysis, not the 
implementation of a constraint in itself.
Now as for "finding ourselves together" with John, there's no shame for me, 
but may I suggest that you just forgot one in the bag: Peter Hovenkamp himself.
For the plain reason that "patterns" cannot "speak for themselves" like 
that (maybe you disagree, but that's a  key point in my view).

Patterns say nothing. Patterns of what, and why? Patterns say nothing for 
themselves (or by, or in themselves?) without a logic, a theory, for 
interpreting the pattern. There is no '"neutral observation of objective 
patterns speaking for themselves". There cannot be, for reasons of logics.
And I already suggested that we better have an a priori idea about how to 
interpret a pattern of some kind, rather than draw a pattern for pattern's 
sake (!) and wonder afterwards what we can do with that. Because patterns 
in themselves are stubbornly dumb.

At the very limit, maybe we could "explore data" by drawing "any possible 
kind of patterns" out of them, and hope that some ideas for setting a 
problematic could emerge... maybe... but I don't think many people explore 
data like that. The current situation is the implementation of explicit 
theories, and frequently implicit theories, so that patterns make sense.

>To both I say: if either tectonics or ecology is really important, you 
>will be able to find traces of it in the result of the analysis.
>To John Grehan: if the Indian Ocean is one tectonic feature, you can 
>expect to find a single consistent pattern - if it is not, you may find 
>different general patterns involving the northern and the southern part...

Agreed, provided that "an ocean basin being a tectonic feature" predicts 
"tracks will be this way". We need an explanatory law, or a predictive 
theory, according to which tracks and patterns of tracks make sense, not 
just "consider a pattern" like that.


>To Pierre: The way across a mountain range may not be the "shortest" way 
>to connect two groups of fish - but if it repeatedly turns up between 
>different pairs of sister-groups of fishes - then maybe the mountain is a 
>relevant feature after all...


OK you can try it this way: trace straight lines through mountain ranges as 
the crow flies between sister groups of fishes and wonder. But at some 
point the fact that you are dealing with FISHES will assuredly be put into 
the debate. If not by you, no doubt by your readers. And geology also can 
fall into the debate. But I'm pretty sure that Peter Hovenkamp himself will 
not reason with fishes as with any other possible kind of living features. 
I suggest that we better face this "background knowledge" explicitly.

For example, redundant "straight lines as the crow flies" for sister groups 
of fishes can also be redundant "shortest connections downstream, then 
upstream, via the river network"; in this case maybe you will not be able 
to decide immediately between:
- a history of capture of an upper river system by the adjacent river basin,
- and a faunistic exchange via a downstream passage between adjacent rivers 
and fishes getting upstream afterwards (e.g. contraction /extension of the 
river system through periods of aridity / heavy rainfalls, hence upstream 
infeodated fishes populations get ecologically connected then separated). 
Then geological and paleoclimatological features may also come into the 
argument: hence using total relevant evidence for reconstructing the past, 
as you "whoopee" it just below...

>>>Whoopee again. I totally agree that explanation of individual taxon
>>>distributions should be based on all relevant evidence.
>>
>>Except spatial evidence through baseline homology apparently does not
>>constitute evidence at all in your view.
>
>(if only I knew exactly what it was...)

Here I agree with Peter. If only I could grasp what a baseline means 
historically.
To answer John's last message by the way: my starting point is precisely a 
lack of explicitation of that in the panbiogeographic litterature (or 
possible lack of understanding of this literature from myself), thus 
sending me back to this litterature for explanations just closes the loop, 
and I think I will stop circling round.

But given the above development, I fear I could have to ask Peter too: what 
does a division between adjacent sister-groups distributions mean 
historically? It simply doesn't "speak in itself" to me. If you tell me of 
a nested pattern of more and more inclusive sister-groups distributions 
indicating a histocrical succession of vicariance events, I guess I will 
find a historical dimension... but only because we would have an explicit 
explanatory model (vicariance is the process, phylogeny gives relative time 
of divergence and so on...), not just consider a "pattern in itself".
Looks like I am now putting Peter an John "in the same bag" some way, to 
their great astonishment... To be fair, I guess Peter will agree that he 
has models in mind, simply he tries to develop an approach independent from 
"tectonics", thus susceptible to be confronted to tectonic models, and that 
it is what he meant by "let the patterns we observe speak for 
themselves"... just relatively to tectonic hypotheses.

Maybe I was thus partly "playing the deaf" with Peter... but I maintain, on 
general grounds, that "patterns in themselves" are definitely dumb, and 
that "ecological screening" is one on the commonest implicit assumptions in 
historical biogeography, which would gain to be explicit and thus open to 
argument and criticism.

Pierre