Thursday, March 6, 2014


Recently I got into a discussion with somebody who dismisses philosophy as useless. What particularly fascinated me was his touching faith in the capability of science (as opposed to philosophy) to "weed out" nonsense. This got me thinking about the apparently never-disappearing nonsense in my area of science.

Looking at the bar on the right, one might suspect that I am referring to "evolutionary systematics", to those colleagues who are still promoting the acceptance of paraphyletic taxa. Yes, of course I think their arguments are wrong, but that is not what I mean now, because whether to classify by descent or by superficial similarity is ultimately dependent on what we want a classification to do, not on evidence. Instead, I am thinking of schools of science in my area that are, at least in my eyes, based on an ideologically motivated rejection of empirically demonstrable facts.

Today I want to introduce one of those that come to mind: panbiogeography.

To cite from the website linked above,
The panbiogeographic approach, originally proposed by Léon Croizat (1958), basically plots distributions of a particular taxon or group of taxa on maps and connects the disjunct distribution areas or collection localities together with lines called tracks. A track is a representation of the spatial form of a species distribution and can give insights into the spatial processes that generated that distribution. Crossing of an ocean or sea basin or any other major tectonic structure (eg. a fault zone) by an individual track constitutes a baseline. Individual tracks are superimposed, and if they coincide according to a specified criterion (eg. shared baselines or compatible track geometries), the resulting summary lines are considered generalized (or standard) tracks. Generalized tracks suggest the pre-existence of ancestral biotas [sic], which subsequently become fragmented by tectonic and/or climate change. The area where two or more generalized tracks intersect is called node. It means that different ancestral biotic and geological fragments interrelate in space/time, as a consequence of terrain collision, docking, or suturing, thus constituting a composite area. A concentration of numerical, genetical [sic] or morphological diversity within a taxon in a given area constitutes a main massing.
All clear? So far, we learn that panbiogeography consists of inventing a whole slew of arcane terminology, drawing lines across a map, and then trying to infer something about ancestral biota and biogeographic history from those lines. One might wonder if it would not be more productive to infer biogeographic history from a combination of fossils, paleoclimatic models, dated molecular phylogenies, and ancestral area reconstruction algorithms, but then again trying to use another tool for comparison would not necessarily be a bad thing.

No, if it were merely a rather odd, idiosyncratic tool among others, panbiogeography would not be a problem. But this is only what we might call panbiogeography in theory; the actual panbiogeographers in practice are, sadly, a different issue. In the quotation above, where proponents of the school explain their approach, one might easily overlook what turns out to be the point of contention:
ancestral biotas [sic], which subsequently become fragmented by tectonic and/or climate change [...] as a consequence of terrain collision, docking, or suturing, thus constituting a composite area
What this is is essentially an exhaustive list of the biogeographic processes that most real life panbiogeographers accept: tectonic movement of land masses and climate change. In other words, all present distributions are explained entirely by vicariance, the splitting of an ancestrally wide distribution area into smaller parts in the descendants. In practice, this means that your average panbiogeographer categorically rejects the possibility of dispersal, and in particular of long distance dispersal.

(Note that I generalize here based on my experiences so far. It is well possible that there are more reasonable representatives somewhere but I simply haven't met any of them yet.)

The first thing that is odd about this is that vicariance as the only process is self-defeating: If it is so damn hard for organisms to disperse across appreciable distances, then how did the common ancestor ever cover the whole area before the vicariance event happened? Conversely, if the ancestor at some point could disperse so far, then why is the same possibility ruled out in the descendants?

In the best case, the panbiogeographer could argue that organisms can disperse slowly across land, and that all wide ancestral distributions go back to the time of larger, subsequently fragmented continents like Pangaea, Laurasia and Gondwana. So one would claim an absence of long distance dispersal "only" for the time since their breakups. Basically, organisms magically cannot cross large water bodies without human assistance (similar to Vampires being unable to cross running water?).

This brings us to all those readily observable examples of long distance dispersal: rafting lizards, ballooning spiders, wind dispersed seeds, bird-dispersed seeds, birds blown widely off-course by strong storms... So there are many processes, often directly observable today, that put the lie to the idea that long distance dispersal does not and cannot play a role in biogeography.

In addition, there are many groups occurring on distant continents that are simply too young to be explained by vicariance. To name just one botanical example, cacti only occur in the Americas - except, of course, for the genus Rhipsalis, which is also found in Africa. There are many other such cases. These days, dated phylogenies provide us with ample evidence for dispersal events that postdate continental breakups, but more on that perhaps in another post.

And then there is the small matter of volcanic islands, the most striking example of which is Hawaii. Here we have a chain of islands that were produced by volcanoes rising above the ocean, thousands of kilometres distant from the nearest continents. If long distance dispersal was impossible, it would have no native plants and animals. In actual fact, it has a rich native flora and fauna (which was even richer before humans came along), whose closest relatives are partly found in East Asia, partly in Oceania, and partly in western North America.

At a conference in Canada I once met a Hawaiian colleague, and when we talked about crank scientists, I asked her whether she had ever run into a panbiogeographer. Yes, they actually had one visit her institute recently, and he gave a seminar on his research. Okay. So. Um. Did anybody ask him what he made of the native flora of Hawaii if long distance dispersal was impossible? Indeed somebody asked, she said, and he answered that there must once have been a land mass connecting Hawaii to Asia and North America, and then most of it sank under the ocean "because the volcanoes got so heavy". She went on to describe the shocked, incredulous silence that greeted this explanation, and we had a good laugh.

But really this is where it stops being funny. If panbiogeographers treat the impossibility of long distance dispersal as gospel - and at least in my experience, they do - then they just aren't bona fide scientists. Some of them twist results to fit their pre-conceived conclusion; some of them reject any methodology as untrustworthy that would lead to results that they find unacceptable; and, as one of my Australian colleagues has found, some of them, as peer reviewers or journal editors, reject papers demonstrating long distance dispersal not because they found a methodological flaw but on the grounds that it simply cannot be.

And pace that guy I was arguing with, the scientific community has so far not been able to ditch this rather dogmatic* school of biogeography. It is alive and kicking, especially in South America and New Zealand (where the controversy about whether the native flora is a relic of Gondwanan times or a mixture of later arrivals is quite heated), but there are also vocal representatives in the USA and Australia. And some of them run influential journals.

Yes, they are a minority, but this shows that it would be naive to expect all scientists to agree on any given issue, and thus it would be foolish to feel too superior to philosophers just because a minority of them are, say, still dualists.


*) In case somebody finds this one-sided, or in case a panbiogeographer reads this and feels offended, sorry, there is no other way of putting it. The controversy does not have two symmetric sides, with the mainstream seeing a stronger role for dispersal, and panbiogeographers seeing a stronger role for vicariance; or a symmetry of the "dispersalist establishment"** explaining everything with dispersal, and panbiogeographers explaining everything with vicariance. Instead we have the mainstream accepting a variety of possible processes which are invoked on a case by case basis depending on the evidence, and panbiogeographers categorically rejecting every process except vicariance. There is only one side that is unreasonable here, and it is not the "establishment".

**) Quoting from the Wikipedia article on Croizat, which was clearly written by his followers.

 Updated on 30 April 2014 to correct two grammatical mistakes and to link to a later follow-up post.

1 comment:

  1. The late Donn Rosen was among the first of the vocal vicariance only advocates. He argued that sympatry of sister species was the only acceptable evidence of dispersal. He was quite effusive about Croizant, but they later came to an acrimonious parting of the ways, as I recall it.