Tuesday, November 2, 2010

Beating a dead passenger pigeon

I don't mean to turn this into Steve Budiansky's Debunking the Species–Area Curve Blog, but my post the other day on species-extinction alarmism has provided yet another opportunity (as if I needed one) to prove the theorem that it is impossible to raise even the mildest questions about the scientific methodology used to generate predictions of imminent ecological doom without being told that:
(a) You are a shill for corporate interests
(b) You lack the qualifications to discuss the issue
or
(c) You are not truly passionate about saving the planet
For the record, I think we should be doing a lot more to save endangered species from extinction. I also think we ought to be doing the things that are actually likely to achieve that objective, and not harm the cause by overstating the case with sweeping predictions based on mathematically and scientifically dubious methodologies. (And as I mentioned in another post the other day, why is it always the ecologists — but never the nuclear physicists, mathematicians, linguistic theorists, astronomers, Wittgensteinian philosophers — who tell me I may not write about their field because I lack the qualifications? I'm pretty sure than ecology is not as hard as any of those other subjects.)

Anyone who has actually read the definitive review article* on the methodological basis for the oft-repeated predictions of mass extinction would have legitimate reason to be dubious about the way the so-called "species–area relation" (or "curve" or "effect") has been employed to generate these scary scenarios.

The history, briefly told, is this: Biologists back in the 1960s and '70s looked at islands and found that the bigger the island, the greater the number of species. They then tried to fit the relationship they had found between area and number of species mathematically, using an extraordinarily simple-minded  formula.

The next step was where the real trouble began: conservation biologists took this extraordinarily simple-minded descriptive statistical tool and had the bright idea of using it as a predictive tool for what happens to the number of species when a habitat (such as a tropical rainforest) is shrunk in size.

It was never intended to be used this way; it is not a mathematical formula that in any way models cause and effect of extinction processes; and indeed the definitive review article I mentioned cautions against using it this way for several reasons:
1. The statistical association between species number and area is likely a "correlation . . . without a functional relationship." Just because two variables tend to vary together does not mean one is causing the other to vary.

2. Only one-half the variation in species number from one unperturbed island to another can even be accounted for statistically by variations in area, meaning that other factors besides habitat area are at least as important in determining species abundance.

3. The species-area curve tends to generate impossibly large species numbers when extrapolated to larger areas, raising doubts as to its realism even as a valid description of biological reality.

4. The parameters in the species-area formula have no biological significance; in other words, they are just fudge factors in a cookie-cutter formula that does not (as any true mathematical model must) incorporate a cause and effect understanding of mechanism.
Vernon Heywood, a well-respected plant biologist, made this point as well in a rare critical review. I've quoted him before but I'll  quote him again for the simple reason that I have never seen any of the propagators of the predictions of mass extinction acknowledge, confront, discuss, address, deal rationally with his key point:
"The species–area curve (in a mainland situation) is nothing more than a self-evident fact: that as one enlarges an area, it comes to encompass the geographical ranges of more species. The danger comes when this is extrapolated backwards, and it is assumed that by reducing the size of a forest, it will lose species according to the same gradient."
As Heywood then went on to point out, there are many reasons why this is not going to happen: species are not distributed at random, conservation measures are already protecting many critical habitats, many species can adapt to other habitats as the original forests are cut down, and preservation of even small bits of critical habitat (ecological "hotspots") may be enough to save many endangered species.

We know that a few species a year are being killed off by human action. The tens of thousands or hundreds of thousands or millions of species that we keep reading about — the mass extinctions, the catastrophes of unprecedented magnitude — all come from a remarkably bad bit of bad theoretical science. And you don't need a Ph.D. in ecology to see that there's a problem here, both with the misuse of scientific methodologies and with the politicization of the subject.

--

* Connor, E.F. & McCoy, E.D. The statistics and biology of the species-area relationship. Am. Nat. 113, 791–833 (1979)