Sunday, September 26, 2010

The teflon doomsayers

In The Rational Optimist, Matt Ridley offers example after spectacular example of a phenomenon that has baffled me ever since I began covering environmental issues in my first job in journalism thirty years ago: to wit, that while the entire presumable goal, purpose, and raison d'ĂȘtre of applied environmental science is to solve environmental problems, any environmental scientist who dares to suggest that problems are being solved is asking for trouble. As Ridley observes, we have arrived at a state where even the most wildly irrational pessimism is treated with reverence, while the most cautiously sober optimism is ridiculed.

Some of this is human nature and was ever thus; intellectuals, as The Rational Optimist reminds us, have been decrying modernism ever since modernism began. Actually, I wouldn't stop there: the belief in a lost golden age is as old as civilization, as is the intellectual vanity of casting oneself as the lone uncorrupted voice in the wilderness. A few thousand years before Dostoevsky, Malthus, George Orwell, and Paul Ehrlich, the Hebrew prophets were pouring out gloom and dismay with the best of them, dismissing the superficial comforts of the civilized world and its material rewards as a fool's paradise. Pessimism is what people with deep minds and deep souls have; optimism is what idiots with vacant grins on their faces have.

Pessimism is of course a proven fund-raising tool; "save the whales!" is always going to bring in more cash than "the whales are being saved!" But much more than that, we have today the amusingly ironic spectacle of tenured professors with salaries, health insurance, lifetime job security, and excellent retirement plans courtesy of TIAA-CREF being showered with worldly rewards (bestselling books, "genius" awards) for telling us that progress is an illusion and the end is near . . . while still preening themselves as daring outsiders courageously taking on the mighty and powerful. The fact that it takes no daring at all to adopt such an intellectual posture these days does not stop any of the practitioners of this business model from invariably announcing themselves to be the bearers of "dangerous" or "heretical" ideas and congratulating themselves for "speaking truth to power."

So there are understandable reasons why it pays to say that things have gone to hell and will continue to go to hell.

What I find almost inexplicable in all of this, however, is how the scientific doomsayers get away over and over again with making predictions that are fabulously, ridiculously — and demonstrably — incorrect, without the slightest repercussions upon their credibility or careers. Predictions of impending doom are published based on absurd methodologies and threadbare evidence of a kind that in the normal course of scientific affairs would be sufficient to ruin careers ten times over, and the authors walk away from them without a scratch.

Ridley has a number of remarkable for-instances in his book, many provided by MacArthur genius award winner Paul Ehrlich — who in addition to insisting in 1971 that the world had already lost the race to feed an expanding population and that mass starvation in the 1970s and 1980s would cause death rates to soar and world population to collapse to 2 billion, also declared around the same time that because of exposure to cancer-causing chemicals that had already occurred, "the U.S. life expectancy will drop to forty-two years by 1980, due to cancer epidemics."

The astonishingly wrong and repercussion-free prediction of imminent doom that first riveted my attention was the claim of the impending mass extinction of the Earth's species. In 1979, the biologist Norman Myers declared that a fifth of all species on the planet would be gone within two decades. This prediction was based upon . . . absolutely no evidence whatsoever. Myers acknowledged that the documented species extinction rate of animals was 1 per year; he then asserted that scientists had "hazarded a guess" that the actual rate was 100 per year; he then speculated that government inaction was "likely to lead" to several thousand or even tens of thousands a year, which would add up to as much as a million species over two decades. (This was when people thought there were 5 million species; the best guess now is at least 10 million.) It swiftly became conventional wisdom.

Subsequently, an attempt was made to give these made-up numbers a patina of scientific respectability that was in many ways an even worse abuse of scientific logic and evidence. In the 1990s E. O. Wilson began citing the so-called "species–area relation" as the basis for predicting that tens of thousands of species were being extirpated a year by habitat loss caused by forest clearing. Wilson popularized various numbers ranging from 4,000 to 100,000 species a year being lost, and these numbers were repeated over and over again in environmental groups' fundraising literature, in congressional testimony, in speeches by Al Gore (who in 1993 said that "one-half of all species" could disappear in our lifetime, apparently an extrapolation of Wilson's and Ehrlich's pronouncement, in a 1991 paper in Science, that as many as a quarter of all rain forest species will disappear in 30 years).

I started to look into the science and mathematics of the species-area relation when my father, who was an applied mathematician at Harvard University, mentioned to me an Op-ed in the New York Times by his fellow Harvard faculty member Wilson that included a description of this formula, which had struck my father as absurd on its face as a mathematical model. The formula most often used is:

             S = CAz

where S is number of species, A is area, and C and z are arbitrary constants tweaked to make the curve try to match the data. Basically, the formula says if you count the number of species on, say, islands of varying sizes, the bigger the island, the more the species. Wilson's argument was that if you start cutting down rainforests, say, you'll shrink the number of species contained in them according to the same curve.

The prima facie problem, which irked my father, is that the dimensions of the arbitrary constant C vary according to the numerical value of the other arbitrary constant z. Without going into the technical details too much, this is (as my father put it) "cockamamie" from any scientific perspective; it means that this is just an exercise in curve-fitting, not a scientific model based on any cause-and-effect understanding or mechanism.

The more I looked into it the more ridiculous it became. The definitive review article on the species–area relation correctly noted that the formula is at heart nothing more than a "sampling phenomenon . . . without a functional relationship." The authors concluded that there was no biological significance to the constants C and z; the fact that z tends to fall in the range of 0.2 to 0.4 when you fit the curve to islands is simply a mathematical coincidence, and the same thing happens when the same formula is used to fit other empirical relationships (e.g., the relation between brain size and body size in mammals).

The much more serious problem, as a few (truly daring) conservation biologists pointed out, is that there is absolutely no reason to think that such an empirical, broad-brush, descriptive formula has any predictive value in the real world at all. As the conservation biologist Vernon Heywood wrote: "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."

Heywood pointed out 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. Rather than seeing species numbers decline in lockstep with loss of forest area, a more biologically realistic model might predict few if any extinctions until habitats are almost completely destroyed, and even then species numbers would certainly not plunge to zero (as the species–area curve predicts), since many species would be able to survive in the secondary forests that regrow or in other habitats still available.

Even more striking is the fact that the predictions from the formula are wildly incorrect in practice where they can be checked. More than 90 percent of the Atlantic coastal forests of Brazil were cut down, mostly in the 19th century; by the species–area relation that means 50 percent of species should be gone. In fact the actual number of animal extinctions has been zero, even though many of the Brazilian species are highly endemic, found nowhere else in the world.

Similarly, the eastern U.S. forests were reduced to less than half their original extent from colonial times to 1900; but instead of 30 extinctions of birds as predicted by the model, there have been 4 — 2 of which (the Carolina parakeet and the passenger pigeon) were wiped out by hunting, not habitat loss, and the other 2 of which (the ivory-billed woodpecker and Bachman's warbler) were restricted to very specific habitats in the southeast that were destroyed by logging and agriculture. The fact is you could have cut down vastly more forest area in the entire region east of the Rockies without losing a single bird species had we protected the small but critical habitats required by the ivory-billed woodpecker and Bachman's warbler and halted the criminally stupid hunting of the passenger pigeon and Carolina parakeet. (Please note: I am not advocating cutting down vast areas of forest east of the Rockies.)

You might think that with such a record of slapdash science and wildly incorrect predictions, conservationist biologists might be a bit sheepish. But there is simply too much invested in a methodology that gives the "right" answer — while nothing else in ecological science (actual field studies, surveys of endangered species, historical evidence) comes close to doing so. A recent peer-reviewed article in a prominent journal referred to the species–area curve reverently as "ecology's oldest law" and even "the periodic table" of ecological science, an assertion more than sufficient to make Mendeleev turn over in his grave. Virtually every article invoking the species–area relation as the basis of a catastrophic prediction of impending species extinction cites the definitive review article I mentioned above — without ever heeding (or even mentioning) its fundamental conclusion regarding the logical and methodological fallacy of using it to make such predictions.

And the circular reasoning/begging of the question that takes place is simply jaw-dropping at times, my favorite example being Thomas Lovejoy's defense (in 2002) of Norman Myers's 1979 prediction of extinction doom (by 2000): "Myers . . . deserves credit for being the first to say that the number was large and for doing so at a time when it was difficult to make more accurate calculations." Another egregious practice in the ecological literature has been to adjust the formula and starting assumptions to minimize the number of predicted extinctions (sometimes by a factor of 10 or 100 or more) when comparing the predictions generated by the formula to the actual evidence in specific cases (thereby "validating" the methodology), but doing the exact opposite when using the formula to produce the alarming worldwide extinction predictions of tens of thousands per year.

There is no scientific dispute that extinctions are occurring, that they are occurring at a rate above the natural level due to human action, and that strenuous efforts are needed to protect critical habitats, to eliminate invasive competitors that threaten species, and to prevent overexploitation.

But the egregiously bad science that is still being invoked to shore up wholly unsubstantiated predictions of catastrophic mass extinctions is only undermining the credibility of environmentalists, and is already causing a dangerous political backlash that has handed ammunition (exactly as in the case of global warming) to those who want to reject any and all evidence of human impacts on the natural environment.

A first step in restoring credibility might be to revive some intellectual opprobrium for those who are flagrantly wrong, even in a good cause.