23 August 2008

Why I'm not a Behe fan, Part IIB: abusing genetics

In a previous post, I started to explain a fact that some people (who don't know me) seem to find surprising or noteworthy. Michael Behe is a Christian who accepts common ancestry and an ancient cosmos, so you'd think I would be excited about the work of a fellow "theistic evolutionist." But I'm not. Two overall problems come to mind. (Basically, I find his conduct as a scientist to be unacceptable, and I find his proposals to be laughable failures.) I'm addressing the second one here. The discussion is quite long, so I divided it into two sections, Part A and this post, Part B, which will have to be split up. I'm sorry about the length; it would really take a whole book to carefully explain how Behe has misused genetics and probability.

1. Behe's fans say that he's a nice guy, and that the evolutionists are "crucifying" him. Both claims seem to be true, but they can't hide some serious problems with his conduct as a scientist.

Those issues are the subject of the first post.

2. Some of Behe's defenders think that he has effectively answered his critics. He has not, nor has he understood or acknowledged the most important criticisms of his crude claims.

Behe's recent book The Edge of Evolution (henceforth EoE) is the focus of this series, and as I exlained in Part A:
EoE makes exactly one specific scientific claim, accompanied by simplistic genetic assumptions and supported by a "case study." The scientific claim is that the mutations that drive large-scale evolution, and that are thought to underlie all evolutionary change (past and present), are non-random. And the "case study" is a long-winded account of the adaptation of the malaria parasite in the face of drugs intended for its destruction.
Part A dealt with the laughable case study. But the heart of EoE is the claim that random mutation rates are insufficient – spectacularly insufficient – to support step-by-step evolution of complex features. The implication, then, is that the mutations that underlie major evolutionary change did not occur randomly.

First, some important points of clarification:
  • Behe is not denying that common descent is true, or that evolutionary change results from mutation. He acknowledges both. He is saying that the most important mutations – those that led to, say, new cell types – could not have been random.
  • Behe is not saying that the combination of random mutation and natural selection (the "darwinian" mechanism) is not a driving force in evolutionary change. He acknowledges the efficacy of the process in explaining "a number of important details of life," such as drug resistance in bacteria or pesticide resistance in insects, and is willing to attribute the differences between widely divergent organisms to the workings of "randomness." Specifically, he writes that "explicit design appears to reach into biology to a certain level, to the level of the vertebrate class, but not necessarily further." (p. 220) This means that Behe claims to be certain that the major distinctions between goldfish and bats are non-random, but that the major distinctions between bats and people could be accounted for by random mechanisms. (He asserts the "edge of evolution" to lie somewhere between the species level and the class level. [p. 201])
  • Behe does not commit himself to a particular mode of divine intervention whereby the supposedly non-random mutations came about, and in fact he seems to favor a front-loading scenario in which God "was able to specify from the start not only laws, but much more." (p. 231)
These clarifications are important, because much of the criticism of EoE has been botched significantly. The book is bad, really bad, but it can't be honestly characterized as an anti-evolution argument. Ultimately, Behe seeks to prove that evolution had to be guided. That's the way to understand EoE, and as Joan Roughgarden wisely noted in her review, there are some "constructive" aspects of the book, including the abandonment of opposition to – or even ambivalence about – common descent.

So what's so wrong with Behe's argument in EoE? Well, first, here's the argument summarized:
  1. Evolutionary changes in the features of organisms require changes in genomes, changes which occur by mutation.
  2. Many of the most interesting evolutionary changes require multiple changes in the same genome, often in the same gene.
  3. Mutation rates, in terms of number of mutations per generation, are known to be on the order of 1 in 100 million.
  4. Based on this mutation rate, the probability of occurrence of an evolutionary change requiring several mutations is vanishingly small, such that the whole of life's history is not nearly long enough for the change to occur via random mutation.
And here are some ways in which Behe's argument is wrong and/or misleading.

I. Behe's assumption of a particular mutation rate is both absurdly oversimplified and inappropriately extrapolated into the entire tree of life.

The basis of all of Behe's calculations is a mutation rate of 1 in 100 million. This is the estimated rate at which misspelling-type mutation occurs in each generation, averaged over the entire genome, in humans. (The number doesn't consider other types of mutation, now known to be more common than previously thought.) Behe uses this number in all of his (flawed) probability calculations. Even if we knew nothing about mutation rates, the notion of extrapolating from an human (or even mammalian) characteristic to the whole of the biosphere (past and present) is ludicrous enough that it would by itself cast doubt on the credibility of the author.

Rates and characteristics of mutation are the focus of active current research, and many important questions remain unanswered. But we know that there is no such thing as "the mutation rate," in the biosphere or even in particular species. In fact, mutation rates can vary significantly, between types of organisms, between organisms in different states of health, in individual subpopulations of organisms, even between regions of the genome of a particular organism.

More importantly, it is ridiculous to assume that "the mutation rate" has always been the same. Consider a flowchart outlining mutation and its effects, taken from a recent review of the evolution of mutation rates:

Image from "Mutation rate variation in multicellular eukaryotes: causes and consequences," by C.F. Baer et al., Nature Reviews Genetics, August 2007. Click to enlarge (opens in new window/tab).

The idea is that mutations are created in at least two ways: 1) damage to DNA from external influences such as radiation; and 2) errors in the replication process. During the evolution of early life, neither of these influences would be expected to be the same as – or even comparable to – similar influences today. And that's just the beginning of the flowchart. There are error-correction systems that erase mutations before they can be passed on to the cell's descendants; again, only a fool would suppose that these systems have been present throughout life's history; indeed, bursts of mutation that occur today are usually caused by deficiencies in DNA repair and the appearance of "mutator lines" is thought to be an accelerating force in adaptation.

My point is not that we know what the genetic landscape was like during the early evolution of life's toolkit, nor am I claiming that we know whether or not certain mutations were "nonrandom." My point is that the extrapolation of estimated mutation rates in modern humans into the deep past is clearly unjustified, a move so foolish that it can only be the product of folk science.

II. Behe's treatment of adaptation always ignores existing genetic variation, and his arguments seem to assume that multiple mutations must occur simultaneously.

I've mentioned these problems before, and they constitute some of Behe's biggest errors. When he envisions the process of adaptation, in which several genetic changes separate one state from another, he automatically assumes that none of the changes exists at the beginning. Yet even Darwin knew that populations of organisms harbor huge amounts of genetic variation, as evidenced by the profound success of domestication (of plants and animals) by human selection. Most of Behe's critics have noted this, and Behe's response was a lame dodge. But perhaps the critics haven't been clear about why superfast evolution under human selection is such a problem for his ideas. Here's why: since organisms are so profoundly diverse genetically, many of the genetic changes that could be exploited by selection already exist. In fact, current theory predicts that rapid evolution, such as that required after significant environmental change, is much more likely in populations with significant standing variation.

With his simplistic view of genetics and variation in mind, Behe then describes how an adaptation that requires two different changes will be extraordinary unlikely, because the probability of each change is one in 100 million, and the probability of each occurring together is one in 100 million times 100 million. His critics argue, correctly, that his calculations assume that the mutations must occur simultaneously, and that is indeed very improbable. (Although maybe not nearly as improbable as we used to think.) In some of the discussions in EoE, he describes sequential acquisition of mutations (e.g., p. 111), but he calculates probabilities according to simultaneous occurrence (e.g., p. 63). Jerry Coyne explains why this is a gigantic error, and Behe seems unable to understand why.

I've written a separate post about Behe's mishandling of probability. It shows that he is not someone to consult when the subject is population genetics.

III. Behe claims that huge population sizes automatically generate more evolutionary opportunity than smaller ones do. This is incorrect.

It seems so obvious. More organisms means more mutations means more beneficial mutations means more and faster evolution. It's the kind of obvious, simplistic, intuitive claim that forms the bedrock of any folk science. But it's wrong.

On the contrary, very large population sizes lead to a so-called "speed limit" on adaptation that results from competition among beneficial mutations. The phenomenon is called clonal interference and it's particularly well understood in asexual organisms such as bacteria. The basic idea has been around for decades, but measurement and modeling of the phenomenon has been increasing in the last ten years. A very recent report, the subject of an upcoming post here, showed that the beneficial mutation rate in bacteria is 1000 times higher than previously thought – and the underestimation is due entirely to clonal interference.

The effect is not limited to asexual organisms; in fact, the problem of clonal interference is thought to constitute one of the major driving forces behind the evolutionary development and maintenance of sexual reproduction. The idea is that the genetic shuffling that accompanies sexual reproduction can bring beneficial mutations together and increase the effectiveness of selection. One of the few studies to examine this experimentally led to the conclusion that clonal interference is a problem for sexual organisms, and that sex reduces the impact of clonal interference and lowers the evolutionary "speed limit." (Interestingly, the malaria parasite is partly asexual, and reproduction inside a human is completely asexual, so clonal interference is probably a very significant "speed limit" on the evolution of P. falciparum – another reason not to use malaria as a benchmark "case study" for the understanding of all of evolutionary genetics.)

In summary, I find Behe's handling of genetics in EoE to be unacceptable. He seems ignorant of basic evolutionary genetics, and is clearly content to create a folk science alternative to modern evolutionary biology. No one has proven that random mutation generated the wonders of biology, to be sure, and so I'm not saying that Behe's conclusion is known to be false. I'm saying that his attempts to establish his conclusion have failed miserably, as have his responses to his critics, and the result is that he cannot be trusted as a careful, thoughtful, knowledgeable critic of evolutionary science. EoE is folk science, nothing more.

My final post in the series will have closing comments and some ideas for how we might go about posing questions about the processes that yield biological design.

3 comments:

Martin LaBar said...

That's a fine summary paragraph. Careful, yet critical.

Anonymous said...

Really enjoyed your recent posts on Behe. Thanks for sharing.

Anonymous said...

Interesting to know.