9 July 2007
Michael J Behe
Today I have continued my response to University of Chicago evolutionary biologist Jerry Coyne, which began yesterday and will conclude tomorrow. Just a reminder that I’m only quoting the portions of his response that I specifically address here because it quickly gets awkward to include all of the context. Readers who want to see the full back-and-forth should read his posted review and response.
1. There is no evolutionary expectation that complex protein-protein interactions will evolve in a parasite adapting to a new drug.
Darwinism purports to account for the complexity of cellular machinery which, along with much else, involves very many protein-protein interactions. Yet if “there is no evolutionary expectation that complex protein-protein interactions will evolve” in any particular circumstance, then for those skeptical of Darwinism, what independent reason is there to suppose the protein-protein interactions we do find in the cell evolved by random mutations? I can’t think of any. So I and a lot of other people want to decide what Darwinian processes can do based on evidence, not supposition. And the evidence is decidedly against it.
It’s myopic to view these results, as Coyne does, simply as “a parasite adapting to a new drug.” Rather, they are data that bear directly on the question, “What can Darwinian processes do given an astronomical number of opportunities?” In the past, we did not have enough data to address that question. Now we do, and observational evidence indicates the answer is, “Not much at all”. And as I show in the book, the results with malaria mirror results with E. coli and HIV, which are very different organisms in very different circumstances. In a truly enormous number of opportunities, nothing much of fundamental biochemical interest happened.
2. Behe’s probability calculations, on which his entire argument rests, are flatly wrong because they assume that adaptation cannot occur one mutation at a time.
Here is where Professor Coyne and other Darwinist reviewers really miss the boat and overlook the considerable power of the malaria results. The number I cite, one parasite in every 1020 for de novo chloroquine resistance, is not a probability calculation. Rather, it is a statistic, a result, a data point. (Furthermore, it is not mynumber, but that of the eminent malariologist Nicholas White.) I do not assume that “adaptation cannot occur one mutation at a time”; I assume nothing at all. I am simply looking at the results. The malaria parasite was free to do whatever it could in nature; to evolve resistance, or outcompete its fellow parasites, by whatever evolutionary pathway was available in the wild. Neither I nor anyone else were manipulating the results. What we see when we look at chloroquine-resistant malaria is pristine data — it is the best that random mutation plus selection was able to accomplish in the wild in 1020 tries.
Let me elaborate that last point. The fact that de novo chloroquine resistance is observed to be an event of frequency 1 in 1020 means that mutational events of greater frequency are of little help, because events of greater frequency would have been expected to occur many times in the same time interval. For example, if a single point mutation such as K76T alone in PfCRT in the wild were sufficient to confer chloroquine resistance, then resistance would occur de novo in virtually every person treated with chloroquine, as it does in almost every person treated with atovaquone. In 1020 parasites that single mutation would have been expected to have occurred about 1010 times or more. What’s more, every other possible single point mutation, at every position of the parasite’s genome, would also be expected to have occurred roughly the same number of times. And enormous numbers of other types of mutations — deletions, insertions, gene duplications, and more — in every gene of the parasite, would also have occurred. The result: a very few mutations helped the parasite a bit; the overwhelming number of mutations did not help at all.
3. The probability calculations are also wrong because Behe’s argument is based on specifying a priori: the identical pair of mutations that occur in chloroquine-resistant malaria. He neglects the possibility (indeed, the certainty) that many other mutations that cause interactions between proteins and other molecules can also be adaptive.
Coyne is wrong again, for the same reason. I did not “specify a priori” exactly which mutations had to occur to be adaptive. Was I somehow out in the wild in Africa and South America telling the parasite which mutations to try? The parasite was free in nature to do whatever it could. The results are not a priori; they are entirely a posteriori, observational data. Moreover, I did not “neglect the possibility” (let alone “the certainty”) of anything. Nobody told the parasite to restrict mutations just to its pfcrt gene. If other mutations could have been adaptive, Plasmodium falciparumhad 1020 chances in the wild to find them, to come up with whatever it could muster. In the malaria data, we simply observe the exceedingly modest results.
Incidentally, this bears on Coyne’s comment on Miller’s review that “one of the two mutations that Behe claims are ‘required’ for CQR is not actually required (Chen et al. 2003, reference accidentally omitted from Miller’s piece).” If you read that paper you see that, yes, A220S is not found in some resistant strains, as it is in most. (By the way, I was always quite careful in my book to state that A220S had been found in most strains, because I was quite aware of the several exceptions.) However, one also reads that the strains missing A220S have several other, novel mutations, which may be playing a comparable role in them that the mutation at position 220 plays in most other strains. My argument does not depend on exactly which changes are needed in the protein. Rather, the important point is that multiple changes appear to be required for resistance in the wild.
And for the life of me, I don’t see why that proposition — that two mutations might be needed for some adaptations, and that that would be a big evolutionary impediment — is being treated by Coyne and other Darwinists with such horror. It certainly has been discussed in the evolutionary literature in the past. In my book I quote Allan Orr remarking, “Given realistically low mutation rates, double mutants will be so rare that adaptation is essentially constrained to surveying — and substituting — one-mutational step neighbors. Thus if a double-mutant sequence is favorable but all single amino acid mutants are deleterious, adaptation will generally not proceed.” All I have done is to point to an example of the situation he envisioned, to quantify it, and to argue that it’s likely to be a fairly general phenomenon. Why the shock?