In the first three parts of this series, molecular biologist, Dr. David Ussery examines, chapter by chapter, the arguments put forward in the book, The Edge of Evolution, by Michael Behe. This book, like Dr. Behe’s previous book, is written in an engagingly accessible style and has been highly acclaimed by many non-specialists who think that Behe has identified the limits of what science can explain without needing to insert an external Intelligence. David Ussery is a Christian molecular biologist who, like all of us at BioLogos, is deeply concerned that other Christians be aware that Dr. Behe has not identified biology’s edge. Furthermore, none of us are sure why anyone should expect to find an Edge—a place where nature ends and God begins. Nature after all—all of nature—is God’s. There is no aspect of creation which is not God’s. Here is David Ussery’s analysis of Chapters six and seven of Behe’s book.
Note: Title of Dawkins' book was corrected on 11/4/10.
Chapter 6 - Benchmarks
This chapter details how Behe decides whether some biological features are unlikely to have been produced by random mutation and natural selection. As an example, he chooses a quote from an article on how to evaluate proposed mechanisms for biological speciation, based on what seems “biologically reasonable.” Behe claims that the idea of whether evolution is “biologically reasonable” has not been fully tested for all of evolution, and proposes to do so in this chapter. To “judge whether random mutation hitched to natural selection is a biologically reasonable explanation for any given molecular phenomenon,” he uses two criteria: how many steps are necessary to create this?, and coherence - the ordering of steps towards a goal. Richard Dawkins goes through both of these steps in his book, Climbing Mount Improbable. I was surprised to find that, although Charles Darwin, Daniel Dennett, John Maynard Smith, Alan Orr, Jerry Coyne, and Francois Jacob are mentioned here, somehow Behe doesn't say anything about Dawkins classic book that deals specifically with the arguments in this chapter, written in 1996, around the same time as Behe’s Darwin’s Black Box. I think that Dawkins scores a valid point in his review of The Edge of Evolution, when he says that unlike Behe's first book, Darwin's Black Box, in the
…second is the book of a man who has given up. Trapped along a false path of his own rather unintelligent design, Behe has left himself no escape. Poster boy of creationists everywhere, he has cut himself adrift from the world of real science.
In this chapter, Behe concludes that evolution is a 'tinkerer', not an engineer. Fair enough. But then he concludes that “If Darwinism is just a tinkerer, then it cannot be expected to produce coherent features where a number of separate parts act together for a clear purpose, involving more than several components.” (Page 119). But what about Dawkin's Mount Improbable? What about the classic example of the eye? There are many books on this, as well as scientific articles. I encourage the interested reader to go to Amazon.com for example, and have a look at some of the books published on the evolution of eyes in animals. One can find exactly what Behe is claiming can never happen, laid out in clear detail, slow, gradual, evolution of complex systems such as the eye. And in my opinion (as a molecular biologist), there's not much difference in the evolution of the eye than the evolution of a complex biochemical system. Certainly there is a difference in scale, but the same principles apply. But please don't just take my word for it. Again, go to PubMed, type in “evolution complex systems,” and see what is there.
Chapter 7 - The Two-Binding-Sites Rule
In this chapter, Behe further explores his claims of incredulity. Now, instead of looking at single mutations within single genes, Behe examines the likelihood of evolutionary mechanisms producing two different proteins with shapes that will fit each other—that is with “binding sites” which are complementary. What are the chances, he asks, of having TWO binding sites evolve at the same time? The probability is so tiny, as to essentially be impossible, he claims. Yet once again, there are problems here with the initial assumptions. I really hate to sound like a broken record, but once again, the interested reader is invited to have a look at the vast literature in this field. I went to PubMed, typed in “evolution protein binding sites,” and got back more than 5000 articles. The title of one recent article was “Using peptide loop insertion mutagenesis for the evolution of proteins,” and another is “Beauty is in the eye of the beholder: proteins can recognize binding sites of homologous proteins in more than one way.” This brings me to one of the (many) flaws in this argument in chapter 7—there is a lot of room for change in the binding site; it does not have to be a 100% perfect match. It only has to be the right shape, and this can be achieved through many many different amino acid sequences. So the probability is not nearly as dire as one might expect from naive and bad first approximations.
Towards the end of this chapter, Behe brings up the work of Richard Lenski, at Michigan State University. Behe claims that, despite having grown E. coli in the test-tube for more than 40,000 generations, “nothing fundamentally new has been produced.” I've known Rich Lenski for many years, and recently he was here as an opponent for a Ph.D. thesis exam. Rich gave a wonderful talk, demonstrating that early on in his experiments, there was a clear, measurable increase in fitness from the [random] mutations generated in his evolution experiments. For example, a set of mutations which altered DNA topology (three dimensional structure) occurred in many of the strains, thereby increasing fitness. (DNA topology is the expertise of both Behe and myself—it is a real shame that Behe no longer works in the lab with DNA structures and evolution!) In some of Lenski’s later experiments, after the cells had been growing for more than fifteen years (!), a strain arose with an increased mutation rate. Following that, the frequency of newly generated mutations and diversity went through the roof. Early on, for the first 20,000 generations (ten years growing in the laboratory), the number of fixed genetic changes was, on average, just a small handful (usually less than ten). After this “mutator” strain arose, however, the number of fixed mutations (new genetic varieties which came to be present in all cells) rose to more than 250, and the number of single changes altogether rose to more than a thousand.
For me, this in a nutshell is what we see from the genome sequences. Lab stains don’t have much diversity compared to what we see in the natural world. On the one hand, we know that outside of the lab, there is an incredible amount of diversity within an organism (like E. coli). On the other hand, when we sequence a genome of a strain that's been grown in the laboratory for awhile, there are often just a small number of changes (a few hundred) associated with property differences. In nature, it is a whole different story. We have a paper that just came out a few weeks ago, comparing the genomes of sixty-one naturally occurring isolates of E. coli. Although some of the E. coli genomes are quite similar, others are VERY different - having more than a MILLION “extra” bases (DNA letters) in one genome, not found in another. The fraction of shared proteins between two strains ranges from nearly all (99.7%) to less than half (48%). Most E. coli genomes contain around 5000 genes, but if we look for all the different genes in all the genomes analyzed so far, we find more than 15,000 different gene families (or more than 3 times the size of any one E. coli genome!). Less than a thousand genes are conserved across all the E. coli genomes sequenced so far. What does this mean? As an example, pick an E. coli genome, and sequence it. Out of those 5000 genes, less than 20% will be found in nearly all other E. coli genomes, and for every one gene in this genome, there are perhaps another nine or ten E. coli genes that are found in other E. coli genomes, but not present in that particular E. coli genome. In addition to the 15,000 gene families discovered so far, we estimate there are probably around 30,000 more E. coli gene families in the intestinal tract of just a single person. This represents a tremendous diversity of genetic information. Since these many E.coli strains can readily exchange genes and parts of genes, there is an absolutely enormous potential to build new varieties of proteins. Behe’s naïve (to be frank) calculations don’t even scratch the surface in calculating this potential to generate new proteins and new protein interactions. He was not aware of any of this.
Take a look at the above figure. Note that the common lab strain of E.coli has 960 families of genes and from that it can build 4144 proteins. But there are other genes, found in some E. coli genomes, missing in others. How many other gene families are available, in nature to add to its protein repertoire? We estimate around 44,000 gene families are out there, in some E. coli genomes, but missing in others, in addition to the 960 present in the above strain.
So my point is, when Behe claims that in the E. coli evolution experiments 'Nothing fundamentally new has been produced.' (page 142), he is ignoring parts of the story which are extremely important. Since most people will not be familiar with the literature, we consider this to be misleading. There is a vast literature which shows just what can be done! Obviously evolution can happen in E. coli, on large scales, and it can be seen to happen under our very eyes, in the laboratory, under the right circumstances. With regard to the Lenski experiments, in my opinion, it is not being honest to only look at the first half of Rich Lenski’s experiments, where he saw little change, and to conclude that evolution does not happen in E. coli. The mutator (which arose halfway through) changed things dramatically.
The Figure above is a comparison of 61 E. coli genomes (each of the concentric circles is one strain of E. coli), showing the conservation of genes; for more details see Figure 5 in Oksana’s Microbial Ecology paper mentioned previously. The point I want to show here is that there are many large gaps (lighter-colored—regions of genes that are missing in many genomes, but present in others. Some of these regions encode novel ‘molecular machines’—or what I think many (but not Behe) might call ‘fundamentally new’ complexes.