metro park paths
Aerial shot of beautiful metropolitan Park with tree paths, sports grounds.
Aerial shot of beautiful metropolitan Park with tree paths, sports grounds.

Irreducible Complexity And Darwinian Pathways

Guest response to article by R.H. Thornhill and D.W. Ussery Published at ARN Forum

It’s official. Behe’s concept of irreducible complexity (IC) has found itself in the peer-reviewed scientific literature. Ironically, it was introduced by two critics of ID attempting to formulate non-teleological mechanisms for spawning IC. The article is: Thornhill, R.H., Ussery, D.W. 2000. “A classification of possible routes of Darwinian evolution.” J. Theor. Bio. 203: 111-116.

First of all, this article shows that Behe’s work has indeed contributed to science. Thornhill and Ussery (T&U) write:

“However, the more theoretical question about the accessibility by Darwinian evolution of irreducibly complex structures of functionally indivisible components, if such exist, has not been thoroughly examined. ….One factor hampering examination of the accessibility of biological structures by Darwinian evolution is the absence of a classification of possible routes. A suggested classification is presented here.”

Although one can argue about it, this can be viewed as a fundamental confirmation of Behe’s thesis that the origin of these IC structures has not been explained by science. However, what should be clear is that Behe’s skepticism has served as an impetus for these scientists to develop a classification that did not exist before. Therefore, Behe has indeed contributed in an indirect way by serving as the stimulus for the creation of such a classification.

T&U then define terms, but strangely, do not use Behe’s definition of IC. Of interest also is their definition of Darwinian evolution. It includes the following: ” no intervention by conscious agent(s) occurs.” Once again, we see how an a priori assumption of science works to exclude a teleological cause (reminding us that science is simply not an authority when dealing with question of teleology vs. non-teleology).

T&U then outline the four possible routes of Darwinian evolution.

The first is: Serial direct Darwinian evolution. This means change along a single axis.

Here again we get to see the contribution of Behe as the authors then note, “Although it can generate complicated structures, it cannot generate irreducibly complex structures.”

So we can see that IC helps to rule out certain evolutionary pathways. This is also very significant in that the most persuasive examples of random mutation and natural selection (RM&NS) entail serial direct Darwinian evolution. The traditional examples of Darwin’s finches (and their beaks), giraffe necks, elephant trunks, darkening wings in moths are all examples of serial direct Darwinian evolution. Thus, this means that evidence for this type of evolution is not evidence that IC can/did evolve via the blind watchmaker mechanism (BWM).

The second mechanism is: Parallel direct Darwinian evolution. This means approximately synchronous changes in more than one component, so that modification to other components always occurs before the total modification to any one component has become significant.

They then cite some examples: Most complex supramolecular biological structures have primarily this type of accessibility by Darwinian evolution, with examples being bat echolocation, spiders’ web construction, honeybee waggle dances, and insect mimicry by orchids (Dawkins, 1986, 1995). Some complex (but not irreducibly complex) molecular systems, such as the globin proteins (Ptitsyn, 1999; Satoh, 1999), could also have evolved in this manner.

But they also write:

Parallel direct Darwinian evolution can generate irreducibly complex structures, but not irreducibly complex structures of functionally indivisible components (Fig. 1), and this is the valid conclusion to draw from Behe’s thesis.

Thus, once again, we can see that when we are dealing with IC molecular machines (which are composed of functionally indivisible parts), the various examples of Darwinian evolution cited by Dawkins et al. are irrelevant. None of it amounts to evidence that Behe’s IC examples evolved by the BWM.

Thus, before we go on, let’s consider that despite all the expressed incredulity that is so common among Behe’s critics, he has indeed contributed to science by forcing scientists to classify routes of evolution and by showing that 50% of the possible routes can’t generate IC machines. This is progress. Without Behe, for example, many would probably still think that classic evidence of RM&NS allows us to think that the bacterial flagellum evolved by the same mechanism.

Next, we turn to the remaining two possible routes in which Darwinian evolution (DE) can generate IC, that of Elimination of functional redundancy and Adoption from a different function. The authors use the analogy of an arch to illustrate how these mechanisms might work to generate IC:

The arch is irreducibly complex, and, assuming that cement does not set instantaneously, any arch one sees must therefore either have been built using scaffolding, analogously to redundancy elimination, or have been built elsewhere, perhaps horizontally, and moved into position when the cement had set, analogously to adoption.

The biological examples used by the authors to support these mechanisms are not very impressive, as the authors shy away from addressing any of Behe’s examples in detail. In the case of functional redundancy, the authors spend most their time on the origin of the mammalian jaw and not a molecular machine. When it comes to the adoption from a different function, the authors write of the origin of feathers from scales, shared domains in different proteins, and two examples where a protein has been co-opted to perform an alternative function (i.e., crystallins). Let’s take a closer look at the two possible Darwinian pathways for generating IC.

Adoption from a Different Function

Ironically, the main problem with this pathway was first highlighted by another Behe-critic, H. Allen Orr. In his critique from Boston Review, Orr writes:

First it will do no good to suggest that all the required parts of some biochemical pathway popped up simultaneously by mutation. Although this “solution” yields a functioning system in one fell swoop, it’s so hopelessly unlikely that no Darwinian takes it seriously. As Behe rightly says, we gain nothing by replacing a problem with a miracle. Second, we might think that some of the parts of an irreducibly complex system evolved step by step for some other purpose and were then recruited wholesale to a new function. But this is also unlikely. You may as well hope that half your car’s transmission will suddenly help out in the airbag department. Such things might happen very, very rarely, but they surely do not offer a general solution to irreducible complexity.

To appreciate why co-option is unlikely to be a general solution to IC, let’s return to Behe’s definition of IC:

“By irreducibly complex I mean a single system composed of several well-matched, interacting parts that contribute to the basic function, wherein the removal of any one of the parts causes the system to effectively cease functioning.”


“An irreducibly complex system is one that requires several closely matched parts in order to function and where removal of one of the components effectively causes the system to cease functioning.”

Since an IC system is built from closely/well-matched parts, it is unlikely that a component shaped to fulfill another function can snuggly plug-in to generate the IC function. In fact, Behe anticipates this solution by writing:

Even if a system is irreducibly complex (and thus cannot have been produced directly), however, one cannot definitively rule out the possibility of an indirect, circuitous route. As the complexity of the an interacting system increases, though, the likelihood of such an indirect route drops precipitously.”

To illustrate this point, let’s consider the bacterial flagellum (perhaps the most well known example of an IC system). A functioning flagellum requires about 30 gene products (components). So what does the co-option hypothesis predict? That prior to the existence of the flagellum, these 30 gene products (or their duplicates) all existed doing something else. Then, they just happened to all fit together by chance to create a flagellum. And afterwards, the other 30 or so hypothetical functions of these original gene products disappeared. Does this really sound like a general solution to IC?

The brilliance of Darwin was to minimize the role of chance in apparent design. But once we turn to the co-option explanation, we leave this explanatory appeal behind, as chance reasserts itself into a place of prominence. For it is chance that determines whether the 30-or-so gene products just happen to come together to form a functioning flagellum, as selection was pruning these gene products in accord with 30-or-so different functions. Thus, the co-option explanation is really a return to using chance as an explanation for apparent design, and just as it was not convincing in pre-Darwinian days, it is not convincing today.

One seeming way around this problem is to imagine 5 or 6 subsystems, each composed of 6 or 5 gene products, all conducting different functions. Thus, we need only imagine that by chance, 5 or 6 subsystems could happen to come together to form a well-matched whole that generates a new function. But I don’t see how this helps much as Orr’s critique still applies. And what’s worse, in this case, the non-teleologist must now explain the origin of 5 or 6 different IC systems and why they too disappeared after the origin of the new IC system.

When we return to the T&U paper, this problem becomes obvious as the examples of adoption that the authors list give us no reason to think IC system arose by this pathway. Their first example is that of the transition from scales to feathers. But it is not clear this is an example of generating IC and this whole topic ignores Behe’s points on pp. 40-41. The fact that various proteins can share domains doesn’t really support adoption from a different function, as there is no reason to think a designer would ensure that each and every part of a system is truly unique. For example, that both a lawnmower and automobile have spark plugs is not evidence that one motor was co-opted to form another by random changes and selection.

The only clear examples, in in my opinion, of adoption from a different function are as follows:

  1. Antifreeze glycoprotein in the blood of Antarctic notothenioid fishes, which enables them to survive in icy seas, is considered to have evolved from a functionally unrelated pancreatic trypsinogen-like protease, and the recent discovery of chimeric genes which encode both the protease and an antifreeze glycoprotein polyprotein strongly supports this theory (Cheng & Chen, 1999).
  2. Crystallins (proteins with refractive functions in the eye lens) are closely related or identical to stress-protective proteins in non-ocular tissues (eg. Drosophila a-crystallins and small heat-shock proteins are homologous).

But note that in both cases, these proteins don’t function as well-matched components in an IC system. Thus, while sometimes a protein might adopt a different function since it is not constrained to interact with multiple partners, an IC component is unlikely to arise in this fashion because it is constrained to interact with multiple partners.

Since the adoption from a different function explanation relies heavily on pure chance, it is unlikely to be a general solution to IC. In fact, as Behe notes, the more complex the system, the less likely that this pathway is what generated IC (since it relies more heavily on converging independent, random events).

But there is another problem with the adoption explanation. To illustrate this, let’s use a simple 3-part IC system, the chaperone machine. The chaperone machine is a protein complex that binds proteins in their partially unfolded state to prevent aggregation in the cell. Each of the three parts fulfills a different role. DnaK is the protein clamp that binds to other proteins. DnaJ is the protein that loads the clamp. And GrpE is the protein that unloads the clamp (the mechanism is more complicated and involves coordinated movement tied to ATP hydrolysis, but this simple description will suffice).

What the adoption explanation assumes is an inherently flexibility/plasticity to IC components. And what this assumption thus predicts is permutations. That is, IC systems should demonstrate much variability. We should see some bacteria with dnaK and two other chaperones (not dnaJ and grpE). We should see others with dnaJ and two other proteins (not dnaK and grpE). We should also see some bacteria with grpE (and not dnaK and grpE). But we don’t. For example, imagine a mutation occurred in a bacterium that disabled dnaJ. IC predicts this would be lethal and selection would prevent this organism from altering the gene pool. But if adoption from a different function is common, the mutation need not be lethal and a new chaperone machine would evolve around the protein substituting for the lost component.

When we look to eubacteria, the three chaperone components are universal. They are found in gm+ and gm- bacteria, thermophiles (Thermotoga and Aquifex), spirochetes, Deinococcus, Campylobacter, cyanobacteria, Neisseria, and even Mycoplasma. Again, this is most likely due to the fact that this chaperone machine is IC (IC predicts functional constraint) and other proteins cannot substitute through adoption from a different function.

Recently, it was determined that most known Archaea lack this chaperone machine. That some have it has been attributed to horizontal transfer. Here was a good opportunity to gather more positive support for the IC status of such an ancient and highly conserved system. I predicted that where we would find the chaperone machine in Archaea, we’d find all three components given that it is IC. Then, not too long ago, I came across a review paper on stress genes and proteins that verified this prediction. The authors noted that genomic data demonstrate the following in Archea:

“whenever hsp70 was present in a genome, hsp40 and grpE were also found if enough sequencing was done; conversely, genome sequencing has demonstrated that if the hsp70 gene is absent, hsp40 and grpE are also absent.”

IC nicely explains this as the lateral transfer of only one or two components of the chaperone machine would be useless and thus degrade. That’s why you see either all three gene products or none. But if adoption from a different function was a common IC generator, we should see permutations in Archaea, where a laterally transferred dnaK gene would find helpers in the archaeal cytoplasm and evolve a new chaperone machine. And all of this is significant because the chaperone machine is a very simple example of IC having only three parts.

Thus, it would seem that the more complex an IC system is, and the less variable it is across phylogenetic lines, the less likely it is that adoption from a different function explains its origin.

One last problem with the adoption explanation is that it ignores another element of Behe’s argument (pp. 43-45):

“To feel the full force of the conclusion that a system is irreducibly complex and therefore has no functional precursors, we need to distinguish between a physical precursor and a conceptual precursor. . . . Darwinian evolution requires physical precursors.”

That is, Darwinian evolution is supposed to be a description of history. Thus, it invokes real-life proteins with real functions that are really co-opted into a conglomerate to perform a new, real function. Darwinists tend to overlook this and prefer to remain in the conceptual realm, where proteins are imagined to have unknown functions such that they somehow get adopted into another conglomerate. And that’s where it all ends with most Darwinists. But as Behe notes, we need physical precursors and this means we need evidence of these physical precursors. Thus, unless the adoption from a different function story is supported by real evidence, it fails to explain the origin of the IC system.

In summary, the adoption from a different function explanation is unlikely to be a general solution to IC (as Orr notes) as it relies too heavily on pure chance as an explanation for apparent design. We thus need independent evidence that this explanation validly applies in any system in question, especially in light of the fact that examples of this pathway do not lend themselves easily to explaining the origin of IC. Furthermore, if we don’t see evidence of permutations that run through out the entire IC system, there is good reason to dismiss this explanation. So let’s turn our attention to the remaining Darwinian pathway.

Elimination of Function Redundancy

The interesting thing about this pathway is that it too robs the standard Darwinian explanation of its appeal. Richard Dawkins presents Darwinism in its most convincing form:

“We have seen that living things are too improbable and too beautifully ‘designed’ to have come into existence by chance. How, then, did they come into existence? The answer, Darwin’s answer, is by gradual, step- by-step transformations from simple beginnings, from primordial entities sufficiently simple to have come into existence by chance.”

Yet elimination of function redundancy is an explanation that does not begin with “simple beginnings,” but instead begins with a state that is more complex than that which is observed. But if simple beginnings are needed to “come into existence by chance,” the complicated beginning, assumed by this pathway, may be too complicated to “come into existence by chance.”

It seems to me that this pathway is not taken seriously when it comes to explaining origins through non-teleological mechanisms. For example, another Behe-critic, Clare Stevens, relies on simple (more direct) beginnings to explain the origin of IC:

So how can the whole process have arisen step by step as required by gradual evolution? It certainly is impossible to believe that these complex sequential processes arose all at once. So we must look for a different format. For a start we must assume that there was a more direct method of synthesis available at one stage indeed that a more direct process might still be available in some creature yet to be investigated (as I have suggested in the case of adenine). Then we must go on to surmise that the intermediate stages are refinements interpolated subsequently.

In fact, even Ussery himself (one of the authors of the paper) uses simple beginnings on his web page to explain the origin of the bacterial flagellum:

“If you look at bacterial flagella, you find that some are indeed quite complicated, but others are more simple. For example, the basal body can vary with species – in E. coli there are four rings, in Bacillus subtilis two rings, and in Caulobacter crescentus five rings. I can easily imagine a scenario where a “primitive bacterium” might have one ring, and then you have a flagellum with two rings, then three, and so on. This is a “gradual, step-by-step” evolution, which is the antithesis of Behe’s argument.”

” It is likely that as new bacterial genomes continue to be sequenced (at the rate of about one a month!), organisms will be found which require even fewer genes to make a completely functional flagella.”

Thus, Dawkins, Stevens, and Ussery (and many more) all go back to “simple beginnings,” the very opposite initial state assumed by this pathway.

However, I personally find this pathway of redundancy elimination to be very interesting, as it may suggest that some originally designed states were much more complex than seen today, such that evolution was essentially rigged to evolve in particular directions. In other words, this pathway does not eliminate the design inference behind IC, but instead, suggests that IC is an indirect indicator of an originally designed state.

Nevertheless, what we need is evidence that the initial state was more complicated than the IC state. For example, are we talking about flagellum that were originally composed of 60 parts? Where is the evidence for such a claim? It is an interesting thought, but without evidence, we can’t take it beyond the realm of philosophy.


Behe’s notion of IC has found itself into the scientific literature and is being taken seriously by scientists. Behe has contributed to science by forcing non-teleologists to once-and-for-all lay the various Darwinian pathways on the table. This is progress as we can now look to the data to determine if there is any evidence that these pathways apply to an particular IC system in question.

Behe’s notion of IC does indeed help us to effectively rule out some of the Darwinian pathways, as admitted by T&U. What is most relevant is that the pathways ruled out by IC are also those best supported by example/evidence and those that are most persuasive in explaining apparent design. The traditional examples of Darwin’s finches (and their beaks), giraffe necks, elephant trunks, antibiotic resistance, and the darkening wings in moths give us no reason to think IC systems were generated by the RM&NS. The remaining explanations for IC are indeed possible, but without evidence to support them, there is no reason to seriously embrace them. Neither explanation constitutes a better general solution to IC than intelligent design. What’s more, both explanations seriously weaken the overall appeal of the standard non-teleological explanations, as they resurrect a prominent role for pure chance in the origin of apparent design and/or rely on complicated initial states that may lend themselves more readily to a teleological cause.

Without realizing it, T&U have made a significant contribution to ID.

For more discussion, go to ARN Forum.