Skip navigation.

Irreducible Complexity Demystified

Pete Dunkelberg's essay will prove helpful to educators, boards of education, legislators and members of the press who may be wondering about "Irreducible Complexity".

Irreducible Complexity Demystified

by Pete Dunkelberg
[Posted: 26 April 2003]
"Evolution is cleverer than you are."
-biologists' proverb
The Argument That Irreducible Complexity Cannot Evolve
How Might Irreducible Complexity Evolve?
Irreducible Complexity in Nature
      Venus' Flytrap
      How to Eat Pentachlorophenol
      Hemoglobin for the Active Life
      The Blood Clotting System: is it IC?
    Swimming Systems
      The Eukaryote Cilium
      The Archaeal Flagellum
      The Bacterial Flagellum
IC Cores
How Does Irreducible Complexity Get Its Charm?
IC, ID, and Creationism


A new term, irreducibly complex, (IC) has been introduced into public discussions of evolution. The term was defined by Michael Behe in 1996 in his book Darwin's Black Box: The Biochemical Challenge to Evolution (1). Irreducible complexity (also denoted IC) has gained prominence as the evidence for the intelligent design (ID) movement, which argues that life is so complicated that it must be the work of an intelligent designer (aka God) rather than the result of evolution. As you may have heard, the ID movement wants this taught in public schools as a new scientific theory. This essay will, I hope, prove helpful to any school teachers, boards of education, legislators and members of the press who may be wondering about it.

The argument from IC to ID is simply:

  1. IC things cannot evolve
  2. If it can't have evolved it must have been designed

This article just looks at the first part, the argument that irreducibly complex systems cannot be produced by evolution, either because they just can't evolve, or because their evolution is so improbable that the possibility can be ignored.

Let's take a look at the definition of IC, and then see if we can figure out its relation to evolution, and why scientists are so unimpressed. Here is the definition, from page 39 (page numbers refer to Darwin's Black Box unless otherwise noted):

"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." [emphasis in original]

IC is now a single defined term. The new definition, not the ordinary meaning of the words, is now our guide. IC refers to an organism doing something (the function) in such a way that the system (that portion of the organism that directly performs the function) has no more parts than are strictly necessary.

How do we decide when the term IC applies? Organisms don't come with parts, functions and systems labeled, nor are 'part', 'system' and 'function' technical terms in biology. They are terms of convenience. We might say, for instance, that the function of a leg is to walk, and call legs walking systems. But what are the parts? If we divide a leg into three major parts, removal of any part results in loss of the function. Thus legs are IC. On the other hand, if we count each bone as a part then several parts, even a whole toe, may be removed and we still have a walking system. We will see later that Behe's treatment of cilia and flagella follows this pattern.

What about the boundary of the system? This too is up to us. Take the digestive system for example. We may be interested only in the action of acids and enzymes in the stomach, or we may include saliva and chewing, or the lower intestine where some extraction of water and nutrients continues.

As a mental exercise, try before reading on to formulate an argument to prove that IC systems cannot evolve. IC is supposed to be the biochemical challenge to evolution, and thus the case when the parts are molecules, usually proteins, is the important case. So of course there may be multiple copies of a part. Losing a part means losing all copies of it, or at least so many that the function is effectively lost.

The Argument That Irreducible Complexity Cannot Evolve

Behe's argument that IC cannot evolve is central to ID, so it deserves our attention. His method is to divide evolution into what he calls 'direct', which he defines in a special way, and 'indirect' (everything else). He finds that direct evolution of IC is logically impossible, and indirect evolution of IC is too improbable. The argument against 'direct' evolution of IC is contained in this long sentence right after the definition:

"An irreducibly complex system cannot be produced directly
(that is, by continuously improving the initial function, which continues to work by the same mechanism)
by slight, successive modifications of a precursor system
because any precursor to an irreducibly complex system that is missing a part is by definition nonfunctional."

The last part of the sentence, "...because any precursor to an irreducibly complex system that is missing a part is by definition nonfunctional." is why we should agree to the rest of the sentence. There are some problems:

  • The first part of the sentence refers to slight changes. Removing a whole part is a major change;
    this is a major 'disconnect' between the parts of Behe's argument.
  • It is not true that a precursor missing a part must be nonfunctional. It need only lack the function we specified. Even a single protein does something.
  • The actual precursor may have had more parts, not fewer.
  • If the individual parts evolve, the precursor may have had the same number of parts, not yet codependent. We will learn more about this possibility shortly.

How can one construct a valid argument that IC cannot be produced directly? ID proponents have not found a way. Yet it's easy (and left as an exercise for the reader) once you realize that a valid argument from definitions requires carefully defining the terms so that the argument becomes a tautology. This may be accomplished by redefining 'direct' or 'IC', or (best, I think) by defining Behe's expression 'be produced' which he uses in place of 'evolve'.

A precursor to IC lacking a part can have any functions except the specified one, which brings us to 'indirect' evolution. Consider a cow's tail. So far as I know, the main thing a cow uses its tail for is to swat flies. Did tails originally evolve for this function? Hardly. There were tails before there were flies. Long ago, tails helped early chordates to swim. Going back still farther, some very early animals started to have two distinct ends; one end for food intake (with sense organs for locating food) and the other end for excretions. As a consequence, this back end, and muscular extensions of it, could also be used to help the animal move. This illustrates yet another important facet of evolution: not only single mutations, but even large organs may begin more or less accidentally. It also illustrates that biological functions evolve. Indeed organisms and ecosystems evolve. It may not even make sense to expect a precursor to have had the same function.

The long term evolution of most features of life has not been what Behe, or indeed most people, would call direct. And even short term evolution can be indirect in Behe's terms. So it is surprising to read, on page 40, Behe's argument against indirect evolution of IC systems. Here is the crux of it:

"Even if a system is irreducibly complex (and thus cannot have been produced directly), however, one can not definitely rule out the possibility of an indirect, circuitous route. As the complexity of an interacting system increases, though, the likelihood of such an indirect route drops precipitously." (page 40)

He simply asserts that evolution of irreducible complexity by an indirect route is so improbable as to be virtually out of the question, except in simple cases. He makes no special connection between indirect evolution and IC. He offers no evidence. He just asserts that it is too improbable.

Actually, a more complex system probably has a long evolutionary history. Since evolution does not aim at anything in advance, the longer the history, the more circuitous it may be. And his very limited meaning of 'direct' renders much indirect that is not circuitous at all. Yet he insists:

"An irreducibly complex biological system, if there is such a thing, would be a powerful challenge to Darwinian evolution." (page 39)

Here's another exercise: before reading on, try to think of ways that IC systems, including biochemical ones, might evolve after all.

How Might Irreducible Complexity Evolve?

How might an IC system evolve? One possibility is that in the past, the function may have been done with more parts than are strictly necessary. Then an 'extra' part may be lost, leaving an IC system. Or the parts may become co-adapted to perform even better, but become unable to perform the specified function at all without each other. This brings up another point: the parts themselves evolve. Behe's parts are usually whole proteins or even larger. A protein is made up of hundreds of smaller parts called amino acids, of which twenty different kinds may be used. Evolution usually changes these one by one. Another important fact is that DNA evolves. What difference does this make, compared to saying that proteins evolve? If you think about it, each protein that your body makes is made at just the right time, in just the right place and in just the right amount. These details are also coded in your DNA (with timing and quantity susceptible to outside influences) and so are subject to mutation and evolution. For our purposes we can refer to this as deployment of parts. When a protein is deployed out of its usual context, it may be co-opted for a different function. A fourth noteworthy possibility is that brand new parts are created. This typically comes from gene duplication, which is well known in biology. At first the duplicate genes make the same protein, but these genes may evolve to make slightly different proteins that depend on each other.

We can summarize these four possibilities this way:

  • Previously using more parts than necessary for the function.
  • The parts themselves evolve.
  • Deployment of parts (gene regulation) evolves.
  • New parts are created (gene duplication) and may then evolve.

The first of these only comes up if we are looking for IC. The others are the major forms of molecular evolution observed by biologists, phrased in terms of parts. They can lead to new protein functions, sometimes slowly and sometimes, especially when parts are redeployed, abruptly. Gene duplication and changes in protein deployment may introduce a new protein 'part' into a system. Then the parts may coevolve to do something better, but in a codependent manner so that all are required, without further change in the number of parts. But what happens in nature?

Irreducible Complexity in Nature

Can evolution lead to IC or not? It is time to look at living examples and let nature decide. Behe's most famous example is a mousetrap. But since a mousetrap is not alive, it doesn't tell us much about whether or how living IC systems might evolve. How about a flytrap instead?

Venus' Flytrap

Dionaea muscipula, the Venus' flytrap

The Venus' flytrap, Dionaea muscipula, is a small flowering plant which grows naturally in acidic wetlands in North and South Carolina. It has a ferocious looking tooth-edged trap for unwary creatures. It traps and digests insects to make up for the lack of nitrogen in the soils of its habitat.

Here's how the trap works. When an insect brushes against the trigger hairs in the center, the lobes snap most of the way shut with surprising speed. If a small insect is caught, it may escape between the teeth, and then the trap reopens without fully closing. If a good sized bug is caught it is digested over the next few days as the trap closes the rest of the way. Then the trap reopens. A trap can only be fully closed about 4 times, so it must be used sparingly.

Do we have an IC system here? We must specify a function and all the parts needed to carry it out (and no extra parts). The function of interest is trapping insects for food in a manner that brings the plant more benefit than the cost of the trap. The parts are the two lobes, the hinge between the lobes (the midrib of the leaf, which anchors the lobes), the trigger hairs, and spines projecting from the edges of the lobes that make a set of bars as the trap closes. The system is just all these parts, and the trap needs all its parts in order to work. Hence it is an IC system.

How might this trap have evolved? I say 'might' have because Venus' flytraps haven't left any fossils that I know of, except a few grains of pollen. Are there any related plants that might provide a clue? Let's look at the well known sundews (Drosera). Sundews trap insects using flypaper traps, slowly closing around insects that get stuck. Darwin, whose book Insectivorous Plants (2) is now available online, made careful observations of these remarkable plants, especially the round leaf sundew D. rotundifolia. As Darwin notes,

If a small organic or inorganic object be placed on the glands in the centre of a leaf, these transmit a motor impulse to the marginal tentacles. The nearer ones are first affected and slowly bend towards the centre, and then those farther off, until at last all become closely inflected over the object. This takes place in from one hour to four or five or more hours. [...] Not only the tentacles, but the blade of the leaf often, but by no means always, becomes much incurved, when any strongly exciting substance or fluid is placed on the disc. Drops of milk and of a solution of nitrate of ammonia or soda are particularly apt to produce this effect. The blade is thus converted into a little cup. The manner in which it bends varies greatly. (2, pp 9, 12)

A sundew catches a fly

Here is D. rotundifolia with a fly; Makoto Honda (3) shows the action with a faster species, D. intermedia. Recent genetic research confirms that Venus's flytrap and the waterwheel plant Aldrovanda are related and are in the sundew family Droseraceae, and that snap-traps very likely evolved from flypaper traps (4) as Darwin thought:


The six known genera composing this family have now been described in relation to our present subject, as far as my means have permitted. They all capture insects. This is effected by Drosophyllum, Roridula, and Byblis, solely by the viscid fluid secreted from their glands; by Drosera, through the same means, together with the movements of the tentacles; by Dionaea and Aldrovanda, through the closing of the blades of the leaf. In these two last genera rapid movement makes up for the loss of viscid secretion. [...] The parent form of Dionaea and Aldrovanda seems to have been closely allied to Drosera, and to have had rounded leaves, supported on distinct footstalks, and furnished with tentacles all round the circumference, with other tentacles and sessile glands on the upper surface. (2, pp 355-6, 360).

How did the Venus' flytrap avoid the argument that IC can't evolve? In two ways. First, rather than gaining a part, it lost a part - the glue that the sundews use. Even more interestingly, the trap was able to evolve because the parts evolved. The trap started out as a Drosera-like leaf, and the parts of the leaf were progressively changed. This makes a striking contrast with the mousetrap which Behe has repeatedly presented to illustrate why IC cannot evolve. As a manufactured item the mousetrap neatly illustrates his definition, but with its static parts it cannot model evolution. With evolving parts, nature can create a snap-trap after all. The mechanical and manufacturing analogies so influential in Behe's thinking miss the flexibility of living things.

How to Eat Pentachlorophenol

Pentachlorophenol (PCP) is a highly toxic chemical, not known to occur naturally, that has been used as a wood preservative since the 1930's. It is now recognized as a dangerous pollutant that we need to dispose of. But how?

Evolution to the rescue! A few soil bacteria have already worked out a way to break it down and even eat it. And conveniently for us, they do it in an irreducibly complex way. The best known of these bacteria is called Sphingomonas chlorophenolica (also called Sphingobium chlorophenolicum).

The PCP molecule is a six carbon ring with five chlorine atoms and one hydroxyl (OH) group attached. The chlorines and the ring structure are both problems for bacteria. S. chlorophenolica uses three enzymes in succession to break it down, as follows: the first one replaces one chlorine with OH. The resulting compound is toxic, but not quite as bad as PCP itself. The second enzyme is able to act on this compound to replace two chlorines, one after the other, with hydrogen atoms. The resulting compound, while still bad, is much easier to deal with, and the third enzyme is able to break the ring open. At this point, what is left of PCP is well on its way to being food for the bacterium.

All three enzymes are required, so we have IC. How could this IC system have evolved? First of all, bacteria of this type could already metabolize some milder chlorophenols which occur naturally in small amounts. In fact the first and third enzymes were used for this. As a result the cell is triggered to produce them in the presence of chlorophenols. The second enzyme (called PcpC) is the most interesting one; the cell produces it in sufficient quantity to be effective all the time instead of just when it is needed in its normal metabolic role. Thanks to this unusual situation PcpC is available when it is needed to help eat PCP.

The inefficient regulation of PcpC is evidently the key to the whole process. So far as biologists can tell, a recent mutation that changed the deployment of this enzyme is what made PCP degradation possible for this bacterium. It also happens that both PcpC and the first enzyme in the process are now slightly optimized for dealing with PCP; they handle it better than the corresponding enzymes in strains of S. chlorophenolica that use PcpC only in its normal role, but not nearly as well as would be expected for an old, well adapted system. These factors, combined with the fact that PCP is not known to occur naturally, make a strong circumstantial case that this system has evolved very recently.

The chemistry and probable evolution of this system are explained in much greater detail in Shelly Copley's article "Evolution of a metabolic pathway for degradation of a toxic xenobiotic: the patchwork approach" in Trends in Biochemical Sciences (5).

Hemoglobin for the Active Life

Hemoglobin is a wonderful protein that picks up oxygen in our lungs and delivers it to the rest of our cells. Oxygen binds to hemoglobin very quickly in our lungs and stays bound. Then in our tissues oxygen is released very quickly. How does this happen? What we call a hemoglobin molecule is a complex of four hemoglobin chains, or subunits. There are two each of two different chains called alpha and beta hemoglobin. The complex binds reversibly to oxygen, one O2 molecule per each subunit. It tends not to bind to the first oxygen until the oxygen concentration is fairly high, which is the usual situation in our lungs. Then the complex changes shape so that the next O2 binds more readily, the third still faster, and the fourth faster yet. Then it holds the oxygen until the surrounding oxygen concentration is quite low, which happens in our tissues. When finally one oxygen is released, the next is released faster and so on. This mechanism for oxygen transport is much more efficient than can be achieved with alpha or beta hemoglobin alone, and allows for our active life style. It takes all four parts to do this; take away part of the complex and it doesn't work (6). So we have another IC system. Behe discusses hemoglobin briefly (pp 206-207), mainly commenting that it makes a poor case for Design. He doesn't mention that it is IC. This post (7) has some sharp commentary on the subject.

The hemoglobins (globular proteins incorporating a heme group, which in turn cradles an atom of iron) turn out to be a widespread protein family with a long history. They occur in plants and bacteria as well as in animals, and have diverse functions including oxygen transport, oxygen storage, scavenging oxygen to protect some metabolic processes from it, and electron transfer. Interestingly, these diverse functions depend critically on when and where the protein is deployed. Commenting on this in his article "The Evolution of Hemoglobin", Ross Hardison says "This suggests that the creation of new protein functions arises as much from changes in regulation as from changes in structure." (8, p 126). Fetal hemoglobin, which must extract oxygen from the mother's hemoglobin, is a good example of this. We always have the genes for it, but only make it at the right time. Gene duplication has also played a key role. Lampreys and hagfish, which don't have jaws, also don't have the alpha and beta varieties of hemoglobin. Instead they have just one variety of hemoglobin in their blood, and not so efficient oxygen transport. The gene duplication which led, after further changes, to our distinct alpha and beta chains evidently happened in the ancestor of all living vertebrates with jaws.

Let's take stock of what we have learned before moving on to more complicated examples. Venus' flytrap makes an instructive comparison with Behe's mousetrap. In one, the parts evolve. In the other, they don't. What a difference this detail makes. The protein parts of biochemical systems also evolve, so the flytrap is a good model for them and the mousetrap isn't. The flytrap and hemoglobin show in different ways that removing a part is often not the same as evolution in reverse. The flytrap has already lost a part (the glue that Drosera use to trap insects). With hemoglobin, taking away either the alpha or the beta chain would be a disaster unless the whole animal could be 'evolved back' to a much earlier stage.

Both hemoglobin and the recent evolution of a way to metabolize PCP show that what we have called 'deployment of parts' is important in evolution. Biologists usually call this regulation of gene expression, or just gene regulation. From another point of view, it is called co-option or recruitment of a protein to a new function. If a protein takes on two roles, any subsequent duplication of the corresponding gene will be subject to selection for both its regulation and the separate functions. Hence this duplication will be more likely to persist and spread in the population.

Here's another interesting thing about the PCP example: it amounts to 'adding a part to a previously non-functional system', which is exactly what Behe thinks cannot happen, because he thinks the organism couldn't have lived without that part. It turns out that a single mutation can create a new function and mechanism, allowing the organism to live better or in a new environment. This is indirect evolution in Behe's terms, but to DNA it is just another mutation.

So far IC seems to be no problem for evolution. Is there anything to the biochemical challenge? Let's look at the impressively complicated examples on which IC's reputation rests.

The Blood Clotting System: is it IC?

Blood clotting is an example of what biochemists call a cascade: one protein does something, which starts another protein doing something, which starts another.... Cascades, and the clotting cascade in particular, are among the favorite examples of ID proponents. Yet giving a precise specification of system, parts, and function so that the specified system is IC turns out to be difficult. Hard to specify or not, it is still one of Behe's favorite examples. He devotes his entire fourth chapter to it. After explaining how it works, he indicates that scientists know almost nothing about how it evolved. His main evidence for this is a nontechnical lecture given by Russell Doolittle. But of course that talk, using analogies to Yin and Yang, was not meant to convey a technical understanding. After several people commented on this, Behe responded with an online essay "In Defense of the Irreducibility of the Clotting Cascade" (9). The defense comes down to saying that evolution of this system would require too many 'unselected steps'. But this is not true, as pointed out by Ken Miller in Finding Darwin's God (10) and in his online article (11) where he gives more details than the publisher wanted in the book.

The clotting cascade is a member of a family of cascades with a long pedigree. Our immune system includes a related cascade which Behe considers to be IC, but see Matt Inlay's article "Evolving Immunity" (12). A recent paper by Krem and Di Cera (13) pursues the evolution of cascades farther down the evolutionary tree. They discuss biochemically similar cascades in horseshoe crabs, fruit flies, and ourselves. They find that "Extensive similarities suggest that these cascades were built by adding enzymes from the bottom of the cascade up and from similar macromolecular building blocks." Behe argues that this type of evolution would not happen because there would be unselected steps. But he thinks in terms of precursor systems with missing parts, not in terms of ancestor organisms in different environments with different problems to solve. This may reflect a difference between thinking like a chemist and thinking like a biologist. Early forms of the cascade occurred in animals without a high pressure circulatory system like ours. In horseshoe crabs, for instance, a simpler form of the clotting cascade serves to entangle invading bacteria. There is no reason to presume unselected steps (other than gene duplication, which may be neutral at first) if the organism and its way of living and its environment are changing.

But have you noticed something missing from our discussion of the clotting cascade? We haven't proven that it is IC. The way to do this, as Behe tells us on page 42, would be to take the parts one by one and show that each is required for clotting. Or point to published research that does this. Surely Behe took care of this detail in the fourth chapter of his book? No. He 'proved' it rhetorically, but not systematically. Well then, when he published a web page several years later entitled "In Defense of the Irreducibility of the Blood Clotting Cascade" (9) he must have filled in the details? No again. He advanced his argument against the evolvability of the clotting cascade, but that has been answered (10, 11, 13, 14). Meanwhile, the little matter of proving that it is IC has been overlooked. And there is evidence to the contrary: whales, mammals like us, lack a key part called Hageman factor but their blood clots anyway (15). Under questioning at a recent meeting (16) Behe finally agreed that the cascade is not IC after all. Indeed, Acton gives reasons why he never should have thought so (14). (As far as I know, Behe has not 'done his homework' on any of his examples except the mousetrap).

Swimming Systems

We come now to what have become the very most important purported examples of IC in nature: swimming systems. These are flexible projections that microbes use to move themselves through fluids. The three main types of microbes, bacteria, archaea, and single celled eukaryotes, use different swimming structures, and there are major differences between species of each type. Some bacteria even manage to swim without flagella, including little understood Synechococcus (17) and much better understood Spiroplasma melliferum (18). Of course microbial motion is not limited to swimming. They also have ways to move along surfaces and maneuver in sand and ooze. Bardy et al. review almost all of the known ways bacteria and archaea move (19).

Swimming systems depend on what are called molecular motors, a favorite topic of molecular biologists. Those who are curious about molecular motors may start here (20). Brownian ratchets, fascinating in their own right (21), are one of the energy sources for these tiny motors.

From a biological perspective, the function of an organism is to live and grow enough to reproduce. The function of any part of the organism is to contribute to this in any ways whatsoever. Appendages can help a cell in various ways such as sensing the environment, finding food or mates or communicating with other cells. It helps if the appendage can move about. This in turn will move the cell a little. (Think of waving your arm under water). In an environment where swimming is advantageous, it is not surprising that the ability to swim would evolve. Never the less, as the evolution of vertebrate systems like the clotting cascade and the immune system has become better understood, ID proponents have come to rely more and more on swimming systems, especially the bacterial flagellum, as the real evidence for Design in nature.

The Eukaryote Cilium

Eukaryotes are any organisms like trees, people, protozoa and amoebae which, unlike bacteria, have their DNA in a separate nucleus within the cell. Many eukaryote microbes propel themselves through water by waving projections called cilia, which they also use to collect food such as bacteria. A bit of terminology: cilia are also called flagella, especially when a cell only has one or two. The microbes are then called flagellates. But eukaryote flagella and bacterial flagella are entirely different structures.

How do we define an IC system in the case of the eukaryotic cilium? Behe first specifies the system as the entire cilium. The function of the system is to move the cell through liquid by a sort of waving action. What about parts? At the level of biochemical machines, one usually thinks of individual proteins as parts. But Behe simply divides a cilium into three large parts, which he calls 'motor, connector, and paddle' (page 65). It is clear that a cilium wouldn't work without each of these big parts, so we have IC. Cilia are many and diverse (for examples see Finding Darwin's God (10, page 142)) and may contain two hundred or so different proteins and various numbers of microtubules. Some proteins are always present; others vary from microbe to microbe. If we take proteins as our parts (the biochemical challenge), then cilia aren't IC; no one has been able to find a real cilium with an 'irreducible' set of proteins. If we take microtubules as our parts as Miller does, the cilium is not IC. But with Behe's parts it is IC. Remember, it's up to us to choose function, system and parts to satisfy the definition. Or not. So it turns out that being IC or not is not a property of the cilium itself. It depends on choices we make.

With the parts (motor, connector and paddle) so removed from the mutation-by-mutation level of change, how does Behe relate the ICness of the system to its evolution? First, with his choice of function, parts and system, it is IC. This rules out 'direct' evolution to his satisfaction. What about 'indirect', i.e. normal evolution, taking into account that everything changes including functions? This is ridiculed on pages 65-67. Although the cilium is an extension of the cell's cytoskeleton, he suggests that a proto-cilium would be disadvantageous. He winds up:

"... but even if [a proto-cilium] were at the cell surface, the number of motor proteins would probably not be enough to move the cilium. And even if the cilium moved, an awkward stroke would not necessarily move the cell. And if the cell did move, it would be an unregulated motion using energy and not corresponding to any need of the cell."

So a proto-cilium would be useless and probably even a harmful waste of resources until it was perfected, in Behe's opinion.

Microbes do not agree, and make use of a variety of projections that have the 'defects' that Behe mentions. The amoeba Raphidiophrys pallida, shown here (22) has projections called axopodia which it uses to capture prey and to move itself along a surface such as a bit of pond weed. The protozoan Actinophrys in this Pond Scum Action Video (23) explores with gently waving axopodia. Foraminifera are very common protists in the oceans and in the ooze beneath. They use projections called reticulopodia to find and capture food, and to maneuver among sand grains (24). These projections, although dependent on many of the same proteins for motion, are not cilia. But they resemble the clumsy cilia that Behe objects to, and show that his objections do not hold up in nature.

Now, what about cilia in the strict sense? The cilium in its early form would have been too short to function as a rowing device. What could it have done? The first flagellates are long gone, but we can still learn from the ones at the base of the family tree as it now exists. The soil dwelling flagellate Phalansterium is about as basal as any. It is hard to watch in action, but it probably uses its cilium to sense the environment and to collect bacteria to eat. The eukaryote family tree has two main branches, leading to plants and animals. At the base of these branches we find water dwelling flagellates that push water in opposite directions (25, 26). Mastigamoeba creep along surfaces and move their cilium to create a slight current toward themselves, drawing in food particles. Choanoflagellates, on the line leading to animals, use their cilium to push water away (27). This draws in more water, and food along with it.

Any projection that could stir the water at all would help to bring more food to the microbe. Gradually improving it naturally leads to a swimming system. Behe's objections overlook evolutionary change of function, which would naturally occur to a biologist but perhaps not to a chemist like him.

The Archaeal Flagellum

Archaebacteria, or Archaea for short, have recently been recognized as an important group of microbes distinct from bacteria and from eukaryotes. Their flagellum is analogous to a bacterial flagellum, but simpler and quite different in detail. As the diagram shows, it resembles another kind of projection called a type IV pilus, to which it is probably related (19). Type IV pili themselves are not used for swimming, but bacteria use them for simpler ways of moving called twitching and social gliding (19, 28, 29). Behe has not discussed archaeal flagella, and I am not sure how he would divide them into parts. They don't appear to fit his preferred three part division (a motor, then a rotor or connector, then a third part which pushes against the medium) on which he bases his statement that "the complexity is inherent in the task itself" (page 65).

The Bacterial Flagellum

Here it is -- the number one argument for design in nature. ID advocates have even made a movie called Bacterial Flagella: A Paradigm for Design. It is on sale at the ARN web site (30) and briefly discussed in (31). Behe said recently:

"If [biologist Jerry] Coyne demonstrated that the [bacterial] flagellum, (which requires approximately forty gene products) could be produced by selection, I would be rather foolish to then assert that the blood clotting system (which consists of about twenty proteins) required intelligent design". (32)

Bacterial flagella are many, diverse, and complicated. Behe concludes that any bacterial flagellum is composed of at least three parts: a paddle, a rotor, and a motor, and so with swimming as the specified function must be IC (page 72). Even at this crude level, the ICness of a flagellum is not so clear. The problem is that there are additional parts to a complete flagellum. For instance, there are proteins at the base that react to external stimuli and turn the motor on and off, and in some flagella cause it to change directions. And there are other proteins that are arranged in rings where the flagellum passes through the cell membrane.

But the more interesting question is: could a flagellum be IC with proteins, not paddles etc. as parts? Remember, IC is supposed to be the biochemical challenge to evolution. We've already seen that it isn't such a challenge after all, but so much has been made of the purported ICness of the flagellum that one should be well informed on the subject just to be more interesting at parties :). In order to decide, one must first choose a flagellum. Even within a single bacteria species, different strains may have different proteins and different numbers of proteins in their flagella. Even a single rod-shaped bacterium may have quite different flagella at its ends and around its sides. Next, discover and list all the proteins in that particular flagellum. This requires deciding just where it begins, and one's decision about this may depend on the exact function one has in mind for 'the' flagellum. Then comes the hard part: proving that every last protein is required for the function. Oddly, ID proponents show no interest in doing any of this work, not even picking a particular flagellum of a particular bacterium to start on. It is as if just asserting the ICness of 'the' flagellum gives them full satisfaction.

What's the answer? Is any flagellum IC with proteins as parts or not? As this would depend on arbitrary criteria, scientists have not pursued this question as such. But quite a bit has been learned about various flagella. It is clear that all of them absolutely require a good many of their proteins in order to function as swimming systems. But not one is yet known to require every last protein, and some are known not to (19, 33, 34). Could a flagellum be IC with proteins as parts? Sure. As we have seen in the much simpler case of hemoglobin, proteins can evolve to become codependent. There may be a perfectly IC flagellum out there just waiting to be discovered.

Even so, it wouldn't be the simplest swimming system. As the diagram shows, a bacterial flagellum is much more complex than an archaeal one. This is in part because it is built from, in fact secreted by what is called a type three secretion system (TTSS). This is a complicated thing in itself. It is a tiny tube which starts below the cell wall and sticks out through it, and serves as a conduit for protein export. The flagellar TTSS (there are other kinds) specializes in secreting the rest of a flagellum. The TTSS base counts as part of the flagellum, and is itself about as complex as an archaeal flagellum.

Since it is more complicated than is required for swimming alone, you might suspect that a bacterial flagellum has other functions. You would be right. These other functions vary from bacterium to bacterium and from situation to situation, and scientists have only recently been able to observe them. First, some flagella also export proteins, including ones that cause sickness (35). This is not too surprising since that's what the other TTSS's are known for.

But spirochetes, the spiral shaped bacteria, use flagella in a way one wouldn't expect. Their flagella don't stick out, yet are used for swimming, burrowing, and maintaining the cell's shape. Flagella are grown at both ends and extend toward the middle under the outer membrane. The flagella maintain the cell's spiral shape, and by rotating can create a moving wave along the cell, causing the cell to move in the opposite direction (19).

It is not easy to observe the behavior of individual bacteria in the wild. Just recently though, Danish researchers noticed some unusual behavior by bacteria living on low oxygen marine sediments. To see exactly what the bacteria were doing, they recreated the ecosystem in the laboratory. Who would have thought that some bacteria, shaped like slightly bowed rods, would tether themselves to the sediment with a mucus stalk secreted from the center and then use flagella at both ends to move like a propeller? But that's what these bacteria do. They create a tiny current, refreshing the water around them much faster than diffusion alone could do it (36).

Bacteria can move across surfaces in organized swarms, and quickly colonize a new food source such as your own much larger cells. When swarming, they often grow more flagella than usual and make cell-to-cell contacts with these flagella (37). Some bacteria also use their flagella to hang on to our cells as they try to break in and eat the cell contents (38).

This brings us to the dark side of design. Flagella participate in the cause of quite a few bacterial diseases, including diarrhea (38), ulcers and urinary tract infections (39). If the Designer is directly responsible for flagella then he is implicated as a cause of human diseases. Diarrhea is no joke; it is a leading cause of infant death in some parts of the world. To make matters worse, one can hardly give the Designer credit for flagella without also crediting him with TTSS's in general (40). This puts the Designer solidly behind Bubonic plague (41, 42) and many other diseases (43). Happily, science makes such beliefs unnecessary.

Swimming systems provide a good illustration of how (not) to think about evolution. Behe argues that evolution can't produce them because they are IC (a dubious claim and not an obstacle to evolution as we have seen). He buttresses this by arguing that is quite improbable that a swimming system good enough to be useful would appear all at once. And it wouldn't evolve slowly, he supposes, because until it became an effective swimming system, there would be nothing for natural selection to select. However, he envisions a part that sticks out, but that has no use at all other than swimming - and at first it can't even do that. But parts that stick out can have a number of functions, and bacterial flagella clearly have several. If there is another reason for it to be there, the sticking out part can gradually evolve more abilities. This involves change of function, or indirect evolution as Behe calls it. He dismisses this possibility, calling it improbable. A closer look shows the opposite.

IC Cores

Have you noticed that none of our complex examples is IC at the molecular level? The argument that IC can't evolve is made in general terms, but it is at the molecular level that the 'biochemical challenge to evolution' is supposed to really count. Granted, we have seen that IC can evolve, but proponents of 'IC implies ID' are able to overlook this. They have not entirely overlooked the fact that not even one impressively complicated molecular system has been shown to be IC. The proposed solution to this problem is that these systems have 'IC cores': if you remove proteins one by one, at some point what is left will be IC. And if you remove parts in a different order, you may find other IC cores. But arbitrarily removing parts that have become coadapted is not "evolution in reverse", so IC cores don't tell us about evolution. And how are they relevant to IC if nature uses more than the core? All I know is that IC cores seem to matter to ID proponents.

Now let's take a more biological look at it. The whole irreducible complexity argument is based on fixed functions, parts, systems, organisms and environments. In nature all these things vary. Evolution, appropriately enough, is all about change. The search for what might be called an evolutionarily irreducible core to any of the complex examples will take you back, back, back to who knows what? The immediate precursor may have had more parts. Or if it had fewer parts, it is probably not appropriate to remove a part before modifying the parts so that they are not coadapted and codependent. Perhaps the whole organism should be modified and placed in a different environment before removing a part. Where do you stop (14)?

Since we found that simple IC systems readily evolve, why are complicated ones hard to find? Some of the same modes of change that can produce IC can just as well add complications that don't fit the definition. Most genes are members of gene families that have grown through gene duplication over time. The time since a particular duplication happened can be estimated by the amount of difference between members of the family. This is a dead giveaway that the organism's ancestors got along with fewer "parts". And given a large number of parts, these parts are likely to have additional functions, which is a good reason for there to be more parts than the minimum needed for what we decide is 'the' function of those parts. For example, we noted earlier that whales lack one of our blood clotting proteins, called Hageman factor. People are sometimes born with a mutation that leaves them with only 40 to 60 percent of the normal amount of Hageman factor. Their blood still clots. But women with reduced Hageman factor tend to have more miscarriages.

How Does Irreducible Complexity Get Its Charm?

Evolution doesn't even notice whether a combination of parts, system and function chosen by an observer happens to satisfy a definition in a book. It just doesn't matter. This is in a nutshell what scientists have been saying since Darwin's Black Box was published (44). Yet the book has been very influential with the public (see for instance the 370 or so reviews at (45)). And it provides the one seemingly scientific reason to teach ID in public school science classes.

How can the book's success with laymen be explained? First, it appears that evolution is hardly taught in the US. Basic knowledge such as the four modes of evolutionary change given at the beginning of this essay would show a reader that evolution is much too flexible for IC to be an issue. Biological basics and careful reading would enable one to see that Behe's theoretical argument that IC can't evolve is unsound.

Without a good basic understanding of biology, a tricky ambiguity sets in. First, there is the definition of IC. Then comes the apparent proof that it cannot evolve. After that, 'unevolvable' is casually used as the meaning of IC. To complete the picture, the subtext all along is that IC is impressively complicated. Thus one's attention is directed away from simple cases which directly show how basic modes of evolutionary change may lead to IC. The definition is used to argue that IC exists, and the other two meanings seal the conviction that IC systems are very unnatural indeed. It is a case of using words which seem to mean what some people want them to mean, but on closer examination don't. This results in the reader losing sight of the fact that none of Behe's examples are in fact IC in biochemical terms, and of the fact that IC doesn't matter for evolution in any case, so that there actually is no 'biochemical challenge to evolution' at all.

IC, ID, and Creationism

Is IC/ID a form of creationism? It looks like it to many people, but proponents reject that label. Let's see if we can sort it out.

You may have heard that ID differs from creationism in not insisting that the earth is only a few thousand years old. It's not quite that simple. Creationists come in both 'young earth' and 'old earth' varieties. So do ID proponents. The difference is that old and young earth creationists are at odds. ID, on the other hand, takes a 'big tent' approach. You are free to accept geological evidence or wave it aside. The Designer could have made the earth look older than it is.

ID mirrors creationist thinking in a fundamental way that you might not notice if you are not familiar with the genre. All creationists agree that there are some inherited genetic changes. The different breeds of dogs, for instance, are not held to be special creations. But creationists always divide evolutionary changes into two kinds: there is a simple kind of change, which they agree evolution can do. But evolution is always somehow blocked from causing the really significant changes, either because evolution just can't do it, or it is so improbable that you can forget about it in practice. Is this beginning to sound familiar? IC biochemical systems are what biochemist Behe has decided evolution cannot produce. According to him, they literally can't evolve directly and their indirect evolution is too improbable.

Creationists often assert that intermediate forms along the way to things they say could not evolve just would not work, and make fun of intermediates (as described by themselves). As we have seen, Behe also does this. In support of this view, he proposes what he calls minimal function as "... another difficulty for Darwin" (page 45). This is explained by imagining being stranded in the middle of a lake in a small boat powered by a propeller that only turns at one revolution per hour. The implication is that a flagellum must have been a quite capable swimming system the first time out, or it couldn't evolve. This idea is another consequence of dismissing change of function, or indirect evolution as he calls it, out of hand. As we already know, parts that stick out, including flagella as the finely adapted swimming organs that they now are, can have other functions. The projection that became the flagellum as we know it may not have have started as a swimming system at all, and is very unlikely to have had that function alone.

Precisely pinpointing a barrier that evolution can't cross is the Holy Grail of creationism. Behe claims to have done it. Naturally creationists are enthusiastic, nor is it surprising that many observers see IC and ID as simply a new version of creationism. Still, leading proponents try to distance themselves from the term. The term 'neocreationism' is a good compromise. It acknowledges new developments and important continuities alike.

Why are biologists never convinced that the barriers claimed by creationists are real? It always comes down to the same things: Given a population with inherited variation and also new variations from mutation or immigration, evolution occurs. Natural selection (instead of only random drift) occurs if some heritable variations are related to reproductive success. This process takes no notice of whether the changes that occur are direct or not, or whether something is becoming IC. Likewise, evolution just doesn't notice the other barriers proposed by creationists.

There is one difference that may be of interest to school boards. In the past, creationists have, naturally enough, formed creationist organizations such as the venerable Institute for Creation Research and the net-based Answers in Genesis. The leaders of the ID movement on the other hand are all high ranking members of a political organization calling itself the Discovery Institute (46).


Irreducible complexity, intelligent design's closest brush with biology, is marked by three ironies.

  • IC is supposed to be important because it cannot evolve. But it can evolve, in the same ways that anything else does.
  • Not one of the impressively complex biochemical systems said to be IC by IC/ID proponents has been shown to be in fact IC and several are known not to be. The known cases of IC are simpler and their evolution is understood.
  • Although the subject is religiously motivated, proponents have focused on bacterial flagella as the last hope for a highly complex IC system. This has the unintended consequence of making The Designer (aka God) responsible for serious diseases.

It is easy to see why scientists are not impressed by the claim that IC cannot evolve. IC is a matter of an observer specifying a combination of function, parts and system so that the specified function requires all the parts. There is no way for evolution to be sensitive to this, no way for it to matter at all. Nor does nature care about 'direct' vs 'indirect' evolution as perceived by us. Indirect evolution is as normal as tails on cows. Evolution merely requires populations with heritable variation. The processes of mutation, natural selection and random drift are not sensitive to whether a change will be deemed direct or not, nor whether a function, system and parts as specified by some observer are changing to meet the 'all parts required' condition.

There was supposed to be a special reason why it was impossible or at least very difficult for evolution to arrive at an 'all parts required' situation, but there is no such reason. The proposed reason was based on overlooking standard evolutionary processes and making analogies to manufactured items. Comparing Behe's mousetrap to Venus' flytrap confirms the reasonable suspicion that analogies and arguments based on manufactured items lead to underestimating nature. Since IC can occur in the ordinary course of events we have a known process, evolution, which is acting in the present and which given time is sufficient to produce the adaptations that Behe finds perplexing. This is like the raising of the Rocky Mountains; a known process acting in the present is sufficient, given time, to produce the result. Of course there is no way to predict all the details in either case, nor is it necessary.

Finally, this version of 'gap theology', basing the Designer on gaps or purported gaps in our knowledge (which is not mainstream religion), ends up implicating the Designer in human disease. This makes ID rather questionable as a public school lesson. Gap theology is bad enough at best, and always has the problem that the gaps keep getting smaller. This new version of it is especially bad. Darwin did theologians a favor by freeing them from this sort of thing.

Despite all this, there is a strong political drive to force public schools to misrepresent neocreationism as science. But misrepresentation is not acceptable. And it would be awkward to tell teachers to teach ID science when there isn't any. If it becomes politically necessary to teach something about the subject, the present essay contains material for several lessons. And if the plan is to teach 'the controversy', it would be proper to tell the students that there is no scientific controversy, although there is a public one. Books like Darwin's Black Box: The Biochemical Challenge to Evolution are surely part of the reason. Yet the widespread public acceptance of Behe's thesis is stark evidence that we need stronger science education, especially about evolution.


I thank Victor Eijkhout, Matt Inlay, Ian Musgrave and especially Nick Matzke, along with the talkdesign crew for helpful discussions. Special thanks to Barry Rice, author of for permission to use photos of Venus' flytrap and sundews.


1. Behe MJ. Darwin's Black Box: The Biochemical Challenge to Evolution (New York: The Free Press, 1996).

2. Darwin, Charles. Insectivorous Plants (New York: D. Appleton & Co., 1875. [first published London: John Murray, 1875]) Insectivorous Plants online.

3. Honda M. (2001). Insectivorous Plants in the Wilderness. Makoto Honda's carnivorous plants web site.

4. Cameron KM, Wurdack KJ, Jobson RW. (2002). Molecular evidence for the common origin of snap-traps among carnivorous plants. American Journal of Botany 89:1503-1509.

5. Copley SD. (2000). Evolution of a metabolic pathway for degradation of a toxic xenobiotic: the patchwork approach. Trends in Biochemical Sciences, 25(6):261-265.

6. Ackers GK, Doyle ML, Myers D, Daugherty MA. (1992). Molecular Code for Cooperativity in Hemoglobin. Science 255:54-63.

7. Hines W. (2002). Hemoglobin and Irreducible Complexity. post.

8. Hardison R. (1999). The Evolution of Hemoglobin. American Scientist 87(2) 126-137.
        - (1998). Hemoglobin from bacteria to man: Evolution of different patterns of gene expression. Journal of Experimental Biology 201:1099-1117.

9. Behe M. (2000). In Defense of the Irreducibility of the Blood Clotting Cascade: Response to Russell Doolittle, Ken Miller and Keith Robison
Doolittle's article is one of several responses to Orr's review of Darwin's Black Box, published in the December 1996/ January 1997 issue of Boston Review. The responses including Behe's were published in the February/March 1997 issue of Boston Review and are online here.

10. Miller K. Finding Darwin's God: A Scientist's Search for Common Ground Between God and Evolution
Cliff Street Books, 1999.

11. Miller K. (2000). The Evolution of Vertebrate Blood Clotting.

12. Inlay M. (2002). Evolving Immunity: A Response to Chapter 6 of Darwin's Black Box.

13. Krem MM, Di Cera E. (2002). Evolution of enzyme cascades from embryonic development to blood coagulation. Trends in Biochemical Sciences 27(2):67-74.

14. Acton G. (1997). Behe and the Blood Clotting Cascade. Talk Origins Post of the Month: February 1997.

15. Semba U, Shibuya Y, Okabe H, Yamamoto T. (1998). Whale Hageman Factor (Factor XII); Prevented Production Due to Pseudogene Conversion. Thrombosis Research 90(1):31-37

16. Miller questions Behe at the Question and Answer session at the American Museum of Natural History meeting, April 23, 2002.

17. Samuel AD, Petersen JD, Reese TS. (2001). Envelope structure of Synechococcus sp.WH8113, a nonflagellated swimming cyanobacterium. BMC Microbiol 1(1):4, online here.

18. Trachtenberg S, Gilad R, Geffen N. (2003). The bacterial linear motor of Spiroplasma melliferum BC3: from single molecules to swimming cells. Molecular Microbiology 47(3):671-697.

19. Bardy SL, Ng SYM, Jarrell KF. (2003). Prokaryotic motility structures. Microbiology 149:295-304.

20. Molecular motors: References for a course on molecular motors at Berkeley.

21. Brownian ratchets.

22. Raphidiophrys pallida. Note the axopodia in the second micrograph.

23. Pond Scum Action Video. An Actinophrys explores its world with gently waving axopodia.

24. Foraminifera.

25. Cavalier-Smith T. (2002). The phagotrophic origin of eukaryotes and phylogenetic classification of Protozoa. Int. J. Syst. Evol. Microbiol. 52:297-354.

26. Stechmann A and Cavalier-Smith T. (2002). Rooting the Eukaryote Tree by Using a Derived Gene Fusion. Science 297:89-91.

27. Leadbeater B and Kelly M. (2001). Evolution of animals - choanoflagellates and sponges. NIWA Water & Atmosphere 9(2) Web article.

28. Thomas NA, Bardy SL, Jarrell KF. (2001). The archaeal flagellum: a different kind of prokaryotic motility structure. FEMS Microbiology Reviews 25:147-174.

29. Wolgemuth W, Hoiczyk E, Kaiser D, Oster G. (2002). How Myxobacteria Glide. Current Biology 12:369–377.

30. Bacterial Flagella: A Paradigm for Design. ARN advertisement for the Flagella video.

31. Discussion of the Flagella video in the news group.

32. Behe MJ. (2001). Reply to my critics: A Response to Reviews of Darwin's Black Box: The Biochemical Challenge to Evolution. Biology and Philosophy 16(6):685-709. Also online.

33. Musgrave I. (2000). Evolution of Bacterial Flagella.

34. Klose KE and Mekalanos JJ. (1998). Differential regulation of multiple flagellins in Vibrio cholerae. Journal of Bacteriology 180(2):303-16.

35. Young GM, Schmiel DH, Miller VL. (1999). A new pathway for the secretion of virulence factors by bacteria: The flagellar export apparatus functions as a protein-secretion system. Proceedings of the National Academy of Sciences of the USA 96:6456-6461.

36. Thar R and Kühl M. (2002). Conspicuous Veils Formed by Vibrioid Bacteria on Sulfidic Marine Sediment. Applied and Environmental Microbiology 68(12):6310-20.

37. Kirov SM, Tassell BC, Semmler AB, O'Donovan LA, Rabaan AA, Shaw JG. (2002). Lateral Flagella and Swarming Motility in Aeromonas Species. Journal of Bacteriology 184(2):547-555.

38. Girón JA, Torres AG, Freer E, Kaper JB. (2002). The flagella of enteropathogenic Escherichia coli mediate adherence to epithelial cells. Molecular Microbiology 44(2):361-379.

39. McGee DJ, Coker C, Testerman TL, Harro JM, Gibson SV, Mobley HL. (2002). The Helicobacter pylori flbA flagellar biosynthesis and regulatory gene is required for motility and virulence and modulates urease of H. pylori and Proteus mirabilis. Journal of Medical Microbiology 51(11):958-970.

40. Foultier B, Troisfontaines P, Muller S, Opperdoes FR, Cornelis GR. (2002). Characterization of the ysa pathogenicity locus in the chromosome of Yersinia enterocolitica and phylogeny analysis of type III secretion systems. Journal of Molecular Evolution 55(1):37-51.

41. Young BM and Young GM. (2002). YplA Is Exported by the Ysc, Ysa, and Flagellar Type III Secretion Systems of Yersinia enterocolitica. Journal of Bacteriology, 184(5):1324-1334.

42. Cornelis GR, Boland A, Boyd AP, Geuijen C, Iriarte M, Neyt C, Sory MP, Stainier I. (1998). The virulence plasmid of Yersinia, an antihost genome. Microbiology and Molecular Biology Reviews 62(4):1315-52.

43. Stuber K, Frey J, Burnens AP, Kuhnert P. (2003). Detection of type III secretion genes as a general indicator of bacterial virulence. Molecular and Cellular Probes 17:25–32.

44. Ussery D. (1999). Review of Darwin's Black Box. Bios 70:40-45. There is an expanded version online.
Cavalier-Smith, T. (1997). The Blind Biochemist. Trends in Ecology and Evolution 12(4):162-163.
Roger Dorit (1997). Molecular Evolution and Scientific Inquiry, Misperceived. American Scientist 85(5):474-475. Also online.

45. Reviews of Darwin's Black Box at

46. Links to The Institute for Creation Research, Answers in Genesis and like minded groups are here.
The leaders of the intelligent design movement are all Senior Fellows or higher at the Discovery Institute.

Photos of Venus' flytrap and sundew © Barry Rice. All rights reserved.