Are some molecular machines irreducibly complex?

In Darwin’s Black Box (1996), the biochemist Michael Behe defined an irreducibly complex system as “a single system which is 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”^{[source 1][source 2]}. His best-known example is the bacterial flagellum, a rotary motor bacteria use to swim; others include the blood-clotting cascade and parts of the immune system^{[source 2]}.

Behe’s argument takes up a test Darwin himself proposed: if any complex organ could be shown that “could not possibly have been formed by numerous, successive, slight modifications, my theory would absolutely break down”^{[source 4]}. Natural selection can only preserve what already works, Behe argues, so a precursor missing a part would give selection nothing to favor; an irreducibly complex system therefore cannot be produced directly by slight, successive modifications^{[source 2]}. He concludes that such systems are better explained by design^{[source 6]}.

Darwin himself anticipated that transitional forms would often be missing: “No doubt many organs exist of which we do not know the transitional grades,” he wrote, especially where related species had gone extinct^{[source 4]}. Behe’s point is different. He argues that for some molecular systems the problem is not missing evidence of intermediates but that no functional intermediate could exist on a direct route, which turns Darwin’s own test against the theory^{[source 1][source 6]}.

Behe pointed to the flagellum because it is a true rotary motor, with a rotor, a stator, a drive shaft and a propeller made of protein, and because removing many of its core proteins leaves bacteria unable to swim, though flagella vary between species^{[source 2][source 3]}. Behe did not claim that indirect routes are impossible, only that they become too improbable to credit as systems grow more complex^{[source 6]}. The Kitzmiller court described his 1996 claim that science would never find an evolutionary explanation for the immune system^{[source 2]}.

Biologists’ main answer is that evolution need not build a system by adding parts toward its present function. Parts can be borrowed from systems that did something else, a process called co-option or exaptation, and parts that once helped can later be lost. Behe’s definition, critics say, rules out only the direct route^{[source 3][source 5]}.

For the flagellum, Mark Pallen and Nicholas Matzke reported that many of its proteins share ancestry with proteins in other systems: its export machinery with the type III secretion system some bacteria use to inject proteins into host cells, its motor proteins with other membrane transport systems, and its ATPase with the F- and V-type ATPases found across life^{[source 3]}. The secretion system shows that a subset of flagellar proteins can do a useful job on its own, though later work suggests it evolved from the flagellum rather than the reverse^{[source 3][source 8]}. At the Kitzmiller trial, the biologist Kenneth Miller presented peer-reviewed studies on possible precursors of the flagellum and evolutionary explanations for the clotting cascade and the immune system, and the court accepted them^{[source 2][source 5]}.

Researchers have also watched a machine of this kind gain a necessary part. In 2012 Joseph Thornton’s group reconstructed ancestral versions of a ring in the cell’s proton pump and showed that a two-protein ring became a three-protein ring, each part essential, through gene duplication followed by complementary losses of function: ordinary, high-probability mutations^{[source 7]}.

The immune system became a turning point at the Kitzmiller trial. On cross-examination Behe was presented with fifty-eight peer-reviewed publications, nine books and several immunology textbook chapters on the evolution of the immune system, and said they were not “good enough” as evidence. The court concluded that the ID argument depends on setting a scientifically unreasonable burden of proof for evolution^{[source 2]}. Behe has long maintained that studies of this kind describe what might have happened without showing, step by step, that it did^{[source 6]}.

Pallen and Matzke argue that the idea Behe’s argument leaves out is exaptation: “the function currently performed by a biological system is different from the function performed while the adaptation evolved”^{[source 3]}. On this view, asking what use half a flagellum would be as a swimming motor asks the wrong question, because its parts may have served export, adhesion or other jobs first.

Proponents reply that shared ancestry is not a pathway: showing that parts resemble parts elsewhere does not show how the whole was assembled, or that each intermediate stage worked and could be selected^{[source 6]}.

The Kitzmiller court observed that, unlike ID as a whole, irreducible complexity is testable, since it is refuted by showing intermediate structures with selectable functions^{[source 2]}. Most biologists regard it as answered in principle for the flagellum and the clotting cascade.

Both sides have proposed experiments. Behe and the microbiologist Scott Minnich suggested that ID could be tested by trying to evolve a flagellum in the laboratory; the Kitzmiller court noted that no one, including those who proposed it, had carried out that test, that Behe conceded it could not approximate real-world conditions, and that Minnich admitted it “would merely be a test of evolution, not design”^{[source 2]}. Biologists, for their part, have pursued reconstructions of ancestral proteins like Thornton’s, which test specific evolutionary steps directly^{[source 7]}.

Detailed step-by-step histories remain incomplete; Pallen and Matzke wrote that the flagellar research community “has scarcely begun to consider how these systems have evolved”^{[source 3]}. Behe maintains that the proposed pathways are speculative^{[source 6]}.