Is Every Trait There for a Reason? A Church Roof Says Otherwise

Setting a round dome on four arches leaves four curved triangular surfaces between the top of the arches and the base of the dome, and those surfaces are an unavoidable consequence of the geometry. A builder choosing arches and a dome gets them whether or not any use is wanted. In San Marco in Venice they are decorated elaborately, each carrying a mosaic of an evangelist fitted precisely to its shape, and the fit is so good that a visitor could reasonably conclude the surfaces were designed to hold those figures. The conclusion would be exactly backwards, since the decoration was adapted to the space and the space came from the structure.

Stephen Jay Gould and Richard Lewontin published the comparison in 1979:

• •Biologists routinely asked what a trait was for and found an answer
• •An answer can always be found, which makes the method unfalsifiable
• •Some traits are by-products of building something else
• •Others are consequences of physics, development or shared ancestry
• •Explaining everything by selection therefore explains too much
• •The paper named the habit after Kipling's tales of how things came to be

The paper became one of the most cited in evolutionary biology and provoked forty years of argument, some of it heated. Critics pointed out that its central complaint was partly a caricature, since serious workers already tested adaptive claims against alternatives and did not simply invent stories. Some argued the architectural example itself is wrong, since the triangular surfaces are not the only way to make the transition and were therefore chosen rather than forced. Defenders replied that the practice being criticised was genuinely widespread, particularly in popular accounts and in claims about human behaviour, where an adaptive explanation for any trait is still produced with very little evidence. The dispute sharpened standards for what counts as demonstrating that something is an adaptation.

The dispute produced a clearer account of what evidence would settle whether a trait is an adaptation, which is the useful legacy of it. Comparing many related species and asking whether the trait appears where the supposed advantage applies is one line, since a trait that is genuinely for something should track the conditions that make it useful. Manipulating the trait experimentally and measuring survival or reproduction is another, and is the strongest where it can be done. Reconstructing when the trait appeared relative to the supposed function, using fossils and family trees, addresses the origin directly. And measuring whether variation in the trait actually affects reproductive success within a population tests whether selection is acting on it now.

The same authors and others developed a companion concept that is now standard, covering a feature that arose for one reason and was later put to a different use. Feathers appear to have evolved for insulation or display long before any animal flew, and were later recruited for flight. Bones in the reptile jaw became bones in the mammalian middle ear. The lungs of land vertebrates derive from a structure used for buoyancy in fish. In each case asking what the feature is for gives a correct answer about the present and a completely wrong answer about the origin, which is precisely the confusion the 1979 paper was attacking, and separating the two questions is now expected.

Curved triangles appear above the arches of a domed church because arches and domes were built, and decorating them beautifully does not mean they were made for decoration. Gould and Lewontin used that in 1979 to argue that biologists too readily found a purpose for every trait, since some are by-products of building something else. The dispute that followed sharpened what counts as evidence that a feature is an adaptation.