Evolution Icon Evolution
Intelligent Design Icon Intelligent Design

From Jerry Coyne, “Evolution-of-the-Gaps” and Other Fallacies

Continuing my review of Jerry Coyne’s Why Evolution Is True. (See my previous posts here and here.)

Chapter 3: Remnants — Vestiges, Embryos, and Bad Design

In Chapter 3 of his book, Coyne turns his attention to the argument from suboptimal design. Whereas the “god-of-the-gaps” fallacy states that “evolution can’t explain this; therefore god must have done it,” Coyne’s argument in this chapter commits a converse “evolution-of-the-gaps” fallacy: “God wouldn’t have done it that way; therefore evolution must have done it.” It is curious that this dichotomous mode of thinking is precisely what Darwinists like Coyne so often like to accuse ID proponents of. Much like “god-of-the-gaps” arguments, the “evolution-of-the-gaps” argument has to retreat with advances in scientific knowledge, as biologists uncover important reasons for the way these features have been designed. Furthermore, a common critique of ID is that it is unscientific because it isn’t falsifiable. But, in this chapter, Coyne purports to have falsified the design hypothesis. One cannot have it both ways — either ID is falsifiable or it isn’t. By offering a critique of design as a hypothesis, Coyne implies that ID is falsifiable, and therefore that it is scientific by that criterion.


The first example of a vestigial trait cited here are the wings of ostriches and other flightless birds such as ratites. Few would doubt that an ostrich’s ancestors were flying birds. This isn’t a very strong piece of evidence, however, if Coyne wants to convince skeptics of the efficacy of modern evolutionary theory — vestigial structures of this kind document, at best, the loss of function. What needs to be demonstrated is that modern evolutionary theory is adequate to account for the gain of new functions. This is true also for the second example Coyne raises: vestigial eyes in cave-dwelling animals such as fish.

Coyne’s next example is vestigial hind limb and pelvic bones in whales. I am very open to the possibility that whales may have possessed hind limbs at some point in their evolutionary history. But, again, this demonstrates, at best, the loss of traits, not their origination.

Coyne’s next exhibit is the human appendix. This is another case, however, of a loss of function, not a gain. Moreover, the appendix is not without important function in humans, as Coyne himself notes. The appendix is thought to be involved in the immune system, and is believed to play a role in B lymphocyte maturation and in the production of immunoglobulin A (IgA) antibodies.

Coyne also makes mention of the vestigial tail bone: the coccyx. The coccyx, however, is functional inasmuch as it provides essential anchorage points for muscles.

The arrector pili which make our hair stand on end, also referenced by Coyne, is another example of a loss-of-function (our ancestors were hairier than we are). In any case, even in humans, goosebumps act to conserve heat thereby keeping us warm.

Coyne’s final example is the muscles for wiggling our ears. This example represents yet another loss, rather than a gain, of function. Perhaps our ancestors used those muscles to move their ears and improve their chances of detecting predators. But that doesn’t necessarily entail that we share a lineage with other extant primate species.

Coyne anticipates the typical response to the argument from vestigiality:

Opponents of evolution always raise the same argument when vestigial traits are cited as evidence for evolution. “The features are not useless,” they say. “They are either useful for something, or we haven’t yet discovered what they’re for.” They claim, in other words, that a trait can’t be vestigial if it still has a function, or a function yet to be found.
But this rejoinder misses the point. Evolutionary theory doesn’t say that vestigial characters have no function. A trait can be vestigial and functional at the same time. It is vestigial not because it’s functionless, but because it no longer performs the function for which it evolved. (p. 58)

But surely, by Coyne’s reckoning, this loose definition of “vestigiality” would entail that every organ and structure is vestigial, since, in Coyne’s view, all traits have evolved from something else. As Jonathan Wells explains in his own review of the book,

[I]f the human arm evolved from the leg of a four-footed mammal (as Darwinists claim), then the human arm is vestigial. And if (as Coyne argues) the wings of flying birds evolved from feathered forelimbs of dinosaurs that used them for other purposes, then the wings of flying birds are vestigial. This is the opposite of what most people mean by “vestigial.”


Coyne’s next subject is evolutionary atavisms, which involve the embryonic appearance of some trait that is thought to have been present in an evolutionary ancestor. Although Haeckel’s theory of recapitulation is now thought to be false, at least as a ubiquitous principle, recapitulation in this more limited sense is still thought to be valid. Coyne explains,

True atavisms must recapitulate an ancestral trait, and in a fairly exact way. They aren’t simply monstrosities. A human born with an extra leg, for example, is not an atavism because none of our ancestors had five limbs. (p. 64)

In other words, when an embryonic abnormality corresponds to a trait in some evolutionary ancestor, the phenomenon may be taken as evidence of common ancestry. But when an embryonic abnormality does not correspond to a trait in some evolutionary ancestor, the phenomenon does not count as a “true atavism.” Indeed, humans are born with abnormalities all the time, including an extra rib or finger. But no one thinks that this provides evidence that our ancestors once had an extra rib or finger. This same circular reasoning is found abundantly throughout evolutionary thinking. When similarity is explicable by common ancestry, it may be classified as “homology” and thus be taken as evidence for common ancestry. But when similarity is not explicable by common ancestry, it is classified as “analogy” and is evidence for convergent evolution.

Examples of evolutionary atavisms offered by Coyne include whales “born with a rear leg that protrudes outside the body wall,” modern horses “born with extra toes,” and birds that can be induced to “produce tooth-like structures” on the bill. All of these examples, none of which I have a problem with, again illustrate the ability of organisms to shed traits, but have little relevance to the efficacy of the neo-Darwinian mechanisms to create fundamentally new structures and traits.

Dead Genes

Coyne subsequently turns his attention to pseudogenes. According to Coyne, evolution makes a prediction:

We expect to find, in the genomes of many species, silenced or “dead” genes: genes that were once useful but are no longer intact or expressed. In contrast, the idea that all species were created from scratch predicts that no such genes would exist, since there would be no common ancestors in which those genes were active. (pp. 66-67)

He continues,

And the evolutionary prediction that we’ll find pseudogenes has been fulfilled — amply. Virtually every species harbors dead genes, many of them still active in its relatives. This implies that those genes were also active in a common ancestor, and were killed off in some descendants but not in others. Out of about 30,000 genes, for example, we humans carry more than 2,000 pseudogenes. Our genome — and that of other species — are truly well populated graveyards of dead genes. (p.67)

The literature of the last decade, however, had unearthed a plethora of functions for pseudogenes. To get a sense of some of the functions that have been identified, take a look at the following papers:

Coyne’s focus in this section is on the well-known L-gulono-1,4-lactone oxidase (GULO) gene, the final enzyme in the biosynthetic pathway of ascorbic acid (vitamin C), and on the olfactory receptor (OR) genes which are involved in the detection of odor molecules.

The functioning GULO gene allows most plants and many animals to produce vitamin C from glucose or galactose. In some taxa, however, the GULO gene does not function in this capacity and is given the “pseudogene” label. The GULO gene is thought to be broken in humans (Nishikimi and Yagi, 1991), primates and guinea pigs (Nishikimi et al., 1994; Nishikimi et al., 1988), as well as in bats of the genus Pteropus (Cui et al., 2011). On the significance of the GULO pseudogene with regard to common ancestry, I refer readers to my previous article on the subject.

What about human olfactory receptor pseudogenes? Are they non-functional? Not necessarily. Indeed, Balakirev and Ayala (2003) note that,

The human olfactory receptor (OR) pseudogenes may be important for the generation and maintenance of receptor diversity. Intensive intergenic gene conversion has been revealed for this multigene family that leads to segment shuffling in the odorant binding site, and evolutionary process reminiscent of somatic combinatorial diversification in the immune system. Although OR pseudogenes have lost full coding function, they are apparently under new evolutionary constraints: OR pseudogenes adopt noncoding functions as CpG islands, enhancers, and matrix attachment regions. [internal citations omitted].

Coyne also mentions the presence of tens of thousands of endogenous retroviruses (ERVs) in our DNA, which make up some 7 to 8% of our genome. Coyne writes off these virus-like sequences as “dead genes.” The research of the last decade, however, has revealed a host of functions for these sequences. For a sample of some of these functions, take a look at the following papers:

Coyne mentions that “some of these [retroviral] remnants sit in exactly the same location on the chromosomes of humans and chimpanzees.” I am not convinced, however, that it can be ruled out that these parallel inserts may reflect preferential integration of ERV sequences. Indeed, such target site preferences have been identified for several classes of mobile DNA (Spradling et al., 2011; Levy et al., 2009; Li et al., 2009; Zou and Voytas, 1997; Wang et al., 2006).

Palimpsests in Embryos

Coyne writes,

Well before the time of Darwin, biologists were busy studying both embryology (how an animal develops) and comparative anatomy (the similarities and differences in the structure of different animals). Their work turned up many peculiarities that, at the time, didn’t make sense. For example, all vertebrates begin development in the same way, looking rather like an embryonic fish. As development proceeds, different species begin to diverge — but in weird ways. Some blood vessels, nerves, and organs that were present in the embryos of all species at the start suddenly disappear, while others go through strange contortions and migrations. Eventually, the dance of development culminates in the very different adult forms of fish, reptiles, birds, amphibians, and mammals. Nevertheless, when development begins they look very much alike. (p. 73)

Vertebrate embryos, contrary to Coyne’s assertion, actually differ substantially at the early stages. They converge somewhat mid-way through development (at the phylotypic or “pharyngula” stage) before diverging again in later development. As embryologist Brian Hall writes in this article, “Despite repeated assertions of the uniformity of early embryos within members of a phylum, development before the phylotypic stage is very varied.” Indeed, even patterns of gene expression exhibit this pattern (Kalinka et al., 2010).

The divergence of vertebrate embryonic development at the earliest stages is, in fact, a fundamental challenge to the conventional evolutionary narrative. Embryonic screening studies reveal that modifications to early development are not tolerated by embryos — the consequences are frequently lethal (Johnston, 2002; Driever et al., 1996; Haffter et al., 1996; Kodoyianni et al., 1992).

Moreover, mutations affecting germ cell development are likely to result in infertility, an evolutionary dead end (Pellas et al., 1991). Yet embryos often differ significantly with respect to primordial germ cell (PGC) formation. As Johnson et al. (2003) explain,

Surprisingly, among the major extant amphibian lineages, one mechanism [of PGC formation] is found in urodeles [frogs and toads] and the other in anurans [salamanders and newts]. In anuran amphibians PGCs are predetermined by germ plasm; in urodele amphibians PGCs are formed by inducing signals.

They further note,

In anuran embryos primordial germ cells (PGCs), the cells that give rise to gametes, are of endodermal origin, and they are specified by the differential distribution of maternally deposited germ cell determinants (known as germ plasm) to the presumptive germ line blastomeres. Thus, from the inception of development, anuran PGCs are considered to be predetermined by germ plasm. In urodele embryos PGCs derive from lateral plate mesoderm. Urodele embryos do not contain germ plasm, and so PGCs are specified later in development than in anurans. In urodeles PGCs form in response to extracellular inducing signals, not unlike those that produce other mesodermal cell types. This is considered regulative germ cell specification.

Now, here’s the conundrum. The difference relates to organs of extreme importance — i.e., the germ cells. The difference is not only substantial, but it occurs extremely early in development. From the standpoint of evolutionary rationale, there seems to be two possible explanations for this:

  1. A very radical change has occurred in the relevant developmental mechanisms.
  2. An early bifurcation has occurred in the phylogeny of amphibians — a remarkable case of convergent evolution, perhaps from different groups of fish.

The second hypothesis seems to be implausible, for it would be a remarkable coincidence for two groups to have evolved independently for such a length of time and yet share so many distinctive features. The first hypothesis seems to be even more implausible still since subsequent development derives from those early processes. As such, modifying these early embryonic stages is likely to cause catastrophic harm to the organism.

Bad Design

The suboptimal design argument is fraught with problems. For one thing, as stated previously, many cases of once-thought-to-be “bad design,” upon closer inspection, turn out to exhibit design trade-offs. This makes it an argument from ignorance, which has to retreat with each advance in our understanding of these systems. The logic is also precarious. Do the design flaws of the Windows operating system justify the claim that it was not engineered by intelligent agents? Clearly not. The ability to detect the products of intelligence does not require that designs be perfect.
Coyne considers “one of nature’s worst designs” to be the recurrent laryngeal nerve (RLN), which loops around the aorta before travelling back up to the larynx in the neck. In giraffes, this detour ends up being 15 feet longer than the direct route. Coyne explains this apparent detour in terms of our evolution from fish ancestors. But could there be good reasons for the apparent detour of the RLN? ENV’s Casey Luskin offers us three potential design reasons for the chosen route:

(1) There is evidence that supposed fundamental evolutionary constraints which would prevent loss of the circuitous route of the RLN do not exist. This implies that there is some beneficial function for the circuitous route.
(2) The path of the RLN allows it to give off filaments to the heart, to the mucous membranes and to the muscles of the trachea along the way to the larynx.
(3) There is dual-innervation of the larynx from the SLN and RLN, and in fact the SLN innervates the larynx directly from the brain. The direct innervation of the larynx via the superior laryngeal SLN shows the laryngeal innervations in fact follows the very design demanded by ID critics like Jerry Coyne and Richard Dawkins. Various medical conditions encountered when either the SLN or RLN are damaged point to special functions for each nerve, indicating that the RLN has a specific laryngeal function when everything is functioning properly. This segregation may be necessary to achieve this function, and the redundancy seems to preserve some level of functionality if one nerve gets damaged. This dual-innervation seems like rational design principle.

Luskin quotes Michael Egnor, who lists some other interesting advantages of the route taken by the RLN:

There is actually a design advantage to the course of the recurrent nerves, if one wishes to pursue this line of argumentation. The course of the nerves brings them through the mediastinum, where the heart and lungs meet. There are many lymph nodes there, and enlargement of these lymph nodes from processes such as cancer or infection (e.g., tuberculosis) often irritates these nerves and causes hoarseness or coughing. The course of the nerves reveals disease in an otherwise hidden part of the body (deep in the chest) by interfering with a process (speech) that is readily evident. It serves as an early warning to get medical care (or, with infectious diseases, as a warning to others that this person is ill), and this early warning has saved many more lives than the redundant course of the nerves has cost lives. The risk/benefit ratio needs to be examined comprehensively before one claims that the course of the nerves is biologically disadvantageous.

Egnor further notes,

Of course ID advocates have never claimed perfect design. But the argument that the anatomy of the recurrent laryngeal nerve is evidence for “bad design” fails on many levels. The descent of the recurrent nerves below the aortic arch and subclavian artery is the result of patterns of coalescence and movements of components of the aortic arch during embryogenesis. It appears that proximity of various layers and structures in the embryo serve to guide embryogenesis (it’s called induction). The details of this process are only beginning to be understood, and the Darwinist argument that the relationship between the recurrent nerves and the aortic arch is evidence of bad design fails to take into account the enormous complexities of embryonic development. It’s analogous to a 3 year old taking apart a computer and asserting that it was designed badly because some of the circuit board patterns were “curvy” instead of straight. The design wisdom of the anatomy of the recurrent nerves can only be judged by someone who knows all of the design specifications necessary for that region of the human body. Even the best embryologists are preschoolers when it comes to that.

In a subsequent post I will turn to Coyne’s Chapter 4.

Jonathan McLatchie

Resident Biologist & Fellow, Center for Science and Culture
Dr. Jonathan McLatchie holds a Bachelor's degree in Forensic Biology from the University of Strathclyde, a Masters (M.Res) degree in Evolutionary Biology from the University of Glasgow, a second Master's degree in Medical and Molecular Bioscience from Newcastle University, and a PhD in Evolutionary Biology from Newcastle University. Previously, Jonathan was an assistant professor of biology at Sattler College in Boston, Massachusetts. Jonathan has been interviewed on podcasts and radio shows including "Unbelievable?" on Premier Christian Radio, and many others. Jonathan has spoken internationally in Europe, North America, South Africa and Asia promoting the evidence of design in nature.