Darwin's Doubt

by Stephen C. Meyer

  This chapter will not only examine the ideas of evolutionary developmental biologists, but three of the other most prominent alternatives to neo-Darwinism, some proposed by members of the Altenberg 16 (see Fig. 16.1). Each of these alternatives emphasizes certain elements of the “triad” at the expense of others. Whereas the self-organizational alternatives that I discussed in the last chapter emphasize the role of law-like processes over random mutations, these other new theories reaffirm the importance of mutations, though each also reconceptualizes how mutations act. One approach falls under the rubric of “evo-devo” and conceives of mutations producing modifications in larger increments. Another, the neutral theory of evolution, sees mutations acting absent selection. Another, neo-Lamarckian “epigenetic inheritance,” envisions heritable alterations in epigenetic information influencing the future course of evolution. Still another, called “natural genetic engineering,” affirms that nonrandom genetic rearrangements drive evolutionary innovation.7

  FIGURE 16.1

  The tenets of neo-Darwinian orthodoxy and the different ways that various non-Darwinian models of evolution deviate from those tenets. The boxes representing new evolutionary models are positioned under the headings of the neo-Darwinian tenets that they challenge.

  Let’s see if one of these proposals solves the twin problems of the origin of form and information and whether, therefore, it might also help to resolve the mystery of the Cambrian explosion.8

  Evo-Devo and Its Proposals

  The neo-Darwinian synthesis has long emphasized that large-scale macroevolutionary change occurs as the inevitable byproduct of the accumulation of small-scale “microevolutionary” changes within populations. The consensus in support of this idea began to fray in evolutionary biology during the early 1970s, when young paleontologists such as Gould, Niles Eldredge, and Steven Stanley realized that the fossil record did not show a pattern of gradual “micro-to-macro” change. In 1980, at a now famous symposium on macroevolution at the Field Museum in Chicago, the rebellion burst into full view, exposing what developmental biologist Scott Gilbert called “an underground current in evolutionary theory” among theorists who had concluded that “macroevolution could not be derived from microevolution.”9

  At the conference, paleontologists who doubted the “micro-to-macro” consensus found allies among younger developmental biologists. They were dissatisfied with neo-Darwinism in part because they knew that population genetics, its mathematical expression, sought only to quantify changes in gene frequency rather than explain the origin of genes or novel body plans. Thus, many developmental biologists thought that neo-Darwinism did not offer a compelling theory of macroevolution.10

  To formulate a more robust theory, many developmental biologists, such as Rudolf Raff, a developmental biologist at the University of Indiana and one of the founders of “evo-devo,” urged evolutionary theorists to incorporate insights from their discipline.11 For example, developmental biologists know that mutations expressed early in the development of animals are necessary to alter bodyplan morphogenesis. Thus, they argue that these mutations must have played a significant role in generating whole new animal forms during the history of life. They assert that this understanding of developmental processes is crucial to understanding animal evolution. Some evo-devo advocates such as Sean B. Carroll and Jeffrey Schwartz have pointed specifically to homeotic (or Hox) genes—master regulatory genes that affect the location, timing, and expression of other genes—as entities capable of producing such large-scale change in animal form.12 These evo-devo advocates have broken with classical neo-Darwinism primarily in their understanding of the size or increment of mutational change.

  Major but Not Viable, Viable but Not Major

  Despite the enthusiasm surrounding the field, evo-devo fails, and for an obvious reason: its main proposal, that early-acting developmental mutations can cause stably heritable, large-scale changes in animal body plans, contradicts the results of one hundred years of mutagenesis experiments.13 As we saw in Chapter 13, the experiments of scientists such as Nüsslein-Volhard and Wieschaus have shown definitively that early-acting bodyplan mutations invariably generate embryonic lethals—dead animals incapable of further evolution. The results of these experiments have generated the dilemma for evolutionary biologists that geneticist John McDonald aptly described as the “great Darwinian paradox.” Recall that McDonald noted that early-acting regulatory mutations do not produce viable alterations in form that will persist in populations, as evolution absolutely requires. Instead, these mutations are eliminated immediately by natural selection because of their invariably destructive consequences. On the other hand, later-acting mutations can generate viable changes in the features of animals, but these changes do not affect global animal architectures. This generates a dilemma: major changes are not viable; viable changes are not major. In neither case do the kinds of mutation that actually occur produce viable major changes of the kind necessary to build new body plans.

  In 2007, I coauthored a textbook with several colleagues titled Explore Evolution. In it, we explained this “either/or” (“major-not-viable, viable-not-major”) dilemma and suggested that it posed a challenge to theories that rely on the mutation and selection mechanism to explain the origin of major morphological changes.14 The National Center for Science Education (NCSE)—an influential activist group that opposes allowing students to learn about scientific criticisms of evolutionary theory—challenged our critique. They charged that our textbook “fails to acknowledge the extensive research on mutations in DNA sequences that do not encode proteins, but which have important morphological effects.”15 In other words, they claimed that some viable mutations do produce major large-scale changes.

  FIGURE 16.2

  Superficial changes in insect wing coloration thought to be caused by mutations in cis-regulatory elements. Such examples show that mutations that affect development and that also result in viable offspring tend to be minor. Courtesy National Academy of Sciences, U.S.A.

  The NCSE cited papers from the “evo-devo” literature claiming that a type of mutation in the regulatory regions of the genome, “cis-regulatory” regions, have been shown to produce large-scale changes in winged insects. According to the NCSE, mutations in these cis-regulatory elements (or CREs) are “considered by many evolutionary biologists to have the greatest potential for generating evolutionary change.”16 What’s more, they insisted that “mutations in CREs play an important role in morphological evolution.”17 The NCSE cited a paper in the Proceedings of the National Academy of Sciences by three developmental biologists, Benjamin Prud’homme, Nicolas Gompel, and Sean B. Carroll.18

  The paper did not show what the NCSE claimed, however. It did assert that changes in regulatory DNA produce “both relatively modest morphological differences among closely related species and more profound anatomical divergences among groups at higher taxonomical levels.”19 But the study only showed how changes in the cis-regulatory elements in fruit fly DNA might have affected the coloration of wing spots in several different types of flying insects. It did not report any significant change in the form or body plan of these insects. Instead, the study highlighted a clear case of a viable mutation generating merely a minor or superficial change (see Fig. 16.2).

  Not surprisingly, many evolutionary biologists recognize that such regulatory mutations do not explain the evolution of new body plans. For example, Hopi Hoekstra, of Harvard University, and Jerry Coyne, two traditional neo-Darwinists, have published an article reviewing various evo-devo proposals in the journal Evolution. They note, “Genomic studies lend little support to the cis-regulatory theory” of evolutionary change.

  They also argue, tellingly, that most cis-regulatory mutations result in the loss of genetic and anatomical traits, including a famous case in which evolutionary biologists attributed the loss of pelvic spines in stickleback fish to mutations in cis-regulatory elements.20 Yet, as they argue, “supporting the evo-devo claim that cis-regulato
ry changes are responsible for morphological innovations requires showing that promoters are important in the evolution of new traits, not just the losses of old ones.” Hoekstra and Coyne conclude, “There is no evidence at present that cis-regulatory changes play a major role—much less a preeminent one—in adaptive evolution.”21 Given their commitment to neo-Darwinism, it’s fair to assume that Hoekstra and Coyne probably did not intend, in making this argument, to refute the NCSE’s criticism of our textbook Explore Evolution. Nevertheless, science, like politics, sometimes makes for strange bedfellows.

  What About Hox Genes?

  When biology students hear my colleague Paul Nelson describe the “great Darwinian paradox” (see Chapter 13) in public lectures on university campuses, they often ask, “What about Hox genes?” Recall that Hox (or homeotic) genes regulate the expression of other protein-coding genes during the process of animal development. Some biologists have likened them to the conductor of an orchestra who plays the role of coordinating the contributions of the players. And because Hox genes affect so many other genes, many evo-devo advocates think that mutations in these genes can generate large-scale changes in form.

  For example, Jeffrey Schwartz, at the University of Pittsburgh, invokes mutations in Hox genes to explain the sudden appearance of animal forms in the fossil record. In his book Sudden Origins, Schwartz acknowledges the discontinuities in the fossil record. As he notes, “We are still in the dark about the origin of most major groups of organisms. They appear in the fossil record as Athena did from the head of Zeus—full-blown and raring to go, in contradiction to Darwin’s depiction of evolution as resulting from the gradual accumulation of countless infinitesimally minute variations.”22

  What resolves this mystery? Schwartz, an evo-devo advocate, reveals his answer: “A mutation affecting the activity of a homeobox [Hox] gene can have a profound effect—such as turning … larval tunicates into the first chordates. Clearly, the potential homeobox genes have for enacting what we call evolutionary change would seem to be almost unfathomable.”23

  But can mutations in Hox genes transform one form of animal life—one body plan—into another? There are several reasons to doubt that they can.

  First, precisely because Hox genes coordinate the expression of so many other different genes, experimentally generated mutations in Hox genes have proven harmful. William McGinnis and Michael Kuziora, two biologists who have studied the effects of mutations on Hox genes, have observed that in fruit flies “most mutations in homeotic [Hox] genes cause fatal birth defects.”24 In other cases, the resulting Hox mutant phenotype, while viable in the short term, is nonetheless markedly less fit than the wild type. For example, by mutating a Hox gene in a fruit fly, biologists have produced the dramatic Antennapedia mutant, a hapless fly with legs growing out of its head where the antennae should be (see Fig. 16.3).25 Other Hox mutations have produced fruit flies in which the balancers (tiny structures behind wings that stabilize the insect in flight, called “halteres”) are transformed into an extra pair of wings.26 Such mutations alter the structure of the animal, but not in a beneficial or permanently heritable way. The Antennapedia mutant cannot survive in the wild; it has difficulty reproducing, and its offspring die easily. Similarly, fruitfly mutants sporting an extra set of wings lack the musculature to make use of them and, absent their balancers, cannot fly. As Hungarian evolutionary biologist Eörs Szathmáry notes with cautious understatement in the journal Nature, “macromutations of this sort [i.e., in Hox genes] are probably frequently maladaptive.”27

  FIGURE 16.3

  Photograph of an Antennapedia mutant with a pair of legs growing out of its head, where antennae would normally develop. Such examples show that mutations that occur early in animal development and that also produce major changes typically result in less fit offspring—in this case offspring that cannot reproduce. Courtesy Elsevier, Inc.

  Second, Hox genes in all animal forms are expressed after the beginning of animal development, and well after the body plan has begun to be established. In fruit flies, by the time that Hox genes are expressed, roughly 6,000 cells have already formed, and the basic geometry of the fly—its anterior, posterior, dorsal, and ventral axes—is already well established.28 So Hox genes don’t determine bodyplan formation. Eric Davidson and Douglas Erwin have pointed out that Hox gene expression, although necessary for correct regional or local differentiation within a body plan, occurs much later during embryogenesis than global bodyplan specification itself, which is regulated by entirely different genes. Thus, the primary origin of animal body plans in the Cambrian explosion is not merely a question of Hox gene action, but of the appearance of much deeper control elements—Davidson’s “developmental gene regulatory networks” (dGRNs).29 And yet, as we saw in Chapter 13, Davidson argues that it is extremely difficult to alter dGRNs without damaging their ability to regulate animal development.

  Third, Hox genes only provide information for building proteins that function as switches that turn other genes on and off. The genes that they regulate contain information for building proteins that form the parts of other structures and organs. The Hox genes themselves, however, do not contain information for building these structural parts. In other words, mutations in Hox genes do not have all the genetic information necessary to generate new tissues, organs, or body plans.

  Nevertheless, Schwartz argues that biologists can explain complex structures such as the eye just by invoking Hox mutations alone. He asserts that “[t]here are homeobox genes for eye formation and that when one of them, the Rx gene in particular, is activated in the right place and at the right time, an individual has an eye.”30 He also thinks that mutations in Hox genes help arrange organs to form body plans.

  In a review of Schwartz’s book, Eörs Szathmáry finds Schwartz’s reasoning deficient. He too notes that Hox genes don’t code for the proteins out of which body parts are made. It follows, he insists, that mutations in Hox genes cannot by themselves build new body parts or body plans. As he explains, “Schwartz ignores the fact that homeobox genes are selector genes. They can do nothing if the genes regulated by them are not there.”31 Though Schwartz says he has “marveled” at “the importance of homeobox genes in helping us to understand the basics of evolutionary change,”32 Szathmáry doubts that mutations in these genes have much creative power. After asking whether Schwartz succeeds in explaining the origin of new forms of life by appealing to mutations in Hox genes, Szathmáry concludes, “I’m afraid that, in general, he does not.”33

  Nor, of course, do Hox genes possess the epigenetic information necessary for bodyplan formation. Indeed, even in the best of cases mutations in Hox genes still only alter genes. Mutations in Hox genes can only generate new genetic information in DNA. They do not, and cannot, generate epigenetic information.

  Instead, epigenetic information and structures actually determine the function of many Hox genes, and not the reverse. This can be seen when the same Hox gene (as determined by nucleotide sequence homology) regulates the development of different anatomical features found in different phyla. For instance, in arthropods the Hox gene Distal-less is required for the normal development of jointed arthropod legs. But in vertebrates a homologous gene (e.g., the Dlx gene in mice) builds a different kind of (nonhomologous) leg. Another homologue of the Distal-less gene in echinoderms regulates the development of tube feet and spines—anatomical features classically thought not to be homologous to arthropod limbs, nor to limbs of tetrapods.34 In each case, the Distal-less homologues play different roles determined by the higher-level organismal context. And since mutations in Hox genes do not alter higher-level epigenetic contexts,35 they cannot explain the origin of the novel epigenetic information and structure that establishes the context and that is necessary to building a new animal body plan.36

  Neutral or Nonadaptive Evolution

  Michael Lynch, a geneticist at Indiana University, has offered a different mechanism of evolutionary change, and a different explanati
on for the origin (or growth) of the genome as well as the origin of anatomical novelty. Lynch proposes a neutral or “nonadaptive” theory of evolution in which natural selection plays a largely insignificant role. His theory is based on contrasting observations about the features and strength of evolutionary mechanisms at work in populations of different sizes.

  He observes, first, that in general, the larger the population of organisms, the lower the mutation rate and (in sexually reproducing eukaryotes) the higher the rate of genetic recombination. He notes that the genomes of organisms in larger populations (such as those of bacteria and unicellular eukaryotic organisms) tend to be smaller and more streamlined—meaning they have fewer intervening nonprotein-coding sequences (i.e., introns). Most important, he notes that in large populations, natural selection tends to be relatively effective in eliminating deleterious mutations and fixing beneficial ones, whereas the process of genetic drift (the tendency for gene variants to be lost through random processes) plays a relatively less significant role.

  By contrast, Lynch observes that small populations—which would include almost all animal groups—are characterized by higher mutation rates and lower rates of genetic recombination. They also tend to have large genomes with a lot of nonprotein-coding DNA—introns, pseudogenes, transposons and various repetitive DNA elements—as well as gene duplicates. Lynch argues that in these small populations, natural selection tends to be weak—unable to remove mildly deleterious mutations or to fix mildly beneficial ones efficiently. As Lynch summarizes, “Three factors (low population sizes, low recombination rates and high mutation rates) conspire to reduce the efficiency of natural selection with increasing organism size.”37 Consequently, nonprotein-coding elements are not removed from the genome, but instead tend to accumulate, causing the genomes of organisms living in small populations to grow—even though these sequences may be neutral or even deleterious. Moreover, in small populations, “neutral” processes such as random mutation, genetic recombination, and genetic drift predominate in their effects over natural selection.