When biologists construct phylogenetic trees based upon anatomical characteristics, they typically group the animal phyla according to the presence or absence of several key characteristics. For example, the standard version of the animal tree, based upon anatomy, groups animals according to their style of body plan symmetry and by their mode of body plan development. As noted earlier, animals with mirror symmetry along their vertical head-to-tail axes all fall within the Bilaterian group. Animals with radial symmetry (or no symmetry) fall outside this group. Within the Bilateria, taxonomists distinguish other main groups—protostomes and deuterostomes—based upon their differing modes of body plan development—i.e., either “mouth first” or “anus first.”
Yet a significant difficulty arises when evolutionary biologists consider how a particularly fundamental characteristic—the mode of germ-cell formation—is distributed among various groups on the canonical animal tree of life (see Fig. 6.2).48 Germ cells produce eggs and sperm (in any sexually reproducing species) or gametes (in any asexually reproducing species), giving rise to the next generation.49 Animals have two main ways of generating germ cells. In one mode of germ-cell formation, known as preformation, cells inherit internal signals from a region within their own cell structure to become germ cells (illustrated by solid black squares on Fig. 6.3a). In the other main way of generating germ cells, known as epigenesis, germ cells receive external signals from surrounding tissues to become primordial germ cells (PGCs, illustrated by solid white squares in Fig. 6.3b).
Germ-cell formation has indisputable evolutionary importance. To evolve, a population or a species must leave offspring; to leave offspring, species of animals must generate primordial germ cells. No PGCs, no reproduction; no reproduction, no evolution.
One might expect, therefore, that if a group of animals is all derived from a common ancestor (with a particular mode of gamete production), then the mode of germ-cell formation should also be essentially the same from one animal species to the next in that group. Further, assuming the common ancestry of all animals, our expectation of homologous modes of germ-cell formation among the animals ought to be higher than for any other tissue type, cell line, or mode of development. Why? Because mutations affecting the developmental mechanisms that govern PGC formation inevitably disrupt successful reproduction.50 Again, if a species cannot reproduce, it cannot evolve.51
Thus, similar groups of animals—indeed, all animals, if they have descended from a common ancestor—ought to exhibit the same basic mode of germ-cell formation. Further, that the evolutionary tree derived from an analysis of the “mode of germ-cell formation” ought to be congruent with the trees derived from other such fundamental characteristics (such as bodyplan symmetry, mode of development, number of primary tissues, and so forth).
But the mode of germ-cell formation is nearly randomly distributed among the different animal groups, making it impossible to generate a coherent tree based on this characteristic, let alone making any comparison between such a tree and the canonical tree. Note also the distribution of the two basic modes of germ-cell development within the animal phyla as depicted on the canonical tree. Figure 6.4, derived from the work of Harvard developmental biologist Cassandra Extavour,52 shows this distribution and provides another way of understanding the incongruence that arises when analyzing different anatomical characteristics.
FIGURE 6.2
The canonical tree of the Metazoa as determined by the analysis of selected anatomical characters and genes.
FIGURE 6.3
Two modes of germ-cell formation. Figure 6.3a (left): In fruit flies as the egg is being formed, the mother’s nurse cells (the four cells on the left of the oval-shaped egg chamber, indicated by large black circles) deposit proteins and RNAs, which are transported to the posterior pole of the egg (indicated by the dark patch to the right of the large cell on the right). These maternally synthesized molecules then trigger the development of the germ cells and sex organs of the fly during embryogenesis. Figure 6.3b (right): In the eggs of mice there are no maternally deposited products that determine germ-cell formation. Rather, as the embryo develops a subpopulation of cells (represented by the triangles on the left) express “germline competence genes.” These cells then “read” signals that arrive from other tissues (see arrows), causing the cells to differentiate into primordial germ cells (as indicated by the stars on the right).
Notice the two modes of germ-cell formation do not cluster together in separate parts of the canonical tree. Instead, they are distributed haphazardly among various phyla on different branches of the tree. In the protostomes, for instance, modes of germ-cell formation wink on and off between preformation and epigenesis. The same is the case within the deuterostomes: germ-cell formation varies almost randomly, and several groups exhibit both modes, rendering it difficult or impossible to determine which characteristic was present at different ancestral branching points. Noting this pattern of distribution, Cassandra Extavour concludes that “the data presently available cannot suggest homologies of the somatic components of metazoan gonads.”53
FIGURE 6.4
The distribution of modes of primordial germ-cell formation (epigenesis, preformation, or both) among various animal groups. The solid thin lines between the boxes on the right and the phyla names on the left show where the different modes of germ-cell formation are present in different phyla. The nearly random distribution of types of germ-cell formation among the various animal groups makes it impossible to generate an animal phylogeny (evolutionary history) based upon this character that will match the evolutionary history implied by the canonical animal tree of life.
FIGURE 6.5
A selection of incompatible (mutually incongruent) phylogenetic trees representing the history of the major groups of animals, drawn from the zoological and evolutionary literature (1940–present). Note: The definitions of some taxonomic groups in some of these phylogenetic trees may have changed significantly since the time those phylogenies were originally constructed. Branch lengths may not always be drawn to scale.
After completing a survey of many such difficulties, University of St. Andrews zoologist Pat Willmer and Oxford University zoologist Peter Holland, experts on invertebrate anatomy, draw this conclusion: “Taken together, modern re-evaluations of traditional evidence support different and mutually exclusive subsets of [phylogenetic] relations.”54 They go on to observe that “patterns of symmetry, the number of germ layers in the body, the nature of the body cavity, and the presence or type of serial repetition [segmentation] have all been used to infer common ancestry.” But, they explain, the phylogenetic story these characteristics tell is “now either unacceptable or at least controversial” because the data are, at best, inconsistent.55
The historical record of ongoing uncertainty about the animal tree of life since 1859 confirms, as one respected textbook on invertebrate animals explains, that “phylogenetic analysis at the level of the phyla is highly problematical.”56 As a result, “the study of higher level animal phylogeny has yielded an expansive literature but relatively little detailed consensus… . In point of fact, there exists no such thing as ‘the traditional textbook phylogeny.’ A diversity of different schemes can be found.”57 To appreciate this problem visually, look at Figures 6.1a and 6.1b as well as Figure 6.5. These show some of the many metazoan phylogenies, based on anatomy, published in the twentieth century. These branching patterns plainly do not agree with one other.
The Assumptions of Phylogenetic Inference
All these problems underscore several fundamental difficulties with the methods of phylogenetic reconstruction. When biologists analyze multiple anatomical traits or genes, the animal phyla consistently defy attempts to arrange them into the pattern of a single tree. Yet if there was a period of hidden Precambrian evolution, and if comparative sequence analyses reveal the actual history of animal life and, by implication, the existence of Precambrian animal forms, phylogenetic studies should converge more and more aroun
d a single tree of animal life. Just as only one possible divergence point could represent the event at which animal forms began to evolve from a common animal ancestor, only one of the many trees produced by phylogenetic analysis can represent the actual Precambrian history of animal life. If, instead, phylogenetic analyses consistently generate different possible evolutionary histories, it’s difficult to see how any one of them could be known to be sending a reliable historical signal. Again, the history of animal life only happened once.
One could argue that these conflicting trees do, at least, show that some treelike evolutionary pattern of common ancestry preceded the Cambrian, since all conflicting trees do affirm that. But, again, they all “show” this because they all presuppose it, not because they demonstrate it.
Convergent Evolution
There is yet another reason to wonder whether studies of anatomical or molecular homology convey anything definitive about the history of life. Many animals have single traits or features in common with otherwise decidedly different animals. In such cases, it makes no evolutionary sense to classify these forms as closely related ancestors. For example, moles and mole crickets have remarkably similar forelimbs, though moles are mammals and mole crickets are insects. No evolutionary biologist regards these two animals as closely related, for understandable reasons.
The theory of universal common descent assumes that, generally, the more similar two organisms are, the more closely related they must be. Assuming common descent, animals with wildly differing body plans should not be closely related. The presence of nearly identical individual traits or structures within organisms exemplifying otherwise different body plans cannot, therefore, be attributed to evolution from a common ancestor. Instead, evolutionary biologists attribute similar traits or structures in such a context to so-called convergent evolution, the separate or independent origin of similar characters emerging on separate lines of descent after the point at which those lines diverged from their last common ancestor. Convergent evolution demonstrates that similarity does not always imply homology, or inheritance from a common ancestor.
For this reason, the repeated need to posit convergent evolution (and other related mechanisms)58 casts further doubt on the method of phylogenetic reconstruction. Invoking convergent evolution negates the very logic of the argument from homology, which affirms that similarity implies common ancestry, except—we now learn—in those many, many cases when it does not. Repeatedly invoking convergence negates the assumption that justified the method of phylogenetic reconstruction in the first place, namely, that similarity is a reliable historical signal of common ancestry.
A Family Reunion?
So what lesson should we draw from these many conflicting trees? Clearly, these contradictory results call into question the existence of a canonical tree of animal life. To see why, imagine being invited to an event billed as an extended-family reunion where you’ve been told you will encounter hundreds of your relatives, most of whom you have never met. Let’s call the description in the invitation the “reunion hypothesis.” The invitation says that a group photograph is planned, for which relatives will be grouped together according to their degree of relationship (first cousins with first cousins, and so on).
You show up and grab a coleslaw, eager to meet these many previously unknown relatives. You see the hundreds gathered and have every reason to believe that the “reunion hypothesis” is true. After all, the invitation in your mailbox described the event as a family reunion.
As the day goes on, however, something seems amiss. Here and there, you see familiar facial features—“Yes,” you think, that person could be my cousin”—but the majority of attendees and all the strangers you engage in conversation exhibit no discernible family resemblances. Nor does anyone seem to share any personal relationships with anyone else at the reunion, no matter how long they chat and try to establish points of commonality. What’s more, each person tells a different story of his or her family history. You try to group the strangers by physical characteristics (height, hair color, body type, and so on), but the characteristics that you find fail to yield evidence of family ties or genealogical connections. Whatever points of commonality you find refuse to assemble into a coherent, consistent story. The pedigrees are unclear. The reunion hypothesis is under major strain.
When the photographer arrives, she gamely tries to draw everyone together for the planned photograph. Chaos ensues. No one knows where to stand, because the family relationships are so unclear, if they exist at all. After milling around hopelessly for half an hour or so, people depart for their cars, wondering why they bothered to attend the picnic, and indeed why they were invited in the first place.
Do you now have good reasons to doubt the “reunion hypothesis”? You do. If the family reunion hypothesis were true, it would have been increasingly confirmed, because the evidence would have converged on a single consistent pattern. Had there been a true pattern of family relationships, the longer everyone talked, the more a single coherent pattern of relationship would have become apparent. But the people at the picnic were not your relatives, at least not in any way you could determine to be true. Instead of converging on a single pattern, the “evidence” was all over the map.
A Forest of Trees
Of course, my family reunion illustration breaks down as an analogy to the history of animal life, because if we could trace the history of all the people at the reunion back far enough we would find that they are all related by common ancestry. Though we can choose to assume that the same is true of the Cambrian animals, neither the fossil evidence nor the evidence of genetics and comparative anatomy actually establishes that. These three classes of evidence either provide no compelling evidence for Precambrian animal ancestors (in the case of fossils), or they provide question-begging and conflicting evidence (in the case of genes and anatomy).
And that is the point of my story. Since there can be only one true history of the Cambrian animals, the evidence should converge on a common family tree—if indeed we are looking at evidence of true history. The picture given by the evidence should be stable, not constantly changing. But the evidence from a variety of quarters has instead continually generated new, conflicting, and incoherent pictures of the history of animal life. As with the “cousins” in my illustration, there seems to be no consistent and coherent way to organize the animal groups into a family tree.
But if the genes don’t tell the story of Precambrian ancestral forms, if they don’t compensate for a dearth of fossil evidence and establish a unequivocal long cryptic history of animal life from an original animal, an ur-metazoan, then logically we are back to taking the fossil record at face value. In that case, the mystery of the missing ancestral fossils remains. If so, is there any way to explain the abrupt appearance of new forms of life in the fossil record within an evolutionary framework? During the 1970s, two young paleontologists thought that there just might be a way to do exactly that.
7
Punk Eek!
Scientific discoveries are rarely made in Laundromats, but at least one great scientific breakthrough—an “aha” moment—occurred in one. The year was 1968, more than a decade before the discovery of the first Chengjiang fossils. The scientist overtaken by the muses was paleontologist Niles Eldredge. One day while standing in a Michigan Laundromat, following months of collecting trilobite fossils for his Ph.D. research, Eldredge happened to reach into his pocket. He removed one of the fossils he had been collecting, a specimen of a trilobite species called Phacops rana. Initially, as he examined the specimen, he felt “depressed.” The fossil closely resembled many others that he had found across layers of strata during his fieldwork in the Midwest. His trilobites showed no evidence of gradual change, as classical neo-Darwinism had taught him to expect.1
As Eldredge explained in a lecture at the University of Pittsburgh in 1983, he then experienced a kind of scientific epiphany. He realized that the “absence of change itself” was “a very interesti
ng pattern.” Or as he later put it, “Stasis is data.”2 “Stasis” is the term that Eldredge and his scientific collaborator, Stephen Jay Gould (see Fig. 7.1), later gave to the pattern in which most species, “during their geological history, either do not change in any appreciable way or else they fluctuate mildly in morphology, with no apparent direction.”3 As Eldredge examined that solitary trilobite, he realized that he had been observing evidence of stasis for some time—however much he might have wanted it otherwise. As he explained, “Stasis … was by far the most important pattern to emerge from all my staring at Phacops specimens.” He continued, “Traditionally seen as an artifact of a poor record, as the inability of paleontologists to find what evolutionary biologists going back to Darwin had told them must be there, stasis was, as Stephen Jay Gould put it, ‘paleontology’s trade secret’—an embarrassing one at that.”4
This embarrassing realization proved pivotal, eventually leading Eldredge and Gould to reject both the gradualistic picture of evolutionary change articulated by Darwin and the neo-Darwinian understanding of the mechanism by which such change allegedly takes place. It also led them to formulate, in a series of scientific papers from 1972 to 1980, a new theory of evolution known as “punctuated equilibrium.”5
Darwin's Doubt Page 15