When a body of evidence supports multiple conflicting historical hypotheses, the evidence cannot be sending a definitive historical signal about what happened in the past. That raises the possibility that it may not be sending a signal at all. Conversely, when the evidence leads investigators to converge around a single historical hypothesis, when one hypothesis best explains a whole group of clues, it is much more likely that the evidence is telling us what actually happened.
Consider, by way of illustration, a case in which we know a true history of ancestor–descendant relationships to see how evidence can converge around a single (unequivocal) history. Between 1839 and 1856, Charles Darwin and his wife, Emma, produced ten children, listed below in alphabetical order:
Anne
Charles
Elizabeth
Francis
George
Henrietta
Horace
Leonard
Mary
William
This alphabetical listing, of course, is not their actual birth order. Instead, it is one of a large number of possible birth orders for Darwin’s children, only one of which is the correct sequence. Indeed, only one of these arrangements can represent the actual Darwin family history.
Now, suppose I gave you and some friends a pile of historical evidence about Darwin’s children, and asked you to “solve for their birth order.” No one would consider the problem solved if you came back with more than one order. On the other hand, if you came back and presented a single coherent hypothesis of the birth order supported by evidence from birth records, family letters, and photographs from the Darwin family archives, that would provide persuasive evidence that you had obtained the correct solution. Since only one true history exists, once you find it the evidence will tend to fall naturally into place.
But does the evidence for a Precambrian animal tree of life fall similarly into place or does it generate multiple conflicting histories? We’ve already seen that fossil evidence does not point to a specific Precambrian tree of animal life, or perhaps to any tree at all. We’ve also seen that genetic evidence by itself does not establish a single divergence point for animal evolution. But what about the genetic and anatomical evidence taken together? Does that evidence converge on a single history of animal life? If so, then it could well make up for a lack of fossil evidence. Otherwise, it would seem to raise an obvious question: Are the observed genetic and anatomical “affinities” among the Cambrian phyla sending reliable historical signals at all?
Conflicting Histories
There are several reasons to doubt that evidence of genetic and anatomical similarity is sending a reliable signal of the early history of animal life. First, comparisons of different molecules frequently generate divergent trees. Second, comparisons of anatomical characteristics and molecules frequently produce divergent trees. Third, trees based solely on different anatomical characteristics often contradict each other. Let’s examine each problem.
Molecules vs. Molecules
Just as the molecular data do not point unequivocally to a single date for the last common ancestor of all the Cambrian animals (the point of deep divergence), they do not point unequivocally to a single coherent tree depicting the evolution of animals in the Precambrian. Numerous papers have noted the prevalence of contradictory trees based on evidence from molecular genetics. A 2009 paper in Trends in Ecology and Evolution notes that “evolutionary trees from different genes often have conflicting branching patterns.”13 Likewise, a 2012 paper in Biological Reviews acknowledges that “phylogenetic conflict is common, and frequently the norm rather than the exception.”14 Echoing these views, a January 2009 cover story and review article in New Scientist observed that today the tree-of-life project “lies in tatters, torn to pieces by an onslaught of negative evidence.” As the article explains, “Many biologists now argue that the tree concept is obsolete and needs to be discarded,” because the evidence suggests that “the evolution of animals and plants isn’t exactly treelike.”
The New Scientist article cited a study by Michael Syvanen, a biologist at the University of California at Davis, who studied the relationships among several phyla that first arose in the Cambrian.15 Syvanen’s study compared two thousand genes in six animals spanning phyla as diverse as chordates, echinoderms, arthropods, and nematodes. His analysis yielded no consistent treelike pattern. As the New Scientist reported, “In theory, he should have been able to use the gene sequences to construct an evolutionary tree showing the relationships between the six animals. He failed. The problem was that different genes told contradictory evolutionary stories.” Syvanen himself summarized the results in the bluntest of terms: “We’ve just annihilated the tree of life. It’s not a tree anymore, it’s a different topology [pattern of history] entirely. What would Darwin have made of that?”16
Other studies trying to clarify the evolutionary history and phylogenetic relationships of the animal phyla have encountered similar difficulties. Vanderbilt University molecular systematist Antonis Rokas is a leader among biologists using molecular data to study animal phylogenetic relationships. Nevertheless, he concedes that a century and a half after The Origin of Species, “a complete and accurate tree of life remains an elusive goal.”17 In 2005, during the course of an authoritative study he eventually copublished in Science, Rokas was confronted with this stark reality. The study had sought to determine the evolutionary history of the animal phyla by analyzing fifty genes across seventeen taxa. He hoped that a single dominant phylogenetic tree would emerge. Rokas and his team reported that “a 50-gene data matrix does not resolve relationships among most metazoan phyla” because it generated numerous conflicting phylogenies and historical signals. Their conclusion was candid: “Despite the amount of data and breadth of taxa analyzed, relationships among most metazoan phyla remained unresolved.”18
In a paper published the following year, Rokas and University of Wisconsin at Madison biologist Sean B. Carroll went so far as to assert that “certain critical parts of the TOL [tree of life] may be difficult to resolve, regardless of the quantity of conventional data available.”19 This problem applies specifically to the relationships of the animal phyla, where “[m]any recent studies have reported support for many alternative conflicting phylogenies.”20 Investigators studying the animal tree found that “a large fraction of single genes produce phylogenies of poor quality” such that in one case, a study “omitted 35% of single genes from their data matrix, because those genes produced phylogenies at odds with conventional wisdom.”21 Rokas and Carroll tried to explain the many contradictory trees by proposing that the animal phyla might have evolved too quickly for the genes to record some signal of phylogenetic relationships into the respective genomes. In their view, if the evolutionary process responsible for anatomical novelty works quickly enough, there would not be sufficient time for differences to accumulate in key molecular markers, in particular those used to infer evolutionary relationships in different animal phyla. Then, given enough time, whatever signal did exist might become lost. Thus, when groups of organisms branch rapidly and then evolve separately for long periods of time, this “can overwhelm the true historical signal”22—leading to the inability to determine evolutionary relationships.
Their article brings the discussion of the Cambrian explosion full circle from an attempt to use genes to compensate for the absence of fossil evidence to the acknowledgment that genes do not convey any clear signal about the evolutionary relationships of the phyla first preserved by fossils in the Cambrian. The logic of their analysis also leads them to a strangely familiar conclusion. Since the analysis of key genetic markers—like the genes tracked in molecular-clock studies that presumably accumulate mutations at a constant rate—shows a low number of mutational differences between the Cambrian animal phyla, Rokas and Carroll conclude from specifically genetic evidence that the phyla must have diverged rapidly. As they put it in another paper, “Inferences from these two independent lines of evidence
(molecules and fossils) support a view of the origin of Metazoa as a radiation compressed in time.”23 Thus, the inability to reconstruct the evolutionary history of the animal phyla from the molecular data not only fails to establish a Precambrian pattern of descent; it ironically also reaffirms the extreme rapidity of the origin of the Cambrian animal forms.
Molecules vs. Anatomy
In 1965, chemist Linus Pauling and biologist Emile Zuckerkandl, often hailed as the fathers of the molecular-clock concept, proposed a rigorous way to confirm evolutionary phylogenies. They suggested that if studies of comparative anatomy and DNA sequences generated similar phylogenetic trees, then “the best available single proof of the reality of macroevolution would be furnished.”24 As they went on to explain, “only the theory of evolution … could reasonably account for such a congruence between lines of evidence obtained independently.”25 By focusing attention on these two independent lines of evidence and the possibility of their convergence (or conflict), Pauling and Zuckerkandl provided a clear and measurable way to test the neo-Darwinian thesis of universal common ancestry.
And according to some scientists, studies of molecular homologies have confirmed expectations about the history of the animal phyla derived from studies of comparative anatomy. After citing Pauling and Zuckerkandl’s test, Douglas Theobald claims in his “29+ Evidences for Macroevolution” that “well-determined phylogenetic trees inferred from the independent evidence of morphology and molecular sequences match with an extremely high degree of statistical significance.”26
In reality, however, the technical literature tells a different story. Studies of molecular homologies often fail to confirm evolutionary trees depicting the history of the animal phyla derived from studies of comparative anatomy. Instead, during the 1990s, early into the revolution in molecular genetics, many studies began to show that phylogenetic trees derived from anatomy and those derived from molecules often contradicted each other.
Probably the most protracted conflict of this type concerns a widely accepted phylogeny for the bilaterian animals. This classification scheme was originally the work of the influential American zoologist Libbie Hyman.27 Hyman’s view, generally known as the “Coelomata” hypothesis, was based on her analysis of anatomical characteristics, mainly germ (or primary tissue) layers, planes of body symmetry, and especially the presence or absence of a central body cavity called the “coelom,” which gives the hypothesis its name. In the Coelomata hypothesis, the bilaterian animals were classified in three groups, the Acoelomata, the Pseudocoelomata, and the Coelomata, each encompassing several different bilaterian animal phyla.28 (See Fig. 6.1a.)
Then, in the mid-1990s, a very different arrangement of these animal groups was proposed based on the analysis of a molecule present in each (the 18S ribosomal RNA; see Fig. 6.1b). The team of researchers who proposed this arrangement published a groundbreaking paper in Nature with a title that surprised many morphologists: “Evidence for a Clade of Nematodes, Arthropods and Other Moulting Animals.”29 The paper noted the conventional wisdom, based on Hyman’s hypothesis, that arthropods and annelids were closely related because both phyla had segmented body plans.30 But their study of the 18S ribosomal RNA suggested a different grouping, one that placed arthropods close to nematodes within a group of animals that molt, which they called “Ecdysozoa.” This relationship surprised anatomists, since arthropods and nematodes don’t exactly look like kissing cousins. Arthropods (such as trilobites and insects) have coeloms, whereas nematodes (such as the tiny worm Caenorhabditis elegans) do not, leading many evolutionary biologists to believe nematodes were early branching animals only distantly related to arthropods.31 The Nature paper explained how unexpected this grouping of arthropods and nematodes was: “Considering the greatly differing morphologies, embryological features, and life histories of the molting animals, it was initially surprising that the ribosomal RNA tree should group them together.”32
FIGURE 6.1
How scientists reconstruct evolutionary history depends on which similarities they regard as revealing the true history of descent (homology) and which similarities they regard as misleading (homoplasy). Advocates of the Coelomata hypothesis (Figure 6.1a) regard the coelom (body cavity) as a homologous feature. Thus, they think the presence of a coelom in both arthropods and vertebrates indicates a common ancestor that possessed a coelom (indicated by the solid horizontal line in 6.1a). But advocates of the Ecdysozoa hypothesis (6.1b) think the coelom evolved at least twice independently (indicated by the two dashed horizontal lines in Figure 6.1b). They regard the presence of the coelom as a historically misleading similarity—one that does not indicate the presence of that feature in the most recent common ancestor of the groups possessing it. These two hypotheses and their implied histories are not congruent, and cannot both be true.
Since the Ecdysozoa hypothesis was first proposed, other scientists have vigorously opposed it, reaffirming the Coelomata hypothesis, based on the analysis of other molecular evidence.33 Advocates of the Ecdysozoa grouping pushed back hard, however,34 contending that, properly interpreted, available genetic evidence supports the Ecdysozoa, not the Coelomata hypothesis.35
My point in summarizing these disputes is simply to note that the molecular and anatomical data commonly disagree, that one can find partisans on every side, that the debate is persistent and ongoing, and that, therefore, the statements of Dawkins, Coyne, and many others about all the evidence (molecular and anatomical) supporting a single, unambiguous animal tree are manifestly false. As can readily be seen by comparing Figures 6.1a and 6.1b,36 these hypotheses—Coelomata and Ecdysozoa—contradict each other. Although both might be false, both cannot be true.
Various papers analyzing other groups have found similar discrepancies between molecular and morphological versions of the animal tree. A paper by Laura Maley and Charles Marshall in the journal Science noted, “Animal relationships derived from these new molecular data sometimes are very different from those implied by older, classical evaluations of morphology.”37 For example, when tarantulas were used as the representative of arthropods, the arthropods were grouped more closely to mollusks than to deuterostomes (animals that develop anuses first and mouths later). This makes sense because both mollusks and arthropods are protostomes (animals that develop mouths first and anuses later). But when brine shrimp were used as the representative of the arthropods, the arthropods became the odd man out. Now mollusks were grouped most closely to deuterostomes, far away from arthropods—a result clearly at odds with the conventional phylogeny based upon anatomical characteristics.38
Examples of similar conflicts abound. The traditional phylogeny placed sponges at the bottom of the animal tree, with progressively more complex phyla (e.g., cnidarians, flatworms, nematodes) branching off. But Valentine, Jablonski, and Erwin note that molecules “indicate a very different configuration” of the tree, where some higher deuterostome phyla branch off very early and some comparatively less complex phyla branch very late.39
Likewise, morphological studies suggest phoronids (see Fig. 3.6) and brachiopods (see Fig. 1.3), both marine filter-feeding animals, are deuterostomes, but molecular studies classify them within protostomes.40 Morphological studies typically imply that sponges are monophyletic (all part of an exclusive branch on the tree of life) because of their distinctive body architecture, but molecular studies suggest that sponges don’t belong to a single unified group, with some sponges more closely related to jellyfish than they are to other sponges.41 Cnidarians and ctenophores have similar body plans, leading many to expect they were closely related on the basis of morphology. But molecular data have distanced these phyla significantly.42 As a major review article in Nature in 2000 observes, “Evolutionary trees constructed by studying biological molecules often don’t resemble those drawn up from morphology.”43 And the problem isn’t getting better over time. A 2012 paper admits that larger datasets are not solving this problem: “Incongruence between phylogenies derived from morph
ological versus molecular analyses and between trees based on different subsets of molecular sequences has become pervasive as datasets have expanded rapidly in both characters and species.”44
Indeed, the widespread discrepancies between molecular data and morphological data and between various molecule-based trees have led some to conclude that Pauling and Zuckerkandl were wrong to assume that the degree of similarity indicates the degree of evolutionary relatedness.45 As Jeffrey H. Schwartz and Bruno Maresca put it in the journal Biological Theory: “This assumption derives from interpreting molecular similarity (or dissimilarity) between taxa in the context of a Darwinian model of continual and gradual change. Review of the history of molecular systematics and its claims in the context of molecular biology reveals that there is no basis for the ‘molecular assumption.’ ”46
Anatomy vs. Anatomy
Attempts to infer a consistent picture of the history of animal life based on analyzing the anatomical characteristics of different animals have also proven problematic. In the first place, there is a general and long-standing problem with attempts to infer the evolutionary history of the animal phyla from similar anatomical traits. At the level of the phyla—that is, when one compares the phyla to each other and tries to determine their branching order—the number of shared anatomical characteristics available for inferring evolutionary relationships drops off quite dramatically. There is an obvious reason for this. For example, an anatomical character such as the “leg,” that is useful for diagnosing and comparing arthropods, which possess legs, proves useless for making comparisons between (for example) brachiopods or bryozoans, which do not. In the same way, basic structural features of human-designed systems, such as the distinctive submarine “trait” of an encapsulating watertight hull, might help to distinguish it from a cruise ship, which is only watertight on its underside. But this “trait” would be irrelevant for comparing and classifying, say, suspension bridges, motorcycles, or flat-screen televisions. In a similar manner, biologists find that there are only a handful of highly abstract characters, such as radial versus bilateral body symmetry, the number of fundamental tissue layers (triploblasty, three layers, versus diploblasty, two layers), or the type of body cavity present (true coelom, pseudocoelom, or no coelom), available for morphological comparisons of the many diverse animal forms. Yet evolutionary biologists have often disputed the validity of these rather abstract traits as guides to evolutionary history.47 In addition, just as trees based upon the analysis of different sets of similar genes or proteins often conflict, trees constructed on the basis of different developmental and anatomical characteristics often conflict.
Darwin's Doubt Page 14