“Was the first craniate on the road to cognition?”
Evolution and Cognition 2003; 9(2):142-156.
 Fredric J. Heeren   (Page 2)
The first generally recognized evidence for possible chordates in the Cambrian came to light almost a quarter of a century ago, when Cambridge paleontologist Simon CONWAY MORRIS tentatively promoted a middle Cambrian species called Pikaia from annelid worm to chordate status (CONWAY MORRIS / WHITTINGTON 1979). Stephen Jay GOULD saved the tiny, simplified eel-shaped animal for the climax of his popular book about Canada’s Burgess Shale animals. On the last pages of Wonderful Life, he called Pikaia “the missing and final link in our story of contingency —the direct connection between Burgess decimation and eventual human evolution” (GOULD 1989, p322). Since then, however, less enthused scientists have questioned Pikaia’s chordate classification because of its lack of chordate features like gills, gonads, and a full notochord (HOLLAND personal communication).

For GOULD, the middle-Cambrian Pikaia best fit what the earliest chordate should look like: simple, sleek and headless. He mentions no brain, eyes, or other sensory organs when describing Pikaia in his popular book; even the possibility of a head seems remote in an animal whose anterior end, in his illustration, splits into two (GOULD 1989).

The next best hope came from animals whose chordate status was disputed or who appeared too late to show that chordates joined in the early Cambrian radiation of new forms. The eel-like conodont, long known only from its teeth, extended back only to the late Cambrian (PURNELL/DONOGHUE 1997). During the 1990s battles ensued over descriptions of two new chordate claims represented by just a few specimens: Yunnanozoon (CHEN et al. 1995) and Cathaymyrus (SHU/CONWAY MORRIS/ ZHANG 1996). The discoverer of Cathaymyrus thought Yunnanozoon looked more like a hemichordate (acorn worm) than a chordate (SHU/ZHANG/ CHENG 1996); and the discoverer of Yunnanozoon opined that his challenger had mistaken Cathaymyrus’s squashed dorsal fin for a notochord (CHEN/ HUANG/LI, 1999b).

After the Cambrian waters had been sufficiently muddied, researchers wondered if any true chordate had ever been found in Cambrian strata. Maybe our own “sophisticated” phylum had not yet evolved. Even GOULD’s Pikaia, though used to illustrate Cambrian chordates in vertebrate textbooks, no longer looked convincing, since it lacked many of the chordate features claimed by the more recent finds.

Thus the significance of the discovery of Haikouella —displayed in over 300 specimens. In CHEN’s description of Haikouella fossils, he pointed out features that not only demonstrate its chordate status, but that shed light on the origin of craniates (biology’s new name for vertebrates) (CHEN/HUANG/LI 1999b). The new nomenclature reflects a new primary diagnostic feature for this taxon: a distinct head enclosing a brain and sensory organs, recognizing that this character should now take precedence over the presence of a vertebral column.

Known for his research on amphioxus, the present-day animal thought to best represent the ancestor of all vertebrates, Nicholas HOLLAND said: “There’s no question these things are chordates” (ENSERINK 1999). He remarked on the great number of specimens with conspicuous gill slits (for straining food out of the water) and other diagnostic characters: “The muscle segments are unarguable, and the notochord’s good too” (HOLLAND 1999). Unlike specimens from other recent finds, both Haikouella and Yunnanozoon exhibit large notochords that clearly run the full length of their bodies. “It’s the earliest known chordate ancestor”, said HOLLAND. “Every zoology student and every paleontology student for many, many generations is going to have to look at that picture. This is going to be page one, two, three and four of vertebrate texts, and paleontology texts, and invertebrate zoology texts” (HOLLAND personal communication).

Since the discovery of Haikouella, Degan SHU et al. (1999) reported their discovery of two new chordates, Myllokunmingia and Haikouichthys, each based on a single specimen. Collaborator Simon CONWAY MORRIS proposes that the animals had skulls made of cartilage (MONASTERSKY 1999). CHEN notes that the specimens display two important features: distinctive fins (large dorsal and possibly paired ventral fins) and zigzag-shaped segmented muscles, similar to the pattern in modern fish (CHEN, personal communication). Though paleontologists of these various discoveries continue to contend with one another over whose specimens are ancestral to whose—and whose are true chordates—all agree that chordates have now been found in the early Cambrian (ENSERINK 1999; DZIK 1995).

What will happen to GOULD’s Pikaia, the animal zoology textbooks presently tout as our earliest chordate ancestor? HOLLAND contends that the textbook writers had no business picking up Pikaia as a chordate ancestor from GOULD’s popular book, since GOULD was not an authority on the animal (HOLLAND personal communication). GOULD had simply made it fit what he needed to relate the Burgess Shale fauna to humans. “Why do humans exist?” asked GOULD on the last page of Wonderful Life. “A major part of the answer, touching those aspects of the issue that science can treat at all, must be: because Pikaia survived the Burgess decimation” (GOULD 1989, p323). GOULD had used Pikaia to relate Pikaia to us and us to his overriding theme: contingency. “What this conference has done”, said HOLLAND at the symposium where Haikouella was announced, “is to pull the rug out from under Pikaia, for sure. Nobody will ever talk about it again” (HOLLAND personal communication).

Shedding Light on Vertebrate Origins
Now that lower Cambrian chordates have been con- firmed, zoologists must deal with the fact that Haikouella —and other early Cambrian chordates—look nothing like what they expected to see in a predecessor of Pikaia. Rather than finding evidence that this complex animal had less sophisticated ancestors, CHEN and SHU instead found examples of more complex, fully formed chordates—fifteen million years earlier. None of these newly discovered chordates have vertebrae or endoskeletons, so strictly speaking, they aren’t vertebrates. But displaying relatively large brains, these animals appear to be in the line to vertebrates, so that at the conference where Haikouella was announced, the strange term “pre-backbone vertebrate” was frequently bandied about. The brain’s early appearance would seem to demonstrate that brain and endoskeleton did not evolve together, as had been assumed, but rather that the brain appeared long before the development of the vertebrate spine.

“The discovery of the first craniate shows that the evolutionary history toward vertebrates had been on track long before the origins of the backbone”, says Taiwanese biologist Chia-Wei LI ( 1999), co-author of the Haikouella description. Haikouella findings run counter to the commonly held notion that the head could not become the dominant body structure until the body’s superstructure was also in place. It now appears that, against externalist expectations, cephalization (when the head became the dominant or controlling body structure) preceded endoskeletization (the development of an internal support structure).

CHEN also identified other important features in Haikouella that preceded the development of a bony skeleton: a neural cord that, like the notocord, runs the length of the body; a heart; a pair of lateral eyes; and tiny teeth. The teeth are located far back in its large pharyngeal cavity rather than in the mouth, indicating that it used them for grinding, not biting. Biologists had assumed that chordates did not develop the ability to accumulate minerals in their bodies to form teeth or bones until about 500 million years ago. But Haikouella and Yunnanozoon demonstrate that biomineralization had begun at least 30 million years earlier. Teeth led the way long before the development of a notochord-protecting, mineralized vertebral column or other bones.

Constraints, Channeling, and Convergence
The sudden explosion of widely disparate Cambrian animal Bauplans, followed by no new body plans throughout the rest of geologic history, fits the picture of a constrained process, the channeling of changes within particular forms. Scientists also find evidence of constraints today in the form of parallelism and convergence,9 both in experiments with living animals and in theoretical modeling. From his research on the development of amphibians, brain researcher Gerhard SCHLOSSER notes trends “where several characters tend to act as a ‘unit of evolution’, i.e., they tend to coevolve repeatedly” (SCHLOSSER 2000).

Evolutionary geneticist Paul RAINEY and his colleagues have also noticed convergence in evolution while experimenting with the bacterium Pseudomonas fluorescens. “These experiments in test-tube evolution”, says RAINEY, “allow us to replay life’s tape, albeit on a small scale, as often as we like” (RAINEY 2003). Their findings? “Evolution repeats itself”. By growing rapidly diversifying strains of the bacterium in test tubes of nutrient broth, they have discovered that “in the face of similar selective conditions, different lineages can find similar solutions to the same problems”. RAINEY is not afraid to find implications from his findings for human evolution: “Replay life’s tape”, he claims, “and while Homo sapiens may not evolve there is a high probability that introspective bipedal organisms with binocular vision will” (Ibid).

Simon CONWAY MORRIS reaches a similar conclusion. Speaking of the property of consciousness, he writes: “Here the reality of convergence suggests that the tape of life, to use GOULD’s metaphor, can be run as many times as we like and in principle intelligence will surely emerge” (CONWAY MORRIS 1998, p14). What about “the numerous entirely plausible alternatives of strikingly different forms” that GOULD expected if the tape should be rerun from the beginning? “Put simply”, says CONWAY MORRIS, “contingency is inevitable, but unremarkable…. There are not an unlimited number of ways of doing something. For all its exuberance, the forms of life are restricted and channeled” (p13). CONWAY MORRIS believes that convergence “effectively undermines the main plank of GOULD’s argument on the role of contingent processes in shaping the tree of life” (Ibid). GOULD, he says, “presupposes that constraints are weak” and makes a “most egregious misinterpretation of the Burgess Shale” (CONWAY MORRIS 1998–1999). His “egregious misinterpretation” —contingency as the major lesson of the Burgess Shale—is a conclusion that GOULD drew from his personal credo, according to CONWAY MORRIS, not from paleontology (Ibid).

Cladistics, a branch of biology that does indisputably draw its evidence from paleontology, hypothesizes relationships between organisms according to shared derived characters (synapomorphies). The distribution of these diagnostic features forms a set of nested groups (clades), in which smaller clades are contained within larger ones. The hierarchic pattern that has become the hallmark of cladistic analysis is related to the lack of transitional forms found between groups. DARWIN expected evolution to leave us with surviving modern groups within groups, but he expected the history of life to proceed in a gradualistic sequence that blurs the lines between groups. The scarcity of such fossil transitions can only be explained in DARWINIAN terms as a sampling problem, an artifact of an incomplete fossil record (DARWIN 2000, p292). Modern paleontologists generally agree, however, that the fossil record is actually robust enough to tell us that the scarcity of transitional forms is real and significant (SIMPSON 1960; GOULD 1977; VALENTINE/ERWIN 1987; DONOVAN/ PAUL 1998; FOOTE 1996; FOOTE/SEPKOSKI 1999), making the hierarchic pattern a genuine aberration in the gradualistic picture.

The priority of typology over continuity has persisted, according to SIMPSON, among “all schools of taxonomy including some that usually oppose typology in principle” (SIMPSON 1961, p49). Haikouella contributes to this crystalizing picture of distinct, fully formed body plans from near the start. Developmentalists observe the same hierarchical processes at work in both ontogenesis and evolution. Biologist Brian GOODWIN writes: “Developmental processes are hierarchical. So are biological classification schemes” (GOODWIN 1994, p234). Wallace ARTHUR agrees: “A theme running through the work of most contributors to what can now be described as evolutionary developmental biology is the relationship between these two hierarchies”, (ARTHUR 1997, p256) and he asserts that “it is informative about the nature of evolutionary mechanisms” (p257).

How much further back can we trace our ancestors? Nicholas HOLLAND, for one, wants to know what preceded these complex, early Cambrian craniates, a question, he says, that remains as big a mystery as ever: “Where are Haikouella’s ancestors? The sixtyfour dollar question is, What is this hooked to? That nobody knows” (HOLLAND personal communication).

In his presentation to an international symposium on Cambrian body plans (1999), HOLLAND gave genetic reasons why the most popular theoretical predecessor for chordates, tunicates (sea squirts), only works in the imagination of the theorists. When chordates are compared genetically with tunicates and fruit flies, he says, “the fruit fly is closer to the tunicate every time” (HOLLAND personal communication).

No obviously ancestral fossils presently exist to support theories about how chordates, or the other phyla, evolved in Precambrian times. “There are a lot of different totally cutup paper doll ideas about where things come from that aren’t based on fossils at all, but people sitting in their armchairs”, says HOLLAND (personal communication). The ceaseless re-interpretation of ancestral lineages for the phyla is easily demonstrated by the relevant literature (ARTHUR 1997, p73; BERGSTRÖM 1994; LYNCH 1999). Wherever the first chordates came from, HOLLAND thinks science must now take seriously the concept of “saltation”, the possibility of evolution in quick jumps. However broadly one defines “saltation”, paleontological evidence for the notion is certainly supportive of the internalist/developmentalist position.

Though opinions vary about the Precambrian antiquity of the phyla, all agree that almost all of these most widely separated animal groups had appeared by the early Cambrian period. Why didn’t new phyla continue to evolve during subsequent eras? Why did such disparate phyla as chordates, mollusks, arthropods, and the 35-or-so others first show up in the fossil record so close to the same time? CHEN places the window of opportunity for the explosive evolution of the majority of body plans within a narrow window of three million years (CHEN 1999), though of course, this is hotly disputed.

Body plans seen in the Precambrian include sponges, annelid worms, and echinoderms (like sea stars), but little else to represent the many lineages expected to lead to the 35 Cambrian groups. Gradualists have claimed that the ancestors of the many other disparate Bauplans must have been too small or too soft to be preserved. But since 1998, phosphate deposits at a Precambrian locale called Weng’an have proved capable of preserving the smallest and softest organisms imaginable (LI et al. 1998). Sponge embryos have been found by the thousands in early cleaving stages, seen under the microscope in groups of 2, 4, 8 cells, etc. (Figure 3). Though small and soft specimens are found in abundance, the number of body plans remains small.

The questions raised by such findings drew sixty scientists to Kunming, China, for a symposium entitled: “The Origins of Animal Body Plans and Their Fossil Records”. Perhaps it took the discovery of our own phylum’s participation in the early Cambrian big bang to bring together such an international gathering to consider a pattern some call “top-down evolution”.
Figure 3. Sponge embryos seen under the microscope at the cellular level in early cleaving stages, well preserved by the thousands from Precambrian deposit at Weng’an.
previous page
page 3 of this paper