[229] The multicellular animals form a logical and elegant assemblage that excites admiration for their effective use of micro- and macroevolution. One can only hope that the generality of their features can someday be tested against another life system.

The origin and early diversification of multicellular organisms do not seem to mark planets in a fashion detectable from afar. Indeed, after generations of effort, the uncertainty associated with the dating of metazoan origins and early radiations on Earth is several hundred million years. Since we have by no means exhausted the possible observations, there is still hope that we shall further delimit the timing of these events from direct evidence. In the meantime we must rely on indirect evidence to support plausible models of the emergence and radiation of multicellular organisms on Earth.

One of the problems has been that metazoans with hard parts burst relatively suddenly into the fossil record, already well-differentiated into distinctive living phyla only distantly allied. Since evolution has been regarded as a gradual process, it has seemed doubtful to many that these phyla arose in a sudden burst of adaptive radiation. The Phanerozoic fossil record suggests a long Precambrian history for multicellular forms. Durham (1971) calculated that if Precambrian evolutionary rates were comparable to those displayed by Phanerozoic metazoans, then the morphological distance achieved by early Cambrian phyla implied that divergence from a founding metazoan lineage began a billion or more years earlier. Where are the traces or remains of those early animals?

Within the past two decades, evidence has accumulated to indicate that there was no such long period of gradual metazoan evolution culminating in [230] the ground plans of Phanerozoic phyla. The sudden advent of fossil phyla near the early Cambrian boundary probably coincides closely with their origin. Also within recent years, work in genetics and developmental biology has rendered plausible the sudden origination of such morphological novelties. Work in Phanerozoic paleontology has shown that even when the fossil record is most complete, most taxa display an abrupt rather than a gradual pattern of change. This has led to a downgrading of the role of microevolution in the origin and success of taxa.

In light of this work, an attempt is made here to assess the factors that led to the origin of major (phylum-level) multicellular grades and ground plans and that account for their diversity. Two particular questions are pursued: First, if time were turned back to the late Precambrian and multicellular grades were allowed to evolve again, how closely would this second fauna resemble the first, and how would it differ? Second, if this experiment were repeated under different environmental conditions, as on another planet, how closely would the results resemble and how differ from those on Earth?

Figure 1 depicts the time of first appearance in the fossil record of each living animal phylum. The figure is fairly conservative, listing those appearances that seem at present to be most definitive. The intent is that most subsequent changes will involve only the discovery of earlier representatives of the phyla. Of course, it is possible that restudy of fossil material may discredit some of these occurrences. In most cases the earliest records are securely based on body fossils, but for a few phyla the remains of secreted structures (such as pogonophoran tubes) or other traces are tentatively accepted.

The earliest certain metazoan fossils occur in the Ediacaran interval, which began 100 million years or more before the start of the Cambrian (Glaessner, 1971). At least some elements of the Ediacaran fauna persisted into Cambrian time (Borovikov, 1976). The characteristic Ediacaran forms are soft-bodied. The only living phyla certainly represented are the Cnidaria and the Echiuroidea. There are also some vermiform fossils that might represent living phyla (perhaps the Annelida) but might equally well be the remains of extinct ones. Finally, there are enigmatic forms (such as Tribrachidium) which cannot be assigned to a living phylum.

Mineralized skeletons and other hard parts do occur during the Precambrian; most are minute denticles, plates, and other skeletal elements that....

[232] ....formed parts of larger organisms but cannot be associated with any living phylum (Matthews and Missarzbevsky, 1975). In the earliest Cambrian stage, the first extensive fauna with mineralized skeletons appears and includes skeletons identified as Mollusca. In succeeding Lower Cambrian stages, five more living phyla appear, so that all living phyla that have easily fossilizable hard parts appear during the Lower Cambrian except for the Bryozoa and the Chordata.

The next first appearances are of four living phyla in a remarkably preserved, soft-bodied fauna that also contains hard-part remains of animals typical of the time. This is the Middle Cambrian Burgess Shale fauna from British Columbia. It includes the first known chordate, a soft-bodied form somewhat resembling cephalochordates (Conway Moins and Whittington, 1979). Bryozoan skeletons first occur in the early Ordovician.

The remaining 13 living phyla are all soft-bodied; eight of them appear as fossils but are exceedingly rare. The other five are not known as fossils at all. Four of these are minute pseudocoelomates (commonly parasitic today), and one is the diploblastic comb jellies. None of these forms leaves characteristic traces, and they would be preserved only under the most unusual of circumstances. Their lack of a fossil record is thus not surprising. It is perhaps more surprising that all the living coelomate phyla, even soft-bodied ones, do have a fossil record of some sort.

The order of appearance of phyla in the fossil record seems to be based more on their fossilizability than on their order of origin. However, there is evidence that the appearance of phyla with mineralized skeletons near the Precambrian-Cambrian boundary is associated closely with their origins. The skeletal ground plans of these phyla are closely coadapted with their body plans, and in some cases the latter depend on the former (Cloud, 1948; Valentine, 1973). In these cases the advent of a mineralized skeleton indicates the origin of the phylum, and the record indicates that these events occurred near the Precambrian-Cambrian boundary.

The two phyla with durable skeletons that do not appear in Lower Cambrian time, the Chordata and Bryozoa, may have been represented then by soft-bodied lineages, for they have body plans that do not require durable skeletons. Indeed, the chordates appear in the Middle Cambrian, while durable chordate skeletons are not known until the Late Cambrian (Repetski, 1978). The Bryozoa first occur as relatively simple, calcified branching tubes, but they could have been represented by soft-bodied forms prior to mineralization of the tubes. Thus either of these phyla could have been present in the Lower Cambrian.

In short, it is possible to hypothesize that the entire array of living coelomate and pseudocoelomate phyla arose near the Precambrian-Cambrian boundary, and that their entry into the fossil record occurred as they acquired hard parts or as they were preserved by a fortuitous concomitance [233] of circumstance. We will not argue this hypothesis at length here, but it does Serve the facts and may be nearly, if not precisely, correct.

In addition to living phyla, the fossil record contains remains of extinct groups different from any living groups, and that probably represent extinct phyla. The anatomy of some unusual soft-bodied fossils from the Middle Cambrian Burgess Shale can be reconstructed in enough detail that we can be reasonably sure we are dealing with extinct phyla. Their body plans sometimes resemble those of living phyla but are different enough to suggest that they evolved separately. Ten forms are now known from the Burgess Shale that are likely to represent extinct phyla (Conway Moins and Whittington, 1979). Other soft-bodied fossils that probably represent extinct phyla are known from later rocks; a particularly distinctive form is found in the Late Carboniferous of Illinois (Richardson, 1966). The times of origin and extinction of these distinctive soft-bodied forms are unknown. It is possible that they all originated during the late Precambrian-Lower Cambrian interval and are part of a general radiation that gave rise to most or all living phyla.

Durable skeletons distinct in architecture from those of living phyla are also known in the fossil record. These fossils are sometimes placed into a living phylum that they resemble, but such assignments are often in dispute. Two of particular interest here are the Archaeocyatha and the Hyolithida. Both appear in the Lower Cambrian, and each has been considered an extinct phylum by some authorities. Since the soft-part anatomies of these forms are largely unknown, and since skeletal ground plans alone are not necessarily definitive of relationship at the phylum level, their status remains unclear.

If a gigantic radiation of phyla did indeed occur near the end of the Precambrian, how can we account for it? What triggered it, and even more importantly, what processes operated to produce so much fundamental morphological divergence so quickly? Reconstruction of the fauna of the Late Precambrian before the suggested radiation is an obvious first step toward an answer.

The heading above is somewhat facetious-because of the convention by which animals are classified, all must belong to one phylum or another. Nevertheless most living phyla represent clades that diversified well after the metazoans had originated. The morphological distinctiveness believed appropriate to phyla is based on experience with these clades, although some, such as the Arthropoda (Tiegs and Manton, 1958), may be polyphyletic. While Cnidarians and Platyhelminthes may be ancestral to other phyla, the fossil record suggests that most phyla have distinctive morphologies because they [234] were adapted to different modes of life rather than because they represent a chain of ancestors and descendants arranged in some scala naturae. Put another way, the ground plans of the phyla probably did not evolve gradually as steps or divisions on a scale of complexity, but rather evolved relatively suddenly in response to broad adaptive opportunities that may all have been present contemporaneously. To be sure, some phyla descended from others, but the pattern of novel morphology results from the quality of adaptive opportunity.

Something of the character of animals ancestral to living phyla can be inferred from the architecture of the latter. The body plans of living phyla fall into several distinctive groups. These are particularly well discussed by Hyman (1940) and Clark (1964) and are summarized in table 1. Note that the times of appearance of these groups as fossils bear little relationship to their presumed phylogenies. Some groups represent different grades of construction: the diploblastic Cnidaria, the triploblastic but acoelomate Platyheminthes, and the pseudocoelomates, for example. Others are on the same grade of construction but have distinctive body plans; groups of coelomates, for example, can be identified on the basis of their coelomic architectures (table 1 ).

Presumably the Late Precambrian fauna included diploblastic and triploblastic acoelomate phyla (Cnidaria and probably Platyhelminthes at least), primitive pseudocoelomates, and a few coelomate groups ancestral to each group of living phyla that differs by coelomic plan. The Cnidaria and Platyhelminthes, probably largely pelagic and epifaunal, would not ordinarily leave much of a record of their presence (we are indeed fortunate to have the Ediacaran body fossils). Pseudocoelomates and coelomates, however, possess body cavities that were probably evolved as hydrostatic skeletons, an important primitive function of which was to aid in burrowing (Clark, 1964). Burrows that disturb bedding fossilize rather easily, so traces of early burrowing phyla might be expected. Both horizontal and vertical burrows are reported from the Late Precambrian. They are not all accurately dated but certainly occur during the Ediacaran interval and possibly earlier. These burrows are relatively rare in the Precambrian, but they increase significantly in kind and number in the Cambrian (Crimes, 1974). It is reasonable that the Edicaran burrows represent pseudocoelomate or (perhaps more likely because of their larger size) coelomate stocks that are ancestral to modern phyla (Valentine, 1973).

The exact timing of the rise of multicellularity as a new grade of organization cannot be deduced from the fossil record. There is little dispute as to the advantages of this condition: increased size, longevity, homeostasis, and ultimately complexity of cellular differentiation can stem from a multicellular condition, permitting development of life modes unavailable to unicells. Multicellularity arose numerous times from unicellular lineages (Stebbins in

[236] ....Dobzhansky et al., 1977). This is clear evidence of its adaptive advantaged although the precise adaptive pathways followed by the earliest metazoans are much disputed.

On present evidence, a plausible descriptive model of the origin of metazoan grades and ground plans has the following features (fig. 2). Multi-

[237] -cellular organisms arose from protozoan ancestors, and it is easy to imagine a wide variety of pelagic and benthic forms evolved at a simple grade of construction (Valentine, 1977). This early multicellular radiation must have occurred some time after the appearance of eucaryotes (perhaps near 1400 million years ago) but before 700 million years ago; very likely it was nearer the younger age since meiosis may have had to be perfected in eucaryotic unicells. From products of this early radiation, diploblastic and acoelomate triploblastic grades arose, possibly independently but probably by evolution of the latter from the former. These events may have occurred very near 700 million years ago. Then body cavities evolved in triploblastic Iineages at least twice and probably a few times, radiating to form a variety of pseudocoelomate and coelomate clades. The coelomate radiation produced the metamerous, oligomerous, amerous, and pseudometamerous vermiform "superphyla" stocks from which the phyla of the Lower Cambrian radiated in turn (or perhaps in some cases the Cambrian phyla represent Late Precambrian holdovers). The coelomate superphyla may have developed during the hundred-million-year interval preceding the Cambrian. The bulk of living phyla originated in radiations from the coelomate superphyla and from pseudocoelomate stocks probably near 570 million years ago at the beginning of the Cambrian The entire sequence of the origin of metazoa and the development of phyla could have been compressed inside the 150 million years preceding the Cambrian (although it may have taken longer). At any rate the radiation associated with the origin of most of the clades that we now call phyla seems to have taken only a few tens of millions of years at most.

Microevolution usually refers to the changes consequent on the selection of one genotype over another within the gene pool of a population or of an entire species. Macroevolution usually refers to the origin of taxa and of evolutionary trends above the level of species. This term will be used here in a somewhat more specialized sense to refer to the changes consequent on selection between gene pools. Genotypes that increase in frequency by microselection are said to be more fit; populations or species that increase through macroselection are said to have adaptive advantage. Both fitness and adaptive advantage are relative measures, the former between genotypes, the latter between taxa. Adaptation or adaptive value is the ability of an individual or taxon to cope with a given environment; it is measured on an absolute [238] rather than a relative scale (Ayala, 1969). Of course, all three of these fee tures-fitness, adaptive advantage, and adaptive value-vary in relation to the environment.

Microevolution can certainly be responsible for the origin of some new species, but it acts too slowly to account for the origin of many species and of most higher taxa (Stanley, 1975, 1976, 1978). Since fitness must by definition be enhanced by new genes or gene combinations for microevolution to occur, large heterozygous genetic changes that upset the biochemistry of development are usually lethal. Eldredge and Gould (1972) and Stanley (1975) have shown that many fossil species (morphospecies) evolved only slightly if at all during their existence; the morphological changes that differ entiate them from their ancestors must have been associated with their origins. The morphological distance achieved during the speciation process must often be considerable, equivalent to the morphological differences between many genera. Indeed, fossil genera appear much too rapidly to have originated through microevolutionary processes alone (Stanley, 1978). They must be products of macroevolution and perhaps often of a single speciation event. This argument has even more force when applied to higher taxa such as orders, classes, and phyla; the time required for microevolutionary processes to produce such distinctive morphologies as distinguish these higher taxa (which sometimes possess different body plans altogether) is far greater than the time available according to the fossil record.

Macroevolution does not depend on fitness and indeed must often circumvent it. If this can be done, then novelties considerable different morphologically from their ancestors can arise rapidly; their survival depends only on their adaptive value. An example is the rapid quantum speciation possible in organisms (chiefly plants) that are self-fertilizing. A large mutation can produce a descendant that is infertile with members of its parental species, including its parents. If it can fertilize itself, however, it may propagate and thrive; whether it does or not depends on its adaptive value. Indeed, if it should happen to have a distinct adaptive advantage over its parent, or any other species for that matter, due to a higher reproductive rate, greater hardiness, or some such feature, it may eventually replace an older species Microevolution has had very little to do with the origin or the success of this "instant species."

In lineages that are not self-fertilizing, the perpetuation of a marked novelty is more difficult, for it must backcross with "normal" members of its species and pass through at least one heterozygous generation to produce homozygotes. Since the heterozygotes would usually have much lower fitness than normal individuals, they must evade the winnowing effects of microselection to produce offspring. That such evasions do happen has long been postulated. For example, Simpson (1944, 1953) and Mayr (1942, 1963) have both appealed to genetic "drift" in small populations as a means [239] of achieving novel gene combinations in macroevolution. Fixation of chromosomal mutations in small populations of mammals, isolated by breeding patterns and subject to drift because of their small effective sizes, has been reported by Bush (1975) and Bush et al. (1977). In these mammalian examples, there appears to be little morphological difference between the normal and mutant phenotypes, but they do indicate how novel genotypes can be fixed despite probable heterozygote disadvantage.

The creation of form through developmental processes has long served as an analogy to the rise of form through evolutionary processes. Even though ontogeny does not recapitulate phylogeny in the classic sense, there is clearly much to be learned of evolutionary processes from the study of individual development. In both cases, a single cell can be elaborated into a complex organism. A complex individual may have 200 different types of cells, such as nerve and muscle cells, each functioning in tissue and organ systems as different as brains and biceps. Each cell (excluding gametes and with unimportant exceptions) contains an identical genotype. Differentiation occurs because different sets of genes function in different cells: the genomes in each cell contain all the information necessary to construct an entire organism, but instead they produce only a particular cell phenotype. Switching of gene activities on and off to obtain products from the appropriate set of genes in the proper amounts and in the correct sequence to make the proper number of cells of each type arrayed in a desirable form requires an elegant regulatory system.

Two main gene functions involve the production of polypeptide products (via structural genes) and control of the pattern of gene activity (via regulatory genes). Results of regulatory gene activity are seen in differentiation among genetically identical cells (Britten and Davidson, 1969). Some genes have both regulatory and structural functions. All eucaryotic cells require a rather similar set of structural genes to produce substances required for cell growth and general cell metabolism. The chief differences between the genomes of morphologically distinctive metazoa seem to reside in the genetic regulatory apparatus. Although some differences in structural functions between members of different phyla are obvious, their distinctive body plans must arise from different patterns of activities of their structural gene; rather than from differences in the structural genes themselves. These Patterns of gene expression are determined by the regulatory gene system (Britten and Davidson, 1971; Wilson, 1975; King and Wilson, 1975 Valentine and Campbell, 1975).

[240] Molecular details of genetic regulation in eucaryotes have not been extensively defined, but there are enough observational data to give some idea of how the regulatory systems operate. In individual organisms, growth and development are mediated by hormones. Some hormones are polypeptides and thus are coded by structural genes, while others are removed from direct gene action along what Stebbins (1968) has called informational relays. Hormones may switch on or off large sets of genes. In some cases, two or more hormones control an array of processes according to their relative concentrations. Matsuda (1978) has reviewed the activity of some hormones in arthropod development. Morphogenesis is triggered by the hormone ED (ecdysone), the effects of which vary according to the availability of a second hormone JH (juvenile hormone). ED induces larval molt in the presence of large amounts of JH, pupal molt with less JH, and imaginal molt in the absence of JH. The inhibiting effect of JH is clear, and it has other well-studied effects. For example, overproduction of JH late in postembryonic development causes precocious egg maturation, resulting in developmental neoteny. Mutations that reduce or stimulate the production of this hormone at unusual levels or times can result in abnormal metamorphosis.

Unusual environmental stimuli can alter the level of hormone secretion and produce abnormalities similar to those resulting from mutations. The morphological consequences of these developmental abnormalities can be gross, ranging from changes in limb size and number and wing reduction or loss to the maturation of larvae with body plans distinct from those of adults. Many examples of regulatory gene abnormalities are known from representatives of other phyla, including humans.

The potential for large and sudden morphological change thus clearly exists within metazoan genomes. The problem is to translate this potential into a rapid evolutionary event that produces descendants quite different from their ancestors but ones that are viable and reasonably well adapted. Genetic drift is a possible mechanism, as we have seen. Genomes suffering extensive changes in developmental pathways to produce "monster" morphologies, however, are likely to produce lethal heterozygotes when crossed with normal genomes; at least this is the experience of laboratory studies (Dobzhansky, 1971). Matsuda (1978) has suggested how this problem may be surmounted. If a population is subjected to a novel environment that induces unusual hormone production, many similar abnormal phenotypes will result. This is a nongenetic origin for a novel morphology. These novel individuals would presumably have no special trouble in reproducing, assuming they happen to be well adapted. The morphology could thus be propagated. Later genetic changes could then fix the new morphology in phylogeny. These genetic changes might be favored by microevolution because they would stabilize a phenotype with a large adaptive value.

There are thus several scenarios to explain the occurrence of sudden large morphological jumps-quantum events-in evolution. One is microevolutionary, when mutations with large morphological effects prove fit enough in heterozygotes that homozygotes, which are fitter still, become common; eventually the mutation is fixed. The case when this seems most likely to occur is neoteny, for the developmental pathways of a mutant and of its normal associates may remain nearly identical up to the point of adventitious reproduction in the mutant. Heterozygotes might therefore develop viably; if mutant homozygotes were fitter they would eventually replace the ancestral type.

The other scenarios are all primarily macroevolutionary in the sense that it is the adaptive value of the novelty that permits it to endure, even though it is less fit than the normal members of its parent species within the parental population. One of these scenarios involves genetic drift in small, reproductively isolated populations. Geographic isolation would be the most common type in invertebrates, and perhaps social isolation (inbred kin groups or harems) would be common among some higher vertebrates. Even though heterozygotes for a mutant would be less fit than the parental genotypes, drift would sometimes permit the appearance of homozygotes by chance. Their subsequent success would depend on their level of adaptation.

A third scenario involves organisms that are self-fertilizing and do not require drift to create homozygotes. For metazoans this scenario must have limited application.

A fourth scenario involves the creation of a new morph by environmental induction of developmental changes, perhaps by the switching of regulatory genes through hormonal activity. Individuals belonging to the new morph should not have unusual genotypes, and if the new morph happens to be well adapted there may be no reduced fertility. Subsequent genetic changes, presumably microevolutionary, are required to fix the novelty.

Whatever the scenario, microevolution and macroevolution play different roles (Stanley, 1975). The more significant novelties, in which the greatest morphological changes are involved, must be a product of a macroevolutionary mode. This should occur most commonly when the population is suddenly faced with a distinctively new environment. Large changes would then be indicated because small ones would not usually suffice to maintain a viable adaptive value. The sorts of genetic changes known to underlie highly distinctive morphologies are those of the regulatory gene system; many structural loci may have their expressions changed by a mutation at a single regulatory locus. The morphological results of regulatory changes will not be scaled to the amount required by environmental change; they could easily be [242] greater than necessary. Regardless of the size of the morphological change the survival of the novelty is first of all a function of its adaptive value when founded. Next it would be a function of the ability of microevolution to maintain the adaptive value at a suitably high level to prevent extinction This may be critical and is the main importance of microevolution. The aura tion of lineages should often be a measure of the success of microevolutionary processes.

As an aid in thinking about the evolution of phyla, it is useful to model the process in terms of a changing adaptive landscape and some simple rules for diversification. Adaptive landscapes have been used for this purpose for many years (see McCoy, 1979, for some early examples); Simpson (1944, 1953) used a model of adaptive zones to excellent effect in considering macroevolution. None of the previous visualizations is quite right for our purposes, however. We shall use a topographically featureless plain demarcated into small irregularly shaped patches. Each patch represents a fairly homogeneous environment. The boundaries between patches represent environmental discontinuities so that conditions change across patch boundaries but are relatively constant between them. There are several classes of boundary strengths; we shall arbitrarily specify that each succeeding increase in boundary strength doubles the environmental change across the boundary. In effect, the plain represents an environmental mosaic, and each patch is an individual tessera.

Through time the environment changes; tesserae enlarge or shrink, or coalesce as boundaries fade. Other boundaries change their positions and intensities, and new boundaries may appear across some tesserae. The environmental mosaic becomes an asymmetric kaleidoscope. Heterogeneous environments have numerous small tesserae separated by a high average boundary class; more homogeneous environments have large tesserae separated by a low average boundary class.

On this sort of kaleidoscopic game board, we shall model the diversification of the animal kingdom. At some location on the board, we position potential descendant lineages) are episodically broadcast into the air to fall back onto the board. The direction in which they travel is selected at random. The distance they travel depends on two factors: their [243] original impetus and the strength of the tessera boundaries they happen to cross. The original impetus may be expressed in boundary units: an impetus of 10 will take a sphere across two class five boundaries, 5 class two boundaries, or some such mixture. The sphere then settles on the last tessera, the airspace of which it can penetrate. The original impetus is selected at random (with replacement) from a pool in which the frequency distribution is previously determined (see below). The spheres are fragile, and there are only certain tesserae scattered randomly across the board on which any given Sphere may settle without bursting. Furthermore, only one sphere can be accommodated on any tessera; late newcomers burst. There is one last rule: each sphere that successfully settles on a tessera becomes a locus for broadcasting new spheres under conditions identical to those at the original locus.

Before starting the game we must choose a pool of impetuses (fig. 3). We shall play three games, each with a different pool. One involves small impetuses (fig. 3(a)); they must be greater than 1, however, or they cannot penetrate any of the tesserae. A second game is played with a pool of very large impetuses (fig. 3(b)). Finally, there is a game from a pool in which most impetuses are small but a very few are large (fig. 3(c)). The spheres are obviously analogous to species. The impetuses are analogous to morphological distances. Clusters of spheres with little morphological distance represent genera or families; spheres that are very different morphologically represent different phyla.

We begin game 1 using pool 1, small impetuses only (fig. 4). By definition, the spheres that successfully colonize tesserae near their source are morphologically similar to their parent sphere. Tesserae distant from the....

---

....original locus can be reached only by the colonization of intervening tesserae since all impetuses are small; the occupied region of the board therefore spreads gradually about the source. Tesserae inhospitable to spheres from the parent locus may accept later ones originating from other loci. When the tesserae near the original parent fill up, the parent can no longer produce viable offspring spheres.

Eventually, the spread of spheres will reach all edges of the board. The game would now be over except that, as the tesserae change kaleidoscopically, some spheres inevitably burst and some new tesserae appear. These unoccupied sites provide fresh opportunity for colonization. A balance is eventually reached between the rate of sphere production and the rate of extinction, establishing an equilibrium population size of spheres or one that oscillates with random accelerations and decelerations in rate of environmental change as the kaleidoscope operates. This recalls the equilibrium species....

....diversity postulated in island biogeographic theory (MacArthur and Wilson, 1967). Morphologically, the entire population of the board is rather similar, since all the spheres are variations on the ancestral theme, with nearly all morphologies connected across small gaps only. A few larger gaps may be created by extinctions. Such an assemblage of morphologies might all be placed within a single phylum, with clusters of similar morphologies (classes, orders, and so forth) defined chiefly by extinction patterns. This is clearly not a close simulation of evolution on Earth.

For game 2 we use only large impetuses (fig. 5). Vast adaptive distances can be crossed rapidly by hopping across the board, many tesserae at a hop, [247] so that the game board would soon contain a scattering of spheres. Gradually, the density of spheres would rise and the frequency of successful colonizations decline. As open tesserae became rare, they would be harder to fill than in the preceding game, for the target circumference of new spheres would be larger with large impetuses than with small. An equilibrium sphere number would eventually be achieved. Morphologically, a given sphere would be very different from its parents and even more so from its neighbors, especially from those originating in sources far from the source of the given sphere. In the extreme case, such an assemblage might have to be broken up into as many phyla as there are spheres. Obviously, this does not approximate evolution on Earth either.

Finally, we use the pool of impetuses that is strongly skewed so that most are small, a few are of intermediate strength, and a rare few are large (fig. 6). As play begins, the original source morphology begins to spread as in game 1, but a few different morphologies occasionally appear on outlying tesserae; in some rare cases, descendants from distant tesserae may appear back near the original source. Each very distinctive morphology becomes surrounded by a spreading zone occupied by spheres with minor modifications, making it increasingly difficult for the rare spheres from distant sources to find a suitable open tessera. Eventually, the board fills to an equilibrium occupation level. As vacant tesserae appear, their chances of being colonized from a distant source are vanishingly small; they might rarely be occupied from a source at an intermediate distance, but will most likely be filled from a closely neighboring source. To increase the colonization chances from a far or intermediate source, a large area of the board comprising many tesserae must be cleared of spheres. Then, although spreading from marginal tesserae will gradually fill the cleared region from the edges, the central part might well receive immigrant spheres from some distance. The larger the cleared patch, the more likely this would be.

Morphologically, the spheres could be divided hierarchically, first into clusters of spheres descended from the founding spheres with large impetuses; then into subclusters descended from separate intermediate impetus spheres, daughters or grant/daughters of the founding sphere; and then perhaps into tertiary clusters within these and quaternary clusters Within those and so on. Clearly, among the games we have played, this most closely represents the hierarchical pattern of morphology on Earth.

Game 3 could be modified in a number of obvious ways to bring it into closer conformity with life history. For example, the density of tesserae that Could be successfully colonized might be decreased with distance from each source tessera, or some spheres could be permitted to occupy more than one tessera at certain times or places, and so forth. However, the game rules already rival those of Monopoly, and such complications do little to clarify the basic questions of the diversification pattern of higher taxa.

The adaptive kaleidoscope involves features that may not closely simulate life history. For example, the morphogenetic events analogous to the large impetuses of the model may not often or ever result from single mutations. They remain the most mysterious parts of metazoan history. Also the model implies that morphological and adaptive distances go hand in hand; this is clearly not the case. Nevertheless, the model does contain some features similar to those indicated by the fossil record, and thus it suggests interesting hypotheses.

The most striking aspect of game 3 is the large role played by chance in determining the morphological pattern of the phyla (and of lower taxa as well). A given tessera, or a given region of tesserae that has low-class boundaries internally but is demarcated by higher class boundaries, may be colonized by a descendant of any number of ancestral spheres; that is, it may support any number of body plans. The potential body architectures that can function in any given habitat on Earth are clearly very great. In the game, the first arrival that does not burst takes possession of a tessera. If the source tessera is somewhat remote, then the immigrant sphere must be rather novel. Empty nearby tesserae are most likely to be colonized by descendants of the novelty since minor novelties are so common. Therefore, the chance success of a major novelty will commonly result in the occupation by spreading of a large area of the gameboard by organisms with a similar body plan. Raup and Gould (1974) have demonstrated that patterns of random spreading of morphologies in situations much like this game result in quite plausible family trees. If the area had been colonized from a very different source, the major novelty of the founding lineage would have been very different, and the subsequent inhabitants of the area, deployed by spreading from this major novelty, would be quite different.

Thus, in the game, the key to body-type pattern is the chance of colonization, which is based on two factors: the arrival of an immigrant from a distance and the viability of the immigrant on the tessera. In nature, the latter is chiefly a matter of chance preadaptation, but microevolutionary processes, operating after colonization, may contribute significantly to continuing success of the immigrant. The origin of the immigrant morphology is not related to the conditions at what becomes the occupied tessera. In addition to the morphological features that prove useful in the tessera, the immigrant may possess many other features that have no special relation to the particular conditions that distinguish the tessera or the neighborhood of the tessera from other places. These features may nevertheless contribute the harmonious functioning of the organism at the time of colonization and may be retained or perhaps modified as the immigrant is engineered to conditions on the tessera by microevolution.

[249] From the foregoing, it appears that prediction from first principles of the morphology of the occupant of any given tessera would not be possible even if one had a knowledge of the pool of morphologies from which it must arise (assuming a reasonably large diversity of morphology within the pool) consider a neighborhood of tesserae that represents infaunal burrowing in marine bottom sediments. Among living phyla, a wide variety of body plans have burrowing representatives. Most use a hydraulic system (usually associated with a hydrostatic skeleton) to burrow. Some Cnidaria (sea anemones) burrow by using their gut as a hydraulic skeleton. Some phyla burrow by puncturing the substrate with probelike organs, pulling themselves after (some sipunculids use probosci operated by coeloms; some bivalve mollusks use a foot operated by the haemocoel). Other phyla use peristaltic waves (some nematodes use a pseudo-coal; some annelids use a coelom). In some phyla, such as arthropods, burrowers use appendages, and still other methods exist.

Some of the morphologies associated with burrowing are elegantly engineered as burrowing machinery; annelids are a fine example (Clark 1964). Others seem to be jury-rigged, with improvised structures and functions tacked onto basic morphologies adapted to some different habit altogether. The burrowing inarticulate brachiopods certainly convey this impression; although equipped with elongate posterior pedicles containing a coelomic cavity, they burrow anteriorly by a scissorslike motion of their shells, aided by setae for transporting sediment particles (Thayer and Steel-Petrovic, 1975). This is hardly an elegant approach to burrowing, yet the fossil record indicates that these forms developed their burrowing behavior early in the Phanerozoic and have maintained it for several hundred million years. A theoretical morphologist might predict that an annelidan sort of burrowing could evolve, but he would hardly predict the inarticulate system There are several phyla that do not and presumably have not burrowed in any significant way (Chaetognatha and Bryozoa, for example). However, it is not safe to conclude that their body plans preclude a burrowing habit.

The convergence of distinctive morphologies on similar functions is not unique to burrowing functions; indeed it is a widespread phenomenon among living phyla. A single subtidal rock may support sessile suspension feeders belonging to several phyla and feeding on roughly similar resources. For example, Mollusca (mussels), Brachiopoda (articulates), Bryozoa Arthropoda (barnacles), and Echinoderma (crinoids) have representatives with such a life mode. Of these, bryozoans and articulate brachiopods are elegantly adapted to this habit, as are crinoids; mussels were nicely pre-adapted, while barnacles have passed along a tortuous adaptive pathway. One would hardly predict a barnacle as an outcome of arthropod radiation, even [250] though they are perfectly plausible. The potential modes of life available to primitive arthropods, given modifications extensive enough to develop barnacles, are so broad as to preclude the prediction of any special one that might be realized.

Clearly, many important functions can be realized by animals with any of a wide array of body plans. The model of game 1, filling the board gradually in small morphological steps so that the entire planetary fauna has in the end the same basic body plan, seems plausible. It requires an evolutionary process that does not produce large viable morphological jumps. The modification of a body plan to permit life in most tesserae would seem to present no special problem, although, of course, there may be internal constraints in some cases.

The model of game 2, filling the adaptive landscape with forms that diverge by large morphological jumps only, seems plausible also. There is certainly no unique fit of body plan to tessera, and if there are general limits on modification to body plans, they have not yet been defined.

The number of major ground plans realized on Earth as viable animals is not known. A reasonable guess might be between 40 and a few hundred on the level of the phylum. An important regulator of this number is obviously the frequency distribution of the distinctiveness of morphological novelties. The number of phyla we have had reflects this distribution as it existed in Late Precambrian and Lower Cambrian times. Another possible regulator of phylum number is the rapidity and extent of environmental change, measured against the effectiveness of microevolution in maintaining adaptation. If environmental change is so rapid as to cause large-scale extinctions, then large areas of the adaptive landscape may become available for reoccupation. A phylum with low diversity may become extinct. The probability of appearance of new phyla in the vacated tesserae then depends on the size of the vacant area and the frequency distribution of novel morphological distance. On Earth we have lost a number of phyla to extinction. New classes or, more commonly, new orders have appeared, presumably filling tesserae vacated by extinction, but it seems that new phyla have rarely, if ever, been evolved after the Cambrian. This does not necessarily indicate that large regions of the adaptive landscape were never emptied by extinction. It may mean, in part, that lineages from which major new body plans could be given off became swamped by an increasing number of lineages that could invade open regions of the adaptive landscape with less significant modification, recognized not as phyla but as lower taxa (Valentine, 1973). Even the invasion of the terrestrial environment failed to produce a novel phylum. The probability of a new phylum arising probably becomes lower as the old phyla diversify.

From the adaptive kaleidoscope game, one can gain insight into general features of the diversification of multicellular animal-like organisms. One approach is to imagine multiple runs of each game. Figure 7 depicts typical life trees resulting from each game. Reruns would resemble each other more.....

---

[252] ....closely in game 1 (fig. 7(a)) than in the others. First, the morphology of the organisms, arising gradually from a similar unicellular ancestor and diverging gradually as descendant lineages spread outward from the ancestral tessera would not diverge as much from the simple multicellular forms. Second although details would vary, the usual pattern of morphological variation would be similar: irregularly concentric with respect to the ancestor. More extreme cases would arise when descendants happened to be given off chiefly in one direction, producing a chain of morphologies extending from the founding ancestor and snaking across the gameboard. The end member of the chain would be more different from the founding member than in other spreading patterns; if extinction eliminated segments of the chain, one might recognize two or more phyla instead of just one.

In game 2 (fig. 7(b)), with only major morphological jumps, the pattern of morphological variation would also tend to be similar from run to run with great heterogeneity among morphologies in any region but with the average degree of divergence increasing away from the founding tessera. Extreme spreading patterns-for example, radial until the board margin is reached, then counterradial back to the inner tesserae, then radial again, and so on -would create different patterns of divergence, but these would be rare. A richness of morphology would appear in all games. Occasional close similarities of body plans in two different games would not be surprising, but the arrays of morphologies in each game might be quite distinctive. It is likely that in separate games the body plans of the occupants of any particular tessera would be derived from different ancestral body plans, even though they would display similar functions. Similarities in results from reruns of game 2 would be greater than expected if strong constraints were placed on the colonization pattern of a given lineage by its morphology. Some such constraints must exist, but whether they would force a great convergence among successive runs is uncertain. This is an important problem.

Game 3 (fig. 7(c)) would also display a similarity of patterns from run to run: a hierarchy of morphologies would appear, varying with the frequency of large vs small branching events (and capable of some extreme configurations). The actual body plans representing the major morphological branches would usually differ greatly from game to game. The major branches that arise earliest would be more likely than later ones to resemble each other in repeated games. The rate of major branching in time should resemble the classic S-shaped logistic curve, but it should be compressed into short time interval early in the game (see Sepkoyski, 1978). In this respect, early game 3 runs resemble game 2 runs. When the tesserae finally fill up, so hat those opened by extinction are chiefly filled from neighboring tesserae ,game 3 runs resemble game 1 runs. The occasional large patches of tesserae cleared by mass extinction permit game 3 to develop its characteristic family tree at lower taxonomic levels.

[253] The richness of variation possible within such trees is indicated by the clade simulations of Raup et al. (1973), in which splitting, continuation, and termination of lineages are determined as random choices but at fixed rates. The patterns of growth and diminution of diversity within these simulated clades resemble patterns documented by fossils in actual clades. Here we are interested in the pattern of distinctness among the branches. The characteristic tree of game 3 (fig. 7(c)) is most like the pattern documented by fossils, and it would display a richness in detailed variation in reruns.

In the kaleidoscope games, the quality of macroevolution is the most important factor in structuring the morphological variation. It is probably a safe assumption that macroevolution is a universal feature of life. Macroevolutionary factors should affect multicellular diversification patterns on other worlds, much as they do on Earth in one or another kaleidoscope game, although the results would, of course, vary with environmental conditions. Microevolution, on the other hand, might be an entirely different matter elsewhere. It would certainly be different in detail. We could not expect the same enzyme systems and probably not the same genetic materials or cell behaviors as we find on Earth. Meiosis as we know it would not be likely to develop in another life system. Earthly microevolution is closely tied to the planet's particular genetic system.

The pattern of morphological variation of animal-like organisms other worlds can run the gamut from the most homogeneous sort resulting from game 1 to the most heterogeneous and chaotic sort resulting from game 2. The character of the morphologies is not strictly predictable. Certain qualities may be likely. Cellular differentiation would seem to be a prerequisite of any true complexity on the multicellular level. Complexity is a property that usually develops when evolving organisms are faced with a series of environmental challenges. It is often easier to modify by adding something to an established structure, thus producing a complex elaboration, than by redesigning the structure from a simpler developmental stage. Also, given a morphology, there are more potential variants of it that are complex than that are simple.

Wherever we find it, complexity seems always to be organized hierarchically (Simon, 1962). Therefore we might expect complex morphologies to be based on a hierarchical organization, with cells or other such modules at the base, and higher levels analogous to tissues, organs, and organ systems. Adaptations for maintaining a spatial place in nature-locomotory, sessile, or floating adaptations-have shaped the body plans of most of Earth' s phyla, and presumably it would be much the same elsewhere. Nearly all such body plans require a contractile system; movement is essential.

[254] Although it is amusing and not too difficult to continue in this vein, listing the attributes required of presentable multicellular organisms, it turns out that one is simply tabulating the basic features of multicellular animals here on Earth. They form a logical and elegant assemblage that excites admiration for evolution and gives aesthetic pleasure in its results. One can only hope that the generality of their features can someday be tested against another life system.

I thank Dr. S. Conway Moins, Open University, for a discussion of the manuscript and for data on the fossil record of certain phyla, and Dr. C. A. Campbell for reviewing the manuscript.

- Ayala, F. J.: An Evolutionary Dilemma; Fitness of Genotypes Versus Fitness of Populations. Canadian J. Genet. Cytol., vol. 11,1969, pp. 439-456.

- Borovikov, L. I.: First Find of Fossil Dickinsonias in Lower Cambrian Sediments in the USSR. Dokl. Akad. Nauk. SSSR, vol. 231, 1976, pp.1182-1184.

- Britten, Roy J.; and Davidson, Eric H.: Gene Regulation for Higher Cells: A Theory. Science, vol.165, July 1969, pp. 349-357.

- Britten, Roy J.; and Davidson, Eric H.: Repetitive and Non-Repetitive DNA Sequences and a Speculation on the Origins of Evolutionary Novelty. Quarterly Rev. Biol., vol. 46,1971, pp.111 -133.

- Bush, G. L.: Modes of Animal Speciation. Ann. Rev. Ecology Systematics, vol. 6,1975, pp. 339-364.

- Bush, G. L.; Case, S. M.; Wilson, A. C.; and Patton, J. L.: Rapid Speciation and Chromosomal Evolution in Mammals. Proc. Natl. Acad. Sci. USA, vol. 74, Sept.1977, pp. 3942-3946.

- Clark, Robert B.: Dynamics in Metazoan Evolution; The Origin of the Coelom and Segments. Clarendon Press, Oxford, 1964.

[255] - Cloud, P. E., Jr.: Some Problems and Patterns Exemplified by Fossil Invertebrates Evolution, vol. 2, no. 4,1948, pp. 351-374.

- Conway Moins, Simon; and Whittington, H. B.: The Animals of the Burgess Shale. Sci. American, vol. 241, no. 1, July 1979, pp. 122-133.

- Crimes, T. P.: Colonisation of the Early Ocean Floor. Nature, vol. 248, March 22,1974, pp. 328-330.

- Dobzhansky, Theodosius: Genetics of the Evolutionary Process. Columbia Univ. Press, N.Y.,1971.

- Dobzhansky, Theodosius; Ayala, F. J.; Stebbins, G. L.; and Valentine, J. W.: Evolution. W. H. Freeman, San Francisco, 1977.

- Durham, J. W.: The Fossil Record and the Origin of the Deuterostomia. Proceedings North American Paleontol. Convention, Chicago, I L, Allen Press, Lawrence, Kansas,1971, pp. 1104-1131.

- EIdredge, N.; and Gould, S. J.: Punctuated Equilibria: An Alternative to Phyletic Gradualism. In: Models in Paleobiology, Thomas J. M. Schopf, ed., W. H. Freeman, San Francisco, 1972, pp. 82-115.

- Glaessner, Martin F.: Geographic Distribution and Time Range of the Ediacara Precambrian Fauna. Bull. Geol. Soc. Am., vol. 82, Feb. 1971, pp.509-514.

- Hyman, L. H.: The Invertebrates: Protozoa Through Ctenophora. McGraw-Hill, New York, 1940.

- King, Mary-Claire; and Wilson, A. C.: Evolution at Two Levels in Humans and Chimpanzees. Science, vol. 188, April 1975, pp. 107-116

- MacArthur, R. H.; and Wilson, E. O.: The Theory of Island Biogeography, Princeton Univ. Press,1967.

- Matsuda, R.: Abnormal Metamorphosis and Arthropod Evolution. In: Arthropod Phylogeny, A. D. Gupta, ed., Van Nostrand, N.Y., 1978, pp.137-256.

- Matthews, S. C.; and Missarzhevsky, V. V.: Small Shelly Fossils of Late Precambrian and Early Cambrian Age: Review of Recent Work. J. Geol. Soc. London, vol. 131, May 1975, pp. 289-304.

[256] - Mayr, Ernst: Systematics and the Origin of Species. Columbia Univ. Press N.Y., 1942.

- Mayr, Ernst: Animal Species and Evolution. Harvard Univ. Press, 1963

- McCoy, J. W.: The Origin of the "Adaptive Landscape" Concept. Am Naturalist, vol.113, 1979, pp. 610-613.

- Raup, D. M.; and Gould, S. J.: Stochastic Simulation and Evolution of Morphology - Towards a Nomothetic Paleontology. Syst. Zool. vol. 23,1974, pp. 305-322.

- Raup, David M.; Gould, Stephen J.; Schopf, Thomas J. M.; and Simberloff Daniel S.: Stochastic Models of Phylogeny and the Evolution of Diversity. J. Geol., vol. 81, Sept. 1973, pp. 525-542.

- Repetski, John E.: A Fish from the Upper Cambrian of North America. Science, vol. 200, May 5,1978, pp. 529-531.

- Richardson, Eugene S., Jr.: Worm-like Fossil from the Pennsylvanian of Illinois. Science, vol. 151, Jan. 7, 1966, pp. 75-76.

- Sepkoyski, J. J., Jr.: A Kinetic Model of Phanerozoic Taxonomic Diversity. I. Analysis of Marine Orders. Paleobiology, vol. 4, 1978, pp. 223-251.

- Simon, H. A.: The Architecture of Complexity. Proc. Am. Phil. Soc., vol. 106, 1962, pp. 467-482.

- Simpson, George, G.: Tempo and Mode in Evolution. Columbia Univ. Press, 1944.

- Simpson, George G.: The Major Features of Evolution. Columbia Univ. Press, 1953.

- Stanley, S. M.: A Theory of Evolution Above the Species Level. Proc. Natl. Acad. Sci. USA, vol. 72, Feb. 1975, pp. 646-650.

- Stanley, Steven M.: Stability of Species in Geologic Time. Science, vol.192, April 1976, pp. 267-269.

- Stanley, Steven M.: Chronospecies' Longevities, the Origin of General and the Punctuational Model of Evolution. Paleobiology, vol. 4, 1978, pp. 26-40.

[257] - Stebbins, G. L.: Integration of Development and Evolutionary Progress. In: Population Biology and Evolution, R. C. Lewontin, ed., Syracuse Univ. Press,1968, pp.17-36.

- Thayer, Charles W.; and Steel-Petrovic, H. Miriam: Burrowing of the Lingulid Brachiopod Glottidia pyramidata: Its Ecologic and Paleoecologic Significance. Lethaia, vol. 8,1975, pp. 209-221.

- Tiegs, O. W.; and Manton, S. M.: The Evolution of the Arthropoda. Biol. Rev., vol. 33,1958, pp. 255-337.

- Valentine, James W.: Evolutionary Paleoecology of the Marine Biosphere. Prentice-Hall, Englewood Cliffs, N.J., 1973.

- Valentine, James W.: General Patterns of Metazoan Evolution. In: Patterns of Evolution as Illustrated by the Fossil Record, A. Hallam, ed., Elsevier, Amsterdam, 1977.

- Valentine, James W.; and Campbell, Cathryn A.: Genetic Regulation and the Fossil Record. Am. Sci., vol. 63, Nov.-Dec.1975, pp. 673-680.

- Wilson, A. C.: Evolutionary Importance of Gene Regulation. Stadler Symp., vol. 7, 1975, pp.117-133.

---