Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
AMER. ZOOL., 32:106-112 (1992) Macroevolution: The Morphological Problem1 KEITH STEWART THOMSON Academy of Natural Sciences, Philadelphia, Pennsylvania 19103 SYNOPSIS. The central question of macroevolution concerns the evolution of major morphological innovations (and major taxonomic groups). It is a matter of scale rather than simply rate of evolution or hierarchical level of mechanism. Through the history of microevolutionary theory there is a constant counterpoint of macroevolutionary questioning: are current versions of microevolution sufficient to explain the data concerning origins of major novelties? Thus, Simpson proposed the term "megaevolution" and "quantum evolution." Mayr partially separated micro- and gradual aspects by proposing "genetic revolution" as a mechanism for rapid change. Evolutionary theory in general suffers because different concepts have incompatible frames of reference. Key innovation and correlated progression are concepts that approach the problem in terms of evolutionary morphology. They have in common with quantum evolution and genetic revolution features such as threshold effect, preadaptation and smooth transition due to change in function and environment. One aspect of paleontology and morphology, however, is to show that some morphologies can only exist in binary states, with no intermediates. This leads us to look away from selection on graded series of phenotypes to development (developmental cascade/threshold models) for new levels of explanation. I have been asked to discuss two interesting concepts of evolutionary morphology—key innovation and correlated progression—in the context of macroevolution and development. The term macroevolution evidently means many things to many people (often at the same time). I believe that the key macroevolutionary question is not rate of evolution but scale of evolution. Macroevolution is a discussion of the origin of major evolutionary (principally morphological) novelties and therefore the origins of major taxonomic groups. It is not simply a question of the hierarchical level of evolutionary process—for example species selection for macroevolution and population genetics for microevolution. It is, instead, about the scale of evolutionary change that can be produced at any level. This wall be the primary focus of the present paper. While the origins of major morphological novelties remain unsolved, one can also view the stubborn persistence of macroevolutionary questioning, and particularly its 1 From the Symposium on Development and Macroevolution sponsored by the Division of the History and Philosophy of Biology of the American Society of Zoologists and presented at the Annual Meeting of the American Society of Zoologists, 27-30 December 1990, at San Antonio, Texas. revival in recent years, as a challenge to orthodoxy: resistance to the view that the synthetic theory tells us everything that we need to know about evolutionary processes? Similarly, the "success" of particular viewpoints has also depended on reaction against another (heroes require villains). Therefore, one can trace out in the history of evolutionary theory a sort of dialectic between micro- and macro- questions and answers while at the same time the scale of questions has changed. The sub-theme of the paper is that throughout the history of the subject people have argued about matters that correctly belong to different frames of reference. They then talk "past each other" rather than together. Starting with Darwin, all evolutionary problems were redefined in terms of the word gradual and it is this word that continually creates a problem. Within a "gradual" context, theories of mechanisms tend to come and go, depending in large part on what subject is currently most accessible to research (i.e., giving decent hard data), especially through the sine qua non of experimentation. For a hundred and fifty years we have seen paleontology, morphology, genetics, and development wax and wane as dominant evolutionary disciplines (the custodians of the answers, as it were), but largely with no effect on the basic macro- 106 THE MORPHOLOGICAL PROBLEM evolution problem of how major events that appear to happen very quickly can also be gradual. In this discussion I will separately discuss three aspects of the problem as it developed in Britain and the United States, showing the ways in which different approaches reached their limits, and how they have converged upon a new kind of solution. (Unfortunately, space does not permit a thorough discussion of antecedent intellectual traditions, for example in the Soviet Union, Adams, 1980.) 107 mechanisms of a macroevolution, which require time on a geological scale." However " . . . we are compelled at the present level of knowledge reluctantly to put a sign of equality between the mechanisms of macro- and micro-evolution, and proceeding on this assumption, to push our investigations as far ahead as this working hypothesis will permit." However, after this date we see the wagons formed into a circle at the center of which is an extreme reductionist gradualism, as exemplified by Fisher's famous microscope metaphor—his reductio ad absurdum in which virtually any variant is THE GRADUALIST'S DILEMMA hopeless (Thomson, 1988a). In this view The basic article of faith of a gradualist the organism-environment interaction is a approach is that major morphological inno- passive filter, whereas, in principle, for any vations can be produced without some sort individual there might somewhere be a of saltation. But the dilemma of the New functioning match—in large part because Synthesis is that no one has satisfactorily each individual creates as well as receives demonstrated a mechanism at the popula- its own niche. tion genetic level by which innumerable very In the face of the strength of the explansmall phenotypic changes could accumulate atory power of the New Synthesis (to say rapidly to produce large changes: a process nothing of the political power of its expofor the origin of the magnificently improb- nents), paleontologists and morphologists able from the ineffably trivial. This leads to had by 1950 largely reframed their work in skepticism about the microevolutionary microevolutionary terms. The watershed approach. Perhaps, as Waddington (1967) event was the coming together of disciplines put it: "the whole real guts of evolution— at the 1948 conference at Princeton (Jepson which is, how do you come to have horses et al, 1949). In this new unity the work of and tigers, and things—is outside the math- Simpson, especially his 1944 Tempo and Mode In Evolution, was obviously critical. ematical theory." In looking back over the literature of the Simpson applies many of the quantitative last 60 years, it is fascinating that through- approaches of the population geneticists to out the whole grand development of the New the fossil record. For example there is his Synthetic theory, the macroevolutionary classic figure superimposing survivorship question remains as a constant undercur- curves of fossil pelecypod and carnivore rent. It persisted even after the mutationism genera and Drosophila individuals. But at of de Vries, Bateson and others had been the same time, Simpson is still obviously thoroughly stamped out: "Bateson . . . was testing the premises of Dobzhansky's unprepared to recognize the mathematical "working hypothesis." He not only distinor statistical aspects of biology, and from guishes microevolution ("changes within this and other causes he was not only inca- potentially continuous populations" and pable of framing an evolutionary theory macroevolution ("the rise and divergence himself, but entirely failed to see how Men- of discontinuous groups"), but also defines megaevolution which is not merely an etydelism supplied the missing parts of the mologically superior version of macrostructure first erected by Darwin" (Fisher, evolution (as many have supposed) but a 1929). phenomenon of "major systematic disconIn one of the most influential books of tinuities in the record." It is, however, never the New Synthetic approach as it evolved clear from Simpson's rambling discussion in Britain and the United States, Dobzhan- of megaevolution whether he is talking about sky (1937) was quite circumspect about "the 108 KEITH STEWART THOMSON a pattern to be explained or a special process that produces discontinuous patterns. Simpson explores the subject of rates of evolution in searching for a way in which evolution could, as it were, "change gears" while still remaining a gradual process in the analysis of rates of evolution. But he still could not completely surrender to the view that all change must be "micro." Finally in his Major Features of Evolution (1953) he declares that "all three monstrous terminological innovations (micro-, macro-, and mega-evolution) have served their purpose . . . clarity might now be improved by abandoning t h e m . . . it is better to recognize that. . . there are innumerable levels of evolution." Here Simpson not only rejects an exclusively "micro" theory, which he refers to as "a myopic outlook"; in invoking different levels of mechanism, he reverts to another view of Dobzhansky's. In 1937 Dobzhansky suggested that evolution operates at three levels: "gene changes . . . the dynamic regularities of the physiology of populations . . . and the realm of fixation of the diversity already attained on the preceding two levels." This is the predecessor of the modern hierarchical approach of Vrba and Eldredge (1984), Salthe (1986) and others. Paradoxically however, Simpson embraces the term quantum evolution, which seems to be nothing but our old friend macroevolution in disguise—"a particular set of evolutionary events.. . that involves an allor-none reaction. They are changes of adaptive zone such that transitional forms between the old zone and the new cannot, or at any rate do not, persist . . . (this is) a special, more or less extreme and limiting case of phyletic evolution" (1953; cf. Simpson, 1944). By this time, of course, Goldschmidt had set himself up as the new "mutationist to be against," supplanting de Vries and Bateson. It is important to note Goldschmidt's definitions of micro- and macroevolution. For him, the former is entirely an intraspecific set of processes and macroevolution starts with the origin of new species. In the process of rejecting Goldschmidt, microevolutionists closed ranks against any broader viewpoint. Goldschmidt's The Material Basis of Evolution (1940) is, however, more interesting than its usual caricature. More revealing than "hopeful monsters" per se is his summary of the mechanism by which he thought "species and the higher categories originate in single macroevolutionary steps as completely new genetic systems." Goldschmidt's mechanism depends on chromosomal rearrangement producing a new "genetic system which may evolve by successive steps of repatterning until a threshold for changed action is reached, (this) produces a change in development which is termed a systemic mutation. Thus selection is at once provided with the material for quick macroevolution. The facts of development, especially those furnished by experimental embryology, show that the potentialities, the mechanics of development, permit huge changes to take place in a single step" (1940). The developmental evidence on which he concentrates includes the now very fashionable homeotic mutations in insects. Goldschmidt was trying to place the question in a different framework. He failed, but that does not mean the attempt itself was invalid. Of all the opponents that Goldschmidt summoned up, none was more virulent than Mayr. His Animal Species and Evolution (1963) is the apotheosis of the New Synthesis and yet even here there is a curious difficulty in dealing with the small remaining resistant germ of macroevolution. Mayr, like Simpson, proposes a "micro" exception to "gradual," this time as genetic revolution. The language sounds familiar: although set in terms of population genetics, it is the same language as Simpson used in the paleontological/morphological context. Thus, under particular circumstances of population biology, "the mere change of the genetic environment may change the selective value of a gene very considerably . . . the most drastic genetic change (except for polyploidy and hybridization) which may occur in natural populations, since it may affect all loci at once. Indeed, it may well have the character of a veritable 'genetic revolution' . . . may well have the character of a chain reaction . . . until finally the system has reached a new state of equilibrium" (1963). THE MORPHOLOGICAL PROBLEM The punctuated equilibrium model of Eldredge and Gould (1972) was an attempt to trace out the effect of Mayrian genetic revolution in peripheral populations at the finest scale resolvable (rarely) in the fossil record. Unsuspectedly, this then set off a renewed interest in matters such as "species selection" that fall outside the generally defined mainstream of the New Synthesis, much to the distress of Mayr and the skepticism of J. M. Smith, and reopened the question of different levels of evolutionary mechanism. In hindsight, the obvious flaw of the otherwise superb population genetics of the New Synthesis is that it failed to take into account the fact that morphologies have their origins not only in evolutionary lineages but in developmental pathways (see later). The other, less obvious, drawback is that hypotheses expressed in the framework of quantitative genetics of populations produced predictions that are difficult to test at the morphological, functional, or developmental level. This does not mean that questions posed at those levels are not valid. However, it should cause us to question whether "micro" and "macro," as applied to evolution, really are opposing or alternative terms, and therefore whether one could ever be explainable in terms of the other. Both might be explainable in terms of something else. 109 seemed to be through functional morphology. Experiments with living organisms can illuminate the evolutionary histories of fossil forms, for example through analysis of feeding mechanics, thence to adaptation and from there to ecology and selection, all changing over broad sweeps of evolutionary time. Surely analysis of changes in function, which should be smoothly continuous, would illuminate such questions as the origin of tetrapods, birds, or mammals. In modern approaches to functional morphology we see explicit attempts to translate the concepts of Simpson, Goldschmidt and Mayr into mechanisms (scenarios) of major morphological (macro) change. The concept of the "key innovation" (Bock, 1965) seems to have arisen as the functional morphologists' answer to an interesting aspect of the gradual origin of major morphological novelties—the question of the adaptiveness of intermediates. Although Darwin in Chapter 6 of the Origin deals squarely with the question of the adaptiveness of "intermediate" stages (for example in the evolution of the vertebrate eye), the question is constantly raised: How can the initial phenotypes of quantum evolution or genetic revolution be adaptive? This is an obvious legacy of Fisher's extreme microevolutionary approach. Miller (1949) and Simpson (1953) proposed that morphological transition might be facilitated if some "key" change occurs THE PALEONTOLOGIST'S DILEMMA that unlocks a suite of following changes. While the gradualist cannot show how you The key innovation need only be small, if get there from here, paleontologists and it is (in hindsight) the right one, but it could evolutionary morphologists have been the open up the possibility of exploitation of a principal custodians of the view that there new habitat. That would largely take care might be a "there" to get to. The two-edged of the "gradual" requirement of any model. problem paleontologists have always had is Bock (1965) added to this the essential (key?) their data (the very rapid appearance of ingredient of preadaptation. According to morphological novelties and major groups) this view, the breakthrough event is not the reveal phenomena that appear to be quan- change in phenotype per se, but a subtle shift titatively different from those observed on in function as Darwin himself proposed. the ecological time scale but at the same Then the change would be accomplished time their research methods and data do not without genetic-developmental/phenotypic fit the experimental scale of population biol- disruption and without loss of adaptation. ogy. They cannot fully test, therefore, Morphological change could be accumuwhether there are quantitative, let alone lated slowly either side of the functional shift. qualitative, differences between underlying An example of such a shift would be the evolution of the feather from a modified mechanisms. One possible way out of this dilemma has scale, the shift in function of a fish airblad- 110 KEITH STEWART THOMSON der to the tetrapod lung, or the different use of the lobed fin of an osteolepiform fish in water compared with the same structure on land in a tetrapod. Bock (1965) proposed the key innovation model in order to account for the origin of major groups. It is a model in which major morphological/taxonomic novelty arises through a succession of key (preadapted) innovations, each followed by a "period of postadaptational adjustment associated with a minor adaptive radiation." The key innovation model works very well to describe adaptive radiations at low taxonomic levels. An excellent example appears to be Liem's analysis of cichlid fish radiations around a subtle change in pharyngeal jaw mechanics. However, a crucial postulate of key innovation in explaining the origin of a major group (for example the wholesale revision of morphology between tetrapods and fishes) is the coincidence not only of a whole host of preadaptive features in the ancestral group that change one after the other in rapid succession, but also of successive diversifications (adaptive radiations) of groups showing "mosaic" features at each intermediate stage. The fossil record may show the former, but not the latter. In fact, the record often shows the opposite for any major transition—very low taxonomic diversity. The "correlated progression" model (Thomson, 1966) is the complement and in many ways the opposite of "key innovation." Key innovation isolates phenotypic features and proposes that they change independently and serially. Correlated progression starts with the fact that features may be functionally connected and proposes that if any one changes, all the others will be affected and may (note, not all must) change also at the same time. The first example was taken from the origin of the tetrapod skull and ear. It proposed that a group of features that are mechanically linked—jaw mechanics, hyoid arch, middle ear, opercular system—responded in concert to a single environmental context. The result is a positive feedback that reinforces change. Preadaption is still important in preserving continuity of adaptiveness. An essential part of the model is that change in any part of the functional complex without change in the others would bar, rather than facilitate, further change of the sort that (in hindsight) produces the major new group. For example if the respiratory mechanism were changed alone, leading to loss of the hyomandibular, then that element would not be available for future transformation into a stapes. Such independent changes would occur in some lineages, of course, leading to a mosaic effect in the record of diversity. In this case the mosaic lineages are dead ends rather than stages in a trend. Both models have the advantage that change is integrated by external environmental factors and adaptiveness is ensured. Neither requires any sort of hopeful monster. Neither key innovation nor correlated progression is strictly a model of evolutionary mechanism. They are models of patterns of evolutionary change in morphology and function. They are useful as interpretations of the fossil record and produce predictions that can be tested against fossil data, for example in predicting the coupling (key innovation) or uncoupling (correlated progression) of taxonomic and morphologic rates. To the extent that they are corroborated or falsified, they then challenge us to find parallel underlying patterns, and eventually mechanisms, in development and genetics. But they belong in a different intellectual framework from models of population genetics. They share with the scenarios of Simpson or Mayr (and Goldschmidt) the notion of threshold, preadaptation, and smooth transition based on change in environment and function. There is, however, a development in paleontology that challenges both models and in a most productive way. Because of the lack of fine temporal resolution in most of the fossil record, it is difficult to know just how quickly most changes in phenotype actually have occurred. But it is beginning to be clear that some "major" phenotypic conditions can only exist in binary states—with no graded intermediate stages at all. Perhaps the best example is in the ankle of some reptiles, where the astragalus or both astragalus and calcaneum have moved from the distal to the proximal side of the ankle joint. No intermediate states are possible. Each bone can only exist on one side of the joint THE MORPHOLOGICAL PROBLEM or the other. And the shift is not a simple one because it is accompanied by matching changes in ligaments and muscle insertions to maintain a functioning joint. Once the shift is made it becomes the basis of a broad diversification (in dinosaurs and birds, for example). The sort of muscle and joint changes that Liem envisages for cichlid "innovation" must involve the same sort of binary state. At one level, such a binary shift in bone arrangement could be considered a "key innovation" because it then allows new adaptive radiations in limb morphology. But it is also an example of correlated progression because functionally it must involve not merely bone arrangement but the simultaneous repatterning of ligaments and muscles. Numerous (now comical-seeming) scenarios have been proposed to explain the situation away in terms of the astralagus slowly becoming vestigial until it can slip unnoticed to the other side of the joint and then swell back up again. Of course, the truth is that such a gradual-morphological explanation is bound to fail because it is set in the wrong terms. The transitional states never existed as adult phenotypes in large (or small) populations acted upon by strong selection for different running mechanisms acting on astragalus size. The transition occurred at an early stage of development and required no intermediates at all, just a single mistake (probably of timing) in the maturation of various blastemata. Here is a case where intermediates would all have been selected against. Is this a case of hopeful monsterism? THE EMBRYOLOGIST'S DILEMMA I conclude that quantum evolution, genetic revolution, key innovation, correlated progression, even hopeful monster, are all ways of trying to solve (or even to state) the same problem, only in different frames of reference. An obvious place to try to find an underlying unity is in development, development being a major site of the causation of phenotypes. The decline of developmental biology as a force in mid-twentieth century evolution has two sources. The first is the emptiness 111 of the ontogeny-phylogeny connection. Here is a wonderful dilemma, in that von Baer's laws appear to reflect a set of fundamental facts of nature that has no causal explanatory power. The second factor was the petering out of experimental approaches in the 1950s. They revealed no data useful to the framing of hypotheses about evolution. However, in recent years this has completely turned around, with the result that developmental biology is now a major player in the subject. The field of development offers great opportunity for macroevolution because the sorts of features that morphologists attempt to link in models such as key innovation and correlated progression are integrated in a more profound way than mechanical functional. Some, at least, are linked morphogenetically and so development ought to have a great deal to say about models like key innovation and correlated progression. Because of the hierarchical nature of the developmental process, all structural features are basically linked together. Therefore it has been hard to see how major changes in any part of the phenotype could be produced with disruption of all—leading to hopeless monsters. One way around the problem would be that at very early stages in development, when pattern formation is initiated (say the preblastema stage of the tetrapod limb), only a very tiny subtle change could produce a significant change at the end of the developmental pathway (see Waddington's various models of the developmental landscape). However, it has long been a major tenet of evolution (and development) that any change to early developmental stages would be extremely disruptive or lethal. This then is another "embryologist's dilemma." The problem may have been reinforced by the false view (see Muller, von Baer, and Haeckel) that early ontogenetic stages within a major group are all very similar. In fact, early stages in development in different lineages differ markedly, and that is where divergences in phenotype are set in place. But the features by which they differ are not a matter of gross morphology but occur at the gene and cell level. They have to do with gene expression and genetic and epigenetic driven processes involving relatively undif- 112 KEITH STEWART THOMSON ferentiated cells. So the dilemma is easily resolved: because early stages have changed, they must be capable of change (Thomson, 1988). As such they are available for the creation of new phenotypes. Much has been written about new models for the introduction of evolutionary innovation at the developmental level. Such models depend on the basic integration of development and threshold effects in developmental cascades. I have proposed (Thomson, 1988Z?) that while minor changes in phenotype might be caused at any stages in development, major changes must be caused at early stages. But developmental cascade/threshold models, as we may call them, do not require that major changes in phenotype be caused by drastic revolutions in development. On the contrary, the new models can be paraphrased in terms that rather well parallel the key innovation model: a small change early in the developmental pathway unlocks a series of subsequent changes. And there are strong elements of correlated progression: developmental pathways are functionally linked, therefore changes caused early in development have the chance to affect several phenotypic features simultaneously. Further, because of the buffered and integrated nature of development (see Waddington's canalized landscape model), it is by no means inevitable that new phenotypes will be dysfunctional. Thus, different variants on the reptilian ankle must be caused developmentally by a relatively small shift in timing of blastema maturation at the prechondrogenic stage of limb development. They could not be caused at the late ossification stage because ligament associations are already set in place. The result is not a monster, but a workable joint that turns out to have possibilities for adaptive radiation of limb function. It was not automatically selected against because it found/created its own functional niche. The most exciting possibilities for evolutionary theory now will exist in attempting to find models that provide a feedback between the functional integration of the phenotype, and the capacity of the underlying developmental systems to change in an integrated pattern. (After all, as Louise Roth once observed [personal communication] Waddington's canalized developmental landscape model is Wright's adaptive landscape seen from underneath!) Then it would be necessary to find the correct linkages of those "causes" to the mechanisms of population biology, species selection, and no doubt other factors yet to be discovered. That would truly be a synthesis. REFERENCES Adams, M. B. 1980. Sergei Chetverikov, the Kol'tsov Institute, and the evolutionary synthesis. In E. Mayr and W. S. Provine (eds.), The evolutionary synthesis. Harvard University Press, Cambridge, Massachusetts. Bock, W. J. 1965. The role of adaptive mechanisms in the origin of higher levels of organization. Syst. Zool. 14:272-287. Dobzhansky, T. 1937. Geneticsand the origin ofspecies. Columbia University Press, New York. Eldredge, N. and S. J. Gould. 1972. Punctuated equilibria: An alternative to phyleric gradualism. In T. J. M. Schopt (ed.), Models in paleobiology, pp. 82-115. Freeman, Cooper, San Francisco. Fisher, R. A. 1929. The genetical basis of natural selection. Clarendon Press, Oxford. Goldschmidt, R. B. 1940. The material basis of evolution. Yale University Press, New Haven, Connecticut. Jepsen, G. L., E. Mayr, and G. G. Simpson. 1949. Genetics, paleontology, and evolution. Princeton University Press, Princeton, New Jersey. Mayr, E. 1963. Animal species and evolution. Harvard University Press, Cambridge, Massachusetts. Miller, A. H. 1949. Some ecological and morphological considerations in the evolution of higher taxonomic categories. In E. Mayr and E. Schuz (eds.), Ornithologie als biologische Wissenschaft., pp. 84-88. Carl Winter, Heidelberg. Salthe, S. N. 1986. Evolving hierarchical systems. Columbia University Press, New York. Simpson, G. G. 1944. Tempo and mode in evolution. Columbia University Press, New York. Simpson, G. G. 1953. The majorfeatures ofevolution. Columbia University Press, New York. Thomson, K. S. 1966. The evolution of the tetrapod middle ear in the rhipidistian-amphibian transition. Amer. Zool. 6:379-397. Thomson, K. S. 1988a. Fisher's microscope, or the gradualist's dilemma. Amer. Sci. 76:500-502. Thomson, K.S. 19886. Morphogenesis and evolution. Oxford University Press, New York. Vrba, E. S. and N. Eldredge. 1984. Individuals, hierarchies and processes: Towards a more complete evolutionary theory. Paleobiology 10:146—171. Waddington, C. H. 1967. Discussion. In P. S. Moorehead and M. M. Kaplan (eds.), Mathematical challenges to the neo-Darwinian interpretation of evolution. Wistar Institute of Anatomy and Biology, Philadelphia.