Neophenogenesis: A Developmental Theory of Phenotypic Evolution By: TIMOTHY D. JOHNSTON AND GILBERT GOTTLIEB

An important task for evolutionary biology is to explain how phenotypes change over evolutionary time. Neo-Darwinian theory explains phenotypic change as the outcome of genetic change brought about by natural selection. In the neo-Darwinian account, genetic change is primary; phenotypic change is a secondary outcome that is often given no explicit consideration at all. In this article, we introduce the concept of neophenogenesis: a persistent, transgenerational change in phenotypes over evolutionary time. A theory of neophenogenesis must encompass all sources of such phenotypic change, not just genetic ones. Both genetic and extra-genetic contributions to neophenogenesis have their effect through the mechanisms of development, and developmental considerations, particularly a rejection of the commonly held distinction between inherited and acquired traits, occupy a central place in neophenogenetic theory. New phenotypes arise because of a change in the patterns of organism-environment interaction that produce development in members of a population. So long as these new patterns of developmental interaction persist, the new phenotype(s) will also persist. Although the developmental mechanisms that produce the novel phenotype may change, as in the process known as "genetic assimilation", such changes are not necessary in order for neophenogenesis to occur, because neophenogenetic theory is a theory of phenotypic, not genetic, change. Article: INTRODUCTION A central problem for evolutionary biology is to explain the origin of phenotypic diversity among organisms. In its early years, before the rediscovery of Mendel's genetic work, evolutionary theory was almost entirely a theory of phenotypic change. Darwin's formulation of natural selection required that phenotypic variations exist in a population, but offered no account of the origin of such variations, beyond postulating "a tendency to vary, due to causes of which we are quite ignorant" (Darwin, 1872: 146). The idea that evolutionary change might involve anything other than change in the observable characteristics of organisms had to await Johannsen's (1909, 1911) distinction between the genotype and the phenotype, and the rediscovery of Mendel's (1866) experiments on inheritance in the early 20th century. As the science of genetics advanced, Darwin's "tendency to vary" became identified with the processes of mutation and recombination. This opened the door for theories of population genetics, which explained evolutionary change in terms of selection among genetic variants, rather than among phenotypic variants as proposed by Darwin. With the discovery of DNA by Watson & Crick (1953), the genetic theory of natural selection was placed on a molecular foundation and the current neo-Darwinian synthesis was completed. In the process, however, something had been lost; namely, the focus of evolutionary explanation on the phenotype. As Ho & Saunders (1979: 575) remark, neo-Darwinism "is primarily a theory of genes, yet the phenomenon that has to be explained in evolution is that of the transmutation of form". NeoDarwinism treats phenotypic change solely as the outcome of genetic change, the result of natural selection among members of the population. Although genetic change is no doubt important in the changing phenotypic makeup of an evolving population, in neo-Darwinian theory it has become the only source of phenotypic evolution (Saunders & Ho, 1982). In this article, we propose that it is more useful to view the natural selection of genetic variants as but one component of a broader process of phenotypic change that we call neophenogenesis, the origination of novel phenotypes that persist over evolutionary time. The changes that evolutionary theory attempts to explain are primarily changes in the phenotype—in the anatomy, physiology, or behavior of organisms over long periods of time. Such change may, of course, be brought about by the natural selection of genetic variants, but there are other mechanisms of neophenogenetic change (Novak, 1982b; Socha & Zemek, 1982), and it is these extra-genetic mechanisms, and their relationship to genetic change, that are our primary concern here. Developmental Mechanisms and Evolutionary Change An important current theme in evolutionary biology is that explaining phenotypic change requires us to pay close attention to the mechanisms of development (e.g. Alberch, 1980; Alberch et al., 1979; Bonner, 1982; Fallon & Cameron, 1977; Gould, 1989; Gustafson et al., 1985; Hall, 1975, 1984; Oster et al., 1988; Raft & Kaufmann, 1983; Shubin & Alberch, 1986). Developmental mechanisms are responsible for producing the phenotype and so, as de Beer (1940) pointed out long ago, evolutionary change in the phenotype can only come about by change in development (for an even earlier statement, see Mivart, 1871). But the developmental theory that underlies much of the current work is deeply problematic, because it accepts a relatively strong version of the distinction between inherited and acquired traits. Inherited traits are attributed to the developmental action of the genes, acquired traits to environmental influences experienced during the course of individual development. There are compelling arguments, summarized below, against this view of development and in this article we show how an alternative, and better supported, developmental theory leads to a quite different explanation of evolutionary change in the phenotype. The inherited/acquired distinction, however, is deeply rooted in the history of modern evolutionary theory, growing out of the division between Darwinism and neo-Darwinism that arose in the late 19th century, and that eventually banished Lamarckian, or quasi-Lamarckian, mechanisms from evolutionary biology. It is a central component of modern evolutionary theory, albeit one that is only rarely made explicit. Darwin's view was that "natural selection has been the main, but not the exclusive means of modification" (Darwin, 1872: 483). In addition to selection among the spontaneous variations that are now attributed to mutation and recombination, Darwin believed that the effects of use and disuse could be inherited, a Lamarckian evolutionary mechanism that came to assume progressively greater prominence in successive revisions of the Origin, and that culminated in his theory of pangenesis (Darwin, 1868). This theory, along with others that proposed the inheritance of acquired characters, was dealt a devasting blow by Weismann's (1893) theory of the germ plasm, which erected an impenetrable barrier between the germ-cell line and the somatic tissues. According to Weismann, the somatic and germ-cell lines are entirely separate; no change in the former can ever be transmitted to the latter. The germ-plasm theory was eventually accepted by biologists and later received confirmation in the "central dogma" of molecular genetics, according to which information flows only from DNA (germ) to protein (soma) molecules, not in reverse. Weismann's theory produced a split among evolutionary biologists, separating those who believed (with Darwin) that processes other than natural selection are involved in evolutionary change from those who followed Weismann in arguing that the isolation of the germ-cell line means that natural selection among spontaneous heritable variants is the only mechanism of evolution. Thus, evolutionary biologists became divided into what Romanes (1897) called Darwinists (such as Darwin, Romanes, and Spencer) and ultraDarwinists (such as Weismann, Wallace, and Lloyd Morgan). In the ensuing years, ultra-Darwinism (or neo-Darwinism as it came to be called) gradually became pre-eminent, incorporating the findings of Mendel and later of molecular biology into the modern evolutionary synthesis. The hallmark of neoDarwinian theory was thus from the beginning a belief that acquired characters cannot be inherited and that belief requires, of course, the assumption that acquired and inherited characters can be distinguished in the phenotype. Starting from that distinction, the neo-Darwinian argument is that natural selection accounts for evolutionary change in inherited characters, selecting among their alternate forms as those forms are made available by mutation and recombination. Acquired characters are not subject to natural selection because they are transitory and have no genetic basis. Since they are not inherited (and cannot become inherited, according to both the germ-plasm theory and the central dogma), they must arise anew in each generation and do not evolve (e.g. Ayala & Valentine, 1979: 19). Thus, in neo-Darwinism, "evolution" has become synonymous with genetic change: "Evolution is a change in the genetic composition of populations" (Dobzhansky, 1951: 16; emphasis in italics added)*. Sometimes, however, evolutionists adopt a different position; namely, that evolutionary theory must ultimately explain phenotypic change, and that although genetic models are an important part of that explanation, they cannot provide the entire account. For example, Simpson (1953a: 5) wrote that "genetic factors are not important to us for their own sake, but only because they are among the various determinants of phenotypic evolution" (emphasis in italics added). More recently, this view has been echoed by Lewontin (1974: 19), who suggests that "the real stuff of evolution" are changes in phenotypic, not genotypic characters. In a recent "postsynthesis clarification," Mayr (1988: 530) has expressed sympathy with the position of evolutionary naturalists that evolution "is not merely a change in the frequency of alleles in a population, as the reductionists asserted, but is at the same time a process relating to organs, behavior, and the interactions of individuals and populations." The position expressed by Simpson, Mayr, and Lewontin, and by some other authors (e.g. Bock, 1979; Futuyma, 1979: 21; Lambert & Hughes, 1984; Ho & Saunders, 1979, 1982), may be summarized as follows: Evolutionary theory must ultimately explain phenotypic change, and * Although Dobzhansky's definition is canonical, essentially the same one can be found in other leading statements of neo-Darwinism spanning 30 years (e.g. Simpson, 1959: 15; Grant, 1963: 125; Mettler & Gregg, 1969: 59; Dawkins, 1976: 48; Dobzhansky et at, 1977: 8; Ayala & Valentine, 1979: 18; Lumsden & Wilson, 1981: 371). although genetic models are an important part of that explanation, they cannot provide the entire account of change in the phenotype over evolutionary time. That task will require a theory that incorporates all of the mechanisms that may produce phenotypic change and, in particular, that explains the relationship between genetic and extra-genetic sources of such change † . The interest in development shown by evolutionary biologists over the past few years is an important step towards bridging the gap between genotypic and phenotypic change in a population. But that bridge will only stand if it is buttressed by a secure developmental theory. In many evolutionary discussions, development is represented as the unfolding of a genetic program (Alberch, 1982; Mayr, 1974; Smith-Gill, 1983). According to this programmatic view of development, some characters (those that evolve) develop under genetic control, whereas others depend on input from the environment. From this perspective, the task of developmental studies is to reveal the mechanics of such developmental unfolding, showing how the genes act on developmental processes rather than directly on adult phenotypic characters, and how development itself is constrained. But the development of evolving characters is always seen as being under tight genetic control, as it must be if the neo-Darwinian distinction between inherited and acquired characters is to be preserved. The importance of the distinction can be further appreciated by noting the existence in the neo-Darwinian lexicon of terms that explicitly distinguish inherited (genetic) traits from acquired (environmental) ones, such as the phenocopy (an environmentally induced phenotypic copy of a mutant genetic trait) and the ecophenotype (a novel phenotype produced by the environment rather than the genes). The existence of such terms presupposes the view that inherited traits can be distinguished from acquired traits (Oyama, 1981). The problem is that the inherited/acquired distinction itself is invalid. It has produced innumerable confusions, errors, and omissions in developmental theory (especially in the development of behavior; see Gottlieb, 1976; Johnston, 1987, 1988; Kuo, 1967; Lehrman, 1953, 1970; Oyama, 1982, 1985; Schneirla, 1956) and its retention in evolutionary biology can only lead to similar problems there. The theory of neophenogenesis is an attempt to incorporate an alternative view of development into evolutionary biology, but doing so will require that we abandon the neo-Darwinian distinction between inherited and acquired characters. CRITICISMS OF THE INHERITED/ACQUIRED DISTINCTION IN DEVELOPMENTAL THEORY Perhaps the clearest and most forceful exposition of the inherited/acquired distinction in developmental theory is to be found in the literature on behavioral development, where it is usually presented as a dichotomy between learned and innate behavior. For example, Lorenz's (1935, 1965) theory of instinct required an absolute distinction between those elements of behavior that are specified by the genes and those that arise in the course of individual experience. The neo-Darwinian origins of Lorenz's distinction can clearly be seen in his treatment of behavioral evolution (Lorenz, 1937), in which he forcefully and explicitly rejects any evolutionary connection between the two kinds of behavior. Lorenz's learned/innate † In a recent "post-synthesis clarification", Ernst Mayr has noted the conflicting views of naturalist and reductionist biologists regarding evolutionary change. According to the naturalists, evolution "is not merely a change in the frequency of alleles in a population, as the reductionists asserted, but is at the same time a process relating to organs, behaviors, and the interactions of individuals and populations" (Mayr, 1988: 530). dichotomy was vigorously criticized by developmentalists such as Lehrman (1953, 1970), Schneirla (1956, 1966), Jensen (1961), and Gottlieb (1970) who, building on Kuo's (1921, 1929) pioneering insights, argued that all behavior, and indeed all phenotypic characters, arises in development as the result of an interaction between the animal and its environment. The genes play a role in this interaction, one that is still hard to specify in any detail, but they do not directly determine any aspect of the phenotype. Lorenz (1965) responded that, to the contrary, the genes encode information that requires only the environmental conditions necessary to sustain life in order to determine in detail those components of behavior called "innate" or "instinctive". This information is in the form of a genetic program (see also Mayr, 1974) that unfolds mechanically in the course of strictly determined maturation. The view that development involves a programmatic unfolding of the phenotype is entirely consistent with the neo-Darwinian account of evolution, because it allows phenotypic characters to be divided into those that are specified (programmed) by the genes and those that depend on the environment. The interactionist view, however, which denies that any such division can be made, is much harder to reconcile with neo-Darwinian thinking because it rejects the distinction between acquired and inherited characters. This may account for the tremendous resistance to interactionist developmental thinking in the behavioral sciences, which in their modern form grew out of the neo-Darwinian evolutionary biology of the late 19th century (see Johnston, 1987, 1988 for documentation of this resistance). None the less, the interactionist position is a powerful and compelling alternative to the dichotomous view characteristic of neo-Darwinian thinking. Our current understanding of gene action in development does not allow for direct genetic specification of any phenotypic character beyond the level of protein structure (and even that specification is influenced by intracellular environmental factors such as pH and temperature; Pritchard, 1986); and the route from protein structure to gross anatomy and behavior is long and exceedingly complex. If the interactionist position (in some version) is accepted as a more adequate account of development than the dichotomous view, then evolutionary biology can hardly maintain the distinction between inherited and acquired characters as it attempts to integrate the results of developmental analyses into its account of evolutionary change. Development and Evolution in Neophenogenesis Any account of change in the phenotype over evolutionary time must recognize that the characteristics that change are themselves the product of development. Thus, to account for phenotypic change, we must consider and integrate all of the ways in which changes in development may be brought about. Neo-Darwinian theory incorporates development by distinguishing two kinds of phenotypic traits (inherited and acquried) and offering an account of evolutionary change in only one of these. Our task, by contrast, is to offer an account that proceeds from the position that no such distinction is possible or necessary. The development of an organism is determined by interactions among the various components of the organism and its environment, in which genes, hormones, diet, physical factors, exercise, sensory experience, social interactions, and numerous other factors play important roles (Bateson, 1987; Gottlieb, 1976, 1981; Lehrman, 1953, 1970) (see Fig. 1). A change in any of these components may modify the phenotype; from the interactionist perspective, there is no justification for making a priori judgments as to which of them are most likely to produce adaptively significant changes in the phenotype. The relevant factors can only be determined by experiment, and will likely be found to vary from species to species and from time to time during development. In particular, there is no warrant for singling out genetic change as being more relevant to the analysis of phenotypic change than Fig. 1. Development of the phenotype results from interactions among numerous components of both the organism and the environment. Altering any of these contributing factors, not only the genes, may produce change in the phenotype; if the alteration persists, the phenotypic change may persist long enough to be evolutionarily significant. are changes in any other of these factors. Neo-Darwinian evolutionary theory, of course, does make such an a priori assessment of evolutionary relevance, in asserting that only genetic changes produce true evolutionary change. So long as evolution is defined as change in the genetic makeup of populations, this assertion is necessarily true (by definition), but since neophenogenesis is defined differently and more broadly, such a priori assessments need not, indeed cannot, be made There is a terminological issue that needs to be addressed directly here, because it may result in the arguments we present being unfairly dismissed as inconsequential. Although formal definitions of evolution in neo-Darwinian theory invariably specify genetic change as being necessary for evolution to occur, less formal use of the term frequently refers to any phenotypic change that persists over a relatively long period of time. This latter sense of "evolution" is in effect when we read a description of "evolutionary change" in the primate brain, for example, based on evidence from comparative anatomy and fossil reconstruction. We have no idea to what extent such changes in phenotype involved genetic change in the populations involved, and so they should really be referred to as "phenotypic changes that may, to some extent, be evolutionary". Of course, no one is likely to use such a clumsy circumlocution, and so changes of this kind are almost always referred to as evolutionary, even though evidence about the mechanisms that brought them about is rarely available (Hailman, 1982). Thus, "evolutionary change" has come to have two meanings that are hardly ever distinguished, except when the explanatory hegemony of neoDarwinian theory is threatened. If we offer an account of some change in the phenotype that clearly (at least by hypothesis) does not involve genetic change, a neo-Darwinian evolutionary biologist is likely to retort that such changes, not being "evolutionary" in the formal sense, need not concern him/her. But that same biologist is likely to turn around and describe as "evolutionary" in the informal sense many phenotypic changes whose origin is in fact unknown. Because of this terminological ambiguity, neo-Darwinian theory succeeds in defining for itself two explanatory domains: a formal domain whose extent is largely unknown because we rarely know what genetic changes have taken place in natural populations; and an informal domain that encompasses all phenotypic changes not specifically shown or assumed to be extra-genetic in origin. Unless this problem is explicitly recognized, a theory of neophenogenesis (such as is proposed here) runs the risk of being dismissed because it fails to address evolutionary problems. This is only true if "evolutionary" is construed in the formal sense; in the informal sense of "evolutionary" . It is clear that neo-Darwinian theory itself fails to address many "evolutionary" problems that might be encompassed by a theory of neophenogenesis. A change in the environment of a population may alter the phenotypes of individuals developing in that environment without being a source of natural selection; that is, without changing the relative reproductive successes of genotypes in the population (see also Novak, 1982a, b; Socha & Zemek, 1978). An environmental change that does have selective consequences may also affect phenotypic development, and its selective and developmental consequences are likely to interact in complex ways that are at present very hard to predict (see further below). Let us illustrate our view of neophenogenesis by considering an environmental change that we presume not to have any selective consequences; later we will add selective consequences to the picture. A NEOPHENOGENETIC SCENARIO-DIETARY CHANGE IN A RODENT POPULATION Suppose that a population of rodents whose diet consists mainly of soft vegetation encounters a new food source in the form of hard but highly nutritious seeds. Evidence from studies of food selection in rodents (Kalat, 1985; Richter, 1947) suggests that the animals will initially sample small amounts of this new food, and then gradually increase its representation in their diet, especially if the seeds provide a rich source of some important nutrient. Because young rodents typically acquire their initial food preferences from their parents, especially their mothers (Galef, 1985), the new food habit will tend to stabilize as it spreads through the population, so long as the seeds remain available. Because animals will be eating these seeds during much of their lifetime, the new diet may have developmental effects on the phenotype that go beyond simply the establishment of a new food habit. Diet has consequences for body size and composition, fecundity, age of sexual maturation, nervous system development, and other aspects of the phenotype with far-reaching consequences for the animal's adaptation to its environment. As well as these direct effects of diet on development, there are also indirect effects produced by the animals' interaction with their new diet. For example, as the diet changes from relatively soft to much harder items, the mechanical stresses exerted on growing jaw tissues during development will change. Patterns of bone growth are partly determined by forces exerted on the growing bone (e.g. Frost, 1973; Herring & Lakars, 1981; Lanyon, 1980), and so the skeletal anatomy of the jaw will be different in animals that experience relatively hard and relatively soft diets during early life. Functional demands such as this, which arise out of the interaction between the developing animal and its environment, are central to the theory of neophenogenesis being presented here. To that extent, the theory resembles Lamarck's theory of evolution, which also emphasized the role of animal-environment interactions in producing phenotypic change. However, whereas Lamarck proposed (following what were then widely accepted beliefs; Burkhardt, 1977; Richards, 1987) that the effects of such interactions could be inherited by subsequent generations, our theory requires no such mechanism. It may be, as we discuss below, that the developmental mechanisms that produce the phenotypic character in question (such as the form of the jaw in our hypothetical example) may subsequently change, perhaps as a result of natural selection in the population. But our account, by denying the distinction between acquired and inherited (or genetic) traits, is not required to postulate "genetic assimilation" of developmental modifications, in the manner of Baldwin (1986), Cope (1887), Matsuda (1982, 1987), Morgan (1896), Osborn (1986), Schmalhausen (1949), and Waddington (1957). In that respect, our account differs from some other recent critiques of neo-Darwinism (e.g. Rosen & Buth, 1980; Steele, 1979, 1981), many of which also require a mechanism by which developmental modifications may eventually become inherited. To reiterate our position: Changes in either genetic or other influences on development may lead to relatively enduring transgenerational change in the phenotype which, in our definition, constitutes neophenogenesis. Before describing how we can incorporate natural selection into our account of neophenogenesis, let us consider some objections that might be raised against our position thus far. OBJECTIONS TO NEOPHENOGENESIS Objection 1 A new functional demand merely elicits a different developmental response from an unchanged organism; it does not produce real change in the organism itself. The cogency of this objection depends on what is meant by "the organism". From the standpoint of neophenogenesis, the organism is the phenotype and new functional demands certainly can produce change in the phenotype. Only if "the organism" is taken to refer to the genotype does this objection carry any force, but as already noted, the aim of neophenogenetic theory is to explain phenotypic, not genotypic change. At any stage in the phylogeny of a lineage, normal development of the individuals that it comprises depends on their having both a normal genotype and a normal functional context for development. Enduring changes in either the genotype or the functional context may produce stable, transgenerational phenotypic change, and the effects of both require explication. All such phenotypic changes are "real" changes, regardless of their source. The appeal of this objection depends quite strongly on one's view of the role played in development by the normal environment. On one view, the normal environment may be seen as having an essentially passive or "permissive" role in development, merely allowing the endogenous maturation of a normal phenotype (Lorenz, 1965; cf. Gottlieb, 1970). On that view, it is the genotype that is primarily responsible for the characteristics of the phenotype, and only genetic changes will appear to be of fundamental importance in producing phenotypic change. Alterations to the environment simply block or interfere with normal development, producing developmental aberrations of little or no interest. As argued above, this view finds little support from modern developmental theory, which emphasizes the paramount importance of functional interactions with a normal environment during development. Although the role of such interactions in particular instances of development continues to be debated, no adequate theory of development can exclude them from consideration. The task of a theory of neophenogenesis will be to work out the implications of this fact for explaining change in the phenotype. It is clear from the outset that if we grant that normal functional interactions play an important role in constructing the species-typical phenotype, then we must also grant that a change in those interactions may play an important role in changing the phenotype, and in maintaining that change in subsequent generations. Those are the defining characteristics of neophenogenesis.