Example: Early Pattern Formation in Amphibia

The Amphibia has been one of the most important animal groups for the study of developmental biology, and a huge descriptive and experimental literature has accumulated over the years. While sea urchins, molluscs, and nematodes, and more recently, Drosophila, have each become an important vehicle for the study of different aspects of development, the Amphibia and chordates in general have been especially important as the vehicle for the study of inductive regulative mechanisms. The early development of all chordates is marked by two revolutions in the control of early pattern formation: dorsalization at the blastula stage and gastrulation—primary induction caused by the “organizer” —which was studied in great detail in Amphibia by Spemann and his coworkers and continues to be the subject of intense scrutiny. The early phases of development in Amphibia exemplify rather well some of the problems in discovering the causal processes in development, whether in the egg, at fertilization, in the blastula, or in gastrulation itself. The short discussion of early development in Amphibia that follows is meant to exemplify rather than catalogue these questions. The oocyte in amphibians is radially symmetrical. A first axis of symmetry is established between a more heavily pigmented animal hemisphere and a less pigmented vegetal hemisphere. Before fertilization the egg is covered with a transparent vitelline membrane. When fertilization occurs, the vitelline membrane lifts from the surface of the egg and the egg is then free to rotate inside it so that the animal hemisphere lies uppermost and the vegetal hemisphere is lowermost. This rotation is apparently a response to gravity, which means that the vegetal hemisphere is heavier, almost certainly a result of the concentration of more and larger yolk granules in the vegetal hemisphere. Therefore, if the egg rotates to a new orientation with the yolk down and the animal hemisphere up, it must be the case that at this point the yolk and other egg contents are not free to be redistributed within the egg but are secured in place. The animal vegetal axis now marks the future anteroposterior axis of the embryo.

Later Pattern Formations Morphogenesis

The early phases of pattern formation, as described in the two previous chapters, set the stage for all that follows—for the whole grand sweep of morphogenesis by which the phenotype is created. Right from the oocyte stage there is an ordering of the embryo, both in the sense of spatial patterning and in the sense of setting in place the components of developmental processes. As pattern formation continues, particularly in “regulative” embryos, the ordering of the embryo becomes more and more specific. A point is reached that is quite impossible to define but nonetheless real, when the embryo is set up in such a way that all the components are in place. At least, everything has been specified. Such an embryo, for example the amphibian neurula, may look very little like the final phenotype, but from this point onward morphogenesis represents a working out of potentials that have already been established. All the basic morphogenetic information is in place. Morphogenesis is both simple and complex. It is simple because relatively few processes will be involved. It is complex because of the diversity of cell and tissue types that is involved, because of the subtlety of control of differentiation and even size and shape in organogenesis, and because of the complexity of both embryonic and subsequent adult function. Finally, it is complex because of the interactivity and “wholeness” of the developing embryo as well as the multiplicity of the parts making up that whole. It is with the simplicity of morphogenetic processes, the relatively small number of cell and tissue-level processes involved (processes that are common to all morphogenetic systems), that we will be concerned first. In order to understand both the basic rules of development and the way in which they relate to mechanisms for the introduction of evolutionarily significant phenotypic variation (Chapter 2), we must understand first these common processes of morphogenesis, and the way in which they operate under the rules and mechanisms of pattern control discussed in the previous chapters. Morphogenesis involves a relatively small number of phenomena characteristic of all cells and tissues, and their relation to features of the extracellular environment.

Early Pattern Formation

The processes of development form a continuum that begins with gametogenesis and ends only with the death of the individual organism. It is therefore artificial to try to define separate phases and stages of these processes, just as it is artificial to try to separate the structural history of the embryo into a series of discrete forms through time with discrete and definable properties (let alone trying to match such artifacts to putative phylogenetic stages). But at the same time, the sequence of mechanisms of development is hierarchically organized. The major early event is the transfer of control over development itself from the purely maternal factors inherited within the egg and particularly in the egg cytoplasm, to the switching on of the zygotic genome and transfer to zygotic control of cell function, interaction and differentiation, morphogenesis and cytodifferentiation. This transfer does not occur at a single instant, nor is it easy to generalize about it even with a single group of organisms. Other landmarks are harder to find, especially ones that can be compared consistently over a range of different organisms. However, one can roughly divide the processes of development, for the purpose of organizing a discussion at least, into two main phases: early and late pattern formation phases. Early pattern formation can be defined as that part of the developmental sequence in which all the major mechanisms that control the shaping of the embryo, both its morphogenesis and cytodifferentiation, are set into place. In a vertebrate, early pattern formation would be everything from gametogenesis up to and including gastrulation, by which time all the essential elements of tissue interaction that will cause the morphogenesis of the embryo have not only been regionally defined and correctly positioned, but have started to function. Late pattern formation comprises the stages of morphogenesis and cytodifferentiation. As we will see, morphogenesis itself can be divided into two stages, early and late: roughly speaking, in the earlier part, morphogenetic pattern-controlling mechanisms are set in place, and in later stages their results are expressed.

Development: Pattern and Process

In a hierarchical system, the manifestations of pattern at one level tend to become the components of process at the next, an alternation that is repeated again and again. The processes of molecular genetics produce, under the special circumstances applying at the genomic level, a pattern of gene activity that is carried forward to the organism level. Here new developmental processes based on these gene-level patterns build new organism-level patterns, and these then become the raw material for deme-level processes acting at the next level. This duality of pattern and process operates in any interactive system and was addressed particularly by Gregory Bateson (1979). It requires that we discuss the phenomena of process and pattern separately, as well as discovering their interdependence especially when, as in biological systems, there is additional complexity present in the form of feedback of causations among levels. Implicit in any hierarchical analysis is the assumption that process always involves a lawful set of mechanisms. In all biological systems this lawfulness will be derived from two sources: from the immanent properties of the systems themselves, and from even more general laws applying across all biological systems. For example, the salivary glands and lungs of vertebrates are both constructed in part according to a strict set of developmental rules applying to mechanics of epithelia; these are examples of developmental constraints (Chapter 7). But the size and shape of the lungs also follow more general physical rules such as the gas diffusion laws or volume-surface area relationships that determine how big a lung is required for an animal of a given size. These are structure-function constraints. Biological systems also derive a major set of consistencies from the historical connectedness, through relation by descent, of the organisms concerned. These consistencies are often called “phyletic constraints” (Chapter 7). It is a basic approach in biology to use the analysis of pattern to approach an understanding of process, often first in terms of deriving the “rules” from study of consistency and regularity. This is where the great power of the comparative-analytical method lies.

Introduction

All of science is fundamentally about cause. It is about explanations of the reasons things are the way they are and the mechanisms that produce them. It is now commonplace to observe that Charles Darwin brought evolution and all of organismal biology into line as a truly scientific subject by discussing evolutionary phenomena in terms of cause, and thus in the same testable, quantifiable frame of reference that applies to other science. Darwin's theory of natural selection as a causal agency for evolutionary change was only the beginning of our problems, not the end. For more than a hundred years, we have sought to find all the layers and intersecting lines of causality that produce natural selection as well as to discover other mechanisms for change that are nonselective in nature—genetic drift or neutral mutations, for example. Natural selection is basically a mechanism that involves two components: the introduction of variants into a system and the subsequent sorting of these variants (Vrba and Eldredge, 1984) so that, over generations, there is a differential contribution of these variants to higher levels such as populations and species. Up to the present time, most attention of evolutionists has concentrated upon two aspects of the problem: the genetic basis of phenotypic variation and the dynamic properties of populations containing the individual variants. The present book is concerned with the mechanisms affecting the expression of variation among individual phenotypes. It has been a surprisingly neglected subject. The New Synthetic theory of evolution and its later modifications have largely been pursued as if the intrinsic mechanisms by which variation is caused among individual organismal phenotypes are less important to the processes of evolution than the extrinsic mechanisms of sorting. If only by default, variation introduced at the level of the individual phenotypes is commonly treated as if it were a simple mapping of variation at the genetic level, or at least were only a very simple function of that. It has seemed not only necessary but sufficient to study genetics in order to understand phenotypic variation.

Some General Properties of Morphogenetlc Systems

J. Maynard Smith (1983) has written that “although we have a clear and highly articulated theory of evolution, we have no comparable theory of development.” I would turn this statement around somewhat and say that until we have a general theory of development we are unlikely to be able to derive a complete theory of evolution. This does not mean that a theory of evolution is wholly contained, in some reductionist sense (see Chapter 2), within a theory of development. However, if developmental processes play a major role in determining the modes and tempi of introduction of new variation at the level of the individual organism, and if they also have roles in upward and downward causation to other focal levels in the hierarchy of evolutionary mechanisms, then at least some of the rules of variation must be contained within the rules of development. If we are to progress in evolutionary biology beyond the study only of the contingent, and of unique empirical events, we will need a general theory, and part of that theory must derive from theories of the developmental processes that drive the introduction of variation. From developmental theory we will be able to make new general statements about how variation can be introduced at the phenotypic level. Although we still lack any such general theory, we can begin the process by using the preceding discussions at least to propose some general properties of developmental systems. The properties and processes of morphogenesis form an extremely complex system. Perhaps the hardest parts to grapple with are those “whole-organism” properties by which any given region of the developing embryo responds to field phenomena created by and expressed within the organism as a single whole rather than as a collection of isolated units, each with their own independent problems, mechanisms, and histories. Although these are vital questions (no pun intended), relatively little experimental work is being conducted in this area for obvious conceptual and technical reasons. Reductionist, functionalist approaches tend to make one concentrate on the parts rather than the whole.

Theory, Reduction, and Hierarchy

Implicit in the reasons given in Chapter 1 for development being ignored until recently as a potential causal factor in evolutionary theory is the general concept of reductionism. It is a strictly reductionistic approach either to believe that phenotypic variation is equivalent to genetic variation, or to act as though this were the case until disproven. Thus, to take but a single example, we find Stebbins (1974), who is avowedly a “strict reductionist,” stating that “in the future all general theories about evolution will have to be based chiefly upon established facts of population and molecular genetics.” Reduction is, of course, a powerful tool, but it is one with which biologists have in general had difficulty, and which in recent years has come under strong attack and defense (see Williams, 1986). The basic reductionist statement with which we are all comfortable is ontological, namely that the processes underlying all living phenomena are reducible to the operation of mechanical causes: there is no irreducible vitalist essence. Reductionism in this sense is unexceptionable and universal in science. The more difficult sort of reductionism to deal with is theory reduction. A simple expression of this would be the statement that the laws of chemistry are all explicable in terms of the laws of physics, or the laws of biology in the laws of chemistry. Nagel (1961) shows that such theory reduction requires that, for example, the laws of chemistry must be deducible from the laws of physics and that the terms and concepts of both sets of laws be “connected” (see, for example, Newton-Smith, 1982; Beckner, 1974). Another way of putting it is that the laws of physics must be of wider scope than the laws of chemistry, which then constitute a series of special cases of the former, under particular boundary conditions. Talking about theory reduction within the biological sciences, where general theories of broad scope are lacking (except the general theory of evolution that all organisms are related by descent), is somewhat pretentious. In the biological sciences we are forced to work more modestly with rules and probabilities rather than grand laws.

Morphogenesis and Evolution

In the preceding chapters we have built up the premises of an argument concerning morphogenesis and evolution. These are as follows: 1.Evolution occurs through processes of introduction and sorting of variation. 2.In a genealogical hierarchy, introduction and sorting of variation can occur at a series of focal levels and it is a property of the hierarchy that there is upward and downward causation between focal levels. 3.The crucial role of developmental processes with respect to evolutionary mechanisms is in the causation of new phenotypes. Phenotypes are always expressed in individual organisms, but the properties of other focal levels in the genealogical hierarchy must also be considered. 4.Development is also essentially hierarchical, involving processes acting at different focal levels. Each level is defined as the place where new gene expression occurs. For simplicity we can divide its hierarchy into stages from early pattern formation to late cytodifferentiation, each including unknown (but very large) numbers of phases of new gene expression. It must be noted that ontogeny of a given individual or given taxon represents a route through the basic hierarchy of developmental stages that, through historical accident or selective bias, may be extremely convoluted and unpredictable. 5.In the systematics of any group, there is a general correlation between taxonomic rank and different grades or ranks of morphological characters. There is a series of levels or grades of generality of phenotypic characters caused at different levels of the morphogenetic hierarchy. 6.The morphogenetic hierarchy that produces the different grades of phenotypic morphology can potentially involve upward but little if any downward causation. 7.The course of evolution appears principally to produce clusters of evolutionarily equivalent species rather than lines of progressive change. At any taxonomic level, diversification within a group has therefore to be distinguished from those rarer phases of progressive evolution leading to establishment of new groups. The two involve quite different processes because they involve the morphogenetic causation of different levels of morphological characters. 8.Phenotypic characters may occur in groups that are linked both in the sense of functional integration of the phenotype itself and/or by virtue of the integration and interdependency of developmental pathways in morphogenesis.

Patterns of Evolution

We now need to look more closely at the evolutionary patterns that we wish to explain. Again, we can use the study of patterns to discover and define important problems to be solved. The methods available for the study of patterns depend on the focal level of the mechanism with which one is concerned. If our interest is in individual variation, naturally we need extensive sampling and laboratory experimentation, looking at individuals within populations. In order to study the biology of populations we must work at the population level, and at such phenomena as gene and character frequencies in the field and laboratory. It is at this level that workers have most readily been able to measure selection coefficients and other quantitative elements of evolutionary science. If our interest is in the processes of speciation we must look as closely as possible at examples of the process in action—at sibling species, at a whole range of hybridization patterns in the wild and laboratory, at the distribution of closely related species in space and time. (It will be noted that between population-level work and species-level work there is an unfortunate gap; one can look at prespeciation and postspeciation situations but it is rare indeed to be confident that one is looking at speciation in flagrante delicto.) We can also look beyond the species level at phenomena of species distribution within higher taxa. Here we progress to using the methods of systematic biology and comparative morphology and paleontology. Each of these different approaches gives a different view of the evolutionary process and, therefore, in terms of the questions discussed in this book, gives us a different view of the role of mechanisms acting at the developmental level within evolutionary mechanisms. Obviously the most immediate element of causality in the origin of adaptive structures resulting from developmental properties is in the matter of individual phenotypic variation, but some of the most interesting questions concern the consequences of developmental properties for higher focal levels—their upwardly causing properties. To examine these we need to look at the species level and beyond (what has often been termed croevolution).