RSS Feed
The Evolutionary Dynamics of Life Cycles
SCHUMACHER COLLEGE
An International Centre for Ecological Studies
Chaos and Complexity Essay :
The Evolutionary Dynamics of Life Cycles
Johannes Jaeger
Written for the Chaos, Complexity and Creativity Course during the MSc programme in Holistic science at Schumacher College. Dartington, 01/03/2000
Introduction
Evolution raises many questions, which cannot be convincingly answered by traditional Darwinian evolutionary theory. The Neodarwinist Modern Synthesis is based on stochastic mutations in the genome, which lead to a corresponding variation in the phenotype of an organism (Dawkins, 1987; Dennett, 1995). Most of these mutations will be detrimental in some or another way, so that they reduce the probability of the mutant to have offspring. Advantageous alleles however will spread through a population and after only a few generations, they will prevail over other alleles following the laws of population genetics. Speciation is thought to occur through geographic or ecological isolation of populations. The gene pools of each of the subgroups gradually drift apart from each other, until they cannot interbreed any more.
Although this theory is very powerful for explaining many evolutionary phenomena, it has some major shortcomings. First of all, it assumes a linear relationship between genotype and phenotype. A specific mutation is supposed to be correlated with a clearly definable phenotype and hence, with a clearly definable fitness. It is common knowledge, especially among developmental geneticists that this is not so. Pleiotropy and polygeny seem to be the rule. Only in rare cases can a mutation in a gene be unequivocally linked to a single specific phenotype. Moreover, expressivity and penetrance of phenotypes most often vary heavily even under constant environmental conditions and mutant phenotypes can even be mimicked through heat shock or ether treatments (see Webster and Goodwin, 1996). Only in specially inbred strains, which are studied as model organisms, can clear genetic results be obtained, a fact all too often neglected by experimentalists.
A second weakness is the obvious absence of gradual evolution in the fossil record. Phases of evolutionary turmoil, with many different novel fossil forms appearing over intervals of only a few million years are followed by long periods of stasis. Moreover, very few ‘missing links’ are known, suggesting that speciation and radiation occur in a very short (geological) time span. Eldredge and Gould (1972) called this phenomenon punctuated equilibrium, as opposed to Darwinist gradualism (Dawkins, 1987). Neodarwinism would explain such changing evolutionary paces by external events like planetesimal impacts or major changes of climate due to periods of intense volcanic activity or continental drift. However, this fails to explain the major absence of interspecies hybrid forms in fossil and extant species, which would be expected to result from a slow and gradual speciation process.
Another pattern in evolution, which is extremely difficult to explain in Neodarwinist terms, is the fact that no new phyla seem to have emerged after the Cambrian radiation. Although the fossil record shows evidence of many mass extinctions after the Cambrian, some of them affecting more that 50% of all species present at the time, no new animal bauplans evolved anymore. The major novelty in the post-Cambrian animal kingdom were legged animals in the phyla of the arthropods and the vertebrates, which occupied new ecological niches on the land (Raff, 1996). This is a phenomenon which cannot be expected in a Neodarwinist model of evolution, but which can easily be reproduced by simple computer simulations of evolving genetic networks (Kauffman, 1993; 1995).
In my opinion, all the above mentioned difficulties arise out of the fixation of Neodarwinist theory on genes as the basic units of natural selection (for an extreme case of genophilia see Dawkins, 1976; 1987). There might be many advantages of genetic theories of evolution, among them the availability of a profusion of laboratory and mathematical methods. However, we have to keep in mind that the scope of such theories must be limited, since they neglect the whole organismic level of organisation, or in other words the life cycles, in which the genes that are studied are embedded in.
Therefore, I would like to propose a new focus in evolutionary biology, which is based on life cycles and I will try to outline possible mathematical methods, which might be useful to study it, without having to consider all the genetic details that are involved in (but not causing!) the observed phenomena. I choose life cycles as my unit of selection because, as I mentioned above, it is the organism (and not its genes) that interacts with its environment and succeeds or fails to have offspring. Evolving organisms are not born as adult phenotypes, as which they are treated in population genetics. They are constantly transforming entities that do not stop to develop when maturity is reached, but go on to reproduce, grow old and die. Senescence and death are as much necessary ingredients for evolution as are reproduction and morphogenesis. If we wouldn’t die and make room for future generations, there would be nothing for natural selection to select. Therefore, genes and genetic networks have to be seen embedded in life cycles, since any kind of genetic network, which does not participate in a closed reproductive cycle will not be able to evolve. It is only in this context that we will be able to gain a deeper insight into evolutionary patterns and dynamics. In my view, such a theory would not stand in opposition to current evolutionary theory, but would try to embrace Neodarwinism, to embed it in a broader theory of evolution and to clarify the limits of Neodarwinist explanations,
However, all of this has to remain highly speculative and the mathematical concepts have yet to be used in a metaphorical way. Currently, we are in a position where we do not know enough about the biological and mathematical problems involved in a theory of evolutionary dynamics. Biological systems consist of innumerable components and it is often very difficult to identify (not to mention to quantify) the ones that are important to a specific problem.
Another major problem is the influence of stochastic events. History and contingence undoubtedly do play an important role in evolutionary dynamics, but below all the random noise seems to be a layer of hidden order. However, it might prove somewhat difficult to separate the chaff from the wheat. Hence, there will maybe never be an appropriate universal description of life and its evolution, be it mathematical or biological, which will apply to all phenomena of living nature. However, important lessons could be learned from a different perspective than the currently dominating genetic paradigm, which could help us bridge the gap between understanding the molecular reactions occurring in an organism and their broader evolutionary context.
Nonlinear Dynamics
Before I go into more detail in describing evolving life cycles, I would like to introduce some basic concepts of nonlinear and complex dynamics. Nonlinear dynamics is the field of mathematics, which deals with systems of nonlinear differential equations and their behaviour. It is a relatively young discipline since high-powered computers are required to obtain numerical or graphical solutions for the equations, for which in most cases analytical solutions are not available. The most widely known insight gained through the study of nonlinear systems is the discovery of chaotic behaviour in even the simplest nonlinear equations (reviewed in Gleick, 1987; Ruelle, 1991; Stewart, 1997). Simple mathematical rules can produce highly unpredictable outcomes when iterated. This is due to a high degree of sensitivity to initial conditions, meaning that the system and its behaviour are determined by the mathematical equations but accurate long-term prediction is forever impossible due to practical limitations of precision. Over the past three decades, methods have been developed to discover the characteristic footprint of chaos in experimental data. In this chapter, I would like to give a short summary of mathematical techniques relevant to this discussion of evolutionary dynamics.
Iteration
Differential equations describe a system by its ways of change. A moving body, for instance, is described by means of its acceleration, which is caused by the influence of forces on it. Thus, the trajectory of a body in space will not only depend on the forces that act on it at every moment but also on its initial position. There is continuity, every current state depends on the previous state of the system. This amounts to applying a certain rule in infinitesimally small steps all over and over again. Dynamic systems are therefore dependent on their initial conditions, unlike the pressure and temperature of a gas, which are state variables that can be reached in infinitely many ways. The behaviour of such a system is determined by its rules, the differential equations, which can be solved analytically or numerically to make observations and predictions about the system. The observed behaviour of nonlinear equations shows certain typical characteristics, which can be classified as different types of attractors.
Phase Space
The mathematical technique best suited to study dynamical systems’ behaviour and attractors is called the phase space of a system. A phase space is constructed by plotting all the independent state variables of a system against each other in a Cartesian co-ordinate system. For the model of a pendulum, you plot the pendulum’s angle against its angular velocity in time, one set of two co-ordinates for each measurement you make beginning with its initial state. Since these two state variables are sufficient to describe the pendulum’s behaviour under all circumstances, the pendulum has a two dimensional phase space. Other systems can have n-dimensional phase spaces, depending on the number of independent variables necessary to describe them (also called degrees of freedom of the system). Different kinds of attractors can now be qualitatively distinguished by their geometrical shape in phase space. A system, which is truly random (i.e. not based on any rules or constraints), will appear as a random tangle in phase space, since it can be in any state at any time.
The Expected
If a system settles into a stable static equilibrium after a while, we can say that it has reached a point attractor, depicted as a point or a sink in phase space. When pushed out of its equilibrium, the system will return to it and stay there if left unperturbed. Rocks fall to the ground and stay there if you don’t pick them up again. They will always fall to the ground unless you leave earth’s gravity field. The region of initial values, from which the system will return to the same attractor is called a basin of attraction. Dynamical systems mostly have more than one attractor. If you move close enough to the moon, the rock will fall towards the moon instead of the earth. The description of all basins of attractions in phase space for a system is called the system’s phase portrait.
Another frequently observed behaviour is a regular oscillation between two or more states of the system. In absence of friction, the pendulum will go on forever if left unperturbed. This is called a periodic attractor, described by a circle in phase space. In more complex systems, you can also observe quasiperiodic behaviour, which is an attractor with the shape of a torus in three-dimensional phase space. However, this is not a stable, typical behaviour for dynamical systems and will eventually settle into a periodic motion.
The Unexpected
However, nonlinear systems can not only show static or periodic behaviour. Even the simplest equations show seemingly random and non-repeating fluctuations over certain parameter ranges. Simple deterministic rules easily give rise to apparently random behaviour. Still, the fluctuations are limited in their range and there are recurring similarities and regularities in the observed patterns. Moreover, if you plot this behaviour in phase space, you will not get the random tangle expected for random fluctuations, but an incredibly complex, literally infinitely entangled structure called a strange attractor. This is deterministic chaos. The most famous example of chaotic behaviour is the weather, which makes meteorologists’ lives difficult in that it makes the more or less regular phenomena of weather (rain, sunshine etc.) appear at unpredictable moments.
Strange attractors have strange properties. They are infinitely stretched and folded back on themselves. If you magnify the attractor you will see the same structures you see in the initial picture on a smaller scale, meaning that strange attractors are fractal or self-similar in their structure. Since there is an endless amount of detail hidden within the attractor, infinite trajectories never cross each other. Intersecting trajectories would give rise to indeterminacy, but strange attractors describe perfectly deterministic behaviour, which never repeats exactly.
Moreover, strange attractors have a property, which is really bad news for the scientist who wants to make predictions about the behaviour of real systems. Neighbouring points on the attractor, however close they might be, can give rise to highly divergent trajectories. This is called ‘sensitivity to initial conditions’ and it implies that we will never be able to predict long-term behaviour of chaotic systems, since we would have to measure the initial conditions to an infinitely accurate degree. Therefore, predicting next years’ weather will always remain impossible.
I would like to emphasise again that chaos is not randomness. There are mostly very simple rules hidden below the seemingly random observed behaviour. Moreover, this is testable in real systems, since chaotic systems show certain characteristics, which make them distinguishable from randomness.
The Footprint of Chaos
Strange attractors can be clearly distinguished from randomness in phase space. Thus, it would be desirable to have a method for creating something comparable to a phase space out of a single data series, in order to reconstruct the structure of the attractor of an observed system. This can be achieved by simply duplicating or triplicating the data series and shifting them by one position in time relative to each other. A data point obtained at time t is then plotted against a data point obtained at t+1. This is called the Ruelle-Takens method for reconstructing attractors and the resulting plot is called a delay map. It has been successfully applied to several systems, among them the Lorenz attractor. As long as unequivocal data is available, delay maps provide an easy and fast way to test data for an underlying chaotic mechanism.
Bifurcation
Nonlinear systems can drastically change their properties over a small range of parameters. A system is in its stable state, when you can fiddle around with the parameters and still observe the same typical kinds of behaviour. However, as the parameters approach certain critical values the system’s properties can change suddenly and unexpectedly. New attractors and basins of attraction appear unexpectedly to replace the ones observed before. Such points of instability in a dynamical system are called bifurcation points, a concept very useful to describe symmetry-breaking the process underlying developmental and evolutionary pattern formation in nature.
Life Cycles
Evolutionary Trajectories
As mentioned above, I propose to base a theory of evolutionary dynamics on life cycles as fundamental units of selection. A life cycle could be seen as a continuous branching trajectory through time. Selection acts by channelling or nudging such trajectories into certain directions. Thus, the evolutionary outcome, i.e. the direction of the evolutionary trajectory, is not only determined by natural selection (the effective cause), but also by the organism itself (the material cause). There are constraints intrinsic within the organism, which need to be considered (see Webster and Goodwin, 1996).
The biggest and most obvious intrinsic constraint for an evolving organism is the presence of a closed life cycle itself. An organism has to be able to reproduce and thereby to have offspring, which resembles its parents again. Without such a stable life cycle, in the absence of evolutionary stability, selection cannot work. However, organisms often show a high degree of polymorphism within their populations. Thus, within the constraint of closure, a life cycle can vary considerably even within a species. Moreover, the life cycle is not exactly periodic, since children never resemble their parents exactly. Still, polymorphisms rarely fall out of their species’ limits altogether. Even badly disfigured mutants can mostly be identified as belonging to a certain species.
Considering all the observed phenomena described above, it might be useful to treat life cycles as chaotic attractors in morphospace. Morphospace is a kind of phase space in which all characteristics needed to describe the morphology of an organism are plotted against each other (see Webster and Goodwin, 1996). Strange attractors are limited to a defined range of values within morphospace. This would correspond to the limited variability of morphologies observed in a species. All organisms in a species resemble each other. However, the strange attractor never repeats itself, there are no two identical trajectories. The same applies to organisms of a species, since no two individuals look exactly the same. The constraint of closure for the attractor suggests that all evolutionary trajectories must pertain to a certain class of strange attractor, which shows considerable, but not exact, periodicity over each generation. Each single life would begin and end in a limited region of morphospace, necessary to maintain the species. Moreover, these evolutionary attractors seem to be quite robust towards parameter changes. It takes quite a lot, to push the system over its limits. If this happens, through drastic environmental changes or a series of mutations for example, a new stable state might emerge. Speciation could be seen as a bifurcation of a chaotic life cycle.
In the Neodarwinist view, species are not seen as true categories, but only as temporary names for something, which has common ancestry. It is widely ignored that this very ancestry can only be established through morphological, or more accurately, gene sequence similarities and is highly prone to error (Webster and Goodwin, 1996). In most cases, it is impossible to decide with certainty, if a certain resemblance in recent species is based on common ancestry or if it is a case of convergent evolution, even if the analysis is based on gene sequence comparisons. If organisms are treated as strange attractors in morphospace, we could classify species as different classes or types of attractors. Such a classification would be a genuine classification of natural types, not only a catalogue of names (Webster and Goodwin, 1996). Each of these different attractors would have a typical topology, although their trajectories might approach each other at certain times, as happens in convergent evolution. Moreover, certain typical patterns or regularities might be identified, which are typical for the whole class of evolutionary attractors, i.e. to all organisms or at least to all animals. In the following paragraphs, I would like to illustrate this with a few examples.
However, before I can go into this, I should try to identify the rules that produce an evolutionary trajectory and which correspond to the equations needed to describe such an evolutionary attractor. I suggest that these rules might correspond to the generative principles, which create a life cycle. What is needed as a first step in the development of an evolutionary dynamics, is a dynamical model of a life cycle, a description based on the rules that make up an organism.
Generic Principles
Nowadays, generative principles in biology are usually described as genetic cascades or genetic networks. This is a very useful simplification to gain first insights into the laws of morphogenesis, since there is ample genetic experimental evidence available, but we have to keep in mind that the genetic network approach is also extremely limited excluding the entire organismic context of the genes. Moreover, producing proteins that make up the interactions in these genetic networks is highly energy intensive (measured in use of ATP). Therefore, an obvious first question to ask is, if there are cheaper ways to break symmetry and to create pattern than by genetic instruction.
What happens if you put a number of cells together and let their surface properties vary a little? Do we have to know all the genetic properties of these cells to understand the process of morphogenesis?
The answer is no.
There are typical or generic forms, which arise from simple considerations about the physical and chemical properties of cells. One example is the formation of whorls in the unicellular alga Acetabularia acetabulum (Goodwin, 1994; Webster and Goodwin, 1996). During morphogenesis, several whorls branch from the main stem. In other related species of algae, these whorls serve as gametophores, but they seem to have lost this function in A. acetabulum. The whorls are simply shed after their cytoplasm is reabsorbed. Fossil evidence indicates that this functionally apparently useless structure seems to have survived millions of years of evolution, although it takes energy to produce it and therefore, adaptation through selection should have favoured whorl-less forms. Another astonishing property of A. acetabulum whorls is that they still form in cells, which were cut in half, isolating the cell nucleus from the site of whorl formation. Gene expression seems not to be involved in the determination of the whorl’s shape, although certain gene products need to be present for the whorls to form at all.
When A. acetabulum development is simulated on a computer, solely based on assumptions about the elasticity of its cell wall and the underlying cytoplasm, whorl formation can be detected. Moreover, the specific patterns of Ca2+ ions predicted by the model could later be confirmed experimentally. Goodwin (1994) proposes to use the term morphogenetic field to describe the interaction of cellular components in morphogenesis (such as the cell wall and the cytoplasm with the local Ca2+ concentration in the above example). Morphogenetic fields are an attempt to comprise and integrate genes in their organismic context, in order to bridge the gap between the genetic and the organismic levels of investigation.
Similar arguments apply to basic morphogenetic processes in animals. Gastrulation seems to be an obvious outcome of putting a mass of cells together forming a sphere, considering that their surface properties and mutual adhesion may vary (Newman, 1990). If you have two populations of cells, which stick to each other more strongly than to cells of the other type, the two populations minimise their common surface. Thus, one of the two populations of cells will get pushed into the interior of the sphere of cells through invagination, involution or ingression. Of course, gene expression is needed to produce the cell adhesion molecules in different levels (as well as to produce a whole lot of other cellular compartments). However, no specialised gene or genetic cascade is needed to produce gastrulation, it simply happens if the above mentioned chemical properties of the cells apply. As a good Darwinian, I cannot imagine that an organism would use a much more energy intensive genetic mechanism instead of this generic strategy. The genetic solution would obviously be culled rapidly by selection.
Patterns in Development
Generic in its mathematical form simply means typical. In biology, it is used to describe properties which are characteristic for a group of species (Webster and Goodwin, 1996). Patterns and shapes, which can be commonly observed over a wide range of taxa are obvious candidates for such common themes in evolution. If we regard life cycles as strange attractors of evolutionary trajectories, then we could consider these regularities to be recurring patterns typically observed in chaotic attractors, similar but never identical to each other (Stewart, 1997).
Similar shapes and patterns in evolution are called homologous or convergent in Darwinian theory. Homologous structures are thought to share the same ancestry, whereas convergent structures (or homoplasies) are thought to be fortuitous coincidences whereby two unrelated species have come up with similar solutions to similar ecological niches. William Bateson (1894) proposed a different definition of homology. He found it difficult to clarify the historical relations of species and their parts beyond doubt and therefore proposed a concept of homology as two structures sharing the same generative principle or equivalence in transformation. This definition of homology includes the commonly used concept of serial homology for segments or limbs in animals, whereas in a strict Darwinian sense, these structures cannot be considered homologous, since they do not necessarily share the same ancestral structure (Gilbert et al., 1996; Webster and Goodwin, 1996).
Such homologies in the sense of Bateson could be considered as recurring patterns in related classes of attractors describing stable life cycles. They might be so widespread in evolution, because they might be based on very simple generic mechanisms and therefore, might have evolved independently a number of times. Let me give a few examples:
Segmentation
There is an ongoing controversy among evolutionary biologists, if segmentation has arisen only once or at various times in evolution (Raff, 1996). A lot of research in developmental biology is focussed on identifying the genetic components of the cascades that lead to segmentation. The data is confusing and not very helpful to resolve the issue. Some genes seem to be commonly used across the whole animal kingdom (e.g. engrailed, signalling cascades like WNT-signalling etc.), whereas other levels of the cascades seem to differ even between closely related species (e.g. the pair-rule genes in insects) (Raff, 1996).
Stuart Newman shows in a very simple model, that segmentation is a necessary consequence of periodic expression of cell adhesion molecules, with a phase different from the one of the cell division cycle (Newman, 1993). Such chemical oscillations are commonly observed in cells and are based on short-range positive feedback coupled to long-range negative feedback. The exact same mechanism underlies many reaction diffusion models of pattern formation as well as the concept of an excitable medium, a system in which symmetry can be readily broken due to very small initial differences in chemical concentrations of certain molecules (Goodwin, 1994; Ball, 1999).
Thus, segmentation seems to be a generic phenomenon in animal development and might easily have arisen many times in evolution. This would explain the high diversity in segmentation mechanisms across animal taxa. The conservation of certain genetic systems, like signalling cascades, could be explained through genetic assimilation (see below) and functional conservation of gene products. It is necessary to set up boundaries during the segmentation process and this is exactly what the product of the engrailed gene seems to do best. Therefore, it might have been easily co-opted into novel roles in segmentation from its ancestral role in neurogenesis (Raff, 1996).
The Zootype
A very intriguing case of homology has been termed ‘zootype’ by Slack and co—workers (1993) since it appears in all animal phyla studied so far. Although earlier and later development show great variability, all animals have a stage in which the homeotic (HOM) genes are first expressed in a very characteristic and highly conserved sequence along the body axis. Within the phyla of insects and vertebrates, this stage also shows great morphological homology as well and was hence termed ‘phylotypic stage’ (Raff, 1996). It corresponds to the germ band stage in insects and the pharyngula stage in vertebrates.
It seems very difficult to find a functional explanation for this. It is not at all clear if HOM genes are responsible for segmentation, since they also occur in non-segmented phyla. Rather, they seem to be involved in specification of the antero-posterior axis of the animals, to divide the embryo in specific domains, which develop into different parts along the body axis. There is no functional reason known for this, since polarity and positional information can be conveyed in many different ways in theoretical models.
I think that it would be worth looking for a generic explanation of this phenomenon. Once again, the HOM genes might not be the cause of the high conservation, but might have been very easily assimilated into a generic process of segment or body region identity determination. The nature of this generic process still eludes us, but I am confident that it might become clearer as knowledge about generic principles as well as genetic data on HOM genes increase.
Genetic Assimilation
As briefly mentioned above, many genetic cascades are highly conserved in evolution. A theory of evolution, which is based on the life cycle, has to be able to explain this conservation by other arguments than common ancestry. If we assume that there are robust generic principles of morphogenesis, underlying the genetic networks, then we have to see the genes not as causes of biological form but as stabilisers of hidden generic processes (Newman, 1990; 1993). Often in the course of evolution, the generic mechanisms were so heavily covered with genetic networks that it is very difficult to see that they exist at all. This is certainly true for most of the model organisms for developmental genetics, since these species were chosen because of their easy-to-handle genetics and their short and highly determined modes of development. Drosophila melanogaster and Caenorhabditis elegans seem to be not only very derived members of their phyla, but also highly determined in their development, due to their very short generation time. The very few data we have on development of other organisms, such as short germ insects or crustaceans do not confirm this clear genetic picture (Raff, 1996).
Waddington has proposed a very interesting mechanism for the adaptation of genetic networks to their organismic environment. He called it genetic assimilation (see Thom, 1989). The concept is based on the idea that genes are not only exposed to the environment outside the organism but also to their cellular context to which they can adapt. This adaptation can occur due to natural selection, no teleology or higher force is needed. In an organism, which forms segments due to its intrinsic tendency to show oscillations of gene products in its cells (see above), mutations of genes which disrupt this tendency can easily be assumed to be detrimental. Only those mutations will survive, which allow the segmentation to occur, because otherwise the body’s fundamental structure becomes disrupted and the organism is not viable anymore. It is a very small step to conclude that mutations, which help to stabilise and increase the robustness of the bauplan towards environmental influences, will be beneficial. Therefore, by interplay of generic processes and Darwinian adaptation, we can explain why certain structures are common in evolution and why certain regularities can be seen in the underlying genetic networks. This is the picture of the trajectory being channelled by selection.
Genetic cascades provide certain functions in a cell. A signalling cascade will transmit information from one cell to another in a certain way and over a certain range. During segmentation, compartment boundaries need to become established to distinguish one segment from another. Now, if there is a signalling cascade, which can greatly stabilise this boundary by amplifying distinct patterns of gene expression in neighbouring cells, we can imagine that the co-option of such a cascade into the segmentation process will be very beneficial. If we assume a certain fluidity of the genome, we can conclude that such a co-option would happen reasonably often and lead to genetic homology without common ancestry. A very good example for this phenomenon is the recurring employment of Notch-signalling in all kinds of contexts, where cell to cell interactions are required (Artavanis-Tsakonas et al., 1995).
This leads to a sort of Lamarckism, where acquired characteristics slowly become assimilated by an organism. Waddington used the example of callosities on the skin, which are produced by genes in reaction to a certain environmental influence (viz. attrition or friction). If the organism is constantly exposed to this influence and so are its offspring, the pattern of gene expression, which leads to the formation of the callosities will remain stable in the population, as long as the environmental conditions do not change. To the outside observer, it seems that this is a case of inheritance of acquired characters, but it can also be explained in purely Darwinian terms if we extend the concept of the environment for the genes to the intracellular and intraorganismic environment, which is commonly neglected in modern genetics. A shift of focus from genes to organism and its life cycle can help to explain these common phenomena in evolution and raise a whole new category of questions to be asked. Genetic networks do not cause evolution, but are embedded in the constraints of a life cycle.
Generic + Genetic = Generative
The aim of the organism-based evolutionary theory proposed here is to understand and classify generative principles in terms of chaotic attractors in morphospace. During the last few paragraphs I tried to clarify, what such generative principles are. If we start from the organism as a unit for evolutionary change, we must distinguish generic and genetic principles. Generic principles are based on physical and chemical properties of the cells in an aggregate. If you put cells together, certain things will be more likely to happen due to the intrinsic properties of the cell aggregate. Once these generic patterns are established, they can become fixed and stabilised through the action of genetic cascades or networks (Newman, 1990; 1993).
The resulting generative mechanisms make up a developmental toolbox of reaction norms and patterns that an organism can show in a certain environment. Depending on environmental conditions, these generative mechanisms will produce different morphologies. However, it takes a major environmental perturbation to disturb an organism’s development severely enough to push it in an altogether different stable or a lethal unstable state. The process of morphogenesis is robust or, to use Waddington’s term, canalised (see Webster and Goodwin, 1996).
Fact-Free Science?
I think that we have to revise our view of the organism as a mere recipient of order from the genome. An organism is an open system with its own properties and constraints, with its own material causes, which are not predictable from its genome and which are important for evolutionary theory. We are not the product of pure chance and some natural selection. There are underlying dynamic principles which channel the evolution of life into certain shapes and forms, the observable discretely distributed bauplans of fossil and extant phyla.
In a review of Depew and Weber’s book (1995) in the New York Review of Books, John Maynard Smith called complex dynamics and its application to biology a ‘fact-free science’. I do not believe that this is true. As I have tried to explain in this essay, there are many open questions in evolutionary theory for which I do not see a Neodarwinist solution. We can no longer ignore that genes act within organisms, which follow their own laws. Every level of evolution has its appropriate methods of analysis. The focus of evolutionary biology has shifted over the last few decades in a disproportionate manner towards the study of genes. Morphological studies are hardly performed anymore, since their capacity to produce interesting results seems to be exhausted. I tried to sketch a possible way out of this dilemma, which may lead to the development of mathematical tools to study the evolution of whole organisms and to make new predictions, which ultimately should be tested in novel kinds of experiments. The idea remains highly speculative and will undoubtedly have to undergo many changes and adjustments in the course of the next few decades. A theoretical framework would have to be elaborated upon which new experiments could be performed and hypotheses tested. However, I believe that it is an idea worth exploring and that it will give us many new insights and approaches to the study of evolutionary dynamics. A lot of work remains to be done and I hope to be part of the project, part of a new view on evolution.
Bibliography
Artavanis-Tsakonas, S., Matsuno, K. and Fortini, M.E. (1995) Notch Signaling. Science, 268, 225-232.
Ball, P. (1999) The Self-Made Tapestry. Oxford University Press, Oxford.
Bateson, W. (1894) Materials for the Study of Variation. Cambridge University Press, Cambridge.
Dawkins, R. (1976) The Selfish Gene. Oxford University Press, Oxford.
Dawkins, R. (1987) The Blind Watchmaker. Norton, New York.
Dennett, D.C. (1995) Darwin’s Dangerous Idea. Simon & Schuster, New York.
Depew, D.J. and Weber, B.H. (1995) Darwinism Evolving. MIT Press, Cambridge, MA.
Eldredge, N. and Gould, S.J. (1972) Punctuated equilibria: An alternative to phyletic gradualism. In Schopf, T.J.M. (ed.) Models in Paleobiology. Freeman Cooper, San Francisco.
Gilbert, S.F., Opitz, J.M. and Raff, R.A. (1996) Resynthesizing Evolutionary and Developmental Biology. Dev Biol, 173, 357-372.
Gleick, J. (1987) Chaos: Making a New Science. Viking, New York.
Goodwin, B. (1994) How the Leopard Changed Its Spots. Charles Scribner’s Sons, New York.
Kauffman, S. (1993) The Origins of Order. Oxford Univerity Press, Oxford.
Kauffman, S. (1995) At Home in the Universe. Oxford University Press, Oxford.
Newman, S. (1990) Generic Physical Mechanisms of Morphogenesis and Pattern Formation as Determinants in the Evolution of Multicellular Oranization. In Mittenthal, J.E. and Baskin, A.B. (eds.), The Principles of Organization in Organisms. Addison-Wesley, Reading, MA.
Newman, S.A. (1993) Is segmentation generic? Bioessays, 15, 277-283.
Raff, R.A. (1996) The Shape of Life. University of Chicago Press, Chicago.
Ruelle, D. (1991) Chance and Chaos. Princeton University Press, Princeton, NJ.
Slack, J.M.W., Holland, P.W.H. and Graham, C.F. (1993) The Zootype and the Phylotypic Stage. Nature, 361, 490-492.
Stewart, I. (1997) Does God Play Dice? Blackwell, London.
Thom, R. (1989) An Inventory of Waddington Concepts. In Goodwin, B.C. and Saunders, P. (eds.), Theoretical Biology: Epigenetic and Evolutionary Order from Complex Systems. Edinburgh University Press, Edinburgh, pp. 1-7.
Webster, G. and Goodwin, B. (1996) Form and Transformation. Cambridge University Press, Cambridge.
For more information about Schumacher College and its courses, please contact :
The Administrator, Schumacher College,
The Old Postern, Dartington, Totnes, Devon, TQ9 6EA.
Tel : +44 (0)1803 865934 Fax : +44 (0)1803 866899
Email : admin@schumachercollege.org.uk
Website : www.schumachercollege.org.uk
Schumacher College is part of the Dartington Hall Trust, a company limited by guarantee, registered in England and as a charity (company no. 1485560, charity no. 279756). Registered office: The Elmhirst Centre, Dartington Hall, Totnes, Devon TQ9 6EL, UK.