Advances in Genetics 13: 115-155 (1965)
EVOLUTIONARY SIGNIFICANCE OF PHENOTYPIC PLASTICITY IN PLANTS
A. D. Bradshaw

Department of Agricultural Botany University College of North Wales, Bangor, Wales
and Department of Agronomy, University of California, Davis, California

    Page
I. Introduction 115
II. Genetic Control of Plasticity 117
III. Fitness, Plasticity, and Selection 124
IV. Conditions Favoring Plasticity 125
  A. Disruptive Selection 126
  B. Directional Selection 136
  C. Stabilizing Selection 137
V. Conditions Disfavoring Plasticity 138
VI. Mechanisms of Plasticity 140
VII. Fixed Phenotypic Variation 114
VIII. Conclusions 145
  A. General Characteristics of Plasticity 145
  B. Interrelationship of Plasticities of Different Characters 116
  C. Open Problems 148
  D. Plasticity in Crop Plants 149
IX. Summary 149
  References 151

I. Introduction

One must take into account the organism's capacity for adaptive plasticity. In this regard the question which will undoubtedly receive especial attention in the future is to what extent different species or forms show different degrees of individual adaptive modification. (Nilsson-Ehle, 1914, p. 549, Trans. S. A. Cook.)

We are becoming increasingly aware that the individual cannot be considered out of the context of its environment. The way in which it reacts to different environments is as much part of its characteristics as its appearance and qualities in a single environment At the present time there is a great deal of interest in the way in which an individual can maintain stability in the face of varying environmental influences. A considerable amount of evidence has shown that this stability is under genetic control. Much of the evidence has taken the viewpoint that stability and adaptation are correlated and that lack of stability indicates lack of adaptation. But as Nilsson-Ehle implies, it seems that plasticity, or lack of stability, can be of positive adaptive value in many circumstances. This essay seeks to explore this viewpoint further.

An individual genotype assumes particular characteristics in a given environment. In a second environment it may remain the same, or it may be different. The amount by which the expressions of individual characteristics of a genotype are changed by different environments is a measure of the plasticity of these characters. Plasticity is therefore shown by a genotype when its expression is able to be altered by environmental influences. The change that occurs can be termed the response. Since all changes in the characters of an organism which are not genetic are environmental, plasticity is applicable to all intragenotypic variability.

If plasticity is given this definition then it is necessary to remember that it can have two manifestations, (a) morphological and (b) physiological. All changes are physiological in origin, so fundamentally all plasticity is physiological. Where physiological changes have predominantly morphological end effects however, we can talk about morphological plasticity. Such changes will occur during the course of development: they are likely to be permanent for the organ involved. Purely physiological changes, by contrast, can occur at any time, even in mature organs: these may be reversible and not permanent. These two sorts of plasticity will be closely interrelated, often reciprocally, since morphological stability may result from physiological plasticity. Because physiological changes are less easy to observe than morphological changes it is inevitable that most of the evidence about plasticity is concerned with the latter.

Plasticity in the sense defined here does not include variation which is directly genetic in origin. It is therefore being used more narrowly than by Salisbury (1940) who included both genetically and environmentally determined variation in his definition. His usage is ambiguous, however, since his subsequent examples were all of environmentally determined modifications.

The concept of plasticity does not also have any implications concerning the adaptive value of the changes occurring, although many types of plasticity may have important adaptive effects. Plasticity is therefore not equivalent to phenotypic flexibility as used fly Thoday (1953), since the latter term is the capacity of an organism to function in a range of environments, and may include plastic and stable responses.

Lack of stability or instability, in the sense of Mather (1953a) and others, is the term used to describe variation which is not genetic in origin and which has no observed environmental cause. Since the cause is unknown the variation appears to be random in direction. It is sometimes considered to be due to developmental errors [or developmental noise (Waddington, 1957)] arising from random changes during development, unconnected with any environmental influences. As such it can be considered to be a different phenomenon from plasticity, where specific modifications are induced by definite environmental effects.

Nevertheless, this distinction seems difficult to maintain. It is clear that in many cases the modifications included under lack of stability have distinct environmental causes, but the observations made preclude the possibility of their determination. It is difficult to believe that any such changes do not ultimately have definite environmental causes. This is supported by the experiments of Went (1953). For this reason lack of stability is considered here to be part of the general phenomenon of plasticity, and evidence concerning it will be considered. For convenience, however, the term may be retained to describe nongenetic changes where the cause is not apparent, and the changes therefore appear random. Similarly, stability can be used to indicate a condition where such changes do not occur. It can also be used generally to indicate any condition where there is lack of plasticity.

Currently the term homeostasis is widely used in many contexts. In the sense of Cannon (1932) it is the tendency for the characteristics of a physiological or morphological system to be held constant. Plasticity of a character can therefore be equated with lack of morphological (or physiological) homeostasis of that character, although plasticity of certain characters may lead to homeostasis of others. The use of the term, however, has recently been confused, and its use will therefore be avoided. More lengthy discussions of the semantic problems involved have been given by Waddington (1957, 1961) and Lewontin (1957).

II. Genetic Control of Plasticity

In order to understand the significance of environmentally induced changes, it is perhaps logical to turn to what is known of the basic causes of such changes and details of the mechanisms involved. It has been argued elsewhere (Allard and Bradshaw, 1964) that at the present, owing to the complexities of the developmental pathways concerned, the interactions between pathways and environment will be so complex that it is unlikely that much progress in understanding can be made until more detailed work on the pathways has been carried out. Considerations of basic causes, however, have led to the argument that the degree of plasticity shown by a character can be related to the basic pattern of its developmental pathway. Such a viewpoint has been taken by many authors, such as Klebs (1909). Stebbins (1950) has argued that characters formed by long periods of meristematic activity (such as over-all size, leaf number, etc.) will be more subject to environmental influences and are likely to be more plastic than characters formed rapidly (such as reproductive structures) or than characters whose pattern is impressed on primordia at an early stage of development (such as bud scales, leaves, etc.). This argument can be supported by evidence of the differences in plasticity shown by different characters in Achillea and Potentilla in the experiments of Clausen and associates (1940, 1948). The plasticity shown by the characters of these species can be related to their general pattern of development, i.e., its duration, complexity, the number of interacting processes involved, etc., as shown in the tabulation:

Plastic Not plastic
Size of vegetative parts Pinnate leaf shape
Numbers of shoots, leaves, and lowers Leaf margin serration
Elongation of stems Shape of inflorescence
Hairiness Floral characters

The contrast of the manner in which plants of determinate and indeterminate growth react to density provides further evidence. Species of indeterminate growth such as Vicia faba tend to respond to density by the number of parts formed, whereas species of determinate growth such as Helianthus annuus tend to respond by changes in the size of the parts (Harper, 1961).

While basic developmental pathways are important, it does not seem possible for them to provide an explanation of all observable differences in plasticity. In an experiment where the annual mediterranean grass, Polypogon monspeliensis was grown under low and high fertility conditions, a hundredfold variation occurred in the numbers of spikelets per panicle, while glume and seed size varied by only 10% (Bradshaw, 1958). Variation in density of plants of linseed (Linum usitatissimum) caused the following changes in different characters (Khan, 1963)

Linseed Character High density Low density L.S.D. at 5%
Capsules/plant 5.6 77 5 22.6
Seeds/capsule 8.1 9.3 0.86
Seed weight/100 seeds 6.0 5.9 0.34

In both of these cases it is difficult to see, using arguments based on development, why seed number should change so enormously with density, and seed size remain so constant. Examination of other investigations, e.g., on barley (Ariyanayagam, 1961), on Vicia faba (Hodgson and Blackman, 1956), on Agrostemma githago (Harper and Gajic, 1961), or on subterranean clover or others reviewed by Donald (1963) leads to similar conclusions.

This problem can be examined in a different manner. If it was assumed that the degree of plasticity shown by a character was the outcome of the basic pattern of its developmental pathway, certain deductions would follow. First, since general pathways of characters cannot be changed readily, it should not be possible for plasticity to change readily. Second, since the same organ, e.g., leaf, usually has the same basic developmental pathway in different species, the organ should show the same plasticity in different species.

That such arguments are not true can be seen from a comparison of the plasticity of a single character in a number of species. A conspicuous type of plasticity is the heterophylly shown by certain water plants. Such heterophylly is rarely characteristic of a whole genus; more commonly it is found only in particular species. Closely related species may differ markedly in the degree of heterophylly they show, as can be seen from the tabulated examples:

Heterophyllous Not heterophyllous
Potamogeton natans P. lucens
Ranunculus peltatus R. hederaceus
Sparganium erectum S. minimum
Juncus heterophyllus J. obtusiflorus

The occurrence of such differences was pointed out 70 years ago by Kerner von Marilaun and Oliver (1895). A detailed consideration of this type of evidence will be made later in this essay.


Fig. 1. The response of four varieties of Linum usitatissimum
to changes in density (Khan, 1963).

Marked differences can similarly be found in the amount of plasticity shown by varieties within species. Linum usitatissimum, in common with many annual plants, shows considerable plasticity in seed production in relation to variation in density. Khan (1963) has shown that linseed (oil flax) possesses a much greater capacity to respond in seed number to changes in density than flax (fiber flax) (Fig. I). Indications of similar differences of plasticity in relation to density have been reported in other crops, e.g., in soybeans (Hinson and Hanson, 1962), in cereals (Engledow, 1925), and in sorghum (Karper, 1929). Differences in plasticity within species in relation to other environmental factors are to be found throughout agricultural literature, e.g., capsule number of different linseed varieties in relation to nitrogen (Blackman and Bunting, 1954). Such differences are difficult to explain unless it is assumed that the plasticity of a character is an independent property of that character and is under its own specific genetic control.

The stability, as opposed to the plasticity, of genotypes has recently received considerable attention. Here again there is evidence that the stability of a character can vary from one genotype to another and is genetically determined. Evidence that the stability of the omnibus character of yield can vary from one genotype to another has been reviewed by Simmonds (1962) under the term "general genotypic adaptation," and the significance of this to plant breeding discussed by Allard and Bradshaw (1964). In tomatoes Williams (1960) has shown that different inbred lines can differ markedly in their stability for a number of characters and that this stability is transmitted to the F1 hybrids. An examination of his data also shows that stability levels are specific for individual characters within a single genotype and are not common for all characters of a single genotype. If the four inbred lines and three hybrids are ranked in order of their standard deviations determined in experiment 1, this is very clear (Table 1). It is confirmed by correlation analysis. None of the correlation coefficients determined between mean standard deviation values for all possible pairs of characters approached significance at the 10% probability level. Ariyanayagam (1961) has shown similar differences in the stability of different characteristics of pure lines of barley in response to various environmental conditions. Evidence of a different nature is given by the studies of chiasma frequency in rye by Rees and Thompson (1956), who showed that control of stability of chiasma frequency between plants, between cells, and between bivalents is not the same at these three levels.


TABLE I

Mean Standard Deviations and Their Standard Deviations of Ten Replicates of Four Inbred Lines
and Three F1 Hybrids of Tomato Ranked in Order of Magnitude*

Character Genotypes
59 XT/1 42 XT/2 5 XT/27 18
Fruit number 2.11 ± 0.25 3.04 ± 0.15 3.97 ± 0.37 3.37 ± 0.33 2.71 ± 0.26 3.49 ± 0.22 2.83 ± 0.21
  1 4 7 5 2 0 3
Fruit weight 0.39 ± 0.04 0.46 ± 0.01 0.28 ± 0.01 0.24 ± 0.02 0.26 ± 0.04 0.40 ± 0.03 0.48 ± 0.03
  4 6 3 1 2 5 7
Weight per plant 7.63 ± 0.83 7.08 ± 0.67 7.68 ± 0.88 8.54 ± 1,27 9.24 ± 0.96 9.46 ± 0.91 7.16 ± 0.58
  3 1 4 5 6 7 2
Flower number 1.81 ± 0.16 2.61 ± 0.21 1.76 ± 0.12 2,40 ± 0.20 2.02 ± 0.22 2.24 ± 0.24 3.05 ± 0.19
  2 6 1 5 3 4 7
Flowering date 0.66 ± 0.06 0.70 ± 0.04 2.16 ± 0.21 1.05 ± 0.05 0.90 ± 0.13 0.82 ± 0.07 1.13 ± 0.09
  1 2 7 5 4 3 6

* Data of Williams (1960)

Very extensive evidence is given by Clausen et al. (1940) in a series of experiments involving the growth of contrasting populations of a number of species under sun and shade, and wet and dry, conditions. Examination of their data shows that different characters, e.g., stem height, stem number, and flowering time, showed markedly different levels of plasticity. Moreover, their data shows that while contrasting populations differed in their plasticity for particular characters, such variation between populations in the plasticity of one character was not necessarily correlated with equivalent variation in the plasticity of another character.

Further evidence is given by Paxman (1956) on stability of various characters in Nicotiana rustica. He demonstrated that, for various leaf and floral characters, different genotypes differed in their stability. Such differences were not the property of the whole genotype but were specific for individual characters. Different characters of a single organ (parts of the flower) did possess related stabilities; this might be expected in characters very closely grouped developmentally. Unrelated characters (leaves and flowers), however, had unrelated stabilities. By comparing the stability of characters in relation to both local fluctuations and gross environmental changes, he showed that these were not necessarily related. While it could be considered difficult to distinguish between these two types of environmental variation, this observation is of considerable significance. It implies that plasticity must be considered in relation to specific environmental effects at specific stages of development. Langridge (1963) takes the same viewpoint. He has shown that stability due to heterosis in Arabidopsis under different conditions of stress is specific for particular factors. It was shown very positively in relation to high temperature, but not shown at all in relation to optimal temperature, low mineral nutrients, or other factors. It is further confirmed by the early work of Crowther (1934) on cotton. He showed that whereas internode number of the main axis was linked with nitrogen level, mean length was positively correlated with availability of water.

Plasticity is therefore a property specific to individual characters in relation to specific environmental influences. This is only an extension of the arguments of Haldane (1946) who pointed out that genotype/environment interactions are ultimately entirely specific.

In more general investigations where single characters have been examined in relation to over-all, often random, environmental influences, a relationship has been found in both plants and animals between stability and degree of heterozygosity (Dobzhansky and Wallace, 1953; Lorner, 1954; Lewontin, 1957; and others). In crop plants the evidence similarly suggests that stability of performance can be achieved by heterozygosity (Allard and Bradshaw, 1964). It is clear, however, that there is an equal amount of evidence that stability may be determined by gene systems unrelated to heterozygosity as such. This is apparent from the evidence already presented. Observations on stability of certain characters in pure lines and hybrids of the inbreeding Nicotiana rustica showed that stability was quite unrelated to the occurrence of heterozygosity in the material (links and Mather, 1955). The same situation was found in tomatoes (Williams, 1960). In maize it is clear from the work of Adams and Shank (1959) that while heterozygosity has considerable effects on stability it is not in itself a sufficient explanation of stability; hybrids belonging to the same level of heterozygosity often differed significantly in their stability. Evidence of direct genetic control of stability in wheat has been given recently (Frankel and Munday, 1963). Flower morphogenesis is very stable in normal types of Triticum aestivum but is unstable in speltoid mutants where Q is deleted. Experiments which are particularly critical are those where it has been possible to modify the level of stability by selection without any alteration in heterozygosity. In Drosophila the degree of stability of bristle pattern, determined by the degree of asymmetry, has been changed by selection and shown to be under genetic control (Mather, 1953a; Thoday, 1955; Reeve, 1960). Similarly, selection for low variance of scutellar bristles in scute flies has been effective and has incidentally caused decreased sensitivity to temperature changes (Rendel and Sheldon, 1960). Stability of development time has been altered by disruptive selection in a manner implying an increase in the environmental components of variance (Prout, 1962). Waddington (1960) has been able to increase markedly the stability of the bar phenotype in Drosophila in relation to temperature variation. In this investigation it was also possible to show that the gene system determining the canalization of the character in question was specific to that character and did not influence the degree of asymmetry of facet number.

The influence that selection may have on the stability or plasticity of a character is perhaps most elegantly demonstrated in the investigations on genetic assimilation by Waddington and others (Waddington, 1961). In Drosophila the ease of production of venation phenocopies by high temperature shock was radically altered by about a dozen generations of selection for ease of production (Waddington, 1953; Bateman, 1959). Similarly the ease of production of bithorax by ether treatment was radically altered by selection. In a character not involving a threshold effect Waddington (1959) was able to improve, by selection, the capacity of a strain of Drosophila to react to increased salt content in the medium by the development of large papillae. This is a particularly significant experiment since in this ease the changes appear to be of adaptive significance. The altered strains show increased survival in high salt media. In all these experiments the environmentally induced effect eventually became assimilated and determined genetically. Nevertheless, the significance of the results to the present consideration is that the plasticity of the genotype in one specific respect and direction could be radically altered by selection in a few generations.

III. Fitness, Plasticity, and Selection

The evidence presented so far suggests that the plasticity of a character can be (a) specific for that character, (b) specific in relation to particular environmental influences, (c) specific in direction, (d) under genetic control not necessarily related to heterozygosity, and (e) radically altered by selection.

Up to this point no attempt has been made to attach any particular adaptive significance to the ability (or otherwise) of a character to be altered by environmental influences. In many cases wide fluctuations in a character owing to environmental influences can be considered to indicate that there is a lack of adaptation and that the genotype concerned is inadequately buffered against the environment. This is particularly true in animals. Wide deviations in characteristics of animals owing to environmental effects are not usually found because of an organization which buffers the individual against the environment. Deviations can usually be considered as indications of lack of adaptation. The classic data of Bumpus (1899) on sparrows substantiate this. The viewpoint colors much of the current investigations on stability, homeostasis, etc., particularly in Drosophila. It can also be true in plants. A plant at the point of death owing to unfavorable circumstances may be very reduced and show considerable distortion. This can readily be seen in the altitudinal transplant experiments of Clausen el al. (1940, 1948). Plants grown in inappropriate environments in their experiments show a great reduction and upset of growth. Such instability can be correlated with lack of adaptation to those environments.

In all organisms, however, maximum fitness does not require the same degree of stability in all characters. As a result of natural selection those characters in which stability is paramount for survival are likely to show greater stability than those in which some plasticity is not a disadvantage. This is apparent from the observed coefficient of variation ascribable to asymmetry in wing length of Drosophila, 0.48%, compared with that for sternopleural chaeta number, 4.84% (Mather, 1953a).

The argument may be taken further. In some organisms plasticity in some characters may not be neutral in adaptive value, but of positive selective advantage, as has been argued by Baur (1930), Schmalhausen (1949), and Mather (1955). This will be particularly true in those types of organisms in which highly complex buffering processes are not present, where the organism grows in accordance with the opportunities provided it by the environment. Plants are essentially such organisms.

It has been pointed out that in studies of environmentally induced variation, it is not sufficient to know that a character is of some significance in determining fitness; the effect of its variation is also pertinent (Lewontin, 1957). In plants more is known of the effects of such variation than in other organisms. Plants therefore provide excellent material in which we may seek to discover the adaptive value of plasticity in evolution. Not all plasticity is adaptive, but some is. The conditions under which plasticity is likely to be selected for, and the nature of the plasticity, must therefore be considered.

IV. Conditions Favoring Plasticity

It is convenient to consider plasticity in plants in relation to the various types of selection that may operate at one time or another on plant populations, using the terminology of Mather (1953b). It will become clear that many different conditions may favor plasticity. If this seems to result in contradictions, it only illustrates the complexity of evolutionary processes in plants.

Two essential general points must be made at the outset. Because of their system of nutrition, plants are essentially static, fixed organisms, incapable of movement except during reproductive processes and then normally only by a rather passive scattering of propagules. The new generation falls to the ground and thereafter must endure the particular environment in which it finds itself, including any fluctuations which may occur subsequently. Because they possess complex behavioral responses and locomotory mechanisms, animals do not of necessity have to do this. The infinity of different behavioral patterns which have evolved in animals relate to the degree animals are often able to move from one environment to another, evading those which are unsatisfactory to them and selecting others (Waddington, 1957). Such behavioral plasticity has little equivalent in plants.

It should not, however, be construed that animals are always able to evade adverse conditions and have in contrast no physiological and morphological plasticity. There are many conditions such as season, chronic shortage of food, etc., which they cannot evade. In relation to these, many different forms of plasticity have been evolved, e.g., in coat color, litter size, etc. Plasticity, however, appears to play a more important part in adaptation of plants.

Plants, except those that are not holophytic, must also live in environments which provide the necessary raw materials for their growth. In other words, they must live in situations where they are likely to be strongly exposed to extremes of climate and other influences. At the same time the holophytic type of nutrition can occur efficiently only in a structure which in many important respects will be more attenuated and physically weaker than that of animals; a structure in which elaborate mechanisms buffering the organization of the individual against the environment cannot easily be developed. The only exceptions to this are plants such as cacti: but these achieve stability only at the expense of rapid growth.

A plant is therefore both vulnerable in organization and unable to move away from an environment which is unsuitable to it. Lacking behavioral plasticity, other types of plasticity are likely to be favored.

A. DISRUPTIVE SELECTION

Disruptive selection can be due to recurrent variation in selection in either time or space. If it acts on populations without reproductive isolation, it can cause either (a) genetic polymorphism or (b) plasticity. The part played by disruptive selection in causing genetic polymorphism has been discussed by Mather (1955), Huxley (1955), Thoday (1959), and others. The occurrence of genetic polymorphisms in the strict sense has been reported in plants, but much less frequently than in animals. This is difficult to understand, since the occurrence of disruptive patterns of selection on plant populations must be no less frequent than on animal populations. It is possible that the scarcity of evidence is due to the lack of observation, for little is known about levels of genetic variability in plant populations; recently, several investigators have recorded the existence of genetic variability in populations not previously suspected (e.g., Allard, 1963; Zohary and Imber, 1963). Nevertheless, the ease with which plants are able to respond to disruptive selection by plasticity is perhaps itself the reason why genetic polymorphisms are not conspicuous. Theoretical support for this argument has been given by Levins (1963).

1. Disruptive Selection in Time

If a population is subject to recurrent changes in its environment, whose duration is the same as or less than its generation time, it cannot easily respond to the contrasting environments by directly adaptive genetic changes. If the duration of the environmental fluctuation is much less than the generation time, any adaptation that occurs can only take place by plasticity. If the duration of a single environmental condition is more or less equal to the generation time, it is possible for adaptation to take place by genetic polymorphism; the particular morph that is adapted in any one generation becomes the most prevalent, the others being reduced in number. But such a situation leads to considerable elimination of individuals; the progeny of the best adapted plants of one generation tend to be the least well adapted in the next. This also applies when the oscillation of the environment is longer than that of the generation time. At the same time, any genetic adaptation that occurs is always too late (Kimura, 1955; Crow, 1955). Therefore adaptation by plasticity is more likely. This possibility was envisaged by Darwin; "I speculated whether a species very liable to repeated and great changes of conditions might not assume a fluctuating condition ready to be adapted to either condition" (letter to Karl Semper, 1881).

The only situation in which adaptation to recurrent changes in environment of shorter duration than the generation time of the population can be achieved by direct genetic changes is if the change is not random but occurs regularly at a particular phase in the life cycle. Then intrinsically timed developmental changes, which are genetically determined, can be evolved, adapting the population to the particular environmental condition concerned. For example, flowering in many annuals occurs after a certain period of growth and therefore coincides with optimal summer conditions. There is, however, a close relationship between such changes and plasticity which is discussed further in Section VII. If the changes are random and not regular, intrinsically determined changes cannot be successful. Only adaptation by plasticity will be effective.

Random changes in environment coincident with, or shorter than, generation changes are almost inevitable in plants because of the passive and therefore unselective distribution of their seeds. Environmental conditions are neither constant in time, nor are they constant in space. The seeds shed from a plant growing in good conditions will not all land in equally good conditions. The significance of this is nowhere more obvious than in annual weedy and other species. During its life an annual suffers changes in its environment, but from generation to generation the environment in which it finds itself may also vary radically. If it is to survive it must be able to grow and reproduce effectively in each condition. There will be selection for the individual which can both survive in a reduced state in difficult conditions and grow vigorously in optimal conditions. The degree of over-all plasticity, particularly of seed production, shown by annuals is in great contrast with that of perennials. The annual grass Polypogon monspeliensis has already been discussed: its plasticity of panicle size is in marked contrast with that of the closely related perennial Agrostis tenuis. If we examine, however, whether panicles are produced or not, the plasticity is reversed. Under very adverse conditions P. monspeliensis still produces panicles, although reduced in size; in contrast A. tenuis fails to produce any panicles at all. In terms of presence or absence of sexual reproduction, the perennial here is more plastic than the annual. A similar contrast can be made between Poa annua and Poa pratensis: and between annual and perennial Bromus species (Stebbins, 1964). Everyone is familiar with the plasticity of weeds of cultivation in relation to soil fertility and density. Sonchus oleraceus has been observed to vary in number of capitula from 1 to 1600 (Lewin, 1948); this is in great contrast to the related perennial Sonchus arvensis. Chenopodium album shows remarkable plasticity in comparison with C. rubrum, which may explain its success as a weed (Cumming, 1959), Annual weeds in particular are subject to considerable fluctuation in density from one generation to another. Agrostemma githago shows considerable plasticity in relation to density, although in contrast Papaver species respond to density by mortality of some individuals (Harper and Gajic, 1961). The contrast in the plasticity that can be found within a single annual species Linum usitatissimum, between flax and linseed, which have become adapted to being sown at different densities, has already been mentioned (Khan, 1963).

An extensive experimental examination of the plasticity of a range of populations within an annual species Capsella bursa-pastoris has been made by Sorensen (1954). The populations were subjected to contrasting environments by being grown in fertile garden soil or in coarse poor sand. The latter condition caused very reduced growth, reduced rate of leaf production, and considerable alteration of flowering time. But the significant fact is that populations differed very markedly in their response to these conditions, not only in amount but also in direction. Thus population 7, collected from a luxuriant cultivated field, was constitutionally early flowering. Its reaction to growth in sand was to increase the number of leaves and relatively retard flowering. By contrast population 22, collected from moist beach sand, was constitutionally late flowering. In the sand treatment its reaction was to reduce the number of leaves and relatively advance flowering. Other populations showed little or no change in leaf number or in flowering time. Such differences in plasticity could be shown to be of adaptive significance. Thus, population 22 was perfectly successful in sand, whereas population 7 failed to flower and set seed satisfactorily. These differences can be related to the original habitats of the populations.

Perennials also may show marked plasticity in particular characters. A species with a long life cycle must of necessity endure variations in environment which cannot be met by genetic changes. Many different types of plasticity may therefore be found in perennials. In contrast with annuals however, perennials do not usually show the same marked plasticity of seed production except in its presence or absence. This can be related to the fact that reproduction by seed does not play such an important part in the survival of a perennial from one year to the next and seed production is spread out over many years.

The most obvious environmental changes affecting a perennial are those connected with season. Although perhaps it would not be thought of as a type of plasticity, the deciduous habit must logically be considered a highly developed form of it. A deciduous species undergoes a startling change in shape, involving not only the loss of leaves but also the development of specialized protective structures such as bud scales, by which it becomes adapted to the severity of winter conditions. Closely related species differ in the degree to which they show this form of plasticity, e.g., Quercus robur which loses its leaves over winter, and Quercus ilex which retains them. In California the same contrast is shown by Q. douglasii and Q. turbinella (Tucker, 1952); these two species form hybrid populations where they meet, in which individuals with all grades of leaf persistence occur. Herbaceous species may retreat into a variety of underground storage organs. Seasonal variations may be related to drought. There again many highly developed types of plasticity have been evolved in adaptation to variation in water supply. For instance, Cercidium floridum, a desert shrub, produces leaves only after rain and these are shed when dry conditions return.

Innumerable other examples of the evolution of plastic responses in relation to seasonal variations are available. Different types of response may be shown not only by closely related species, but also by different populations of a single species. This is apparent from the work of Clausen et at. (1940, 1948) on Potentilla glandulosa and Achillea millefolium. One of the most conspicuous differences between the ecological races of each of these species is in the occurrence of vegetative dormancy. The high altitude populations are all winter dormant, the low altitude populations from dry areas are summer dormant, and the low altitude populations from coastal, moist habitats have no dormant period at all. Such differences can be related readily to the habitats concerned. A second particularly significant set of experiments carried out by these investigators was that already mentioned involving the growth of contrasting populations of a number of species under conditions of sun and shade, and dryness and wetness. They were able to show considerable differences in response between populations. Thus, in general, the alpine populations showed little change in general morphology, whereas those from lowland areas showed marked plasticity, e.g., Potentilla glandulosa ssp. nevadensis and ssp. typica, respectively. As has already been pointed out, however, it must be remembered that the plasticity involved only certain characters and not others. The different levels of plasticity between populations and between characters can be related to their adaptative significance in the environment concerned.

In Lolium perenne Cooper (1963) has shown that populations from different regions of Europe differ markedly in their rate of leaf expansion with temperature. The northern populations are very sensitive to temperature, showing very reduced rate of expansion at low temperatures. Southern populations are markedly less sensitive showing little reduction in rate of expansion at low temperatures. This difference can be shown to be of adaptive significance; the plasticity of leaf expansion of the northern populations gives higher survival under severe winter conditions. In Solidago virgaurea leaf expansion in response to light intensity has been found to differ in populations from habitats of differing seasonal light regimes, in a manner suggesting adaptive significance (Bjorkman and Holmgren, 1963).

The photosynthetic mechanism itself is able to alter in adaptation to variation in environment. Measurements of photosynthesis over a range of environments indicated that both rate and position of optimal photosynthesis can be altered by previous exposure to particular environments. Such changes, termed acclimation, are in the direction that implies increased adaptation to the new environments. Recently, Mooney and West (1964) have shown that five Californian species differ in the degree of acclimation of photosynthesis exhibited to temperature and suggest this may be related to the natural distribution of the species. Bjorkman and Holmgren (1963) have shown that populations of Solidago virgaurea differ markedly in their ability to acclimate photosynthesis to changes in light intensity; the differences appear to be the result of genetic adaptation. Physiological acclimation to drought, frost, etc., is also well known. Both physiological and morphological plasticity therefore occur in adaptation to seasonal fluctuations. Despite the complexity of the responses it seems that such plasticity can be evolved quite readily.

Seed dormancy and germination, whether of annuals or perennials, show very clearly the influence of disruptive selection and the development of plasticity in response to it. Seeds are likely to be most successful if they germinate when conditions will continue to be satisfactory for their subsequent development into mature plants. This will happen when the season is appropriate, when they are near enough to the surface to be able to reach it without exhausting their reserves, or when there is little competition from other vegetation, etc. If every seed that fell to the ground germinated in its own inexorable manner without relation to external circumstances, a large proportion would do so in unfavorable situations and fail to survive beyond the seedling stages. This is well shown by the breakdown of dormancy in Papaver hybrids (Harper and McNaughton, 1960). Absence of dormancy in interspecific hybrids permits the seeds to germinate in the field in autumn. These seedlings are then commonly killed by winter frosts. There are a large number of mechanisms of enforced dormancy which prevent or reduce the likelihood of this happening. In many species, e.g., Juncus effusus and Betula pubescens, only those seeds which are in the presence of light will germinate; in others the only seeds that will germinate are those which have experienced a period of low temperature, especially summer annuals, or a period of high temperature, especially winter annuals. Some require particular oxygen or nitrate concentration. Others, e.g., Stellaria media, have as yet undefined mechanisms, perhaps combinations of the preceding ones, which result in only those seeds germinating which are close to the soil surface. The mechanisms are many. They must all be considered examples of physiological plasticity, because the environment is able to alter markedly the physiological state of the seed and the occurrence of germination. In most cases the adaptive significance of such plasticity of germination is very obvious and can be related to the particular habitat requirements of the species concerned. Closely related species may differ enormously in their germination behavior, as has been described by Crocker and Barton (1957), Harper (1959), and Thurston (1960).

An important effect of variation in environment in time concerns reproductive processes, particularly pollination. A common adaptation to difficult conditions is cleistogamy, where flowers are self-fertilized without opening. Some species are permanently cleistogamous, but species are more commonly facultatively cleistogamous (reviewed by Uphof, 1938). The occurrence of cleistogamy in such species is determined by various ecological factors. The flowers of Ranunculus moseleyi and R. aquatilis (S. A. Cook, 1964), produced under water are cleistogamous, whereas those above water are chasmogamous. In Dicliptera assurgens and other species in the West Indies cleistogamic flowers are produced in the dry season and normal flowers at other times. In Bromus carinatus cleistogamic flowers are produced in dry years and chasmogamous flowers in wet years (Harlan, 1945). In Viola odorata cleistogamic flowers are produced under summer conditions. Other factors are important in other species.

These species therefore exhibit distinct plasticity in this character. In most cases the occurrence of cleistogamy can be inferred to be of adaptive value, enabling the species to overcome conditions adverse to chasmogamy. Closely related species not subject to such adverse conditions are usually found not to possess this plasticity. In some species, e.g., Sporobolus subinclusus, cleistogamy is obligate; these are then not plastic.

Agricultural situations provide interesting conditions of disruptive selection in time. Variation in crops from one season to another causes variation in environmental conditions for associated species. This is well exemplified by the variation found in Camelina sativa (Zinger, 1909 and Sinskaia and Beztuzheva, 1931, in Stebbins, 1950). This species is a weed of arable fields, and has two subspecies. The typical form is found in many crops, including flax. It has a wide branching growth form which is very plastic. When growing with flax it assumes a taller, less branched form similar to flax. Where flax is grown intensively this form is replaced by subspecies linicola, which is found only as a weed of flax. It is not as plastic and under all conditions possesses the taller, less branched growth form. It appears that the typical form has evolved a plasticity enabling it to survive in a variety of crops, whereas subspecies linicola has evolved specifically in relation to flax and therefore has lost this plasticity. Other species, e.g., Chrysanthemum segetum, Papaver rhoeas, and Chenopodium album, which are weeds of several contrasting crops, possess a high degree of plasticity and like Camelina sativa are able to compete more effectively with the crop in which they are growing.

Grazing may be considered a marked environmental effect, the incidence of which varies enormously with time. It is noticeable that species vary considerably in their response to the many different consequences of grazing. Some species tiller strongly at the base and regrow rapidly, e.g., Poa trivialis, whereas others do not, e.g., Poa pratensis. The over-all height of vegetation may fluctuate markedly, and it is interesting that some species possess the ability to adjust the height of their leaves in relation to it. The petiole of Trifolium repens must be an example of one of the most plastic plant organs known, an idea implied by Kerner von Marilaun and Oliver (1895). Plantago lanceolata achieves the same adjustment by remarkable plasticity in leaf length. Recent work on Trifolium species (Harper, 1961) suggests that species differ in this plasticity, the petiole of Trifolium fragiferum possessing greater plasticity than that of Trifolium repens. Black (1960), working with strains of Trifolium subterraneum, has shown recently that these differ remarkably in their ability to elongate petioles. The adaptive significance of this was shown in uncut swards containing a mixture of two strains. The strain with the greatest potentiality for petiole elongation rapidly became dominant.

Other factors of the environment which may vary in time are many. It would appear that examples of plasticity may be found in relation to all of them, but space does not permit their treatment in this essay. Further examples are given by Kerner von Marilaun and Oliver (1895), Davy de Virville (1927-1928), Salisbury (1940), Clements et al. (1950), and others.

2. Disruptive Selection in Space

The environment of a plant species may not only change in time; it may also change in space. If changes in space occur over reasonable distances the plant species usually adapts by the formation of localized races or ecotypes. The distances necessary for such differentiation are now known to be as little as 10 meters (Bradshaw, 1963). Changes in environment, however, can occur over much smaller distances than this, distances which may preclude the formation of genetically different populations.

The most elegant example of this is provided by areas of shallow water. Here three different environments occur in close proximity, under water, at the water surface, and in the air above the water. Although specialized plants such as Lemma species can grow in one of these environments, the water surface, without growing in the others, species which root in solid material can only grow in the upper environments by also growing in the lower. These plants are therefore subject to a violent and unchanging form of disruptive selection, which can only be met by changes occurring within each individual. For this reason some of the most remarkable types of plasticity are to be found in water plants. This has been argued by Arber (1919).

Species of Ranunculus, subgenus Batrachium, show a remarkable degree of heterophylly. Submerged leaves are finely dissected, adapted to the conditions of flowing water, whereas floating leaves are entire or lobed. The species differ in their plasticity, in whether they produce leaves which are of both types. The behavior of the British species (Clapham et at., 1962) is given in. Table 2. Experimental studies have shown that these differences in plasticity are permanent and therefore genetic characters of the species. The species that are plastic for leaf shape, that are able to produce both sorts of leaf, are all typically species of shallow water; whereas those that can produce only floating leaves are species of mud or very shallow water; and those that can produce only submerged leaves are those of deep or swift flowing water.

Similar variation in plasticity is shown by species of Potamogeton (Clapham et at., 1962; Dandy, 1961). All species produce submerged leaves, but some can produce both submerged and floating leaves. Innumerable hybrids are known; their plasticity can be related to the plasticity of the particular parents and emphasizes its genetic determination. The plasticity, as far as it is known, of the British species and their hybrids is given in Fig. 2. Further work on this group is, however, necessary.

TABLE 2
The Occurrence of Heterophylly in British Species of Ranunculus Subgenus Batrachium*

Species Habitat Leaves
Floating Submerged
R. hederaceus L. Mud or shallow water Many  
R. omiophyllus Ten. Streams and muddy places Many  
R. tripartitus D.C. Muddy ditches and shallow ponds Some Many
R. fluitans Lam. Rapidly flowing rivers and streams   Many
R. circinatus Sibth. Ditches, streams, ponds, and lakes   Many
R. trichophyllus Chaix Ponds, ditches, slow streams   Many
R. aquatilis L. Ponds, streams, ditches, river Some Many
R. peltatus Shrank ssp. peltatus Lakes, ponds, slow streams Many Many
R. peltatus Shrank ssp. pseudofluitans Fast-flowing streams Rare Many
R. baudotii Godr. Brackish streams, ditches, ponds Some Many

*From Clapham et al. (1962).


Fig. 2. Heterophylly in British species of Potamogeton subgenus Potamogeton, and in their naturally occurring putative hybrids (Clapham et at., 1962; Dandy, 1961). Circles, species; lines, hybrids: white, heterophylly present; black heterophylly absent.

This type of plasticity may be quite complex in some species and involve more than one character. Proserpinaca palustris, for instance, is an aquatic with submerged dissected leaves and entire aerial (not floating) leaves. The dissected leaves are borne on horizontal stems, whereas the entire leaves are borne on erect stems (Burns, 1904). Similar complex responses can be found in Ranunculus species. It is difficult to believe that the character of dissection of leaves and the tropic response of the stems are physiologically connected. It would appear that two characters are involved independently in the one response.

Many other aquatic species show similar plasticity, whereas other related species may not. Some examples of this have been given earlier. Further examples are discussed by Glück (1905-1924). Since it is clear that the character of heterophylly is under genetic control and can be of selective advantage, it would seem likely that variation in this character should be found within species and vary from one population to another. There appear to be no records of this in the literature. In Polygonum amphibium, however, a species which shows marked heterophylly (Massart, 1902), populations have been found which differ markedly in the degree to which they can show heterophylly (Turesson, 1950). Recently, a similar situation has been found in Ranunculus baudotii and R. trichophyllus (C. D. K. Cook, 1962, 1964), and in Ranunculus flammula and R. aquatilis (S. A. Cook, 1964).

In many other habitats similar extreme variations in environment occur over distances which are too short for adaptation to occur by genetic differentiation. Gregor (1956b) discusses an example of this in Plantago maritima. In an exposed area occupied by a fairly dwarf population, shallow depressions occurred affording some protection from the wind. In these depressions the individuals of Plantago maritima in situ were taller than those of the general area. Upon cultivation, however, in five out of the six samples examined the difference disappeared. Elsewhere in the same locality where the exposure was uniformly less, populations of a genetically taller type did occur. In the mosaic of communities on the exposed summit of Monte Maiella in Italy, Whitehead (1954, 1956) has shown the significance of modifications induced by exposure. He suggests there has been selection for single genotypes with high plasticity, rather than different genotypes adapted to different habitat conditions. He points out that species with low plasticity appear to be limited to one community type.

Another apposite example is provided by work on Ranunculus hirtus (Fisher, 1960). This is a plant which occupies the variable habitat conditions of cool forest margins in New Zealand. Its leaves show a greater degree of dissection in the pastures outside the forest than within the forest. This appears to adapt it to the relatively dry cool conditions of the pasture. The modification is controlled by temperature. Experimental studies have shown that not all populations possess the same degree of plasticity. Populations from the South Island of New Zealand show the greatest plasticity, those from the North Island less plasticity, and those of a closely related species from southeast Australia, R. plebius, least plasticity (Fisher, 1964). There appears to be an adaptive explanation for this difference in sensitivity, since the most marked variation in climatic conditions occurs in the South Island habitat and the least in the Australian habitat. Under the same forest margin conditions H. hirtus also shows plasticity in hairiness, degree of prostration, and over-all size. These characters are determined by humidity and light intensity, as well as by temperature, and appear to be independent of the plasticity of leaf dissection.

It is clear that many other examples are to be found. Some that have already been discussed in relation to disruptive selection in time are appropriate here. But only in a few cases have they been adequately investigated, and the adaptive significance of the changes been determined.

B. DIRECTIONAL SELECTION

If directional selection is very severe and the normal, directly adaptive, genetic variation is limited, further adaptation may be afforded by increased plasticity. Experimental evidence for this in Drosophila comes from the work of Thoday (1955, 1958). It is common experience that plant populations collected from severe habitats grow more luxuriantly under garden conditions, although they still retain a considerable degree of distinctness (smaller size, etc.) indicating genetic differences from other populations (Kerner von Marilaun and Oliver, 1895, Turesson, 1922, 1925; Clausen et al., 1940, 1948). This suggests that plasticity is supplementing the genetic adaptation already present. Further evidence comes from the fact that an adaptation observed in the field may be found to be genetically determined in some cases, but environmentally determined in others. Turesson, in particular, has recorded a number of examples (1920, 1922, 1925):

Prostrate maritime forms Lax shade form
Atriplex latifolium Dactylis glomerata
Atriplex patulum Dwarf subalpine form
Chenopodium album Ranunculus acris

The response to local variations in habitat by phenotypic change in Plantago maritima (Gregor, 1956b), already discussed, is another example.

Plastic response is able to provide adaptation to directional selection in some populations which in others is provided by genetic change. The individuals in the former populations may properly be considered phenocopies of the individuals in the latter populations. From the magnitude of such environmentally determined modifications, it is clear that this type of plasticity may be of considerable significance, although there is no indication how much it may vary from species to species or how much it is present in species not subject to intense directional selection.

In most cases the genetically determined forms, would appear to be relatively simple permanent changes, e.g., in Ranunculus acris and Dactylis glomerata. In the case of the genetically determined prostrate forms of Atriplex latifolium and A. patulum, however, Turesson (1920) showed that this was not so. In these forms the prostrate habit was not permanent and was lost at low light intensities. The prostration under normal conditions was due to greater sensitivity. These forms therefore retain a certain degree of plasticity which will appear under extreme conditions of low light intensity, for instance under conditions of severe competition, where it will be of adaptive value.

C. STABILIZING SELECTION

Not only can plasticity permit a single genotype to assume different phenotypes, it can also enable different genotypes to assume a single phenotype. It can therefore cover up variation in a manner analogous to dominance. Where selection is acting in a stabilizing manner it can permit the population to assume a uniform phenotype, while retaining a considerable degree of genetic variation. Many of the arguments that apply to the evolutionary significance of dominance are applicable to plasticity.

Few clear-cut records of plasticity acting in this manner are available, although common sense suggests that it could be found commonly if appropriate measurements were made. The most critical records are those of Gregor (1956a). A population of Plantago maritima, growing wild in Iceland, showed less than one-quarter the variation in plant size of a sample of the same population grown in an experimental garden. Other populations which were uniform in growth form in their natural habitat also showed considerable variation in growth form on cultivation. Turesson (1922, 1925) records a number of similar cases in which populations which were more or less uniform in their natural habitat showed considerable variability when brought into cultivation. These are listed below:

Succisa pratensis, maritime dwarf population from Hallands Vadero
Centaurea jacea, maritime dwarf population from Torekov
Atriplex littorale, unspecified maritime population
Atriplex sarcophyllum, maritime population from East Coast
Sedum maximum, maritime cliff population from Kallen and Varberg
Hieracium umbellatum, sand dune population from Vietmolle

V. Conditions Disfavoring Plasticity

Not all environmental conditions favor the development of plasticity in plants. Any condition of stabilizing selection, where deviations from the optimum are selected against, will tend to increase canalization and stability, apart from the effects it may have on genetic variance, as has been considered earlier. Thus for some plant characters constancy is of adaptive significance, and deviations lead to reduced fitness. This seems to be true of seed size. It is interesting that in comparison with other characters it shows great phenotypic stability, as has already been shown in Linum and Polypogon. It would seem reasonable to argue that constancy of flower size and shape, in insect-pollinated plants particularly, is of considerable adaptive significance; wide deviations in size could be expected to lead to failure of pollination. General observations show that flowers show much less phenotypic variability in size than other organs. Thus in the hemiparasite Euphrasia presence of a host causes considerable increase in vegetative size (x2.2) but only a very small increase in flower size (x1.1) (Wilkins, 1963). The significance of stabilizing selection in plants, however, has been discussed by Berg (1959) and will not be considered further here.

But there are other conditions where stabilizing selection is not occurring, where plasticity also may not be favored by natural selection. For instance, under some conditions of disruptive selection a system of plasticity might be adaptive in itself but only achieved at too high a cost to the organism. This is very likely with a character that is permanent, and the only means of modification is by the loss of the organ concerned and the regrowth of a new one. This will be true in perennials when adaptation to seasonal environmental variation can only be produced by many modifications of an individual during its lifetime. In these cases, spectacular plasticity is only achieved when it is accompanied by other processes which minimize its deleterious effects, e.g., in the deciduous habit where shedding does occur, but is accompanied by translocation of important materials out of the leaf before shedding.

Waddington (1957) has suggested that it may also be difficult for plasticity to be of value in annuals if the modification is not easily reversible. An early adverse period in the life cycle of an individual might so modify it that it was unable to take advantage of succeeding good periods or vice versa. Early waterlogging can cause a plant to have a shallow root system which is ill-adapted to later drought. Such environmental variations are again those of short duration relative to the total life-span. We have already seen that if the variations are of the same length as the lifespan, then spectacular types of plastic response may occur, e.g., in annual weeds. Under these conditions the irreversibility of the plasticity is no embarrassment.

It may also be difficult for plasticity to be of adaptive value if the environmental change, to which the plasticity is adaptive, occurs suddenly. The plant may be unable to assume its new state rapidly enough so that it suffers damage before it becomes fully adapted. This must be very frequent in plants, e.g., in relation to frost, storm, drought, etc. In this case, as Turesson (1922) suggested, the species may only be able to adapt by permanent genetic changes, so that it is already in the appropriate state before the critical environmental changes occur. It is difficult to see the validity of the arguments of Underwood (1954) who suggests that such situations are uncommon.

Finally, it may be biologically impossible for the plant to achieve the necessary degree of adaptation by a system of plasticity, owing to the mere limitations of the potentialities inherent in the genotypes. In other words, it may be beyond the capacity of the species to evolve an effective system.

Such practical considerations go a long way to explain why, in species that occupy a range of environments, a single genetic type with infinite plasticity is not found, but rather a myriad of genetically distinct individually adapted populations. Indeed, although adaptation by plasticity plays an important role in many situations, permanent adaptation by genetic change is more common. Most plant species consist of separately adapted genetically distinct populations whose characteristics are determined, not by the optimal conditions of growth, but by the most adverse. This can be seen from an examination of any work on the pattern of differentiation within species (e.g., Clausen et at., 1940; Bradshaw, 1959). Waddington (1957), following Warburton (1955), has suggested that such genetically adapted populations are "sewn into their winter underwear" to stop it from being blown away. Rather it would appear that either they cannot afford to possess other underwear or they are unable to change it quickly enough.

Adaptation by plasticity is exogenous; adaptation by genetic change is endogenous. The latter changes often have effects resembling the former; in most of the examples that have been discussed this is true. In this case such genetic changes can be termed pseudoexogenous (Waddington, 1957). Arguments based on adaptive value suggest reasons for this resemblance. If the same factor is involved, even if in one situation adaptation by plasticity is more successful whereas in another situation permanent adaptation by genetic change is more successful, this common factor is good reason why the two types of adaptation should resemble one another. Waddington has produced extensive evidence (review, Waddington, 1061) showing that such resemblance can arise as a result of genetic assimilation. While it is clear that this is an important phenomenon, it seems logical that the resemblance can also be due to parallel adaptation.

VI. Mechanisms of Plasticity

The development of a character can be considered to be determined by an epigenetic landscape (Waddington, 1957). The stability of a character depends on the degree of canalization. The manner in which a character may be modified by the environment is dependent on the pattern of the epigenetic landscape. The many ways in which plants may react to variations in environment show that the patterns of the epigenetic landscape are many and various.

At the one extreme is the character which shows a continuous range of modification dependent on the intensity of the environmental stimulus. This has been termed "dependent morphogenesis" by Schmalhausen (1949). The pathway of development can be considered to be a broad flat delta, so that the individual or character can follow any number of different paths to the final adult state. Such characters as height and seed number follow this sort of pathway. In these characters adjustment can be continuous. This is shown by the remarkable manner in which many species can adjust seed output per plant to density so that the seed output per unit area is constant over a wide range of densities (Harper, 1961).

At the other extreme is the character which shows a series of discrete modifications, often only two, with no intermediates. In this ease the pathway can be envisaged as two divergent steep-sided valleys with the environmental stimulus merely acting as switch. This type of plasticity has been termed "autoregulatory dependent morphogenesis" by Schmalhausen (1949). He justifiably argues that it is due to the occurrence of stabilizing selection in each phase of the regularly varying environment. Such a character as leaf form in Ranunculus peltatus follows this pattern of development. The canalization of the two types of leaf shape is considerable; there are no intermediates. Where apparent intermediate leaves do occur, it will be found that these are in fact composites of the two patterns of development and that within the leaf a sudden switch from one to the other has occurred. This has been described for British species of Ranunculus (Cook, 1963). Not all water plants with heterophylly show such a degree of canalization. In Ranunculus flabellaris all gradations of dissection of leaf may be found (Bostrack and Millington, 1962). In Sagittaria sagittifolia, where failure to form lamina occurs in submerged leaves, all grades of reduction of lamina may be found. Similar gradation can be found in Oenanthe aquatica, Sium latifolium, and Rorippa amphibia. Water plants, however, tend to show a higher degree of canalization of alternate growth forms than most other plants, which can be related to the constancy of the distinctness of the contrasting habitats available. An equivalent high degree of canalization of each of two forms is shown in the deciduous habit, which again can be related to the regularity of the environmental variation concerned.

A similar degree of canalization occurs in facultatively cleistogamous species already discussed (Uphof, 1938). The cleistogamous and chasmogamous forms are usually quite distinct in the degree of exsertion of anthers. It seems reasonable to believe that this is because intermediate forms are unlikely to be effectively pollinated. Nevertheless, there are all grades of distinctness in associated structures. In the grass Scleropoa rigida anthers and lodicules of cleistogamous flowers are little different from normal flowers. In Avena scabrivalvis, Bromus carinatus (Harlan, 1945), and other grasses, however, the cleistogamous flowers have very reduced anthers, stigmas, and lodicules. In Mayaca fluviatilis the anthers of the cleistogamous flowers are spoon-shaped and reach over the stigma effecting fertilization easily. In Viola odorata cleistogamous flowers are geotropic from the beginning, whereas chasmogamous flowers are orthotropic until fertilized. This plasticity may therefore involve a highly complex set of modifications in some species, a point which has been made earlier.

A further distinction in mechanism that can be made is whether the plasticity is between plants or within plants. This will depend in particular on the duration of life of the plant. Perennials will inevitably show within plant plasticity, whereas short-lived annuals are more likely to show between-plant plasticity. It will also depend on the reversibility of the response. Plasticity in irreversible response involving the whole plant can only be expressed between plants. Thus plasticity of seed germination is only expressed between seeds; plasticity of seed number in many annuals tends to be expressed between plants since seed formation takes place over a short period. By contrast plasticity of petiole length in Trifolium species is reversible (although it involves the production of new leaves) since a single plant, whether annual or perennial, produces leaves over a long period and the response brought about by one set of conditions can be replaced by that brought about by another. There are, however, only conceptual and not fundamental differences between these two types of plasticity. In many cases the same plasticity can be expressed both within and between plants.

It would be reasonable to expect that all adaptive modifications are determined by the occurrence of the environmental factor to which adaptation occurs. In this manner the intensity of the stimulus determines the degree of response, and the degree of modification can be precisely related to the environment, e.g., in responses to density. Goebel (1893), however, realized that water level was not itself directly responsible for changes in leaf form in some aquatic Ranunculus species. Cook (1963) has shown that change of leaf form of Ranunculus aquatilis is dependent on temperature and photoperiod and not directly on water level. In Ranunculus flabellaris Bostrack and Millington (1962) have shown that the same is true but that there is also a profound effect of water level. Therefore there may be differences between related species in the type of mechanism. The same lack of direct causal connection between the plastic modification and the environment to which it is adaptive is readily seen in the deciduous habit, for leaf fall is determined by an internal rhythm controlled primarily by photoperiod and only secondarily (if at all) by temperature and exposure. Further discussion of mechanisms is given by Sinnott (1960).

Such cases, where an adaptive modification is determined by an indirect stimulus, may have various explanations. The first is that of reliability. In many circumstances the response would be most valuable if it occurred with complete reliability, even if the stimulus were feeble. This will be particularly true of adaptations connected with seasonal changes, which are usually regularly repeating oscillations, yet which may not always be clear-cut and reliable. This is perhaps also the explanation of the mechanism in Ranunculus aquatilis and related species. The stimulus provided by the water surface will not be clear-cut, particularly in moving water, where the tips of the plant oscillate in the current. The pattern of growth and plasticity is not directly connected with it. The plant begins growing in spring from its overwintering rootstock and grows for a period producing submerged leaves; during this time it reaches the surface. But it does not change leaf shape necessarily on reaching the surface and may continue extending for a time, submerged just below the surface. Then in response to the stimulus of temperature and photoperiod it suddenly produces floating leaves on the surface. This occurs with considerable reliability.

The second possible explanation is that of preadaptation or anticipation. If a plastic response is initiated by a particular stimulus, response to that stimulus cannot occur immediately; it will take place only after an appreciable period of time. This is a necessary outcome of the normal mechanisms of growth. The only cases where response can be rapid are in various nastic movements such as those shown by stomata or the leaves of Mimosa pudica and Oxalis acetosella. Such mechanisms are, however, uncommon. Therefore, in many cases, adaptive modifications would be much too late if they were initiated by the environmental change to which they are adaptive. This will be particularly true of environmental changes that begin very suddenly, such as the onset of winter or wet season. It has already been argued that, in this case, adaptation may be possible only by permanent genetic change.

If the adaptive modification, however, can be initiated by a minor environmental change occurring regularly in advance of the major change, then the plant can be already fully adapted by the time of onset of the major change. The change will be anticipated. It would appear that exactly this has occurred in a number of eases. The most obvious stimulus of this nature available to the plant is that of photoperiod. It would appear that this is concerned in many cases of adaptive response where the stimulus is secondary in nature. Not enough is known of plasticities of this type to be able to generalize about the mechanisms involved, but it would appear that other factors, e.g., temperature, may also be involved where appropriate.

The mechanisms of adaptive modification involving secondary stimuli can be considered to be the most elaborate and advanced of the mechanisms studied. Although they may be elaborate, however, it is impossible to say they are more advanced than other mechanisms. They have, for instance, inefficiency in that since the major environmental change is not the stimulus, the degree of modification cannot be related to its severity.

Another inefficiency of this type of mechanism is shown when the pattern of the operative stimulus breaks down. Then the adaptation may not be appropriate. In Ranunculus aquatilis and R. peltatus the plants found growing out of water on mud in midsummer possess dissected leaves and not laminate leaves as would be expected. It has been shown (Cook, 1963) that this occurs because laminate leaf production is normally initiated in stem apices under water and that when those are raised above the water (a condition that would not normally occur) they will only produce dissected leaves. Ranunculus flabellaris, in which the degree of submergence does control leaf shape, does not appear to show this anomalous behavior; the loaves of terrestial plants are not capillary. The more complex mechanism may therefore lead to anomalous behavior when normal environmental patterns are disturbed, whereas the simpler mechanisms will not. But in the absence of disturbance the more complex mechanisms are extremely effective.

It would therefore appear that no one mechanism can be considered the most advanced. Each has its own advantages and disadvantages. Each appears to be related to the particular disruptive situation involved and gives a fine degree of adaptation to the different environments concerned.

VII. Fixed Phenotypic Variation

We have seen that for various reasons an adaptive modification can become determined by an indirect stimulus. For the same reasons it can be argued that an adaptive modification can be determined autonomously by the stage of growth or some other character of the plant itself.

If a seed always germinated under a specific set of conditions which are different from those experienced by the adult, modifications in relation to these could be of adaptive value. These modifications could be directly determined by the effect of the environment on the seedling. But if the particular environment always occurred at the seedling stage of development, the stage of development itself could determine the modification. This would lead to both regularity and anticipation.

It would appear that this happens in a number of cases (Goebel, 1900; Sinnott, 1960). Thus Acacia, Adenostoma, and Ulex species develop expanded leaves in juvenile stages which are not found in adults in nature. The seeds of these species develop under moist conditions not necessarily experienced by the adults and an adaptive value can be inferred. Such a type of development, termed heteroblastic, has been discussed extensively by Goebel (1900).

In water plants the early development of the seedling is inevitably below water. Correlated with this the juvenile leaves of some species, e.g., of Sagittaria sagittifolia and Alisma plantago-aquatica are always of submerged type, thin and ribbonlike, in contrast to those of the adults. These modifications can be considered to be fixed and to be related to the stage of development since they occur whether the plants are grown in water or not.

In these cases there is an apparent plasticity which is not true plasticity, since the modification occurs independently of the environment. The phenotypic variation is in fact fixed, being endogenous and not exogenous. By definition such fixed phenotypic variation must not be considered plasticity at all: it is ontogenetic differentiation. It has a close connection with true plasticity, however, for the boundary between it and true plasticity is not a clear one. The same response may at some stages be fixed, but at other stages be determined by environment. Thus, in Campanula rotundifolia adult linear leaved plants can return to the juvenile condition of rounded leaves under conditions of low light intensity.

The linear leaved juvenile state of Sagittaria sagittifolia and the juvenile states of various other species are similarly reversible in the adult (Arber, 1919). Many other cases are known (Goebel, 1900).

Such fixed phenotypic variation occurs in characters other than morphology. While some species may be facultatively cleistogamous and others obligately cleistogamous, there are some species which are permanently both cleistogamous and chasmogamous (Uphof, 1938). Panicum clandestinum bears cleistogamous flowers at the base of side tillers, as well as normal panicles; Amphicarpum floridanum bears cleistogamous flowers at the ends of rhizomes, as well as normal panicles.

Seed dormancy also involves fixed phenotypic variation. Despite the presence of mechanisms involving plasticity of seed behavior already discussed, causing seeds to germinate in conditions likely to be satisfactory for subsequent development, on many occasions these conditions might not be maintained and all the seeds that germinated would die. Therefore a species is likely to be most successful if all its seeds do not germinate at any one time in a particular habitat, however appropriate the immediate conditions may be.

It is found that many species show a between-seed variability or polymorphism in dormancy. In some cases this seems to be due to genetic variation. But in a number of cases there is no genetic component; it is determined developmentally by the parent plant. In Xanthium there are two seeds in the fruit, the lower of which germinates readily, the other only after a period of dormancy. In Avena ludoviciana there is a similar difference between the seeds of upper and lower florets of the spikelet. This is not found in the related species Avena fatua except in the subspecies septentrionalis (Thurston, 1960). In other species, particularly of the Compositae, similar differences between seeds are found.

Such variation in seed dormancy is very similar to that already considered as plasticity, but it is permanently fixed and is independent of the external environment. Like the other examples of fixed phenotypic variation it may be of considerable adaptive value.

VIII. Conclusions

A. GENERAL CHARACTERISTICS OF PLASTICITY

The many different sorts of evidence show unequivocally that the ability of plants to be modified by the environment is genetically determined. The general pattern of development of a character will exercise some control over the degree of modification possible. But it is clear that a large measure of control is afforded by the detailed developmental pathway, the epigenetic landscape, which has its own genetic determination. The degree of control is considerable. There can be rigorous canalization into one pathway, giving stability; canalization into two distinct pathways, giving the possibility of switching between two precisely determined forms and therefore discontinuous plasticity; or broad control with canalization only between wide limits, giving continuous plasticity. This control is not general to the whole genotype, but is specific for individual characters, and usually specific for individual environmental influences. All this is apparent from a study of the modifications that can be induced in the characteristics of plants and animals.

Since the degree of plasticity of a character is under genetic control, it must follow that it can be influenced by natural selection. There are many situations where the plasticity of a character can have considerable adaptive significance. As we have seen, this is particularly true in plants, because they lack the animals' complex response mechanisms of movement and behavior. As a component of evolutionary adaptation in plants, plasticity appears to play a considerable part. It also plays a considerable part in the processes of competition. Although competition may express itself by differential mortality of the interfering components, it more commonly expresses itself by differences in response (Harper, 1961; Donald, 1963).

B. INTERRELATIONSHIP OF PLASTICITIES OF DIFFERENT CHARACTERS

The fitness of an organism in a varying environment will be maximized by phenotypic changes if these (a) minimize any deleterious effects of the environment and (b) maximize any advantageous effects. Such fitness, dependent on both vegetative and reproductive characters, is the outcome of the interaction of all the component characters. We have seen that these characters can each have their own plasticity and response to environmental variation. It must therefore follow that maximal fitness can be obtained in a number of ways, by response in some characters and stability of others. If a number of interacting characters are all plastic, it is possible that maximal fitness can be obtained by adjustment of one character in one species and by adjustment of another character in a second. An analysis of such interrelationships for yield in cereals has been presented by Grafius (1956).

Nevertheless, essential attributes and balance of organization must be maintained and limits will be set by particular characters which lack the ability to respond or in which response could only be deleterious. We may therefore find increased vegetative growth that may not be achieved by over-all increase in size of all organs equally because of structural limitations, but by increase in their number. There may therefore be stability of certain characters which is compensated for by plasticity of others.

Maximal fitness will also involve interaction between physiological and morphological characters and their plasticities. Physiological plasticity may permit morphological stability. But if plasticity in a particular physiological system is not possible, the fitness of the organism may be maintained by morphological adjustment.

In reproductive growth the same complex interactions will be found. For various reasons constancy of flower and seed size is important. Maximization of fitness therefore usually occurs by variation in the number of flowers. In some organisms, however, there may be variation in some floral characters, such as numbers of seeds per flower.

Interaction between vegetative and reproductive characters will also clearly occur. Maximization of fitness will not necessarily occur by unlimited response of both. After the onset of adverse conditions to an annual plant, further vegetative growth may seriously reduce the possibility of subsequent reproductive growth. Under these circumstances we may find time development of a mechanism diverting vegetative growth into reproductive growth.

The occurrence of such interactions is well exemplified by the responses of populations of Capsella bursa-pastoris, studied by Sørensen (1954), already discussed. Poor growing conditions cause a reduced rate of growth and leaf production which the species cannot overcome apparently by physiological plasticity. As a result flowering, which usually occurs after production of a certain number of leaves, tends to be later. Some populations can, however, under poor conditions, reduce the number of leaves produced before flowering and thereby maintain the date of flowering under poor conditions that they usually show under good conditions. Populations not possessing such plasticity of leaf number are unsuccessful under poor conditions.

There is therefore a hierachy of plasticities. In the evolution of processes maximizing fitness a variety of different solutions may be developed in different plants. This can be seen by comparison of plant species. The essential common character of all such solutions will be that some characters of the plant will for various reasons be held constant, whereas others will be permitted to vary, and therefore show high plasticity. Characters which are held constant can be properly said to show homeostasis (in the sense of Cannon, 1932) or canalization (in the sense of Waddington, 1957). For such constancy can be considered to be at least in part the outcome of the plasticity of the other characters. Which characters are held constant and which vary depend on the exigencies of the structure, physiology, and environment of the species in question.

C. OPEN PROBLEMS

In view of the integral and complex part played by plasticity in adaptation it is remarkable that there are so few critical analyses of specific eases of plasticity in plants, particularly of its genetic control and response to selection. It can be argued that plasticity is under genetic control. An attempt has been made in this article to stress the eases where marked differences in plasticity exist in closely related species or varieties, since this is circumstantial evidence of the ease with which plasticity can be changed by selection. But the direct experimental evidence is meager. In order to fit the concept of plasticity into our framework of evolutionary principles, we need to know the amount of genetic variability for plasticity that is available in natural populations, its genetic control, and the ease with which it can be selected. The same point has been emphasized recently for animals (Mayr, 1963).

It seems also necessary that we should have more primary evidence on the occurrence of different types of plasticity and their underlying mechanisms. The spectacular types of plasticity involving a number of major characters are well documented and are becoming well understood. The simpler types of plasticity are more common, yet have rarely been studied critically except in crop plants. The evolutionary relationship between the two types of mechanisms is not clear. Much of the recent work on plant adaptation has eschewed very carefully any consideration of plasticity. Any modifications induced by the environment during the course of an experiment are usually considered only an embarrassment.

In this connection it would be particularly valuable to know the mechanism by which several unrelated characteristics can be associated in one particular plastic response. We are familiar with the fact that in genetic polymorphisms several characteristics can be determined by one switch gene. This can be due either to a common system of canalization so that a number of different processes all take place as the result of a single switch. It can also be due to the switch being in effect a set of ganged switches operating separately on a set of independent characters (Mather, 1955). It seems reasonable to believe that the same alternative systems exist in mechanisms of plasticity, with the environment and not a gene being the switch. But there is little evidence giving proof of this. The deciduous habit seems to be an example of the first system. The second system seems to occur in Ranunculus hirtus where the characters associated in the one response appear to be controlled by separate environmental factors. An analysis of the plastic response of segregating hybrid populations derived from crosses between individuals with contrasting plasticities and a search for mutants affecting single components of a response would be valuable.

A third aspect of mechanisms about which we need information is the degree to which the direction of response can be controlled. An essential part of the arguments that have been presented is that the direction of response can be controlled by the organism. The evidence already presented unequivocally shows that responses can be in particular directions, e.g., in the adaptation of Drosophila to salinity (Waddington, 1959) or in the flowering of Capsella bursa-pastoris (Sørensen, 1954). Indeed, without such control of direction, the evolution of adaptive responses would be impossible. But we know very little about this control. Initially, plasticity may be undirected: that which is normally considered lack of stability or developmental noise is perhaps of this nature. But at some stage in the evolution of a plastic response randomness in direction and in extent must be replaced by fixation of direction and extent.

D. PLASTICITY IN CROP PLANTS

Finally, it must be realized that an understanding of plasticity is important not only to the framework of evolutionary theory, but also to the practical problems of plant improvement. The environment of any crop plant varies from season to season, from locality to locality, and even from place to place in the field, and much of this variation cannot be controlled by the farmer despite cultivation (Allard and Bradshaw, 1964). The plant breeder is therefore concerned with the stability of final yield. Stability of final yield may be due to inherent stability of the crop; but it may also be due to plasticity of the components of final yield (Grafius, 1956). This is very clear from the reaction of crop plants to density (Donald, 1963). Plasticity is therefore as much an essential part of the adaptation of crop plants as of wild species.

IX. Summary

The expression of an individual genotype can be modified by its environment. The amount by which it can be modified can be termed its plasticity. This plasticity can be either morphological or physiological; these are interrelated.

It can be argued that the plasticity of a character is related to the general pattern of its development, and apart from this, that plasticity is a general property of the whole genotype. A review of the evidence suggests that neither of the conclusions is tenable. Plasticity of a character appears to be (a) specific for that character, (b) specific in relation to particular environmental influences, (c) specific in direction, (d) under genetic control not necessarily related to heterozygosity, and (e) able to be radically altered by selection.

Since plants are static organisms, plasticity is of marked adaptive value in a great number of situations:

  1. Disruptive selection in time. Where changes in environment are of the same or shorter duration than its generation time, a species cannot adapt by genetic changes, but only by plasticity.
  2. Disruptive selection in space. Where changes in environment occur over very short distances, adaptation by the formation of genetically different populations may be precluded. In these conditions very spectacular types of plasticity may be evolved.
  3. Directional selection. If selection is very severe and directly adaptive genetic variation is limited, further adaptation may be afforded by plasticity.
  4. Stabilizing selection. Plasticity permits different genotypes to assume the same phenotype. Under certain circumstances, this may permit the retention of genetic variation in a population in a manner analogous to dominance.

Examples of all these situations in plant species are discussed. They indicate that adaptation by plasticity is a widespread and important phenomenon in plants and has been evolved differently in different species. There can be a considerable interrelationship between the plasticities of different characters; plasticity of one character can allow stability of another. The plasticity of a species must therefore only be considered in terms of the plasticity of its individual characters.

The mechanisms involved are varied. At one extreme the character may show a continuous range of modification dependent on the intensity of the environmental stimulus. At the other the character may show only two discrete modifications. The stimulus causing these modifications may be direct, i.e., the environmental factor to which the adaptation occurs, or it may be indirect, i.e., an environmental stimulus not part of that to which adaptation occurs. Under certain circumstances the modifications may be determined by the stage of growth of the plant itself. This is no longer true plasticity, for it is fixed and independent of the external environment: the adaptation is endogenous and not exogenous. The mechanisms found can be related to the particular environmental situation involved.

ACKNOWLEDGMENTS

I am grateful to the Trustees of the Leverhulme Research Awards for the award of a Fellowship and to Professor B. W. Allard of the Department of Agronomy, University of California, Davis, for his hospitality and stimulus during the period of the fellowship.

I would like to thank the many friends with whom I have discussed this problem on many different occasions, Mr. J. E. Dandy for the information concerning Potamogeton, Dr. S. A. Cook for the quotation of Nilsson-Ehle, and Drs. C. D. K. Cook, S. A. Cook, F. J. F. Fisher, Professor J. L. Harper, Mr. J. G. Pusey, and Professor G. L. Stebbins for their critical reading of the manuscript and valuable suggestions. I owe much to their guidance and help. The views expressed in this article must, however, remain my, and not their, responsibility.

REFERENCES