Unravelling phenotypic plasticity – why should we bother?



The ability of a genotype to change its phenotype was once considered rather a nuisance – making it difficult to define a genotype. This led to the idea that there was a problem called ‘instability’. But quite early it was recognized that stability was under genetic control, and was a character like other attributes of an individual. From this realization came the idea that there were two sides to the character of ‘instability’, and that the ability to change could be important. This ability was thus given the title of ‘plasticity’. Once recognized, it became clear from surveys of different species and populations that plasticity can (i) be a complex character, and (ii) be selected to fit species to the particular demands of different environments. For plants, which cannot meet variations in environment like animals by behavioural responses, phenotypic plasticity can be very important. Plants should therefore be valuable tools for unravelling the mechanisms of plasticity whilst also demonstrating its contribution to fitness experimentally. We ought also to be able to demonstrate that appropriate genetic variability is available through which complex responses can be built up by selection. Genes must exist not only to determine character means, but also to determine character response, which adds interesting complexity to our ideas about evolution.


Upon receiving an invite to open the 14th New Phytologist Symposium on ‘Plant ecological development’ (see accompanying meeting report; Ackerly & Sultan, this issue, pp. 648–653) it seemed plain to me that the last time I considered the topic of phenotypic plasticity properly was a long time ago. Yet this topic has never really left me since the time when I was studying adaptive evolutionary differentiation in the 1950s, and kept finding that about half of what I saw was phenotypic – which few people seemed to be bothering about. Today, of course, the study of phenotypic plasticity is considered central to ecology and evolution. So to have been in amongst a lot of people at the cutting edge of a revitalized topic at this meeting was a great delight; here I outline some of my thoughts on this topic as presented in my keynote address.

Early ideas – on stability

In my early days as a scientist, the ability of a genotype to change its phenotype was considered just an annoyance – a confusion that made it difficult to know what a genotype ‘really’ was – in other words how its characteristics could be defined. From this came the idea that there was a phenomenon that could be called ‘instability’– a source of error, as when we refer to an unstable person as not always to be relied upon.

But this instability can be a serious matter. At its simplest, it can imply lack of adaptation – indicating that a genotype or a character can easily be pushed around. This is nowhere more important than in crop varieties, where variation in yield can be a source of great trouble – scarcities leading to serious supply difficulties, and gluts leading to collapse of market prices. Because of the postwar concern to produce enough food, a lot of work was being done on this issue of stability, leading to sophisticated systems of cultivar assessment (e.g. Finlay & Wilkinson, 1963; Eberhart & Russell, 1966). The wheat cultivar Capelle, for instance, was a very popular cultivar of the 1960s. It was not as high yielding as some of its contemporaries, but was popular because its yield was very reliable. Quite early, geneticists working with both Drosophila (Mather, 1953; Waddington, 1960) and crop plants (Paxman, 1956; Williams, 1960) started to show that stability was under genetic control like other attributes of an individual. Thereafter, in the 1960s, plant breeders started to pay attention to stability as a character that could be selected for, and academics began deedy discussions about fitness, and how stability might be made up (e.g. Waddington, 1957). Stability in functional performance and fitness is often given the grander Greek term ‘homeostasis’, while the developmental processes that cause trait stability are termed ‘canalization’. It remains an important topic for anyone interested in adaptation.

Later ideas – on plasticity

With this work on stability came the realization that there were two sides to this important character, and that the ability to change – (which I termed ‘plasticity’) could be important as well as fascinating, especially in plants. This plasticity, of course, has a very significant relationship to stability. It can be a simple sign of weakness – of lack of fitness – mentioned by many authors, but it can also be a sign of strength, reflecting mechanisms maintaining fitness. If a plant grows smaller in a windy habitat it might be about to die; but it could be demonstrating that it has a special mechanism for avoiding wind damage. Interestingly, botanists had recognized the phenomenon and its evolutionary significance quite a long time previously (e.g. Kerner, 1895; Nilsson-Ehle, 1914). Even Darwin had thoughts on it which he described in a letter to Karl Semper (in 1881). Because plants cannot move to avoid stressful conditions, as can animals, they have to put up with what comes. Phenotypic changes that minimize stress therefore seem understandable.

Indeed, once the significance of plasticity was recognized, it was realized that it should be a character capable of contributing to evolutionary adaptation like any other character. To investigate it further, comparative studies are a good place to start. These involve looking at related species and populations occupying different environments. Related populations are perhaps best, because they will have close genetic affinities, and any observable differences will be more likely to be related to present environments. Differences between species may have evolved in past environments experienced by the species, and may not be easily related to present conditions. Nevertheless, so long as this is appreciated, species differences can be very valuable.

The outcome of comparative surveys has been the revelation of a wealth of examples, involving all sorts of different species and different mechanisms. These have been reviewed by many authors (such as Kerner, 1895; Salisbury, 1940; Bradshaw, 1965; Sultan, 2000). The wealth and diversity of examples are still worth contemplating – they have a lot to teach us. Perhaps one of the best examples to make one think is the phenomenon of etiolation. We all know that most plants get long and thin in shade – and it may seem pretty obvious that they should. But is it obvious? With the production of less photosynthates as a result of shading, surely it would be more likely that plants would grow less. But they do not – at least not in length, and we now know this reaction is controlled by a fairly subtle bit of machinery (Smith, 1990; Schmitt, 2003). Equally importantly, not all species etiolate to the same extent (Morgan & Smith, 1979). Some woodland species, such as dog's mercury (Mercurialis perennis), hardly etiolate at all – which, when you come to think of it, is highly sensible for a woodland plant. Then, when we examine the machinery involved, it is interesting to discover that both the receptor and the responding organ can be in surprising and different places in different species. You will all know the plastic response to shading of the petioles of Trifoliun repens, recognized first by Kerner. In other legumes, stems and not petioles are the place of response. All this may seem curious. But in fact it is logical, relating to the morphology of the species and the environment in which it grows.

I once worked on the taxonomy of the two British species of hawthorn, Crataegus monogyna Jacq, the common species found in hedgerows and scrub, and Crataegus laevigata (Poir.) DC, the closely related Midland hawthorn (with which it hybridizes) typically found in fairly dense woodland. In scrub conditions they look the same. But sometimes they can be found together, caught up in an area of young woodland or coppice. Under these circumstances C. monogyna grows up vertically with its green branches high up in the woodland canopy and all its lower branches shed, with its leaves in a higgledy-piggledy arrangement. So it looks rather like a lavatory brush chasing the light. In the same conditions, C. laevigata produces a series of spreading branches near the ground with its leaves in a more or less perfect mosaic. Evidently these closely related species have evolved to express quite different architectural plasticity in shaded conditions.

The plasticity of growth form is particularly fascinating in water plants, where it can be related to the fact that, if a species that roots in bottom muds is to exploit the complete range of environments available to it, it has to cope with three environments – underwater, water surface, and aerial conditions – for which different leaves are necessary. Trout fishermen can see this necessity if they watch the ways in which the surface leaves and the underwater fronds of water crowfoot, Ranunculus peltatus, cope (or fail to cope) with the current of a chalk stream. It is equally interesting to contemplate that the switch mechanism for these distinct leaf types depends on the species. It can be the presence of the water surface or a fixed switch related to the time of the year.

A plethora of such observations is possible. More generally, we can conclude that plasticity entails the following characteristics:

  • 1specificity for a particular character;
  • 2specificity in relation to particular environmental influences, including patterns of environmental variation;
  • 3specificity in pattern and direction;
  • 4having specific adaptive values or maladaptive effects;
  • 5being under the control of quite elaborate mechanisms of perception and expression;
  • 6being under genetic control; and therefore
  • 7being able to be radically altered by selection to fit the demands of different environments.

All this is what we might expect of a character or property that contributes to evolutionary adaptation. But it is always possible to think that plasticity is an accident – just a by-product of the way plants are made up, and just an incidental part of other more important characters of an organism. Geneticists and others, such as Via (1993), suggest that this nonadaptive plasticity is a correlated response resulting from pleiotropy, with the implication that it is some of the froth of evolution. Some perhaps is. However, other aspects of plasticity, such as the etiolation response, are not the predicted outcome of, for example, low light but rather something more positive and organized. Greater understanding of the mechanisms of plasticity may allow us to better interpret individual cases as either adaptive or accidental.

Some thoughts on future directions

It seems to me that we should be thinking/working on the occurrence and properties of plasticity in different situations to see what more we can discover and unravel. This unravelling will involve many different approaches, as this plant eco-devo symposium has amply shown (Ackerly & Sultan, this issue, pp. 648–653).

Our starting point must be the realization that knowing the mean expression of a character is no longer good enough. Firstly we need to know its variance, its eco-devo, which means we will need to know its response to environmental conditions. But these conditions can be of different sorts – light, temperature, drought, etc. This should lead to some complicated multifactorial experiments, the value of which has been shown by Pigliucci et al. (1995), and we have to stop thinking that there is a single condition called ‘stress’ (Bradshaw & Hardwick, 1989). To understand how organisms cope with various demands, responses to each element or aspect of stress must be separately investigated.

Then there is the need to relate the plasticity, the eco-devo response, to details of the environment (the type of variation, its spatial scale, its timing, the cues available and their reliability, the speed and cost of the response, and its reversibility), and to its effects on fitness, as has been well argued by Dudley (2004).

The need to know more about the nature of the mechanisms involved will, to many investigators, be the essence of work on plasticity. Such work is crucial to our understanding. But I hope that it can put plasticity in perspective in terms of adaptation so as to reveal the subtleties of this remarkable plant character. There are a number of fascinating early examples of such work, such as that on Capsella (Sørenson, 1954) and Solidago (Bjørkman & Holmgren, 1963), as well as more recent studies on the whole matter of response to shading (Schmitt et al., 2003). But when looking for explanations, we must beware of falling into an a posteriori trap, and just look critically at what is involved.

The genetic underpinnings of plasticity remain a poorly understood subject. The surveys involving species or population comparisons already mentioned provide crude evidence of genetic control. Specific mechanisms, however, can only be revealed by proper genetic experiments. One approach is to look for ‘plasticity’ genes and QTLs, using molecular techniques. A more old-fashioned yet still extremely informative method is traditional crossing experiments from which both the genetic components and the heritability of plastic responses can be determined. For instance, in the 1970s Khan did some elegant work on flax density response which has not received the attention it deserved (Khan et al., 1976). He examined 60 F3 families from two different crosses between flax and linseed. These he subjected to two different environments (6-inch and 1-inch spacing). Figure 1 shows the results of one of his experiments. If you follow these and the further details in the published paper, this single experiment shows that plastic responses can (i) vary between genotypes, (ii) be highly heritable, (iii) have different unconnected components, and (iv) that these can be unconnected across environments. It also shows (v) that in a cross, transgressive segregants can turn up. Such a Mendelian approach allows the components of the plastic response as well as the genes involved to be sorted out.

Figure 1.

The correlation of height and capsule response of a random set of 60 F3 lines, from crosses between cultivars of Linum usitatissimum, to high (1-inch spacing) and low (6-inch spacing) density. The distribution of responses and the relationship between the responses of the two characters are shown, where response was measured as the ratio of the value at wide spacing to that at close spacing. Cultivars: Redwing O, Maroc X and Weira V. The error is indicated as the least significant difference at 5%. For more detail, refer to Khan et al. (1976), from which the figure was adapted.

The last property of plasticity that is crucial is whether or not it can be selected. In many ways, if we knew enough about what I have already discussed, we could happily conclude that plasticity was a character that must be able to play a significant part in evolution. Differences found between populations are a fortiori evidence that the character can play its part in evolution by natural selection. The wide range of variability that can arise in crosses is even better. But the best evidence is the progress that can be achieved by direct selection over several generations, such as in the work by Scheiner & Lyman (1991). More such experiments are needed.

Similarly, studying cultivars produced by plant breeding can reveal the potential for selective differentiation within a species. For example, the two types of Linum usitatissum– flax and linseed – have extraordinary differences in density response, which must have been brought about by selection out of the ancestral Linum material (Khan & Bradshaw, 1976). The differences between wild and cultivated sunflowers are an excellent similar example (M. I. Khan, in Bradshaw, 1973).


Clearly, plants can have different and complex patterns of behaviour, of ecological development or ‘eco-devo’. What still needs to be done, however, is to disentangle more of the underlying mechanisms and the genes responsible. If genes exist that not only determine character means, but also determine character response, we have something very interesting to contribute to ideas about evolution and adaptation. Perhaps, as suggested by Trewavas (2005), we shall have to be talking about plants having intelligence.


I am very grateful for the invitation of the organizers to stir from my retirement and deliver this paper at the 14th New Phytologist Symposium, and to Sonia Sultan for nursing this written version into a better shape.