SEARCH

SEARCH BY CITATION

Keywords:

  • communication network;
  • intelligence;
  • intention;
  • purpose

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

The nature of plant behaviour is discussed, and it is concluded that it is best described as what plants do. The possibility that plant behaviour is simply signal-induced phenotypic plasticity is outlined, and some limitations of this assumption are considered. Natural environments present many challenges to growing plants, and the consequent signalling that plants perceive is becoming extremely complex. Plant behaviour is active, purposeful and intentional, and examples are discussed. Much plant behaviour, concerned with stress and herbivory, is also based on an assessment of the future likelihood of further damaging episodes and is therefore predictive. Plant behaviour involves the acquisition and processing of information. Informational terminology provides a suitable way of incorporating the concepts of learning, memory and intelligence into plant behaviour, capabilities that plants are rarely credited with. Finally, trade-offs, cost–benefit assessments and decision making are common plant behavioural attributes. It is suggested that intelligent assessments that involve the whole plant are essential to optimize these adaptive capabilities.


BEHAVIOUR IS WHAT PLANTS DO

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

Defining plant behaviour

The life cycle goal of any individual plant is optimal fitness, usually equated to maximum numbers of viable seedlings or more conveniently, for experimental purposes, the numbers of seeds. Seed yield is known to be dependent on lifetime acquired resources (carbohydrates, minerals and water, i.e. food), extent of predation and success in reproduction. Similar fitness requirements exist for animals – acquisition of adequate food, avoidance of predators (or catching prey) and successful reproduction. In animals, all these behavioural processes involve movement, and movement is recognized as the basis of animal behaviour. Higher plants spend their life cycle rooted in one position and, to the casual observer, exhibit no movement, with only rare exceptions like Mimosa. How then can plant behaviour be described?

‘Among plants, form may be held to include something corresponding to behaviour in the zoological field. The animal can do things without inducing any essential change in its bodily structure. When a bird uses its beak to pick up food, the beak remains unchanged. But for most, but not all, plants, the only available forms of action are either growth or discarding of parts, both of which involve a change in the size and form of the organism’ (Arber 1950, p. 3). This statement identifies phenotypic plasticity as a form of action in plants, that is, plant behaviour. The Latin word habere, from which the word behaviour is derived, means ‘having’ or ‘being characterized by’. Arber's statement indicates that plant behaviour is action, that is, doing something. Behaviour is then what a plant does, rather than something it is characterized by or has.

Behaviour and plasticity

In a seminal chapter, Silvertown & Gordon (1989) defined plant behaviour as the response to signals, and this, along with Arber's description, equates plant behaviour with the phenomenon of phenotypic plasticity (Trewavas 2003; Sultan 2005; Karban 2008). Ecologists describe plasticity in terms of ‘norms of reaction’ that specify the boundaries of plastic variation to individual signals (Schlichting & Pigliucci 1998; Sultan 2000). Not all tissues exhibit plastic responses (Bradshaw 1965).

Phenotypic plasticity is not unique to plants however. Plant behaviour can, and indeed should, express a phenotypically local response to local signalling, but so can that of other organisms. For example, human weightlifting specifically increases the development of the muscles most involved. The real difference between plant and animal behaviour was again indicated by Arber (1950, p. 136). ‘The individuality of the mammalian body is of a much more fixed character; that body consists of a limited number of organs which were once and for all marked out in the embryo. With its parts arranged in an ordered hierarchy there is no such thing as indefinite succession of limbs, of branches of limbs, numerically unfixed and liable to impede one another but this is what we find among plants’. Movement is essential for the higher animal lifestyle. Only with accurate replication of limb numbers and complex coordination among them could this be reproducibly achieved. Thus, crucial embryological tissue specification is limited to the protected environments of the egg or uterus, and subsequent plasticity is constrained to more marginal changes in already specified organs. The potential for plasticity is considered to have a genetic basis, but its realization must be epigenetic.

Most higher plants have a modular structure, and the plant body is plastically constructed from variable numbers of leaves plus buds and branch roots. Plasticity enables the phenotype to accurately occupy local space, change its phenotype as it grows, forage accurately for resources, competitively exclude neighbours and construct, within genetic/environmental limitations, its own niche. The niche concept involves little understood competitive and cooperative two-way signalling between individual and environment that is important in community structure (Muller-Landau 2003; Uriarte and Reeve 2003; Silvertown 2004; Donohue 2005; Kelly et al. 2008; Liebold 2008). Badri & Vivanco (2009) in this issue reviewed recent information on root exudates that contribute to niche construction.

Predation is inevitable for wild plants, but numerous dormant meristems, regrowth and often extraordinary regenerative capacities can diminish but not eliminate the potential reduction in fitness. It is also the reason that plants do not place critical functions in one or two tissues as animals do with heart or kidneys. Such specialization would make the individual extremely vulnerable to even slight predation. However, the phenotype is holistically determined. Excision of either a whole shoot or root inhibits further plasticity changes until regeneration of the lost organs is completed. Moreover, fitness itself is a function of the integrated phenotype, not just the behaviour of individual tissues.

Limitations to equating behaviour just to signal-induced changes

There are several problems with equating plant behaviour only with plasticity. The term ‘reproductive behaviour’ is often used to distinguish whether reproduction is sexual or vegetative without regard to plasticity changes. Some species do have separate male and female flowers, and environmental conditions can change the proportions of each, implicating plasticity, too (Trewavas 2007b). Tsvi Sachs objected to essential developmental processes like germination being classed as behaviour (quoted in Silvertown 1998). On the other hand, there is certainly inherent plasticity in the germination phenotypes of almost any species.

A further complication is the potential lack of reversibility in phenotypic responses raised in this issue by Metlen, Aschehoug & Callaway (2009). As they pointed out, behavioural plasticity in secondary metabolite production is reversible. But in contrast, abscission can be used to substantially reverse phenotypic changes, and innate animal behaviour certainly appears irreversible.

A further issue is whether there is an intrinsic control phenotype, generated in the absence of signalling and often assumed to be a growth-room phenotype. This perception is compounded by the relatively uniform appearance of many field crops, a situation that has arisen because crop plant breeders have (1) eliminated much signal-induced behaviour and (2) selected only a few individuals (genotypes) for subsequent breeding out of a much greater range of behavioural and genotypic variation (Lewontin 2001). But all phenotypes are constructed from a complex two-way conversation between genes and the total environment. Growth-room environments are perceived and used to specify only one out of a range of phenotypes. All that genes can ever do is specify a norm of reaction; they are not invariant determinants of phenotype or behaviour (Lewontin 2001).

ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

Many signals are perceived

What growth rooms cannot mimic is the enormous complexity of the external environment experienced by the wild plant. Behaviour is inextricably linked to environmental signalling. Because plants are sessile organisms, they may perceive more environmental signals and with greater sensitivity and discrimination than the roaming animal. ‘If etiolated seedlings are placed between two sources of light differing so slightly that the differences cannot be detected by ordinary photometric methods, the seedling always bends promptly towards the source of the more intense light’ (Palladin 1918, p. 246) is certainly indicative.

In this special issue on plant behaviour, many articles deal with particular kinds of environmental signalling. Foraging is described as the behavioural alterations that enhance the uptake of essential resources and De Kroon et al. (2009) highlighted both the local and integrated signalling that underpins these vital processes. Mott (unpublished data) described the systems behaviour of complexes of stomatal cells that are crucial for foraging for carbon dioxide. Forde and Walch-Liu (2009) also reviewed the important role of amino acids and nitrate in constructing the root phenotype. The shoot phenotype is dependent on the presence, absence and crucially the identity of neighbours (see pictures in Bazzaz 2000, p. 114), and these may reflect the ubiquity of competition. Ballaré (2009) emphasized the critical role of phytochrome in both light foraging, overall resource allocation, herbivore defence and thus shoot phenotype construction.

Volatile (gaseous) chemicals are increasingly seen as important in plasticity. Well-established information has already reported the phenotypic alterations induced by humidity (water vapour), carbon dioxide, oxygen (in flooding responses and 10% oxygen environments), ethylene, ozone and, more recently, nitric oxide. But as Galis et al. (2009) and Dicke (2009) reviewed here, signalling via volatile chemicals is of crucial importance in herbivory resistance. Particular combinations of volatiles overcome vascular constraints on systemic signalling (Frost et al. 2007), while other combinations signal adjacent plants (Karban 2008). Different groups of volatiles again can be herbivore specific, attracting predatory, parasitoid wasps (De Moraes et al. 1998). Dodder, a parasitic plant, also homes in on its host by detecting stem-released volatiles (Runyon, Mescher & De Moraes 2006), observations critical to understanding how Dodder maximizes the energy gained from the host while minimizing its energy investment in coiling and haustoria formation (Kelly 1992). Apart from ethylene, specific receptors for these volatiles await characterization.

Other kinds of signalling have been detected, but their current molecular basis remains unknown. As both Hodge (2009) and Novoplansky (2009), again in this issue, indicated, root systems are not only able to sense the soil volume in which they grow but can recognize and discriminate against the roots of adjacent conspecifics and thus possess self-recognition. [Astonishingly bacteria have self-recognition (Gibbs, Urbanowski & Greenberg 2008), indicating perhaps the ubiquity of self-recognition processes.] Potentially, the roots of any individual plant avoid each other as far as possible to improve the extent of soil space occupied and exhibit a kind of territoriality (Schenk, Callaway & Mahall 1999). Furthermore, the root system exhibits holistic responses to the patchy environment experienced. These observations do imply complex signalling below ground; could these root signals be presently unknown volatiles, too (Erb et al. 2008)?

Other biotic signalling results from competition for soil resources, from mycorrhizal and cooperative bacterial interactions and from allelopathic chemicals, disease, mutualism, trampling and, finally, plant cooperation (Kelly et al. 2008). Some of these signals, like disease and bacterial cooperation, are relatively well understood; the others are less well characterized.

What makes for much greater complexity is that many of these signals arrive coincidentally. Decisions among often conflicting signals have to be made and priorities determined on phenotypic change. Leyser (2009), in this issue, described the role of auxin in leaf and branch initiation in which a coherent model is beginning to emerge. The abiotic signals of light, gravity, mechanical signals, soil structure and composition, minerals and water availability add to the difficulties for the growing plant because each, like the biotic signals, varies in direction, length of signalling and intensity. This enormous complexity of signalling ensures that no plant behavioural response is autonomic, a kind of behaviour that requires complete replication under all environmental circumstances. Selection will favour individuals that can best assess the probabilities of particular kinds of behavioural action and optimize their fitness.

This enormous signalling/environmental complexity is best conceived by the reader as a complex but changing topological surface composed of hills and valleys, and the successful plant (from seed to flower) must navigate its way through this topological and hazardous environmental terrain, which keeps changing in structure, with minimal expenditure of energy. Bazzaz (2000, pp. 91, 168) illustrated striking, complex topological surfaces involving the influences of only two environmental parameters. How much more complex with 20 or more?

BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

McDougall (1924) described behaviour in the following way. Animals are behaving if they actively resist the push and pull of the environment, exhibit persistence of activity independently of the impression (signal) that may have initiated it and exhibit variation in the direction of persistent movements. This definition would characterize plant behaviour, too.

Movement and behaviour

Movement would seem to be the simplest criterion of behaviour, and movement has always been an essential part of the animal lifestyle to find food, avoid predators (or catch prey) and find mates. Predator–prey relations among animals accelerated the evolutionary specialization of sensory organs and muscles to respond to signals. The nervous tissue, a rapid information transmission system, then evolved, to link these two together. The faster the prey responded and moved, the faster any effective predator had to evolve in turn. Animal behaviour tends to be equated with movement because we ourselves are animals, because our perception/response system works at the rate of transmission of the nervous system (like most other animals) and because we regard our own movement as behaviour.

Multicellular organisms that lack a nervous system can be expected to operate on a very different timescale, and higher plants are no exception. This change of timescale creates problems for recognition of behaviour. Pfeffer (1906, p. 2) early on recognized the problem. ‘The fact that in large plants the power of growth and movement are not strikingly evident has caused plants to be popularly regarded as still life. Hence, the rapid movements of Mimosa pudica were regarded as extraordinary for a plant, and the same applies to the spontaneous movements performed by Hedysarum gyrans (telegraph plant). If mankind from youth upwards were accustomed to view nature under a magnification of 100 to 1000 times (seeing streaming or lower plant sperm swimming) or to perceive the activities of weeks or months in a minute as is possible by the aid of a kinematograph, this erroneous idea would be entirely dispelled’. Pfeffer (1906) thus predicted time-lapse facilities that brings plant behaviour, in some sense, to a more familiar human timescale. The web site (http://plantsinmotion.bio.indiana.edu/plantmotion/starthere.html) constructed by Roger Hangartner contains many excellent time-lapse examples that show behaviour that complies with the aforementioned McDougall (1924) definition. One fundamental difference between plant and animal behaviours is therefore in their respective time frames.

The real value of time-lapse records is to uncover behaviour either difficult to record or missed by previous recording procedures. For example, the Attenborough (1995) time-lapse films record a kind of rapid vertical/horizontal shaking behaviour by a growing bramble stem that so far has no explanation. Other revealing time-lapse movies are to be found in Massa & Gilroy (2003) on root behaviour encountering soil obstructions and Runyon et al. (2006) on Dodder locating its prey by detecting host volatiles. Time-lapse recording needs to be focused on the behaviour of wild plants as well, because the timescale difference with human observers implies that much novel behaviour may simply have never been seen.

Purposeful behaviour

In a seminal paper, Aphalo & Ballare (1995) indicated how plants were commonly perceived as ‘passive organisms’ undergoing a predetermined programme whose culmination was occasionally slowed by poor environments. They argued instead that plant behaviour is both active and predictive. The ‘passive plant’ attitude almost certainly results from experimental experience of plants in which signals are imposed by the investigator to make plants perform in controlled conditions, perhaps, similar to the way circus animals are made to perform. But in the wild, it is plants that must compile environmental information and make active decisions to change development, in order to optimize life cycle behaviour and eventual fitness.

Active behaviour may be more simply defined as a dependence on metabolic energy (Rosenblueth, Weiner & Bigelow 1943). True passive behaviour is then simply limited to processes, like the explosive distribution of seeds, that depend only on unequal drying of dead tissue or the floating of seeds in the wind.

Active behaviour is most easily defined as purposeful when it is goal oriented (Rosenblueth et al. 1943; Russell 1946). The goal is often achieved by some complex form of negative feedback, and obvious examples (out of many) are the adaptive responses of tropic bending to gravity or light. In negative feedback, an information loop is constructed from the signal to the responding cells to indicate the margin of error from the goal and adjust behaviour accordingly (Trewavas 2007a). The clearest indications of a kind of negative feedback control are the damped oscillations around the goal that can sometimes be observed in tropic bending (Trewavas 2003).

Other examples of more complex and less understood, purposeful (goal-directed) behaviour are (1) the stem thickening that accompanies wind sway; (2) leaf abscission that rebalances the water relations of a whole plant when water supplies are diminished; (3) the (indeterminate?) elongation of the leaf petiole in water plants like Nymphaea, which only stops when the leaf breaks surface; and (4) the seasonal, average tree-leaf temperature that remains remarkably uniform at about 21 °C from trees ranging from the subtropical to the arboreal (Helliker & Richter 2008). This unusual form of long-term homeostasis, which undoubtedly benefits photosynthetic processes, is suggested to result from an interaction among internal leaf cooling, leaf structure, branch structure and leaf distribution among others, all important behavioural traits that have clearly been optimized. Russell (1946) included several other good plant examples. The molecular mechanism underpinning goal-directed behaviour is clearly dependent on growth substance involvement (auxin, ethylene, etc.) that is only partly understood.

Certain forms of purposeful behaviour seem overwhelmingly controlled by one signal; the extreme sensitivity of etiolated seedlings to unidirectional blue light that can override opposing gravity signals is an obvious example. The most obvious purposeful behaviour, however, arises from an integration of different signals. Charles Darwin (1880) showed experimentally how seedling roots sensed the signals of touch, light, moisture and gravity simultaneously resulting in sensory integration (Trewavas 2007c). Furthermore, he showed that growing roots could distinguish between these signals and decide which was the most crucial to respond to. Both touch and humidity can override the gravitational signal if applied in a different direction (Eapen et al. 2003; Massa & Gilroy 2003), in a recent excellent expansion of Darwin's observations on soil obstructions, indicating how the root response is integrated between touch and gravity. Natural soil is very heterogenous both in texture and in the distribution of resources (Bell & Lechowicz 1994). Signal integration is therefore necessary. The successful plant must more correctly assess the probabilities of appropriate action in constructing the root phenotype.

Intentional behaviour

Piaget (1979, p. 1) described behaviour as follows: ‘By behaviour, I refer to all the actions directed toward the outside world in order to change conditions therein or to change their own situation in relation to these surroundings’. This definition is equally applicable to plant behaviour but implies intention, usually defined as goal-directed behaviour. Do plants intend to resist herbivores; do they intend to respond to gravity; do they intend to resist the common stresses they experience? The description of the behaviour of individual root systems as growing to actively deny resources to competitors certainly implies intention (Maina, Brown & Gersani 2002; Gruntmann & Novoplansky 2004).

The issue of intentional behaviour was raised and discussed at length by Scott-Turner (2007) in the context of social insect colony behaviour. Relatively simple interactions between the individual insect organisms construct a communication network with complex and some, recognizably, intentional properties. In analogous fashion, the communication network of cells and tissues that construct the individual plant may be the mechanistic basis of intention in plant behaviour.

Does intention imply cognitive involvement? Both Maturana and Varela (1980) and Bateson (1985) indicated that cognition, defined as the act of knowing, is implicit in all life, constructed as it is from complex hierarchical network structures (Trewavas 2007a). These authors stated that even organisms such as plants without nervous systems perceive, respond and thus know about their environment; they are therefore capable of cognition. ‘It is not too much to say that a plant is capable of cognition in much the same way that a human being is. The plant gathers information about its surroundings, combines this with internal information about its internal state and makes decisions that reconcile its well-being with its environment’[Trewavas 2005a; modified from an original social insect quote by Seeley & Levien (1987)].

The notion of intentional behaviour might also conflict with the Neo-Darwinist view of natural selection that suggests organisms as passive in the face of random selection. The alternative to simple selection from a systems framework and permitting intention is powerfully argued by Gould (2002, p. 614 onwards).

MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

The quality of biological information is determined by constraint

Information theory was first posed by Shannon & Weaver (1949). However, their concept of information based on entropy has been difficult to apply to biology (Trewavas 2007a). Biological information can be equated to meaningful communication. This definition of information implies that the quality of information gained is proportional to the constraint with which it is sensed and transduced.

Meaningful plant signals (properly called semeotics) are first distinguished by specific receptors; the greater the selectivity exerted by the receptor, the higher the quality of information conveyed to the cell. However, single receptors can do little more than provide an all-or-nothing signal to the cell. To provide further essential information on the time length of the signal and its direction for example, other related receptors are needed. Thus, families of receptors are common, for example, the phytochromes, cryptochromes, nitrate reductases, auxin receptors, calmodulins (calcium receptors) and so on. Holistic integration of the information provided by the receptor families and integrated with other families helps provide a kind of ‘three-dimensional’ signal perception in both space and time.

Constraint is equally important during signal transduction chains that depend on protein–protein interactions. Cell cytoplasms contain anywhere from 20 to 40% protein, and some membranes, such as the mitochondrial membrane, are 80% protein. Protein–protein interactions of all kinds will therefore be common and without some discrimination, cellular responses will simply be destroyed by informational noise. Transduction sequences are constructed from proteins whose adjacent members exhibit complementary surface topologies often induced by a prior signal. For example, signal-induced changes in the surface topology of protein kinases may now enable specific interaction with a protein substrate. Specific surface interactions between these proteins, when they occur, will exhibit relatively strong binding and will last a relatively long period, and information exchange will be high, as a result of constraint. In contrast, noisy interactions although frequent, will be only weakly binding, short-lived and convey little or no information. GTP-binding proteins are commonly used by cells to ‘time’ the lengths of protein–protein interactions. Only if the protein–protein interaction lasts longer than the hydrolysis rate of GTP, do cells regard the interaction as providing a high quality of information. It is in part for these reasons that Nurse (2008) called for much greater research emphasis on how information is gathered, processed, stored and used, and how this generates higher-level phenomena.

Information terms for learning, memory and intelligent behaviour

In informational language, memory is simply information stored for later use, learning is simply acquisition of information and intelligent behaviour is the assessment of information that leads to adaptive, problem-solving responses. Figure 1 indicates the various kinds of information categories that underpin plant behaviour with their different but equivalent terminologies. The value of using information terminology is that there is no implication concerning the mechanisms involved in transmitting information (Box 1). Nervous systems process information by mainly different mechanisms to those used by cells of all kinds, but the analogous learning, memory and intelligence behaviours can be recognized in cells, tissues, whole plants and other organisms.

image

Figure 1. Information flow through biological networks that initiates eventual behavioural changes. Biological networks are hierarchical; they are constructed starting with cellular molecules, cells, tissues, individual plant (genet), population, niche or community. Each of these levels in the hierarchy forms a network in its right. However, these networks differ markedly in the strength of connections that hold them together, reflecting probably the fidelity with which information is perceived and interpreted between the constituents of the network and the increasing noisiness of the information communication channels. The connections are strongest inside plant cells and then weaken progressively: molecules > cells > tissues > plant > population > niche > communities. As the connection strengths weaken, greater plasticity is experienced, but even in communities, there is some fidelity in communicated information. The psychologist Schull (1990) argued strongly for a species population to express intelligent behaviour. The figure is presented in informational terms, and alternative commoner words that describe analogous processes are indicated in brackets. However, one less than usual feature in individual plants is the loop (earlier indicated as niche construction) that connects the adaptive response back to subsequent signalling. As plants grow, they continually modify their own environment and the characteristics of future perceived signals.

Download figure to PowerPoint

Box 1. Common misconceptions of the use of words involved in information processing

Bruce et al. (2007) suggested that the use of the term ‘memory’ in plants implies a cognitive function. However, neither learning, memory nor indeed intelligence are words limited to biological, let alone, cognitive processes. For example, computers possess memory, and the more advanced ones can learn. Some molecules and steels are described as possessing a shape memory induced by a particular annealing regime (e.g. Sehitoglu et al. 2001). Intelligence has been used to refer to many biological processes and organisms (Trewavas 2005b) and machines. Plant biology has borrowed many words that were originally designed to describe purely human characteristics because the botanical process examined was analogous. The plant ‘vascular system’ containing ‘veins’ and ‘vessels’ is analogous in function to the human vascular system, but the mechanisms whereby each works are entirely different. Hairs, stress, arms race, battles, pathways, cross talk and foraging are a few other borrowed words, and there are many more. However, when defined in information terms, memory, learning and intelligence are suitable terms for any organism regardless of the precise mechanisms involved.

Memory

Memory has been described by authors as contributing to a variety of plant behavioural situations. Memory is probably essential in all plant behaviours. Only a few examples are indicated here.

Several weeks of cold temperature can create a mitotically stable memory (vernalization) that can last for 300 d. Chromatin remodelling may be the molecular underpin of memory here (Goodrich & Tweedie 2002; Amasino 2004; Sung & Amasino 2004), and phosphorylation pathways in animals lead to chromatin remodelling (Stipanovich et al. 2008). Herbivory establishes a long-term memory of previous attacks that primes defence mechanisms, increasing resistance against further attacks (Karban and Niiho 1995; Baldwin & Schmelz 1996; Ruuhola et al. 2007; Karban 2008). Priming can last for years. Targeted illumination generates a spatial memory of phototropic signals lasting several hours that can override a shorter-term gravitropic memory interpolated before or after illumination (Nick, Sailer & Schafer 1990). The memory of mechanical stimulation and mineral imbalance can last for many days after the signal has ceased (Desbiez et al. 1984; Verdus, Thellier & Ripoli 1997). Tendril coiling requires both coincident blue light and mechanical signals, but either signal, given separately, can be remembered for several hours (Trewavas 2005b). Stress effects from numerous treatments [cold, heat, salinity, drought, ultraviolet (UV) light, mineral imbalance, disease, etc., including, surprisingly, ABA] can be remembered and influence not only a later response (Goh, Nam & Park 2003) but memory of some can also be passed to subsequent generations (Durrant 1962; Molinier et al. 2006). The effect of previous plant neighbours on phenotypic development can be remembered for up to a year after transplantation elsewhere (Turkington, Hamilton & Gliddon 1991). In contrast, hypo-osmotic shock memory can last about 20 min (Takahashi et al. 1997). Finally, the Venus flytrap requires stimulation of two hairs to effect closure; these have to be stimulated within 40 s of each other (Shepherd 2005). One stimulation of a single hair is thus remembered for 40 s.

No wild plant could survive without some memory of its current perceived signals or without a cumulative memory that collates its past information experience and integrates it with present conditions so that the probabilities of potential futures could be assessed. The mentioned examples indicate memory lasting from seconds to minutes, to hours, to days, years or even longer. These surely reflect, in part and rather crudely, the various known stages of many signal transduction pathways. Generally, the earliest intracellular events are ion flux changes and protein phosphorylation modification (seconds, minutes, hours); slower events are modification of gene expression and transacting factors (hours, days). Even longer-term events are intercellular signalling modifications (hours, days or even months and years) and, finally, chromatin remodelling (days to years).

Each signal leaves an imprint specific both to the characteristics of the signal and the current state of the intracellular molecular network, thus modifying the topology of the network. Additional signals within the appropriate time frame are thus interpreted differently as information flows through the modified structure.

Learning

Information storage, (memory) cannot be constructed without information acquisition first (learning). Learning can be simple; for example, heliotropic plants learn the optimal direction of the sun and maintain that position of leaf orientation. More complex learning involves reinforcement, and good examples are the plant responses to what are commonly called the stress conditions of cold, drought, heat, heavy metals, soil minerals, salinity, wind sway, flooding, excess/UV light, oxidative stress, herbicides, herbivory, disease and even ABA (McKersie & Leshem 1994; Knight, Brandt & Knight 1998;Goh et al. 2003; Trewavas 2005b; Bohnert 2007; Dinenny et al. 2008; Frost et al. 2008; Voesink & Pierik 2008). In all these cases, eventual resistance to a severely damaging stress can be gained by a progressive application of a milder but increasing strength of stress. This learning response enables a quicker, more aggressive, adaptive resistance to subsequent stress episodes. This is clearly a kind of trial-and-error learning (often called Thorndikean learning) and has been called priming in the case of herbivory and disease. Priming is, however, straightforward learning leading to a long-term memory that can last for months (Frost et al. 2008). Another term applied only to the abiotic stress stimuli is acclimation, a term that reflects the passive (laboratory) control of plant behaviour and ignores the clearly active role played by the wild plant in assessment and response.

The behavioural responses to an increasing imposition of stress, indicate that in the wild, this is really predictive behaviour – a preparation for likely, more severe episodes in the future. The progressive nature of the learning indicates that there is a trade-off between a commitment to a full-blown resistance mechanism, which can be costly, balanced against an assessed probability that further episodes might recur in the near future and ensuring a quicker response if it does. The emphasis here is on probability; certainty only occurs in the laboratory. The skill with which these cost–benefit assessments can be made contributes directly to ultimate fitness and requires an intelligent assessment. Prediction of future loss of photosynthetic light leading to shade-avoidance phenotypes is well established (Aphalo & Ballare 1995).

The basis of this kind of learning (priming) in plants is not understood, but in molecular terms multiple, interlinked, positive feedback loops might offer the least cost of a long-term memory (Ingolis & Murray 2002; Bisjman & Groisman 2003; Xiong & Ferrell 2003; Acar, Becskei & Van Oudenaarden 2005; Brandman et al. 2005). Learning would involve the construction of these feedback processes. Such regulatory, molecular-design circuits can produce graded responses as well as complete conversion (Becskei, Seraphin & Serreano 2001).

An alternative mechanism is the possibility that the progressive nature of the response may simply reflect the numbers of cells that have passed a threshold in an all-or-nothing response to the stress. On this basis, as the stimulus is increased, more cells would undergo this all-or-nothing change. This mechanism implies a substantial variation in sensitivity of individual cells to the mentioned stress signals. A possible indicant of such cell sensitivity variation is to be found in guard cell closure. Whereas some guard cells close with very low concentrations of ABA, a closure of over 90% required four to five orders of magnitude increase in ABA concentration (Trewavas 2003). Variation to the mentioned stress signals between individual plants might also generate the progressive nature when results are expressed as averages if again there is a plant all-or-nothing response.

COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

Intelligence is the capacity for problem solving

What is meant by intelligence? For historical reasons only, some mistakenly identify intelligent behaviour as being a uniquely human characteristic. This perception has arisen because of the importance of intelligence in education and the dominance of educational psychologists in discussion of intelligent behaviour from 1900 onwards. Psychology is, by its nature, concerned predominantly with human behaviour, but it is quite clear that many, if not most psychologists, do not think that intelligence is something limited to human beings or indeed organisms with brains. Fitness and natural selection are the arbiters of all kinds of organism behaviour with the sole exception possibly of recent mankind. It is the failure to recognize this crucial point that leads to all kinds of controversies as to the nature of human intelligent behaviour among psychologists. Intelligent behaviour evolved to increase fitness.

Sternberg, a psychologist who has written more extensively than others on intelligence (e.g. Sternberg 2006, Ciancolo & Sternberg 2004 and references in these), solicited the opinions of some 20 psychologists as to the meaning of intelligence (Sternberg & Detterman 1986). This was a repeat of an investigation first carried out in 1921. In his analysis of these articles, Sternberg (1986) indicated that descriptions involving cognition or adaptation were equally acceptable and identified problem solving as the commonest descriptor. The well-known IQ test is simply an adaptation to the presented situation of the test. Warwick (2001), an Artificial Intelligence expert, generalized intelligent behaviour to be the capacity for problem solving and emphasized that intelligent behaviour in organisms other than humans must be judged in terms of the capabilities of the organism in question. To do otherwise is to be subjective and anthropocentric. Sternberg (1986) himself identified intelligence as existing between species, within species, within organisms, etc., and therefore broadly present in life itself. Gardener (1983) and Steinhardt (2001) indicated that even in humans, intelligence is a whole organism property. Psychologists therefore do not limit intelligent behaviour to humans or other advanced animals or even to cognitive processes (Stenhouse 1974; Griffin 1976; Sternberg 1986; Schull 1990; Warwick 2001).

Intelligent behaviour cannot be divorced from the situation that elicits it. All wild organisms face highly variable situations in which they must attempt to optimize their survival and produce the maximal numbers of siblings. The capacity to solve the environmental problems that vary enormously and threaten the optimization of fitness requires intelligent solutions. The key term that underpins intelligent behaviour is assessment. Optimal assessment using stored information, interacting with the stage of development and current acquired information, leads to problem solving, successful adaptive responses and thus increased fitness. Plant intelligence, like plant behaviour itself, has suffered from an inability of easy human observation, leading to a common assumption that both must be absent.

Problem solving and trade-offs

In the attempts by plants to optimize fitness, numerous problems interfere. The individual plant has to accommodate the uneven distribution of light, minerals, soil structure and water, competition, along with variation in rainfall, wind and damage by disease pests and herbivores. Flowers need to be positioned where pollination is optimal. The costs and benefits of any behavioural change in growth and development and the resources to back it up require assessment (Bazzaz 2000; De Jong and Klinkhamer 2005). Decisions need to be made about how best to redistribute the limited internal resources among competing tissues to try and provide ultimate success. As resource limitation increases, it becomes increasingly crucial to make the right decisions to increase the probability of success. The mechanisms used in decision making and trade-offs are currently only weakly understood, and most research still uses well-nourished laboratory plants. Selection, however, will not allow such decisions to be made at random!

Trade-offs of resources are known to occur between root and shoot, between different shoots, roots, branches or leaves, between vegetative and reproductive growth and between vegetative growth and herbivore/disease resistance (Hutchings 1997; Lerdau & Gershenzon 1997; Bazzaz 2000; Weaver & Amasino 2001; De Jong and Klinkhamer 2005; Trewavas 2005a, 2007a; Frost et al. 2008). Natural pesticide levels in wild plants occupy 1–5% of their dry weight, sufficient to reduce both growth and seed yield; any increased synthesis as a result of attack will diminish growth further. There will also be trade-offs in resources devoted to different abiotic stress conditions that will need careful assessment because an excess resistance response to one will almost certainly diminish the capability to respond to another. The multitude of problems requires intelligent, adaptive responses.

A potential mechanism for plant intelligence; a communication network

‘Plants have evolved an integrated complex of hormonal systems – a coordinated but non-centralized intelligence system . . . that manages bioenergetic resources’ (La Cerra & Bingham 2002, p. 11). An adaptive representational network has also been proposed to underpin intelligent responses (Trewavas 2005a) and is also suggested for bacterial learning (Tagkopoulos, Liu & Tavazoie 2008). Although the interactions among the seven or eight known hormones are beginning to be understood, these on their own may not provide a sufficiently complex network to deal with the variety of presented problems experienced by wild plants. The communication network may be made more complex by including other molecules such as proteins, peptides, nucleic acids, small RNA's oligosaccharides, sugars, minerals, etc., that are known to move between cells (Trewavas 2003). As such networks must self-assemble, they will also self-orchestrate, seeking their most stable configuration (Trewavas 2007a). Orchestration is commonly called downward causation (Trewavas 2007a); an alternative term is circular causality (Scott Kelso 1995). The importance of downward causation in biological networks is also discussed at length by Noble (2006) and Kauffman (2008).

Changes in orchestration will be continuous as the individual plant progresses from seed to flower and accommodates a changing panoply of signals that elicits adaptive behaviour. Any signal, as information, contributes to its own orchestration. Assessment of any signal arises naturally from the present whole network structure that is itself a compilation of present and past environmental and developmental history. Intelligent behaviour (adaptive problem-solving behaviour) is thus a property of the whole plant and not individual tissues.

Plants that can place a root or shoot in the best position to gain resources, as against indifferent or resource-absent places, act intelligently. Those plants that most quickly estimate which branches or leaves no longer provide adequate resource-gathering potential and block them from further root resource access, have a higher capacity for problem solving. Those plants that more accurately predict the future resource availability or herbivore damage and decide on resource distribution appropriately are smarter than others, and the reward is a likely gain in fitness.

CONCLUSIONS

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES

Plant behaviour is best described as what a plant does – doing rather than having. The commonest misapprehension about higher plants is that they are simple organisms. The ‘still life’ description indicated by Pfeffer (1906) is undoubtedly the major cause of that perception. In complexity of signalling and problem solving within their own environmental context, plants lack nothing compared with animals in skill and behavioural complexity. There is an undoubted need for plant biologists to investigate the design characteristics of plant cell protein and phosphorylation networks and how these generate purposeful and potentially intentional goal-directed behaviour. While plant gene identification has been very successful, systems design is presently under-investigated. Ecologists and molecular biologists are needed to construct fruitful interactions centred around wild behaviour, trade-off assessments, communication networks, assessment and thus intelligence.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. BEHAVIOUR IS WHAT PLANTS DO
  4. ENVIRONMENTAL SIGNALLING COMPLEXITY AND BEHAVIOUR
  5. BEHAVIOUR, MOVEMENT, PURPOSE AND INTENTION
  6. MEMORY, LEARNING AND INTELLIGENT BEHAVIOUR
  7. COST–BENEFIT ASSESSMENT, DECISION MAKING AND INTELLIGENCE
  8. CONCLUSIONS
  9. REFERENCES