Plant growth forms: an ecological and evolutionary perspective

Introduction

Some of the main aims of recent evolutionary developmental research have been to discover the genetic and functional organization of the stem, leaf, root and reproductive organs of plants. While increasing efforts are made to understand the development of these organs in model organisms and now a greater variety of plants, few studies have focused on their ecological development and fewer still on how the entire growth form of plants might be altered by the genome and modified by the environment. The extension of evolutionary developmental studies to include the effects of ecology on developmental variation is nowhere more applicable than for understanding growth form diversity of plants.

In this review we discuss how ecology and environment can modify the overall growth forms of plants with particular reference to the mechanical constraints of stems and branches during development and growth. We discuss the main differences between self-supporting plants and climbers, how such forms vary with the environment and how phylogenetic constraints can influence different developmental patterns. We briefly summarize recent hypotheses on the evolution of some major innovations in plants – secondary growth and the appearance of leaves – and broadly discuss how these were affected by environmental change and, indeed, how they themselves changed the environment. We argue that growth form diversity is an obvious but possibly understudied aspect at opposite extremes of biological research, from censoring natural habitats to choice of model organisms in molecular research. We argue here that growth form eco-devo and evolution is, however, a major factor underpinning ecosystem structure and diversity.

Growth Form Variation: Ecological and Evolutionary Contexts

What ecological conditions favour the tree habit and how do trees vary in different conditions? How do climbing plants differ from trees or shrubs? Can some species grow as shrubs or trees as well as climbers? What ecological effects might mediate these changes? How, and to what extent, is plasticity of growth form governed by phylogenetic constraints? When and how did the vascular cambium evolve and why did it become a key innovation for seed plants and their putative ancestors? The evolution of growth forms since the early terrestrial radiations is a complex history of innovation, complexification, simplification, conservatism, radiation and extinction. Trees, shrubs, climbers and herbs have different ecological preferences and different mechanical attributes. Are certain types of growth form highly constrained and immovable in evolutionary terms? Are some plant groups more ‘flexible’ in their capacity to evolve widely differing growth forms and is this capacity related to the evolutionary age or complexity of the group? What are the evolutionary developmental processes (e.g. heterochrony, heterotopy and homeosis) and what are the ecological factors that coerce to either canalise or facilitate growth form variation and evolution?

Self-supporting growth forms

All growth forms are essentially ‘ecological’ in design. There is a wide range of adaptive constraints influencing the mechanical architectures of growth forms but in this review we focus our attention on self-supporting and climbing growth forms. A self-supporting organization must physically support its own static weight and resist higher forces and dynamic loadings from a range of natural factors, particularly high winds. Older, basal parts of the plant body are relatively stiff whereas younger apical axes, as well as leaves and petioles are often more compliant and bend and twist in high winds, reducing drag and potentially destructive loading by wind. From a developmental point of view, this adaptive strategy is complex. Single species show ranges of variation during ontogeny and under different ecological conditions. Recent studies have shown such mechanical variation for mosses (Biehle et al., 1998; Ross et al., 1998), pteridophytes (Niklas, 1990), gymnosperm (Brüchert et al., 2000) and angiosperm trees (Speck & Rowe, 1999) and herbs (Crook & Ennos, 1996).

The notion of ‘safety factors’ plays a role in many biological systems (Wainwright et al., 1976; Niklas, 1998), particularly in terms of self-supporting plants. Trees and shrubs can vary in overall size, branching and mechanical properties with respect to the wind. Low safety factors with slender, tall stems are observed in sheltered conditions. The same species growing in open conditions exposed to the wind will have a shorter, stouter morphology. Changes in geometry and mechanical properties also occur during ontogeny; young individuals can show relatively low safety factors and mature or aged individuals high ones.

From a mechanical perspective, wind is a critical factor in constraining size, shape and retaining an upright growth form (Ennos, 1997). Responses to physical perturbation are known as thigmomorphogenetic reactions and overall morphological responses have been known for many years (Jaffe, 1973; Telewski & Jaffe, 1986). Responses to mechanical perturbation are now known to at least partly involve TCH (touch) genes (Braam & Davis, 1990; Braam, 1992). Recent experiments on ethylene-insensitive mutants of Arabidopsis suggest that the phytohormone ethylene is not necessarily required for responses to mechanical stimulation; interestingly, at least one touch gene, TCH4 is initiated by a range of ecologically important stimuli, including mechanical perturbation, darkness, heat and cold shock (Iliev et al., 2002). While increasing studies are concentrating on thigmomorphogenesis in Arabidopsis (Pigliucci, 2002) as well as trees (Pruyn et al., 2000), the signalling and transduction pathways remain incompletely understood (Johnson et al., 1998; Jaffe et al., 2002).

Plants have probably been buffeted by wind since the earliest radiations of land plants, an influence that increased as plants evolved higher growth forms. As an ecological factor it must have shaped the underlying genetic and developmental trajectories formed during this period when the major clades were accumulating genetic, physiological and structural innovations (Bateman et al., 1998). The view that early plant evolution was marked by strong selective pressure for increasing height, light capture and efficient propagule dispersal, should take into account the effect of wind into the equation (Niklas & Speck, 2001).

Self-supporting architectures are mechanically optimized from the base to the apex of the plant (Niklas, 1999; Speck & Rowe, 1999); they can also show geometric and mechanical changes in exceptionally sheltered conditions. In understory niches of tropical forests, young trees and saplings often reach a point of instability and lean against surrounding trees. A simple interpretation is that competition for light is so fierce that self-supporting architectures are pushed to the upper limits and sometimes beyond their critical buckling length in the need to reach light. Such examples are seen commonly where clearing has removed neighbouring trees and exposed formerly sheltered plants to wind (Fig. 1a). Simply put, many plants, if they are not mechanically perturbed will produce less sturdy mechanical architectures and sometimes growth forms that are no longer self-supporting. This can apply to both large woody growth forms (trees) and to herbs, which can show extreme ecophenotypic variation between sheltered and mechanically perturbed sites.

Climbers

Many dicotyledonous woody lianas show entirely different mechanical architectures compared with self-supporting plants (Fig. 1b). The mature or aged liana can have utterly different mechanical properties from that of a self-supporting plant, reflecting profound differences in ecological function (Gartner, 1991b; Putz & Holbrook, 1991; Speck & Rowe, 1999). These differences have been studied increasingly at an ecological level over recent years (for a general review see Schnitzer & Bongers, 2002). In many lianas, young, distal parts of the plant are composed of relatively stiff tissues and act as ‘searchers’ spanning gaps between host supports (Rowe & Speck, 1996; Speck & Rowe, 1999). A great many different lianoid species also show a juvenile phase of growth where the young plant is self-supporting before locating a host support (Caballé, 1998). In tropical rainforest understory vegetation, such stems are difficult to distinguish from young self-supporting species. This early growth can represent a ‘waiting phase’ before appropriate environmental cues, such as light availability or proximity to a support, after which the self-supporting architecture can drastically change. Alternatively when lianoid species grow in open conditions, some species do not develop a climbing phase (Hegarty, 1991; Gallenmüller et al., 2004) and may die back after a given time; others will, nevertheless, begin to climb and scramble on themselves or anything nearby.

The transition to a climbing mechanical architecture can show profound and abrupt changes in secondary cambial development, particularly among twining lianas. The well-known lianoid species of Bauhinia (monkey ladder) can show a well-developed self-supporting phase (Fig. 1c) with a cylinder of dense, stiff wood (Fig. 1d) and heights of up to 2–3 m (Speck & Rowe, 1999; Hoffman et al., 2002). At the onset of climbing, cambial growth is highly modified (Fig. 1e) producing a ribbon-like stem with highly compliant wood. Plants climb via tendrils and woody, hook-like leaf bases. Later development can involve further anomalous cambial activity and huge flexible ‘trunks’ which can literally swing freely from the canopy of their self-supporting hosts (Fig. 2a). Can the lianoid species Bauhinia guyanensis produce a large-bodied tree under certain ecological conditions? The answer is no. Whatever the internal developmental constraints during the ontogenetic trajectory, the self-supporting phase is confined to the ‘juvenile phase’, which may be expressed for a considerable time before attachment to a support.

Other species of Bauhinia do grow as trees. We have observed at least one species in south-east Asia that, having reached a tree-sized self-supporting morphology, will extend branches into the surrounding canopy, which are supported on branches of self-supporting neighbours (Fig. 2b). While not showing the same ribbon-like form, the response to neighbouring support is not consistent with the ‘crown shyness’ (Putz et al., 1984) commonly seen in mature trees, which often maintain discrete gaps around the branches and canopies of neighbouring dominant trees (Fig. 2c). In Bauhinia and possibly many other clades of plants there is an underlying tendency to produce non-self-supporting traits that goes beyond the kind of thigmomorphogenetic reaction shown in Fig. 1a. Are the reactions controlling ‘safety factors’, the type of thigmomorphogenetic reaction described above, and this tendency or ‘prepatterning’ in many groups of angiosperms towards lianescence somehow linked?

The Appearance and Consequences of the Cambium

The size and overall growth form of a species represents a critical interface with the environment and selection. This is largely confirmed by the observation that survival of potentially lethal environmental perturbation or climatic change can depend on body size (McKinney, 2001); at the risk of over generalization – big organisms die, small organisms live. We argue elsewhere (Rowe, 2000; Rowe & Speck, 2003) that the bifacial vascular cambium represented a key innovation in lignophytes and established a developmental template for plants to modify hydraulic and mechanical attributes as well as size and range of growth form. We also proposed that the combination of both secondary xylem and phloem was more adaptable than the almost entirely unifacial meristems in stems of extinct tree-sized lycopsids and sphenopsids and some other groups in the fossil record. Furthermore we suggested that the ability to change size and shape of the growth form via cambial activity offered better chances for later seed plant clades to survive periods of ecological perturbation.

Recent biomechanical models of phylogenetically basal lignophytes with a bifacial cambium from the Middle Devonian (c. 360 million years ago (Myr)) indicate that relatively young stages of growth were self-supporting, but that oldest stages were not optimized for a self-supporting growth habit (Rowe & Speck, 2003; Speck & Rowe, 2003), principally because of the wide thickness of soft bark-like secondary tissue around the outside of the wood cylinder. Thus, the oldest stages of growth were relatively flexible and would not have supported the weight of tall, branched stems. However, this architectural constraint is not inconsistent with a perennial, and larger-bodied procumbent stem system, which would have, hydraulically and mechanically supported younger upright axes. This mechanical strategy is observed among woody plants growing in unstable ecological conditions (Isnard et al., 2003). It favours survival in disturbed sites, producing upright shoots over a wide area and avoiding or traversing inhospitable microhabitats. The biomechanical models of early lignophytes (Rowe & Speck, 2003; Speck & Rowe, 2003) therefore suggest that the initial advantage of the bifacial vascular cambium was more probably one of potential plasticity of growth variation, above that provided by a mechanical architecture, constrained by primary growth and dominated by the outer fibrous hypodermis alone. Such growth forms could have occupied space without the physiological cost of an extensive self-supporting structure. Overall, the evidence suggests that the appearance of the bifacial vascular cambium did not initially confer the ability to produce large-bodied plants with high self-supporting stems. Greater height and size would have required the addition of further novelties with a mechanically optimized ratio of wood to compliant periderm, the presence of leaves and a more indeterminate growth pattern. The appearance of the bifacial cambium in earliest lignophytes can be viewed as a ‘nascent innovation’ that, after accommodating further added novelties, represented a key innovation, which permitted wide morphological and architectural diversifications. Later evolutionary patterns involving the cambium in lignophytes are a complex history of complexification (e.g. appearance of vessels and libriform fibres in angiosperms) as well as reduction (e.g. loss of the vascular cambium in monocotyledons). These had far-reaching effects on subsequent growth form plasticity and ecological preference within different clades: the paucity of growth form diversity in extant gymnosperms (with the exception of the Gnetales) and extreme growth form diversity in angiosperms.

Profound alterations in growth form such as transitions to climbers or herbs can canalise subsequent evolution, setting different limits on growth-form plasticity in different clades and species. A pine tree cannot grow as a highly compliant liana whatever the ecological circumstances; the poison oak can grow as fairly modest-sized shrubs and up to 30-m high climbers – albeit not highly compliant specialized ones; a lianoid species of Bauhinia can grow as an upright sapling in sapling-dominated forest understory but eventually swings freely from the canopy as an adult. From a mechanical perspective these constraints result principally from phylogenetic constraints centred on wood development.

Phylogenetic studies exploring growth form evolution in plants are few and far between both at the macro and micro evolutionary scale (Funk, 1982; Bateman et al., 1992; Bateman, 1994; Böhle et al., 1996; Civeyrel & Rowe, 2001; Speck et al., 2003). Recent studies suggest a complex evolution of growth forms within the Aristolochiaceae (Speck et al., 2003); the basal taxa, Asarum and Saruma are herbs, whereas the genus Thottea, which is sister to Aristolochia s.l., is a small-bodied shrub. Most species of Aristolochia s.l. are lianoid with herbs appearing in the mostly lianoid Aristolochia s.st. and Howardia. Three rare, woody shrub-like species, as well as lianas, occur in Isotrema, which, interestingly is sister group to another occurrence of a diminutive herb. While woody shrub-like forms are rare in the group, the prevalence of lianas and herbs suggest an underlying development constraint, which does not readily accommodate self-supporting woody architectures. In the young searcher phases of the climbers tested, stiffness is imparted via an outer ring of thick-walled hypodermal tissue (Fig. 3d), possibly reflecting that of putative ancestral herbs (Speck et al., 2003). The wood produced is of a single highly compliant type: once secondary growth has ruptured the outer wall of stiff material (Fig. 3e), the stiffness of the stem is irreversibly reduced. This does have ecological implications on the physical distance and longevity of the ‘searching’ phase and quite possibly constrains the ‘type’ of climbing behaviour.

In other groups, such as the trees, shrubs and lianas of Strychnos (Loganiaceae), the architecture is predominantly woody. Climbers show an initial stiff searcher phase, which can extend from one host plant to another over a matter of metres (Fig. 3a) and is characterized by a cylinder of dense, stiff wood (Fig. 3b). When a support is secured via the woody hooks on apical branches, a second type of compliant wood is produced (Fig. 3c). This development can vary the amount and degree of dense or compliant wood, probably related to the degree of support from neighbouring plants. The type of mechanical architecture might also be coupled with different modes of attachment to the host vegetation such as by anchoring wide-angled branches or placing hooks, as in Strychnos, into the surrounding vegetation. Different lianoid species can therefore inherit different types of mechanical architecture, which can directly influence the ecology and thus plasticity of their climbing strategy and overall growth form.

Wood, Leaves and Atmosphere: Functional Links and Ecological Feedback Loops

Plants can respond to lower atmospheric ρCO₂ by phenotypically increasing stomatal densities of leaves and thus increasing net diffusion of CO₂ into the leaf. Recent studies have documented this general tendency from: (1) experimental studies of living plants growing under differing CO₂ levels; (2) analyses of historic herbarium material and sub-fossil leaf material that can be correlated with anthropogenic/industrial aged changes in atmospheric CO₂ recorded from ice cores (Woodward, 1987), and analyses of fossil stomatal densities from older geological sediments in Mesozoic and Palaeozoic plants, which can be compared with estimates of ancient atmospheric CO₂ content based on geochemical analyses and computer modelling of global carbon cycling (McElwain, 1998; McElwain et al., 1999). A number of difficulties are centred on the interpretation and significance of these changes particularly when applied to stomatal changes in deep time (Royer, 2001). They include the type of stomatal parameter quantified for fossil leaves combined with the effect of local fluctuations of CO₂-levels, the variation of stomatal numbers on sun and shade leaves of the same organism (Poole & Kürschner, 1999), the sensitivity of stomatal density (number of stomata per mm² on the leaf surface) as opposed to stomatal index (the number of stomata in reference to the number of leaf epidermal cells), the relative sensitivity of stomatal parameters with other environmental effects such as light intensity, temperature and humidity, and the relative sensitivity and variation observed between changes in stomatal parameters by different phenotypes and genotypes (Royer, 2001; Beerling & Royer, 2002; Woodward et al., 2002).

Despite the uncertainties surrounding the actual cause, effect and variation underlying stomatal parameters, early land plants in the Silurian to Middle Devonian lack laminate foliage and recent studies indicate that these have low stomatal densities, coinciding with relatively high atmospheric CO₂ concentrations. Laminate leaves are generally perceived to have appeared rather ‘late’ in plant evolution and recent observers have linked their appearance with lowered atmospheric ρCO₂ towards the Late Devonian c. 360 Myr (Beerling et al., 2001; Osborne et al., 2003). These authors propose that archaic land plants were constrained from developing laminate leaves in the early Devonian because high ρCO₂ precluded high stomatal densities and sufficient cooling via transpiration to dissipate heat from solar radiation and avoid lethal overheating of the laminate leaf (Beerling et al., 2001). By the late Devonian and Early Carboniferous, ρCO₂ had dropped by 90%, and planate leaves were a more common though by no means exclusive component of terrestrial plants. The authors suggest that lower ρCO₂ elicited a response to higher stomatal densities, permitting the evolution of laminate foliage after enhancing stomatal conductance and cooling. If true, this scenario is an interesting case where phenotypic plasticity in relation to a specific environmental factor – stomatal densities in relation to ρCO₂– could have played a role in unlocking a developmental constraint for a major innovation.

The ‘delay’ in laminate leaf evolution could have also resulted from the fact that requisite hydraulic traits are highly integrated with many other aspects of the plant's physiology and morphology: from water uptake in the root, water storage, water conduction and safety factors in terms of hydraulic stress/failure and trade-offs with mechanical functioning (Rowe & Speck, 2003). This is contrary to the view that the evolution of the leaf lamina might have been ‘developmentally simple’ (Kenrick, 2001; Osborne et al., 2003). Increasing hydraulic conductance would have been a requisite upgrade in complexity to accommodate lamina evolution and would have required significant alteration of the hydraulic and mechanical architecture of possibly the entire plant body. The most important innovation, which succeeded in doing this, in addition to enlarging primary xylem conduit diameter, was probably the appearance of wood – which brings us back to the cambium and growth forms.

Secondary growth of xylem occurred in at least five lineages by the end of the Devonian during the crucial period leading to the establishment of laminate foliage. In the lignophyte lineage, the appearance of the cambium in the middle Devonian before the appearance of laminate leaves in the group represents an important preaptation for the appearance of leaves and large self-supporting trees of all other gymnosperms and angiosperms. Whatever the exact combination of ecological factors linking the evolution of secondary growth and leaves, the two innovations marked major turning points in how plants would be able to respond to the environment and indeed change it. Observers, in reference to evolutionary innovations (Cronk et al., 2002) as well as palaeontologists (Algeo & Scheckler, 1998) have remarked on the importance these have had in actually changing the ecology of the planet: turning the land green, forming soils, fixing carbon, sequestering carbon, increasing clastic weathering, and altering atmospheric ρCO₂. Interestingly, some of these effects initiated by increasing growth form size towards the middle Devonian, reduced atmospheric ρCO₂, and might actually have actually initiated the appearance of leaves. A two phase feedback loop whereby: (1) increasing growth form size and complexity in the lower to Middle Devonian led to increased sequestering of carbon, soil formation, clastic erosion and lowered atmospheric ρCO₂: (2) lowered ρCO₂ possibly released a phenotypic constraint from low to high stomatal densities, thus providing the requisite preaptation for lamina-cooling and leaf evolution. In the seed plant lineage, the bifacial cambium was already in place for potentially modulating the mechanics and hydraulics necessary for leafy growth forms.

Growth Forms, Ecosystems and Responses to Environmental Change

Climbers can show markedly increased rates of photosynthesis (Greaves & Buwalda, 1996) and higher biomass (Granados & Korner, 2002) when grown in elevated atmospheric ρCO₂. Recent studies based on long-term censused tropical forest sites report an increase in abundance and dominance of lianas, especially in the neotropics (Phillips et al., 2002; Wright et al., 2004; but see Caballé & Martin, 2001) for the paleotropics of West Africa. Increases in liana abundance have been discussed with reference to the fact that climbers respond particularly quickly to elevated CO₂ levels as well as forest disturbance. Either or both of these factors could explain increases in liana abundance. What is of great interest is whether the increase of lianas does either locally or more generally raise atmospheric ρCO₂ either by increased turnover – sequestering more biomass into leaves relative to wood – or by slowing wood growth and increasing mortality rates of self-supporting plants. Of further interest is whether this causes a positive feedback loop (Phillips et al., 2002) where increased lianoid growth would elevate atmospheric ρCO₂ and thus further increase lianoid growth to the detriment of trees. Whatever the reality in this complex issue – most commentators state that more information is required – changes in ecological composition in tropical forests do appear to affect varying growth responses to the environment. A key factor being different environmental effects on different growth forms.

Growth form diversity has been, until relatively recently, an arguably understudied aspect of land plant biodiversity and evolution. Especially so compared with recent emphasis on systematic diversity via nucleotides of genes that are not implicated in growth form development and differentiation. However, phylogenetic studies are essential for placing patterns of growth form in an evolutionary context. Genomic studies on the secondary vascular cambium have been relatively slow to develop compared with those based on primary meristems, leaves and floral development for a variety of reasons (Chaffey, 2002). While studies on model organisms such as Arabidopsis are providing a way forward for how growth forms might respond differently to the environment, much of the development crucial for growth form variation arguably concerns the vascular cambium. Although Arabidopsis can produce a limited amount of wood in the hypocotyl (Chaffey et al., 2001), and may be useful for some aspects of wood development (Bhalerao et al., 2003), it is limited for exploring changes in cambial activity of large-bodied woody plants with long life trajectories and wide ranges of wood development variation (Cseke & Podila, 2004). Genomic studies on woody plants are now providing more insight on developmental variations during wood formation (Hertzberg et al., 2001) and comparative genomics of wood-forming tissues in pine and Arabidopsis (Kirst et al., 2003). The latter suggests a high similarity of long-chain sequences related to wood development between Populus and Arabidopsis, suggesting long-term conservation of coding sequences since the estimated last common ancestor of c. 300 Myr. The plant fossil record can go some way to substantiate such inferences, if not by molecular evidence, then by exquisitely preserved cellular preservation of tissue development. Figure 4a illustrates the detailed organization of the cambial tissue preserved in a c. 340 Myr old seed plant with many of the cell types and differentiating zones present in gymnosperms and angiosperms today.

Observers of evolutionary developmental approaches have encouraged more research into phylogentically distanced organisms in order to understand key evolutionary processes and innovations (Cronk, 2001). A similar suggestion can be voiced for genomic investigations of widely ranging growth forms, particularly the differing wood types between young and old development. Such changes are implicit to many self-supporting and climbing woody plants.

The genomic and environmental controls of these transitions are of great interest for both ecology and evolution. Heterochronic mechanisms are increasingly being seen as important for evolutionary change in plants (Kellogg, 2002). We suspect that evolutionary transitions involving heterochronic modification of the developmental plan can have a significant effect on the resulting mechanical architecture. Lianas can range from saplings to highly flexible, suspended trunks. Retention of juvenile characteristics (stiff juvenile wood) into maturity could result in a self-supporting plant, whereas rapid expression of adult characters (compliant wood) would result in high flexibility from an early age with little or no self-supporting phase and possibly an entirely different growth form. Such scenarios are at least partly consistent with recent discussions on saltational evolutionary change in plants through mutations of key developmental genes (Bateman & DiMichele, 1994, 2002; Bateman, 1994). Some growth forms can show profound morphological changes during the growth trajectory, which could be viewed as a type of age or ontogeny-related phenotypic plasticity. Environmental effects play an important role in modifying these changes. We propose that this kind of phenotypic and evolutionary flexibility centred on the growth form and, in particular, the vascular cambium, at least partly explains the survival of the lignophyte clade across periods of perturbation and extinction, leading to their dominance today.