Both within and between species, leaves of plants display wide ranges in structural features. These features include: gross investments of carbon and nitrogen substrates (e.g. leaf mass per unit area); stomatal density, distribution between adaxial and abaxial surfaces, and aperture; internal and external optical scattering structures; defensive structures, such as trichomes and spines; and defensive compounds, including UV screens, antifeedants, toxins, and silica abrasives. I offer a synthesis of selected publications, including some of my own. A unifying theme is the adaptive value of expressing certain structural features, posed as metabolic costs and benefits, for (1) competitive acquisition and use of abiotic resources (such as water, light and nitrogen) and (2) regulation of biotic interactions, particularly fungal attack and herbivory. Both acclimatory responses in one plant and adaptations over evolutionary time scales are covered where possible. The ubiquity of trade-offs in function is a recurrent theme; this helps to explain diversity in solutions to the same environmental challenges but poses problems for investigators to uncover numerous important trade-offs. I offer some suggestions for research, such as on the need for models that integrate biotic and abiotic effects (these must be highly focused), and some speculations, such as on the intensity of selection pressures for these structures.

The mechanical behaviour of large foliage leaves in response to static and dynamic mechanical forces is reviewed in the context of a few basic engineering principles and illustrated in terms of species drawn from a variety of vascular plant lineages. When loaded under their own weight or subjected to externally applied forces, petioles simultaneously bend and twist, and thus mechanically operate as cantilevered beams. The stresses that develop in petioles reach their maximum intensities either at their surface or very near their centroid axes, where they are accommodated either by living and hydrostatic tissues (parenchyma and collenchyma) or dead and stiff tissues (sclerenchyma and vascular fibres) depending on the size of the leaf and the species from which it is drawn. Allometric analyses of diverse species indicate size-dependent variations in petiole length, transverse shape, geometry and stiffness that accord well with those required to maintain a uniform tip-deflection for leaves with laminae differing in mass. When dynamically loaded, the laminae of many broad-leaved species fold and curl into streamlined objects, thereby reducing the drag forces that they experience and transmit to their subtending petioles and stems. From a mechanical perspective, the laminae of these species operate as stress-skin panels that distribute point loads more or less equally over their entire surface. Although comparatively little is known about the mechanical structure and behaviour of foliage leaves, new advances in engineering theory and computer analyses reveal these organs to be far more complex than previously thought. For example, finite-element analyses of the base of palm leaves reveal that stresses are decreased when these structures are composed of anisotropic as opposed to isotropic materials (tissues).

Leaf expansion rate varies with leaf temperature, photon flux density (PPFD), evaporative demand and soil water status. In most simulation models, it is calculated every day by multiplying the amount of carbohydrate available to leaves by specific leaf area (SLA). However, leaf expansion rate is considerably reduced by mild water deficits which do not affect photosynthesis, and is not affected by a reduction in the PPFD intercepted during rapid leaf expansion. Specific leaf area undergoes a several-fold variability depending on PPFD, soil water status and time of day. It is increased when environmental conditions have a greater depressive effect on expansion rate than on photosynthesis, and is decreased in the opposite case. It is therefore appropriate to model leaf expansion independently of the plant carbon budget. Consistent characteristics can be deduced from a series of experiments, allowing a model of leaf expansion to be proposed. (i) Time courses of relative leaf expansion rate and of epidermal cell division rate are well conserved within a plant and across a large range of environmental conditions, provided that durations and rates are expressed in thermal time. Maximum relative rates are common to all zones of a leaf and to all leaves of a plant, in maize and sunflower. (ii) A water deficit, or a reduction in intercepted PPFD, imposed in the first half of the period of leaf development affects the relative expansion rate in the deficit only, but permanently affects the absolute expansion rate. In contrast, a reduction in PPFD causes no effect on leaf expansion if imposed in the rapid expansion period when the leaf is autotrophic. (iii) Expansion rate is related to evaporative demand and to the concentration of ABA in the xylem sap with relationships that apply under both field and laboratory conditions. (iv) Tissue expansion and epidermal cell division behave as independent processes which determine epidermal cell area at each time.

We have explored the relationships between specific leaf area calculated for a whole plant and its individual leaves. Barley was grown in hydroponics in controlled environment cabinets. Plants were harvested on the basis of physiological age (defined as the number of days after full expansion of leaves on the main stem) and the area and weight of whole, fully expanded, leaves measured and specific leaf area (SLA) of individual leaves or whole plants calculated. Specific leaf area calculated for individual leaves (SLAL) varied with leaf position and with leaf age after full expansion whereas SLA calculated for whole plants (SLAP) varied with plant age. The same conclusions were reached whether the results were based on total dry weight or dry weight minus soluble carbohydrates (‘structural weight’). Transferring plants to shade on the day of full expansion of the third leaf on the main stem increased the SLAP, and also SLAL of leaves 3 and 4 on the main stem (leaf 4 being the younger leaf of the two), because of a decrease in the ‘structural weight’ of these leaves. However SLAL of leaf 2 (which was older than leaf 3) was not affected by shading; the effect was confined to leaves developing in the new conditions.

The contribution of the starch and soluble carbohydrate pools to the diurnal variations of leaf mass per unit area (LMA) has been investigated in tomato leaves. A glasshouse experiment was carried out with plants pruned to two or five fruits per truss. Leaflets were sampled at sunrise, noon and sunset at different positions within the leaf (basal or terminal), and on different sympods along the stem. Carbohydrate contents and LMA were significantly higher in the terminal than in the basal leaflets, except at sunrise. During the day, differences in starch accumulation between terminal and basal leaflets increased with leaf height on the plant. Among sympods, the soluble carbohydrate content of the terminal leaflets did not vary significantly, whereas at 13.00 h the LMA was minimum in the middle of the plant and maximum at the top, and the leaf starch content significantly increased half-way up the plant. The plant fruit load had only small and non-significant effects on the LMA and carbohydrate contents. The response of LMA and carbohydrate contents to changing source activity was observed under controlled climatic conditions. The starch pool of fully expanded leaves was rapidly filled and emptied under increasing and decreasing source activity. In young expanding leaves, this pool was hardly filled during daylight. On average the soluble carbohydrates did not contribute significantly to the diurnal variations in LMA, whereas fluctuations in starch explained c. 70% and 44% of these variations in the upper and lower leaves, respectively. The results are discussed with respect to the modelling of LMA at the level of individual tomato leaves or sympods.

The composition and morphology of leaves exposed to elevated [CO2] usually change so that the leaf nitrogen (N) per unit dry mass decreases and the leaf dry mass per unit area increases. However, at ambient [CO2], leaves with a high leaf dry mass per unit area usually have low leaf N per unit dry mass. Whether the changes in leaf properties induced by elevated [CO2] follow the same overall pattern as that at ambient [CO2] has not previously been addressed. Here we address this issue by using leaf measurements made at ambient [CO2] to develop an empirical model of the composition and morphology of leaves. Predictions from that model are then compared with a global database of leaf measurements made at ambient [CO2]. Those predictions are also compared with measurements showing the impact of elevated [CO2]. In the empirical model both the leaf dry mass and liquid mass per unit area are positively correlated with leaf thickness, whereas the mass of C per unit dry mass and the mass of N per unit liquid mass are constant. Consequently, both the N[ratio ]C ratio and the surface area[ratio ]volume ratio of leaves are positively correlated with the liquid content. Predictions from that model were consistent with measurements of leaf properties made at ambient [CO2] from around the world. The changes induced by elevated [CO2] follow the same overall trajectory. It is concluded that elevated [CO2] enhances the rate at which dry matter is accumulated but the overall trajectory of leaf development is conserved.

Genotypic variability was studied in two Mediterranean grass species, Bromus erectus and Dactylis glomerata, with regard to the response to CO2 of leaf total non-structural carbohydrate concentration ([TNC]lf), specific leaf area (SLA), and leaf carbon and nitrogen concentrations ([C]lf and [N]lf, respectively). Fourteen genotypes of each species were grown together on intact soil monoliths at ambient and elevated CO2 concentrations (350 and 700 μmol mol−1, respectively). In both species, the most consistent effect of elevated CO2 was an increase in [TNC]lf and a decrease in leaf nitrogen concentration when expressed either as total dry mass [Nm]lf, structural dry mass [Nmst]lf or leaf area [Na]lf. The SLA decreased only in D. glomerata, due to an accumulation of total non-structural carbohydrates and to an increase in leaf density. No genotypic variability was found for any variable in B. erectus, suggesting that genotypes responded in a similar way to elevated CO2. In D. glomerata, a genotypic variability was found only for [Cst], [Nm]lf, [Nmst]lf and [Na]lf. Since [Nm]lf is related to plant growth and is a strong determinant of plant–herbivore interactions, our results suggest evolutionary consequences of elevated CO2 through competitive interactions or herbivory.

Oxygen evolution was measured from mesophyll tissues in spinach leaves using a photoacoustic technique. The photosynthetic capacity of individual cell layers was measured by directing microscopic beams of light, 40 μm wide, to cells exposed within a leaf cross section. The resulting profile for oxygen-evolution potential was relatively flat, indicating a uniform capacity for photosynthesis in leaf mesophyll tissues. Two experimental approaches were used to estimate the photosynthetic performance of individual mesophyll cell layers when white light was applied to the adaxial leaf surface. These experiments indicated that oxygen was produced relatively uniformly across the mesophyll and that oxygen evolution increased with irradiance of the white light applied to the leaf surface. The measured profiles for oxygen evolution and capacity are flatter than previous measurements of profiles of fixed carbon and estimates of profiles for absorbed light within spinach leaves.

The function of a leaf is photosynthesis, which requires the interception of light and access to atmospheric CO2 while controlling water loss. This paper examines the influence of leaf anatomy on both light capture and CO2 diffusion. As photosynthetic metabolism is spread between many chloroplasts, a leaf faces the challenge of matching light capture by a given chloroplast with the metabolic capacity of that chloroplast. Chloroplasts nearest the leaf surface receive the greatest irradiance and therefore absorb more light per unit chlorophyll than chloroplasts in the centre of a leaf. Electron transport and carbon fixation capacities per unit of chlorophyll decline with increasing depth in the leaf, to compensate for the decline in light absorbed per unit chlorophyll. Many key photosynthetic protein complexes in chloroplasts have nuclear encoded genetic information. Consequently, all chloroplasts within a given cell have a similar metabolic complement, which limits the potential gradient of photosynthetic capacity per unit chlorophyll across the leaf. A simple model couples light absorption through the leaf (based on the Beer–Lambert law) with the profile of chlorophyll through a leaf and the gradient in photosynthetic capacity. It is validated by comparison with 14CO2 fixation profiles through spinach leaves obtained in various studies. The model can account for published 14C fixation profiles obtained with blue, red and green light of different irradiances and white light applied in different combinations to the adaxial and abaxial surfaces of spinach leaves. The model confirms that spongy mesophyll increases the apparent extinction coefficient of chlorophyll compared to palisade tissue. The palisade tissue nearest the surface which receives light facilitates the penetration of light to a greater depth, while spongy mesophyll promotes scattering to enhance light absorption, thus reducing the gradient in light absorbed per unit chlorophyll through a leaf. CO2 fixation faces a diffusional limitation, which necessitates Rubisco to be spread evenly across the cell walls exposed to intercellular airspace. Mesophyll cell structure reflects the need to have a large cell surface per unit volume exposed to airspaces. The regular array of columnar cells in palisade tissue, or cell lobing in monocot leaves, results in greater exposed surface per unit tissue volume than spongy mesophyll. The exposed surface area per unit leaf area scales with photosynthetic capacity such that the difference in CO2 partial pressure between substomatal cavities and the sites of carboxylation within chloroplasts is, on average, independent of photosynthetic capacity of the leaf. However, Rubisco specific activity declines as the Rubisco content per unit leaf area increases due to greater internal diffusional limitations.

This study explored reflectance indices sampled with a ‘leaf reflectometer’ as measures of pigment content for leaves of contrasting light history, developmental stage and functional type (herbaceous annual versus sclerophyllous evergreen). We employed three reflectance indices: a modified normalized difference vegetation index (NDVI), an index of chlorophyll content; the red/green reflectance ratio (RRED[ratio ]RGREEN), an index of anthocyanin content; and the change in photochemical reflectance index upon dark–light conversions (ΔPRI), an index of xanthophyll cycle pigment activity. In Helianthus annuus (sunflower), xanthophyll cycle pigment amounts were linearly related to growth light environment; leaves in full sun contained approximately twice the amount of xanthophyll cycle pigments as leaves in deep shade, and at midday a larger proportion of these pigments were in the photoprotective, de-epoxidized forms relative to shade leaves. Reflectance indices also revealed contrasting patterns of pigment development in leaves of contrasting structural types (annual versus evergreen). In H. annuus sun leaves, there was a remarkably rapid increase in amounts of both chlorophyll and xanthophyll cycle pigments along a leaf developmental sequence. This pattern contrasted with that of Quercus agrifolia (coast live oak, a sclerophyllous evergreen), which exhibited a gradual development of both chlorophyll and xanthophyll cycle pigments along with a pronounced peak of anthocyanin pigment content in newly expanding leaves. These temporal patterns of pigment development in Q. agrifolia leaves suggest that anthocyanins and xanthophyll cycle pigments serve complementary photoprotective roles during early leaf development. The results illustrate the use of reflectance indices for distinguishing divergent patterns of pigment activity in leaves of contrasting light history and functional type.

The relationships between leaf structure, nitrogen concentration and CO2 assimilation rate (A) were studied for 14 grass species grown in the laboratory under non-limiting nutrient conditions. Structural features included leaf thickness and density, and the proportion of leaf volume occupied by different types of tissue (mesophyll, epidermis, vessels and sclerenchyma). Relationships were assessed for data expressed per unit leaf area and fresh mass. The latter was found to be closely related to leaf volume, which allowed us to use A per unit leaf fresh mass (Afm) as a surrogate of A per unit leaf volume. Assimilation rate per unit leaf area (Aa) was positively correlated with leaf thickness and with the amount of mesophyll per unit leaf area; the relationship with leaf nitrogen content per unit area was only marginally significant. Afm was negatively correlated with leaf thickness and positively with fresh mass-based leaf organic nitrogen concentration. A multiple regression involving these two variables explained 81% of the variance in Afm. The value of Afm was also significantly related to the proportion of mesophyll in the leaf volume, but surprisingly the correlation was negative. This was because thin leaves with high Afm and nitrogen concentration had proportionally more mechanically supportive tissues than thick ones; as a consequence, they also had a lower proportion of mesophyll. These data suggest that, in addition to leaf nitrogen, leaf thickness has a strong impact on CO2 assimilation rate for the grass species studied.

This study examines interrelationships between eight leaf attributes (specific leaf mass, area, dry mass, lamina thickness, mesophyll cell number per cm2, mesophyll cell volume, chloroplast volume, and number of chloroplasts per mesophyll cell) in field-grown plants of 94 species from the Eastern Pamir Mountains, at elevations between 3800 and 4750 m. Unlike most other mountain areas, the Eastern Pamirs, Karakorum system, Tadjikistan provide localities where low temperatures and radiation combine with moisture stress at high altitudes. For all the attributes measured, significant differences were found between plants with different mesophyll types. Leaves with dorsiventral palisade structure (dorsal palisade, ventral spongy mesophyll cells) had thicker leaves with larger but fewer mesophyll cells, containing more and larger chloroplasts. These differences in mesophyll type are reflected in differences in the total surface of mesophyll cells per unit leaf area (Ames/A) or volume (Ames/V). Plants with isopalisade leaf structure (palisade cells under both dorsal and ventral surfaces) are more commonly xerophytes and their increased values of Ames/A and Ames/V decrease CO2 mesophyll resistance, which is an important adaptation to drought. Path analysis shows the critical importance of mesophyll cell volume in leading to the covariance between the different leaf attributes and hence to specific leaf mass (SLM), even though mesophyll cell volume is not itself strongly correlated with SLM. This is because mesophyll cell volume increases SLM through its effects on leaf thickness and chloroplast number per cell, but decreases SLM through its negative effect on mesophyll cell density.

An appendix is available for this article which was not included in the print edition of the Journal. The appendix may be found at the rear of the full text, PDF version of the article.

A popular conceptual model asserts that shade tolerance is characterized by morphological and physiological traits that enhance the net rate of carbon capture in low light. We tested this model by quantitatively reviewing growth, leaf lifespan, CO2 exchange and morphological data from 76 studies on woody seedlings grown under conditions of low light. Data were placed into three tolerance categories (intolerant, intermediate, tolerant), two light categories (less than 4% and 4–12%) and two leaf phenology categories (broad-leaved evergreen and winter deciduous). For both evergreen and deciduous groups, intolerant species had traits conferring greater growth potential than tolerant species in both light categories. These traits included greater leaf mass ratio, leaf area ratio, specific leaf area and mass-based photosynthetic rates above light compensation. However, in 0–4% light, growth rates were similar for intolerant and tolerant species, because low light together with higher respiration rates for intolerant species limited the expression of their growth potential differences. Deciduous and evergreen intolerant species were similar in many respects. However, both intermediate and tolerant deciduous species had markedly lower leaf mass ratios and higher root mass ratios than intermediate and tolerant evergreen species. In addition, deciduous species and intolerant evergreens must cope with as much as sixfold higher leaf turnover rates than tolerant evergreen species. Thus, rather than maximizing growth rates in low light, tolerant evergreen species minimize biomass loss through long leaf lifespans and low respiration rates. Tolerant deciduous species also minimize biomass losses by minimizing whole-plant respiration rates but they accomplish low biomass turnover though low leaf mass ratio and not low leaf turnover rates. Furthermore, unlike most tropical evergreens, tolerant deciduous species can gain large fractions of their total growing season carbon during short periods when the overstory is leafless and then allocate this carbon to storage (as reflected by high root mass ratios) rather than new leaves. In conclusion, we found no support for the low-light-enhanced carbon capture model of shade tolerance as viewed strictly from the perspective of physiological growth capacity. This can be explained by the disadvantages to net growth and survival of maintaining a high growth potential at low light, because high growth potential results in greater rates of whole-plant respiration, tissue turnover, herbivory and mechanical damage and in decreased storage. Thus, shade tolerance can be characterized by traits that maximize survival and net growth, where net growth includes losses to all agents.

A key element of most recently proposed plant strategy schemes is an axis of resource capture, usage and availability. In the search for a simple, robust plant trait (or traits) that will allow plants to be located on this axis, specific leaf area is one of the leading contenders. Using a large new unpublished database, we examine the variability of specific leaf area and other leaf traits, the relationships between them, and their ability to predict position on the resource use axis. Specific leaf area is found to suffer from a number of drawbacks; it is both very variable between replicates and much influenced by leaf thickness. Leaf dry-matter content (sometimes referred to as tissue density) is much less variable, largely independent of leaf thickness and a better predictor of location on an axis of resource capture, usage and availability. However, it is not clear how useful dry matter content will be outside northwest Europe, and in particular in dry climates with many succulents.

Laboratory experiments have shown a large difference in specific leaf area (SLA, leaf area: leaf mass) between species from nutrient-poor and nutrient-rich habitats, but no systematic difference in the construction costs (the amount of glucose required to construct 1 g biomass). We examined how far these patterns are congruent with those from field-grown plants. An analysis was made of the vegetation in a range of grasslands and heathlands differing in productivity. The SLA of the dominant species in 15 different habitats was determined, as well as chemical composition and construction costs of bulk samples of leaves. SLA in the field was generally lower than in the laboratory, but showed consistency in that the ranking across species remained the same. Species from highly productive habitats had higher SLA than those from sites of low productivity, although individual species sometimes deviated substantially from the general trend. Construction costs were similar for plants from different habitats. This was mainly due to the positive correlation between an expensive class of compounds (proteins) and a cheap one (minerals).

Nutrient conservation plays an important role in plants adapted to infertile environments. Nutrients can be conserved mainly by extending the life span of plant parts and/or by minimizing the nutrient content of those parts that are abscissed. Together these two parameters (life span and resorption) define the mean residence time (MRT) of a nutrient. In this review we summarize available information on nitrogen resorption and life span, and evaluate their relationship to the MRT of nitrogen, both between and within species. Abundant information with respect to nitrogen resorption efficiency and life span is available at the leaf level. By definition, woody evergreen plants have a much longer leaf life span than species of other life-forms. Conversely, differences in resorption efficiency among life-forms or among plants in habitats differing in soil fertility appear to be small. Inter-specific variation in leaf life span is much larger than intra-specific variation (factor of >200 compared with 2, respectively), while resorption efficiency varies by about the same magnitude at both levels (factor of 3.8 compared with 2.7, respectively). The importance of resorption efficiency in determining leaf-level MRT increases exponentially towards and above the maximum resorption efficiency observed in nature. This effect is independent of leaf life span, which may explain the lack of life-form related differences in resorption efficiency. When scaling up from the leaf to the whole-plant level, fundamental differences in turnover rate among different plant organs must be considered. Woody species invest c. 50% of their net productivity into their low-turnover stems, while in herbaceous species the life span of stems is only slightly longer than that of leaves. As a result, nutrient turnover of woody (evergreen and deciduous) plants is generally lower than that of herbaceous species (herbs and graminoids) on a whole-plant basis. At the intra-specific level empirical data show that both biomass life span (i.e. the inverse of biomass loss rate) and resorption efficiency are important sources of variation in MRT. However, we argue that the relative importance of resorption efficiency in explaining variation in MRT is lower at the inter-specific level, whereas the reverse is true for life span. This is because variation in MRT and life span is much larger at the inter-specific level compared with variation in resorption efficiency. Plant traits related to nutrient conservation are discussed with respect to their implications for leaf structure, plant growth, competition, succession and ecosystem nutrient cycling.

There is some evidence that traits of fresh leaves that provide structural or chemical protection (‘defence’) remain operational in the leaf litter and control interspecific variation in decomposition rate in or on the soil. We tested experimentally whether the negative relationship between foliar defence and litter decomposition rate is fundamental, i.e. whether it is seen consistently across higher plant species and life forms, and whether it is repeated in the floras of geographically and climatically distinct areas separated by an ocean. We employed the published results of two outdoor litter bag experiments, in which we simultaneously compared the relative mass losses (‘decomposibility’) of leaf litters of a wide range of plant species. One experiment was in Córdoba, Argentina, and included 48 Argentine species typical of the dry, subtropical landscapes along a steep altitudinal gradient. The other was in Sheffield, UK, and hosted 72 British species typical of the temperate–Atlantic landscape there. We linked the two experiments through a similar experiment in Sheffield that hosted litters of subsets of both the Argentine and British species. We also tested fresh leaves of all species from the same areas for tensile strength (‘toughness’) and relative palatability to generalist herbivorous snails in multi-species ‘cafeteria’ experiments. Both in Argentina and in Great Britain there were highly significant correlations between leaf palatability (r=0.61; 0.73) or leaf tensile strength (r=−0.60; −0.60) and litter mass loss across all species. These relationships could be explained by variation both between and within broad life-form groups. Specific leaf area (area[ratio ]dry mass) of fresh leaves was consistently correlated only with litter mass loss within British life form groups. We illustrated the possible ecosystem consequences of these relationships by comparing functional traits of British species differing in leaf habit. In comparison with deciduous species, evergreens generally had innately slow growth, which corresponded to their longer-lived leaves of lower specific leaf area, higher tensile strength and lower palatability to generalist invertebrate herbivores. Correspondingly, evergreens produced more resistant leaf litter. Thus, slow-growing evergreens might maintain their position in infertile ecosystems through leaf traits that help them to conserve their nutrients efficiently and to keep nutrient mineralization low, thereby not allowing potentially fast-growing deciduous species to outcompete them.

Models have been formulated for monospecific stands in which canopy photosynthesis is determined by the vertical distribution of leaf area, nitrogen and light. In such stands, resident plants can maximize canopy photosynthesis by distributing their nitrogen parallel to the light gradient, with high contents per unit leaf area at the top of the vegetation and low contents at the bottom. Using principles from game theory, we expanded these models by introducing a second species into the vegetation, with the same vertical distribution of biomass and nitrogen as the resident plants but with the ability to adjust its specific leaf area (SLA, leaf area[ratio ]leaf mass). The rule of the game is that invaders replace the resident plants if they have a higher plant carbon gain than those of the resident plants. We showed that such invaders induce major changes in the vegetation. By increasing their SLA, invading plants could increase their light interception as well as their photosynthetic nitrogen-use efficiency (PNUE, the rate of photosynthesis per unit organic nitrogen). By comparison with stands in which canopy photosynthesis is maximized, those invaded by species of high SLA have the following characteristics: (1) the leaf area index is higher; (2) the vertical distribution of nitrogen is skewed less; (3) as a result of the supra-optimal leaf area index and the more uniform distribution of nitrogen, total canopy photosynthesis is lower. Thus, in dense canopies we face a classical tragedy of the commons: plants that have a strategy to maximize canopy carbon gain cannot compete with those that maximize their own carbon gain. However, because of this strategy, individual as well as total canopy carbon gain are eventually lower. We showed that it is an evolutionarily stable strategy to increase SLA up to the point where the PNUE of each leaf is maximized.

The coupling between leaf structure and function is illustrated with reference to two examples, the C4 photosynthetic pathway and leaf pubescence. A distinction is made between function and functional significance. The latter is defined as the role, significance or consequence of a structure, whereas the former is more simply the action that a structure is capable of performing. Using the two examples, four generalizations are made concerning the relationships between structure, function and functional significance: the functional significance of leaf structure is environment-dependent; the relationship between functional significance and structure is sometimes non-intuitive; functional equivalency means that there is often more than one ‘solution’ to the same ‘constraint’; and the consequences of leaf structure can exert profound effects at levels of organization beyond those of the individual organism and may play a critical role in determining community structure and function, through interactions with other species and trophic levels. The importance of understanding the consequences in variation in leaf structure at the global scale is illustrated with reference to the issue of global climate change.