Key Plant Structural and Allocation Traits Depend on Relative Age in the Perennial Herb Pimpinella saxifraga

Abstract

• Background and Aims Perennial plant formations always include a mixture of various-aged individuals of community-creating species, but the physiological and competitive potentials of plants of differing age and the importance on whole community functioning are still not entirely known. The current study tested the hypothesis that ontogenetically old plants have limited biomass investments in leaves and enhanced foliage support costs.

• Methods Leaf structure, size and biomass allocation were studied in the perennial herb Pimpinella saxifraga during plant ontogeny from seedling to senile phases to determine age-dependent controls on key plant structural traits. The average duration of the full ontogenetic cycle is approx. 5–10 years in this species. Plants were sampled from shaded and open habitats.

• Key Results Leaflet dry mass per unit area (M_A) increased, and the fraction of plant biomass in leaflets (F_L) decreased with increasing age, leading to a 5- to 11-fold decrease in leaf area ratio (LAR = F_L/M_A) between seedlings and senescent plants. In contrast, the fraction of below-ground biomass increased with increasing age. Leaflet size and number per leaf increased with increasing age. This was not associated with enhanced support cost in older plants as age-dependent changes in leaf shape and increased foliage packing along the rachis compensated for an overall increase in leaf size. Age-dependent trends were the same in habitats with various irradiance, but the LAR of plants of varying age was approx. 1·5-fold larger in the shade due to lower M_A and larger F_L.

• Conclusions As plant light interception per unit total plant mass scales with LAR, these data demonstrate major age-dependent differences in plant light-harvesting efficiency that are further modified by site light availability. These ontogenetic changes reduce the differences among co-existing species in perennial communities, and therefore need consideration in our understanding of how herbaceous communities function.

INTRODUCTION

Plant structural and physiological characteristics are tightly coupled (Wright et al., 2004; Niinemets and Sack, 2005). Several studies have determined the key functional attributes of a large number of herbaceous species to understand species competitive potential in the community and mechanisms of species co-existence (Grime et al., 1997; Huovinen-Hufschmid and Körner, 1998; Wohlfahrt et al., 1999; Craine et al., 2002; Storkey, 2004; Villar et al., 2005). In particular, leaf dry mass per unit leaf area (M_A), fraction of total plant dry mass in leaves (F_L) and leaf area ratio (total leaf area per unit total plant mass; LAR = F_L/M_A) have been identified as the major structural traits determining the potential for light harvesting and accordingly species status in the community (Poorter and Garnier, 1999; Poorter and Nagel, 2000). The studies have demonstrated that these important traits vary widely among species from a given community (Huovinen-Hufschmid and Körner, 1998; Villar et al., 2005), potentially leading to more efficient usage of resources by multi-species communities (Tilman et al., 1997; Reich et al., 2001).

Most of the detailed information of species key traits has been derived from relatively short term studies, often lasting only a few weeks to one growing season. This may be a significant limitation given that the average time for completion of the full ontogenetic cycle is on average 5–20 years, and 50–70 years at extremes in perennial herbs (Gatsuk et al., 1980; Falinska, 1985). This large potential longevity implies that all self-sustaining herb communities consist of plant populations of varying age (Falinska, 1985; Rabotnov, 1985; Moloney, 1988). The population age state structure has been characterized repeatedly (Falinska, 1985; Rabotnov, 1985; Zhukova and Ermakova, 1985; Carlsson and Callaghan, 1991), but the influence of ageing on plant competitive potential within the community is largely unexplored. It is currently widely believed that plant vigour decreases with increasing age, but quantitative data are scarce (e.g. Falinska, 1985; Carlsson and Callaghan, 1991).

In trees, M_A strongly increases with increasing tree height and age due to a juvenile to mature phase change (Greenwood, 1995) and possibly also due to decreases in water availability with increasing tree size (Niinemets, 2002). Age-dependent increase in M_A and decrease in LAR have also been observed in the perennial herb Leontodon hispidus (Niinemets, 2004), suggesting that the age-dependent trend of increasing M_A, that results in decreased light intercepting leaf surface area, is more general across different plant functional types.

Apart from these changes, plant maturation and increase in size are associated with increasing average leaf size (Gatsuk et al., 1980; Zhukova, 1987). Because larger leaves require a greater fraction of supporting biomass within the laminas due to extended lever arms, as well as greater fractions of biomass in petioles and stems (Givnish, 1978; Niklas, 1999; Niinemets and Fleck, 2002b), the overall biomass investment in support may significantly increase with increasing plant age, further reducing the biomass for construction of photosynthesizing and light-intercepting surfaces.

Biomass allocation and leaf structure and shape was studied in the perennial herb Pimpinella saxifraga (Umbelliferae) to determine the age-dependent controls on key structural and allocation traits, and to compare the within-species variability due to age with reported data of interspecific trait variability. Because many plant qualitative traits, e.g. formation of taproot, stem and flowers depend on specific plant ontogenetic status rather than on absolute age, plant age was characterized by a relative ten-level age scale (Gatsuk et al., 1980; Niinemets, 2004). Although the age-dependent changes in plant traits have been associated with modifications in accumulation of total plant mass with calendar age (Coleman et al., 1994), ontogenetic changes in plant structure and biomass allocation also occur at a common plant size (Niinemets, 2004). Using biological age is pertinent as previous studies have demonstrated that plant responses to environmental stresses and modifications, as well as the plant role in the community, may be more strongly controlled by the specific ontogenetic state rather than by the calendar age (Lewis et al., 2002).

Apart from the age, light availability may significantly modify foliar and plant allocation traits (Meziane and Shipley, 1999). The study was conducted in two sites of varying above-community irradiance to generalize the structure/allocation versus age relationships and test for the interaction between ontogeny and irradiance.

MATERIALS AND METHODS

Species and the age states

Pimpinella saxifraga L. is a widespread perennial species in temperate meadow communities that reproduces exclusively in a generative manner in natural sites. It may grow on sites with varying soil water availability and acidity, as well as at different above-ground light availabilities. However, it is rarely a dominant species, especially in sites with well-developed turf. Its large ecological potential and diffuse dispersal contribute significantly to the mosaic structure of many meadow communities.

The methodology for age state (relative age) separation, using a series of qualitative and quantitative plant characteristics, has been developed by Rabotnov and colleagues (for reviews, see Gatsuk et al., 1980; Rabotnov, 1985), and the descriptions of relative plant ages for many perennial species are provided by Serebryakova et al. (1983) and Zhukova (1987). According to this classification, the plants are separated between vegetative and reproductive individuals. The vegetative plants are either pre-reproductive (seedlings, juvenile, immature, and virginile) or post-reproductive (subsenile and senile) depending on the presence and size of the taproot and stumps of flower stalk (Fig. 1; Gatsuk et al., 1980; Zhukova, 1987). Seedlings are still attached to the seed coat and partly rely on seed reserves. Juveniles have simple leaves and lack a taproot. Immatures have a weakly developed taproot, and have pinnate compound leaves with two or three leaflets. Virginiles already have adult foliage and a taproot, but lack generative organs. Young generative plants (G1) have a flower stalk with weakly developed stem leaves, and lack the stumps of flower stalk. Mature generative individuals (G2) possess stems with vigorous leaves and several umbel-type inflorescences, and have stumps of flower stalks from previous seasons. Old generative plants (G3) are characterized by less-vigorous stems with lower stature and with only one weakly developed umbel, and an extensive taproot with many stumps of previous flower stalks. Senile individuals differ from subseniles primarily by a more extensive taproot that possesses fewer lateral roots as well as less deeply toothed and rounder leaflets, resembling those in juveniles (Gatsuk et al., 1980). As only two senile plant individuals were found in the current study, plants from subsenile and senile phases were pooled.

Fig. 1.

Examples of representative plants of different relative age of the perennial herb Pimpinella saxifraga. The age state classification is explained in detail in Gatsuk et al. (1980). According to this schema, plant ontogeny is divided between seedling (Pl), juvenile (Juv), immature (Im), virginile (Vir), reproductive (young, G1; mature, G2; old, G3), subsenile (Subs) and senile (Sen) phases. Each ontogenetic phase has a specific combination of qualitative and quantitative traits (presence of flowers and number of inflorescences, stumps of previous flower stalks, size, shape and number of leaves, balance between living and dead structures). Age states were numbered consecutively (0 = Pl, 8 = Sen) resulting in a relative age scale.

Studies demonstrate large site-to-site differences in the rate of ontogenetic development (Zhukova, 1987), implying that it may be difficult to directly relate ontogenetic age to calendar age. Long-term field studies demonstrate that the whole ontogenetic cycle takes on average 5–15years in temperate perennial species including Pimpinella saxifraga (Falinska, 1985; Zhukova, 1987). According to Zhukova (1987) and the observations made at the site, the juvenile phase takes 3–10 growing months (air temperature above 5 °C), the immature phase 3–9months, the virginile phase 4–15months, and the length of generative phases is 8–20months, depending on site conditions. Because of the long duration of the ontogeny, the overall effect of seasonal changes in environmental conditions and plant development on the age state spectrum is minor.

Study site

The study was accomplished at Kõpu, Estonia (58°55′N, 22°12′E), at two sites on a gleyed pseudopodzolic soil formed on calcareous parent rock, in August during peak flowering. The shaded site was a sparse forest dominated by Pinus sylvestris L. in the overstorey and by Calluna vulgaris (L.) Hull., Dactylus glomerata L., Deschampsia flexuosa (L.) Trin., Festuca rubra L., Fragaria vesca L., Melampyrum pratense L. and Veronica chamaedrys L. in the herb and dwarf-shrub layer in the understorey. The open site was a woody meadow with a few P. sylvestris trees and Deschampsia flexuosa (L.) Trin., Galium verum L., Rhinanthus minor L., Lathyrus pratensis L., Potentilla argentea L. and Trifolium repens L. in the herb layer. Pimpinella saxifraga was a relevant component of the herb layer in both habitats (about 10% of total coverage).

Mean ± s.d. proportion of open sky (relative irradiance) was 0·30 ± 0·07 in the shaded site and 0·75 ± 0·06 in the open habitat (means are different at P < 0·005 according to a t-test). Water content of the mineral soil (2- to 10-cm soil layer) tended to be higher in the shaded habitat (15·1 ± 3·6 %) than in the open habitat (10·9 ± 1·8 %), but this was not statistically significant (t-test, P > 0·1, n = 4).

Plant sampling and structural and chemical analyses

Individuals of P. saxifraga were diffusely distributed in both sites, and four sample plots of 0·25 m² (0·5 × 0·5 m) were randomly allocated in each site. All plants of P. saxifraga were carefully excavated from the plots and divided between the ontogenetic states. Altogether, 52 plants grew in the sample plots (eight seedlings, 20juveniles, seven immatures, seven virginiles, three young generative plants (G1), three mature generative plants (G2) and four subseniles and seniles).

All plants were divided between leaves, stems, roots and flowers, and the leaves between leaflets and rachises. As used in the current study, the term rachis embraces both the petiole (non-leafed support) and rachis sensu stricto (leafed support structure). Every leaflet was measured for width, length and area, and every rachis for length and basal diameter. The exposed area of the rachis was calculated by approximating the rachis to a frustum of a cone with an apical diameter of 0·1 mm.

All plant fractions were oven-dried at 70 °C for at least 48 h before determination of dry mass. Leaflet, rachis and stem carbon and nitrogen concentrations were determined with an elemental analyser CHN-O-Rapid (Foss Heraeus GmbH, Hanau, Germany).

Data analysis

Averages of all characteristics for every age state and site were calculated. As the leaflet area was non-normally distributed (Kolmogorov–Smirnov test), it was log-transformed before calculating the mean. The ontogenetic phases were numbered consecutively (0 = seedlings, 8 = seniles; Fig. 1) to obtain a quantitative relative age scale as in Niinemets (2004). The relationships of leaf structural characteristics with age were generally curvilinear and were fitted by second-order polynomial regressions for each site. To compare the sites, plant age, A, was transformed as log(A + 1) to linearize the relationships, and log(A + 1) was used as the covariate in covariance analysis (ANCOVA). First, separate slope analysis, including the interaction term, site × log(A + 1), was carried out to test for the slope differences. As the slopes appeared not to differ in any of the comparisons (P > 0·15), the significant site differences always refer to intercept differences (common slope model that lacks the interaction term; Sokal and Rohlf, 1995). Allometric relationships between leaf size characteristics (log–log scale) for different ontogenetic stages were also compared by ANCOVA. All relationships were considered significant if P < 0·05 (Sokal and Rohlf, 1995).

Because plant structural and functional characteristics change continuously with relative rather than with absolute age (Niinemets, 2004), the relative age was considered as a continuous variable in the current analyses to highlight the trends in plant functioning with ontogeny. Alternative analyses using Spearman rank correlation, and Kruskal–Wallis non-parametric ANOVA resulted in identical conclusions with respect to the statistical significance of the results and, thus, the conclusions are not biased by the way plant age was treated in statistical analyses.

RESULTS

Effects of plant age on foliar morphology and plant biomass partitioning

Leaflet dry mass per unit area (M_A) increased 2·5-fold (shaded habitat) to 3-fold (open habitat) from juvenile (Pl) to subsenile (Subs) age state (Fig. 2A; see Fig. 1 for definition of age states). Total leaf dry mass (mass of leaflets and the central leaf support structure bearing leaflets, i.e. the rachis) per unit total area (rachis + leaflets) also increased with increasing age (r² = 0·95 for the shaded; r² = 0·90 for the open habitat). The contribution of rachis area to total leaf area decreased with increasing age (Fig. 2B). Therefore, the total leaf dry mass to area ratio increased more than M_A with increasing plant age (Fig. 2A), 3-fold in shaded (from 27·9 g m⁻² in Pl to 82·5 g m⁻² in Subs age state) and 3·3-fold (from 32·8 g m⁻² in Pl and 108·3 g m⁻² in Subs) in the open habitat.

Fig. 2.

Effects of relative age on (A) leaflet dry mass per unit area (M_A), (B) the ratio of leaf rachis area to total leaf area (rachis + leaflets), (C) the fraction of total plant mass in leaflets (F_L), (D) and the leaflet area ratio (LAR = F_L/M_A) in P. saxifraga. The measurements were conducted in both the open (open symbols) and shaded (filled symbols) habitats. Data were fitted by second-order polynomials and are all significant at P < 0·001, except for F_L for the shaded habitat (P < 0·03). Plant relative age is defined in Fig. 1. Rachis is defined here as the entire non-foliated support structure within the leaf, and embraces the petiole (the basal-most non-foliated section) and rachis sensu stricto (e.g. Niklas, 1994).

Biomass partitioning in relation to plant age state

The leaflet mass ratio (F_L, dry mass of leaflets per total plant dry mass) decreased 2- (shaded) to 4-fold (open) from seedling to subsenile age states (Fig. 2C). This change together with age-dependent modification in M_A (Fig. 2A), resulted in 5-fold (shaded plants) to 11-fold (non-shaded plants) decrease in leaf area ratio from seedling to subsenile age class (Fig. 2D).

The root mass ratio (F_R) increased 3-fold in both shaded (r² = 0·78, P < 0·03) and non-shaded plants (r² = 0·87, P < 0·01). Root length increased from approx. 1·5–2 cm in seedlings to 10–12 cm in generative and subsenile plants (r² = 0·71, P < 0·03 for shaded; r² = 0·73, P < 0·02 for non-shaded plants). Thus, the increase in root mass fraction was not only associated with taproot thickening, but partly also with more extensive root systems in older plants.

Changes in leaf size with leaf age

Estimates of leaf size, leaflet area (Fig. 3A), rachis diameter (Fig. 3B), rachis length (r² = 0·74, P < 0·02 for the shaded and r² = 0·58, P < 0·05 for the open habitat), and leaflet number per leaf (Fig. 3C) increased with increasing plant age state. These relationships tended to have a maximum in generative plants (Fig. 3A and C).

Fig. 3.

Age-dependent changes in (A) leaflet area, (B) rachis diameter, (C) and the number of leaflets per leaf in P. saxifraga from open (open symbols) and shaded (filled symbols) habitats. The inset in (C) demonstrates the correlation between the number of leaflets and rachis diameter (all data pooled). The data were fitted by second-order polynomial regressions in main panels, and by linear regression in the inset. All regressions are significant at P < 0·005, except for leaflet area at the open habitat (P < 0·05). Because leaflet area was log-normally distributed, it was log-transformed before calculating the averages. Error bars indicate ± s.e. Plant relative age is defined in Fig. 1.

Average leaflet area increased 5- to 6-fold (Fig. 3A), and the number of leaflets increased 8-fold from seedling to subsenile stage (Fig. 3C), while the corresponding change in the rachis length was only 5- to 6-fold. This suggests that spacing limitations of leaflets increased with increasing plant age. However, leaflet length to width ratio (L:W) gradually increased with increasing plant age (Fig. 4A), demonstrating that the spacing requirement of the leaves of the same area was smaller in older plants. This suggestion was partly corroborated by a strong correlation between the number of leaflets per leaf and L:W ratio (inset in Fig. 4A), while the rachis length was only weekly associated with the number of leaflets per leaf (r² = 0·37, P < 0·05 for all data pooled).

Fig. 4.

Correlation between plant age (Fig. 1) and leaflet length : width ratio in P. saxifraga plants from open (open symbols) and shaded (filled symbols) habitats (A), and the allometric relationships between leaflet area and leaf width separately for Pl, Juv, Im and Vir (filled symbols, n = 201) and G1, G2 and G3 (open symbols, n = 108) phases pooled (B). The inset in (A) demonstrates the correlation between the number of leaflets and leaflet length:width ratio. Data were fitted by second-order polynomial regressions (L:W ratio versus age) or by linear regressions. All regressions are significant at P < 0·05. Error bars indicate ± s.e.

Because the correlation between L : W ratio and age may be biased by leaflet size as the smaller leaflets that are mostly found in younger plants may be more circular, allometric relationships between leaflet area and leaflet width were examined to rule out this possibility. Allometric relationships did demonstrate larger leaflet elongation in generative and subsenile plants relative to younger plant age states (P < 0·005 for the comparison in Fig. 4).

There was a positive correlation between the efficiency of leaf area support (leaf area per unit rachis length) and age state in the shaded plants (r² = 0·86, P < 0·01; r² = 0·51, P = 0·07 for open-grown plants), further corroborating the suggestion that leaf area was more tightly packed in older plants.

Biomass investment in leaf support in dependence on age

The biomass investment in support within the leaf, i.e. the ratio of rachis to total leaf mass, M_R/(M_R + M_L), decreased marginally significantly with increasing plant age in the shaded site (Fig. 5A, P = 0·053) and was independent of plant age in the open site (Fig. 5A). However, the biomass investment in the rachis was large, up to 50% of total leaf mass. Biomechanical principles suggest that the biomass investment in the support within the leaf scales positively with leaf length and total leaf mass, but these correlations were not significant (r² = 0·00, P > 0·8 for rachis length; r² = 0·22, P > 0·1 for total leaf mass). In fact, M_R/(M_R + M_L), scaled negatively with leaflet L:W ratio (r² = 0·58, P < 0·001) and with leaf support efficiency (leaf area per unit rachis length; r² = 0·44, P < 0·02), suggesting that relatively sparser distribution of leaf area along the rachis was the primary driver of larger support investments.

Fig. 5.

Influence of plant age (Fig. 1) on (A) the within-leaf support cost (rachis dry mass, M_R, divided by total leaf dry mass, M_R + M_L), the above-ground leaf support cost (mass of stem and rachises divided by total above-ground mass; inset in A), and (B) on the total leaf support cost (mass of stem and rachises per total plant mass) in P. saxifraga plants from open and shaded habitats. Data were fitted by second-order polynomial regressions. Non-significant regressions (P > 0·05) are shown as dashed lines.

The total above-ground biomass investment in support (mass of stem and rachises) relative to total above-ground biomass, (M_S + M_R)/(M_S + M_R + M_L), was independent of plant age (inset in Fig. 5A). Total biomass investment in leaf support (M_R + M_S) relative to total plant mass (M_T) was largest in seedlings and smallest in subseniles (Fig. 5B), primarily because of greater leaf mass fraction in younger plants and greater root fraction in older plants.

The average (±standard error) carbon percentage of the rachis (40·7 ± 0·4 %) was marginally significantly different (P = 0·073) from that in the leaflets (42·35 ± 0·25 %), while N content of the rachis (1·305 ± 0·015 %) was significantly lower (P < 0·05) than that in the leaflets (1·77 ± 0·11 %). Stem carbon percentage (44·4 ± 0·5 %) was larger and N percentage (1·07 ± 0·02 %) lower than the corresponding values in either rachis and leaflets. Given that N is the primary mineral element limiting photosynthesis, these data suggest that the rachis may also play a role in carbon assimilation, but also that its physiological activity is lower than that of leaves.

Site effects on age-dependent changes in leaf structure and plant biomass partitioning

According to separate slope analyses of covariance [site as the main effect, log(A + 1) as the covariate], the interaction term, site × log(A + 1), was never significant (P > 0·3), indicating that the plant age state did not interact with the light acclimation. Common slope analyses demonstrated several statistically important differences between shaded and open habitats. In particular, leaflet dry mass per unit area (Fig. 2A) was on average 15 % lower in the shaded habitat (P < 0·02), while the leaflet mass ratio (Fig. 2C) was 1·4-fold larger in the shaded plants (P < 0·01). As a combination of these effects, the leaflet area ratio (LAR = F_L/M_A; Fig. 2D) was approx. 1·5-fold larger in the shaded habitat (P < 0·01).

Leaf size characteristics (Figs 3 and 4) were not significantly different between the sites (P > 0·1). Within-leaf (P < 0·03; Fig. 5A) and above-ground biomass investment (P < 0·01; inset in Fig. 5A) in leaf support were larger in the open site, but the overall fraction of plant biomass in leaf support (Fig. 5B) was independent of the site (P > 0·5).

DISCUSSION

Plant age and site effects on light harvesting efficiency

Leaflet dry mass per unit area (M_A) increased (Fig. 2A) and the fraction of total plant dry mass in leaflets (F_L) decreased (Fig. 2C) with increasing plant age state (relative age, Fig. 1), leading to a strong decrease in the leaflet area ratio (Fig. 2D, LAR = F_L/M_A). Given that the light interception per unit dry mass scales with LAR (e.g. Niinemets, 2004), these data suggest a major change in plant capacity for light harvesting during plant ontogeny. In addition, M_A decreased and F_L and LAR increased with increasing light availability (Fig. 2). Such changes are common plastic responses to low irradiance that enhance light harvesting (Meziane and Shipley, 1999), and further demonstrate the enormous potential of phenotypic changes of key functional traits.

Meta-analyses (Poorter and Garnier, 1999; Wright and Westoby, 2001) as well as the studies in the field (Villar et al., 2005) have demonstrated that M_A and LAR are the most important plant characteristics in determining the plant-to-plant differences in relative growth rate. Several studies have provided data of these main characteristics for a large number of species (Poorter and Remkes, 1990; Grime et al., 1997; Huovinen-Hufschmid and Körner, 1998; Craine et al., 2002). Although it is widely believed that the functional traits of plants of different functional type are very different, available data actually demonstrate a continuum of plant characteristics among different functional types (Huovinen-Hufschmid and Körner, 1998). In addition to this continuum among graminoid, forb and legume functional types, the current study further suggests that the species are less distinct entities in the community due to modifications in the key traits during the ontogeny and differentiation due to plasticity to the environment. The ranges of variation of 4-fold for M_A and F_L, and 5- to 11-fold for LAR (Fig. 2) are essentially as large as the ranges of variation in these characteristics observed in multi-species comparisons (Grime et al., 1997; Huovinen-Hufschmid and Körner, 1998; Craine et al., 2002).

Plant biomass investments in support

Important plant biomass characteristics such as biomass investment in support are not routinely measured. As the present study demonstrates, plants may include a major fraction of biomass in support. In perennial P. saxifraga, on average approx. 40–50 % of total leaf mass (Fig. 5A) or 10–30 % of total plant mass (Fig. 5B) is invested in support. These support investments are much higher than the values of 5–25 % observed for within-leaf support investments in tree leaves (Niinemets and Kull, 1999; Niinemets and Fleck, 2002a, b). It may be argued that the green rachises and stems also participate in photosynthesis and thereby also serve as assimilative tissues (Hibberd and Quick, 2001). However, in temperate trees, it has been estimated that the rachis mass-based photosynthetic activity is approximately an order of magnitude less than that of the leaflets (Niinemets, 1999), and similarly low photosynthetic activities have been measured for the green stems of herbaceous plants (den Dubbelden, 1994). In fact, nitrogen concentration was 1·4-fold less in leaf rachises and 1·7-fold less in stems than in leaflets of P. saxifraga, suggesting that leaf rachises and stems primarily function in mechanical support and in transport of assimilates, water and nutrients.

Age-dependent variation in support investments

In addition to changes in leaf morphology and biomass allocation, ontogenetic development also significantly modified leaf size (Fig. 3). In particular, the leaves were longer with more numerous and larger leaflets in older plants. To maintain a specific leaf orientation and light interception efficiency under the self-load of leaflets and rachis, biomechanical models predict that the support investments scale with the cube of leaf length and linearly with the mass of the leaf (Niklas, 1999). However, the fractional investment in support within the leaf was actually not larger in older plants (Fig. 5A), suggesting age-dependent compensation. In fact, leaflets were more elongated (Fig. 4) and more tightly packed along the leaf rachis in older plants, signifying that the mass was distributed closer to the axis of bending in older plants. Thus, due to efficiently shorter lever arms, the hypothetical leaves of common length and mass are less support-costly in old than in young plants (for a detailed discussion on the shape effects on leaf bending, see Niklas, 1999; Niinemets and Fleck, 2002b). The importance of leaflet shape and spacing was further underscored by significant negative correlations between the fraction of leaf mass in support and leaflet length:width ratio and leaf area per unit rachis length. Taken together, these data demonstrate that enhanced support requirements due to increased size may be compensated for by changes in leaf mass distribution along the leaf.

Although the relatively shorter support structures and greater packing of foliar elements decrease the support requirements, allocation of foliage closer to the attachment decreases light harvesting efficiency due to enhanced self-shading (Pearcy and Yang, 1998). Thus, the apparently large support investments in small leaves of young plants may represent a strategy of maximizing light harvesting of plants of a given stature. Thus, greater light harvesting efficiency may involve a significant cost in terms of greater support investments. This is especially evident at a whole plant level, where biomass investment in leaves differed 3- to 4-fold between plants of varying age (Fig. 2C), and this was accompanied by 3-fold variation in whole plant biomass investments in support (Fig. 5B).

CONCLUSIONS

Screening studies have been conducted to determine key plant ecophysiological attributes and species competitive potential in the community (Grime et al., 1997; Huovinen-Hufschmid and Körner, 1998; Wohlfahrt et al., 1999; Craine et al., 2002; Storkey, 2004). The ad hoc hypothesis of these experimental approaches is that plant species are discrete entities with their specific invariable trait combinations. The current study demonstrates that the age-dependent variation in key plant structural and allocation traits—leaf dry mass per unit area (M_A), fraction of plant biomass in leaves (F_L) and leaf area ratio (LAR = F_L/M_A)—within the same species is as large as previously explored variation in these traits among different plant species. In addition, species-specific variation ranges are even larger due to plasticity to environmental factors.

Several studies have demonstrated the caveats of extrapolating from species functional traits determined in short-term experiments to plant performance in the field (e.g. Villar et al., 2005). The current analysis further suggests that for full consideration of species potentials, ontogeny should be not left out. The large age-driven modifications in key plant traits must be included in quantitative models of the functioning of perennial herb communities.

Although the support investments were not larger in older plants, the overall plant biomass investment in support within the leaves was close to 50 %, and the fraction of leaf support biomass was 10–30 % of total plant mass. Fractional biomass investments in support are conventionally not measured in plant growth and biomass allocation studies; but they should be, given their large contribution to total standing biomass and specific function.