Leaf area and growth of juvenile temperate evergreens in low light: species of contrasting shade tolerance change rank during ontogeny


†E-mail: clusk@udec.cl


  • 1Interspecific variation in leaf life span of woody plants should have important consequences for ontogenetic variation in biomass distribution, and hence carbon balance, in low light. This proposal was examined by measuring foliage turnover, growth, biomass allocation and biomass distribution of juveniles (25–1180 mm tall) of four evergreens differing in shade tolerance, growing in low light (2–5% canopy openness) in temperate rainforest understoreys in south-central Chile.
  • 2As predicted, ontogenetic trends in leaf area ratio (LAR) of two very shade-tolerant species with long-lived leaves (Aextoxicon punctatum and Myrceugenia planipes) contrasted strongly with those of mid-tolerant Eucryphia cordifolia and intolerant Aristotelia chilensis. Low-light LAR fell rapidly with increasing plant size in Eucryphia and Aristotelia, but was size-invariant in the two shade-tolerant taxa. As a result, although small seedlings of Eucryphia and Aristotelia displayed more leaf area than shade-tolerant seedlings of similar size (largely because of differences in specific leaf area), this relationship was reversed for plants over ≈500 mm tall.
  • 3Although the slower foliage turnover of the shade-tolerant species at least partly explains this reversal, allocational differences may also be involved. Root mass fraction of both shade-tolerant species was significantly negatively correlated with size, possibly reflecting declining allocation to roots in larger plants. In contrast, root mass fraction was not correlated with size in Eucryphia or Aristotelia.
  • 4Above-ground relative growth rate in low light followed similar ontogenetic trends to LAR, declining more quickly in light-demanding species than in shade-tolerant associates. As a result, large seedlings (>400 mm tall) of shade-tolerant Myrceugenia and Aextoxicon grew more quickly than their light-demanding associates at the same height. The steeply declining LAR and growth of Eucryphia and Aristotelia suggest that these taxa will eventually die of energy starvation in low light.
  • 5Results are consistent with the idea that accumulation of an extensive leaf area eventually gives juvenile shade-tolerant evergreens a net carbon gain advantage over their light-demanding associates in low light. Awareness of long-term ontogenetic trends will enhance understanding of relationships of shade-tolerance variation with morphology and growth of woody evergreens.


Since the 19th century, ecologists and foresters have recognized the importance of shade-tolerance variation in driving forest succession (Warming 1909). However, physiological ecologists’ explorations of the determinants of interspecific differences in low-light survival have to date produced a rather fragmentary understanding of this phenomenon (Naumburg, Ellsworth & Pearcy 2001; Lusk 2002).

The first attempt at a general theory of adaptation to sun and shade was based on the premise that shade tolerance is determined mainly by traits that enhance net energy capture in low light (Givnish 1988). However, many empirical studies have highlighted the low carbon gain potential of first-year seedlings of shade-tolerant trees. When grown in low light, these species often initially develop a lower ratio of leaf area to plant biomass (leaf area ratio, LAR) than their light-demanding associates, mainly as a result of their low specific leaf area (SLA) (Kitajima 1994; Veneklaas & Poorter 1998; Walters & Reich 1999; Lusk & Del Pozo 2002). These findings have led to an emphasis on resource conservation traits as determinants of shade tolerance. It has been suggested that seedling survival in low light depends mainly on the possession of tough, durable tissues resistant to physical stresses and unattractive to natural enemies, and on sacrificing growth in favour of storage (Grime 1979; Coley et al. 1985; Kitajima 1994; Canham et al. 1999; Walters & Reich 1999).

Nevertheless, the main threats to survival in low light may change during ontogeny. Increasing size seems likely to engender greater resilience to accidental damage, drought and attack by herbivores or pathogens (Niinemets 1998), but seedlings must also maintain a positive carbon balance if they are to survive for long in the shade (Givnish 1988; Walters & Reich 1999; Naumburg et al. 2001). Protection of seedling integrity must therefore somehow be harmonized with mechanisms promoting carbon gain, to meet the demands imposed by maintenance respiration, growth and tissue turnover (King 1994; Walters & Reich 1999; Naumburg et al. 2001).

In the understoreys of evergreen forests, differences in leaf life span should have important implications for biomass distribution and carbon gain beyond the first year of a seedling's life. Shade-tolerant evergreens usually have slower foliage turnover than light-demanding associates (Williams, Field & Mooney 1989; King 1994; Walters & Reich 1999; Lusk 2002). As woody tissues accumulate steadily in all taxa, with negligible turnover in juveniles (King 1994), differences in leaf life span should give rise to different ontogenetic trends in LAR: although seedlings of shade-tolerant species start at a disadvantage, accumulation of several leaf cohorts could eventually enable them to develop larger leaf areas than do light-demanding taxa (Lusk 2002). This is most likely to be true in shaded environments where light-demanding species are unable to compensate their rapid foliage turnover with high leaf production rates (King 1994).

Although the likely influence of leaf life span on ontogenetic trends in biomass distribution seems straightforward (Reich 1998), this idea assumes that any ontogenetic variation in biomass allocation is similar across species. First-year seedlings of shade-tolerant taxa generally have low biomass allocation to foliage (Kitajima 1994; Walters & Reich 1999), but little is known about later ontogenetic stages. At least two studies have shown that ontogenetic trends in biomass allocation can differ among coexisting species (Naumburg et al. 2001; King 2003), although it is not known if such variation is systematically related to shade tolerance.

The present paper is intended to show how an ontogenetic perspective can enhance understanding of the linkages among morphology, growth and survival of juvenile trees in the shade. A previous paper (Lusk 2002) compared physiological and morphological traits of large seedlings of eight evergreen tree species across a range of light environments in a Chilean temperate rainforest, and related these traits to differences in low-light survival. Here I complement that earlier work by comparing low-light LAR and growth across a wide range of sizes in four of those species, and relating ontogenetic trends to variation in leaf life span and biomass allocation.

Methods and materials

study site

The study was carried out in the lowland forest of Parque Nacional Puyehue (40°39′ S, 72°11′ W) located in the western foothills of the Andes in south-central Chile. This area experiences a temperate maritime climate, with an average annual precipitation of around 3500 mm (CONAF, unpublished data), a small fraction of this total falling as snow during the winter months. A marked summer rainfall minimum is present, although no month receives an average rainfall of <100 mm (Almeyda & Saez 1958). Soils, derived from recent volcanic ash, are deep and well drained, but acidic.

Most lowland sectors of the national park are covered by old-growth Valdivian rainforest, dominated by broad-leaved evergreen tree species such as Laurelia philippiana, Aextoxicon punctatum, Eucryphia cordifolia and Nothofagus dombeyi (Ramirez 1982). However, this study was carried out in two 40-year-old, even-aged forest fragments located on an alluvial terrace at 350 m a.s.l. Although the overstorey of both fragments was dominated almost exclusively by N. dombeyi, seedlings and saplings of a wide variety of other species were common in the understorey (Lusk 2002).

study species

Four common species of widely differing shade tolerance level were chosen (Table 1). Myrceugenia planipes is a very common small tree or large shrub in the understorey and subcanopy of lowland rainforests in south-central Chile. Its seedlings and saplings grow in deeper shade than those of any of its associates (Figueroa & Lusk 2001; Lusk & Kelly 2003), suggesting that it is probably the most shade-tolerant woody plant of temperate South America. The long-lived overstorey dominant A. punctatum is also very well represented as a juvenile in shaded understoreys (Figueroa & Lusk 2001; Lusk & Kelly 2003), and is probably the most shade-tolerant canopy tree of the region. Aristotelia chilensis is a small, fast-growing, light-demanding species associated mainly with forest margins and large canopy openings. As its seeds also germinate abundantly in the understorey (Figueroa & Lusk 2001), seedlings can also be found in fairly low light (2–5% canopy openness), although these suffer high mortality rates (Lusk 2002). Finally, the long-lived E. cordifolia, although common as a canopy tree or emergent in old-growth forest, regenerates only sporadically in undisturbed stands (Veblen 1985), suffers high seedling mortality in low light (Lusk 2002), and is regarded as a mid-tolerant species (Donoso 1989).

Table 1.  Life-history traits of four temperate evergreen tree species, Parque Nacional Puyehue, Chile
SpeciesSymbolFamilyMaximum height (m)Maximum diameter (cm)Shade toleranceSample size
  1. From Figueroa & Lusk (2001); Lusk & Kelly (2003); Parada et al. (2003).

Aristotelia chilensisAcElaeocarpaceae   10    20Intolerant27
Eucryphia cordifoliaEcCunoniaceae>40>150Mid-tolerant23
Aextoxicon punctatumApAextoxicaceae   35   100Very tolerant29
Myrceugenia planipesMpMyrtaceae   20    40Very tolerant29

As all four species belong to different genera, they are referred to by their generic names from here on.

sampling design

Juveniles <1 m tall were sampled at random points along two parallel transects laid out in the interior of either stand (Table 2). For each species the nearest juvenile to each sample point was located and tagged. Plants with extensive branching or multiple stems were excluded, in order to simplify modelling of above-ground growth and biomass allocation. Individuals with extensive browse or other damage were also excluded.

Table 2.  Sample parameters of seedlings monitored for foliage turnover, allocation and growth, Parque Nacional Puyehue
SpeciesSample sizeMass (g)Height (mm)
RangeGeometric meanRangeGeometric mean
  1. One-way anova did not detect significant interspecific variation in mean (log) seedling mass (F = 2·0, P = 0·13) or (log) height (F = 0·2, P = 0·89).

Aristotelia chilensis270·008–12·40·6525–1180209
Eucryphia cordifolia230·004–17·10·7426–860183
Aextoxicon punctatum29 0·07–42·21·9039–940200
Myrceugenia planipes29  0·1–11·51·4938–665174

A pair of canopy analysers (LAI-2000, Li-Cor, Lincoln, NE, USA) was used to quantify canopy openness within a quasi-hemispherical field of view (148°) immediately above the apex of each tagged seedling. Percentage canopy openness was calculated by referring measurements made within the forest to simultaneous readings taken in a clearing of ≈1 ha. In order to minimize the influence of variation in irradiance on the study parameters, plants growing outside the range of 2–5% canopy openness were excluded from the study.

biomass and growth measurements

Above-ground growth, biomass allocation, and foliage turnover were recorded over an 11 month period. At the end of April 2002, initial stem length, basal diameter and apical diameter of each seedling were measured, and the number of live leaves present on the central axis was recorded. At the end of March 2003, all surviving seedlings were remeasured. In order to include data on seedlings <1 year old, measurements were also made at this date on several new seedlings of each species that had germinated during the study period. After the loss of about 30% of marked plants during the study period, data were eventually obtained from a total of 108 juveniles (Table 2).

Leaf mortality during the 11 month period was used to estimate annual foliage turnover rates. Many of the leaf cohorts under study probably had not reached an age where significant mortality would be expected, especially in Aextoxicon and Myrceugenia. As the data therefore probably do not permit reliable estimations of leaf life span for all plants, it was preferable to report the results simply as foliage turnover. Because leaf longevities were <1 year on most individuals of Aristotelia, abscission scars were counted to determine the mortality of new leaves appearing after the start of the study period. Foliage turnover (percentage year−1) was estimated as:

(no. of leaves shed year−1 /no. of leaves present at the outset) × 100

Immediately after remeasurement, all seedlings were harvested, roots included, to calculate biomass allocation and distribution parameters. Plants were excavated carefully to minimize losses of fine roots. Plant material was separated into four fractions: new leaves, old leaves surviving from previous cohorts, stem and roots. A Li-Cor 3100 leaf-area meter was used to measure the leaf area of each plant. Plant material was then dried for 72 h at 70 °C and weighed. Specific leaf area (SLA) was determined by dividing the area of new leaves by their dry mass. Leaf mass fraction (LMF) was determined by dividing foliage dry mass by total plant dry mass. Leaf area ratio (LAR) was determined by dividing total foliage area by total plant dry mass.

Changes in stem dimensions and foliage biomass during the 11 month period were used to estimate above-ground relative growth rates (RGR) and allocation of above-ground production between leaves and stem tissue. The central axis of each plant was modelled as a truncated cone, in order to estimate increase in stem volume. As wood density showed no significant overall relationship with stem size for any species, initial dry mass of each stem was estimated by multiplying its initial volume by its final density. Initial foliage mass could not be estimated accurately simply from the net change in number of leaves (cf. Lusk 2002), as the average mass of leaves (foliage mass/number of leaves) was strongly positively correlated with stem height, and therefore certain to change for most individuals during the study period. Instead, these species-specific relationships of leaf mass to plant height (Table 3) were used to back-estimate the initial average mass per leaf at the outset of the study, this figure then being multiplied by the number of leaves present initially.

Table 3.  Allometric equations for average mass per leaf (y, mg) as a function of plant height (x, mm), in juveniles of four temperate rainforest tree species
Aristotelia chilensisy = 0·0062x1·500·97
Eucryphia cordifoliay = 0·007x1·550·88
Aextoxicon punctatumy = 0·0914x1·290·90
Myrceugenia planipesy = 0·0226x1·460·88


As age could not be determined accurately for all plants, height and mass were used as scalars of ontogeny. Resting bud scars and growth rings permitted age estimates for 18 Aristotelia and 14 Eucryphia, and age was strongly correlated with height in both these subsamples (P < 0·0001 and P = 0·001, respectively). Although age is strictly more relevant than size to the central prediction of this study, these results suggest that height is a reasonable proxy of age within the narrow range of light environments in which the plants grew (2–5% canopy openness). There was less reason a priori to expect the correlation to hold for global comparisons including all four species. Nevertheless, ancova eventually showed only weak evidence for species differences in mean growth rate, although ontogenetic trends did differ appreciably (Table 4).

Table 4. ancova examining effects of plant height and species on leaf longevity, biomass distribution, biomass allocation and growth, for juveniles of four temperate evergreens growing in low light
Response variableEffectdfF ratioP
Leaf longevityHeight1 41·3<0·0001
Species3 60·2<0·0001
Height × species3  5·0   0·0035
Allocation to foliageHeight1 47·9<0·0001
Species3  3·2   0·027
Height × species3  2·0   0·12
Specific leaf areaHeight1133·1<0·0001
Height × species3  2·9   0·04
Leaf mass fractionHeight1  1·1   0·31
Species3  6·8   0·0003
Height × species3 28·5<0·0001
Root mass fractionHeight1 60·5<0·0001
Species3 10·2<0·0001
Height × species3 17·3<0·0001
Leaf area ratioHeight1 59·5<0·0001
Species3 19·0<0·0001
Height × species3 20·6<0·0001
Relative growth rateHeight1 26·6<0·0001
Species3  2·5   0·06
Height × species3  3·1   0·03

Light environment (canopy openness) was not correlated with any dependent variable of interest, and so was not used as a predictor variable in any analysis. This probably reflects the narrow range of light environments included in the study (cf. Lusk 2002). Neither was light environment correlated with plant height. When ancova was used to explore the effects of species (factor) and seedling height (covariate) on mean light environment, there was evidence of statistically significant variation among species (P = 0·008, F = 4·1). However, these differences were small and therefore probably of minimal biological significance, with mean canopy openness ranging from 3·6% in Myrceugenia and Aextoxicon to 4·2% in Aristotelia. Light environment did not vary systematically in relation to seedling height (ancova: P = 0·26, F = 1·3).

ancova was used to examine the effects of plant size on leaf life span, growth, biomass allocation and biomass distribution parameters, and to test for interspecific variation in elevation and slope of these relationships. Rate parameters (foliage turnover, allocation and RGR) were analysed as functions of initial plant height. On the other hand, biomass distribution parameters calculated from one-off measurements made after harvest (SLA, LAR, etc.) were referred to final plant height or mass. As the plants were obtained from two stands, ‘stand’ was initially included as a factor, but was later removed when found to be non-significant for all analyses. For most analyses, data were log-transformed in order to normalize distributions, reduce heterogeneity of variance and/or improve model fits. All analyses were carried out using JMP Statistical Software (SAS Institute, Cary, NC, USA). Individual data points are shown only where minimal scatter and/or low overlap among species enables trends to be readily compared by eye (Figs 1–3, 6). In the other figures, trend lines only are shown.

Figure 1.

Relationships of foliage turnover rates with initial height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ec) y = 0·309x0·437, R2 = 0·45; (Ap) y = 0·117x0·714, R2 = 0·79; (Mp) y = 0·309x0·577, R2 = 0·32. No significant relationship exists for Ac.

Figure 2.

Allocation of above-ground production to foliage in relation to initial height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = −0·0003x + 0·71, R2 = 0·29; (Ec) y =−0·0003x + 0·61, R2 = 0·26; (Ap) y = −0·0006x + 0·70, R2 = 0·71; (Mp) y = −0·0004x + 0·68, R2 = 0·26.

Figure 3.

Relationship of specific leaf area (SLA) with height in juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = 536·7x−0·104, R2 = 0·40; (Ap) y = 313·3x−0·160, R2 = 0·66; (Ec) y = 537·9x−0·205, R2 = 0·63; (Mp) y = 347·9x−0·192, R2 = 0·56.

Figure 6.

Relationship of leaf area ratio with height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = 926·8x−0·457, R2 = 0·67; (Ec) y = 490·2x−0·365, R2 = 0·69. No significant relationship exists for Ap or Mp.


foliage turnover

Foliage turnover rates varied with both species and plant size (Table 4; Fig. 1). Leaf loss rates of shade-tolerant Myrceugenia and Aextoxicon were somewhat lower than those of mid-tolerant Eucryphia, and far lower than those of intolerant Aristotelia, which averaged ≈130% year−1. Whereas Aristotelia's foliage turnover did not vary systematically with plant size, leaf loss rates of the other three species decreased with increasing height (Fig. 1).

biomass allocation

The proportion of above-ground production allocated to foliage during the study period tended to decrease with initial plant height in all species (Fig. 2). There was significant interspecific variation in mean allocation to foliage (Table 4), but this did not appear to be systematically related to shade tolerance. Although there was no significant interspecific variation in the slope of size-allocation relationships (Table 4), regression coefficients suggested steeper size-related declines in the two shade-tolerant species than in Aristotelia and Eucryphia (Fig. 2).

biomass distribution

Specific leaf area was consistently highest in Aristotelia, and lowest in the two shade-tolerant species (Fig. 3); it declined with increasing plant size in all species (Fig. 3), and there was marginally significant evidence of interspecific variation in the slope of this relationship (Table 4), Aristotelia having a flatter slope than the other three species.

Although there was no overall relationship between LMF and plant final height (Table 4), all species showed individually significant ontogenetic trends. Whereas the LMF of Aristotelia and Eucryphia was negatively correlated with plant height, the two shade-tolerant species showed the opposite trend (Fig. 4). This led to a rank reversal: although young seedlings of Aristotelia and Eucryphia initially had a much higher LMF than their shade-tolerant associates, this relationship was reversed at heights over ≈250 mm (Fig. 4).

Figure 4.

Relationship of leaf biomass fraction with height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = 0·9072x−0·197, R2 = 0·29; (Ap) y = 0·1362x0·194, R2 = 0·57; (Ec) y = 0·9114x−0·160, R2 = 0·41; (Mp) y = 0·1979x0·152, R2 = 0·46.

There were also divergent ontogenetic trends in RMF, as indicated by the highly significant interaction term in the ancova (Table 4). Whereas the RMF of Myrceugenia and Aextoxicon declined steeply with increasing plant size, this parameter was not significantly correlated with size in Aristotelia or Eucryphia (Fig. 5). Thus the shade-tolerant species initially had proportionally more of their biomass in roots than the light-demanding species, but this relationship was reversed with increasing size.

Figure 5.

Relationship of root fraction with height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ap) y = −0·103 Ln(x) + 0·890, R2 = 0·74; (Mp) y = −0·105 Ln(x) + 0·847, R2 = 0·69. Relationships are not significant for Ac or Ec (broken lines).

Ontogenetic trends in LAR differed between species of contrasting shade tolerance (Table 4). Whereas LAR fell steadily with increasing height in Aristotelia and Eucryphia, it was not significantly correlated with plant size in either of the shade-tolerant taxa (Fig. 6). Thus, although small seedlings (<100 mm tall) of Eucryphia and Aristotelia displayed more leaf area than shade-tolerant seedlings of similar size, Myrceugenia and Aextoxicon had slightly higher LAR towards the upper limit of the size range examined here (>500 mm tall).

A similar reversal was evident when absolute leaf area was plotted as a function of plant mass (Fig. 7). Leaf area predicted from these highly significant relationships shows that the rank order of the four species at 0·1 g is almost completely reversed by the time juveniles attain a mass of 10 g (Table 5).

Figure 7.

Relationship of leaf area with mass of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = 72·52x0·77, R2 = 0·95; (Ap) y = 50·42x1·014, R2 = 0·99; (Eu) y = 67·29x0·83, R2 = 0·98; (Mp) y = 57·17x0·967, R2 = 0·98.

Table 5.  Predicted leaf area (cm2) of seedlings of four temperate evergreens at two different sizes. Note changes in rank order
SpeciesLeaf area at (mass)
0·1 g10·0 g
Aristotelia chilensis12·4430
Eucryphia cordifolia10·0455
Myrceugenia planipes 6·0530
Aextoxicon punctatum 4·9521


Above-ground RGR followed similar ontogenetic trends to those of LAR, declining more quickly in light-demanding species than in shade-tolerant associates (Fig. 8). Thus, although small seedlings of Eucryphia and Aristotelia grew faster than those of Myrceugenia and Aextoxicon (Fig. 8), this advantage was reversed in the largest size classes (>400 mm). As a result there was evidence of significant slope differences among species, despite only marginal evidence for differences in mean growth rate (Table 4).

Figure 8.

Relationship of above-ground relative growth rate (RGR) with initial height of juveniles of four temperate evergreens growing in low light (2–5% canopy openness). Regression equations: (Ac) y = 17·434x−0·614; R2 = 0·44; (Ec) y = 10·153x−0·545, R2 = 0·42 (Mp) y = 3·1401x−0·293, R2 = 0·33. Relationship is not significant for Ap (broken line).


As predicted on the basis of leaf life span, shade tolerance differences were associated with different trends in low-light LAR (Fig. 6). In agreement with many experiments with first-year seedlings in artificial environments (Kitajima 1994; Veneklaas & Poorter 1998; Walters & Reich 1999; Lusk & Del Pozo 2002), small seedlings of light-demanding Aristotelia and Eucryphia developed higher leaf areas than their shade-tolerant associates, mainly as a result of differences in SLA (Fig. 3). However, while LAR declined steadily with increasing size in Aristotelia and Eucryphia, it was size-invariant in juvenile Aextoxicon and Myrceugenia, giving rise to rank reversals at heights over ≈500 mm (Fig. 6).

Does the extensive leaf area of large Aextoxicon and Myrceugenia necessarily equate to an advantage in carbon gain over similarly sized Aristotelia and Eucryphia in low light? The oldest leaves on large Aextoxicon and Myrceugenia seedlings might contribute little to whole-plant carbon gain (cf. Hom & Oechel 1983), due to self-shading, physiological deterioration and/or accumulation of epiphylls. However, leaves will be retained only as long as their carbon balance remains positive (Ackerly 1999), and the lower light compensation points of leaves of the shade-tolerant species (Lusk 2002) at least partly explain their longer retention times. Furthermore, a recent study of saplings of 11 sympatric species of Psychotria in a tropical forest found that the more shade-tolerant species of the genus had slightly more efficient leaf area display on average than their light-demanding relatives (Pearcy et al. 2004). The sum of evidence therefore suggests that large juveniles of the shade-tolerant species will hold a net carbon gain advantage over similarly sized light-demanding associates in low light. As we shall see below, this impression is reinforced by the growth rate data (Fig. 8).

Leaf loss rates were lower in the shade-tolerant species (Fig. 1), in agreement with several previous studies in evergreen forests (Williams et al. 1989; King 1994; Lusk 2002). Large differences between turnover rates of foliage and woody tissues therefore at least partly explain the ontogenetic decline in LAR in the two light-demanding species (Fig. 6), and the resulting rank-order reversals in leaf area (Table 5). However, this explanation is of most relevance to Aristotelia, which had much shorter-lived leaves than any of the other taxa. Aristotelia also differed from the other three species in showing no relationship of turnover rates with plant size (Fig. 1).

Biomass allocation also influenced ontogenetic trends in LAR. However, variation in root : shoot allocation appeared to be more important than partitioning of above-ground production between foliage and stemwood. The data set permits calculations of root : shoot allocation only for seedlings <1 year old, which had undergone little no or foliage turnover. However, some broader inferences can be made tentatively on the basis of biomass distribution. The low LAR developed initially by Aextoxicon and Myrceugenia in low light reflects high allocation to roots in the first year of life (Fig. 5), plus a relatively inefficient development of photosynthetic surface area owing to the production of thick and/or dense leaves (low SLA; Fig. 3). Although the root fraction of small Aextoxicon and Myrceugenia was smaller than that of their light-demanding associates at the same size, the ontogenetic decline in this parameter was so steep in these species that 1 m tall juveniles growing in low light can be expected to have only 15–20% of their biomass in roots (Fig. 5). Thus, although initially relatively large, root mass of the shade-tolerant species increases more slowly during seedling development than that of the light-demanders, despite their broadly similar above-ground RGR (Fig. 8). This could reflect either faster root turnover in the shade-tolerant species, or else reduced below-ground allocation after initial establishment. The former seems unlikely, given the long leaf life spans of Aextoxicon and Myrceugenia (Fig. 1), and evidence from other studies that root and leaf traits tend to be positively correlated, albeit rather weakly in some cases (Comas et al. 2002; Craine et al. 2002). Low below-ground allocation, coupled to progressive accumulation of foliage mass due to long leaf life spans, is therefore the most likely explanation of decreasing root fraction in the two shade-tolerant species.

As structural and storage components of roots were not separated in the present study (cf. Canham et al. 1999), it is difficult to interpret the allocation patterns of small seedlings of the shade-tolerant species (Fig. 5). The low initial LAR and high root allocation of young Aextoxicon and Myrceugenia might reflect carbohydrate storage in roots, and minimization of exposure of resources to above-ground herbivores (Kitajima 1994; Walters & Reich 1999). Alternatively, it could indicate that dessication is an important selective pressure on very small seedlings in the understorey. Even in humid forests, penetrating the unconsolidated litter layer and reaching the more reliable moisture source afforded by the compacted underlying horizons, may be vital for surviving the first year (cf. Ng 1978). Figueroa & Castro (2000), working at a somewhat drier temperate forest site in southern Chile, suggested that dessication was a major cause of mortality of young seedlings, after finding that most mortality occurred during summer when rainfall is lowest. Although the small leaf area of shade-tolerant seedlings during their first year of life will greatly restrict carbon gain, it also implies low transpiration rates. Regardless of this uncertainty about structural vs storage components of the roots of germinants, it is clear that large juveniles of the two species which survive well in low light have only a small fraction of their biomass directly involved in water and nutrient acquisition (Fig. 5). This is consistent with evidence that light is the dominant resource limiting juvenile tree growth in the understoreys of other mesic forests (Canham et al. 1996; Finzi & Canham 2000).

Ontogenetic variation in RGR (Fig. 8) more-or-less paralleled that of LAR, underlining the fundamental importance of LAR for net carbon gain in low light (Poorter 1999). The initial growth-rate advantage of the light-demanding taxa, similar to that often seen in glasshouse studies of germinants (Veneklaas & Poorter 1998; Walters & Reich 1999), was reversed in the larger size classes. Although interspecific variation in low-light growth per se probably has little influence on forest dynamics, the relationships of growth with carbon balance and survival can be most informative (Kobe et al. 1995; Walters & Reich 1996; Canham et al. 1999). Although carbon gain was not quantified, the growth rate data suggest that the eventual accumulation of an extensive leaf area gave large seedlings of Aextoxicon and Myrceugenia a healthier carbon balance than similarly sized Aristotelia and Eucryphia. Shade-tolerant species suffer less mortality at a given growth rate than do their light-demanding associates (Kobe et al. 1995), probably reflecting greater storage allocation in the former (Kobe 1997). The slight growth rate advantage of large juveniles of the shade-tolerant species in low light (Fig. 8) therefore indicates major differences in understorey survival probability, as has been shown previously for these taxa (Lusk 2002).

The results are consistent with the idea that increasing plant size is associated with changes in the nature of the principal threats to survival in the shade (Niinemets 1998). Pathogens and herbivores have been attributed a leading role in the high mortality suffered by germinants of pioneer trees (Augspurger 1984; Kitajima 1994). However, the data presented here suggest that, even in the absence of natural enemies, Aristotelia and Eucryphia growing in low light would eventually die of energy starvation. In low light, where their relatively fast foliage turnover cannot be compensated by high leaf production rates, increasing age and size of these species must inevitably be associated with an increasing ratio of heterotrophic to autotrophic tissues (Fig. 4), implying a deteriorating carbon balance (Fig. 8).

This study shows that choice of size or age class will shape scientists’ answers to questions about adaptation to shade in evergreen forests. The relationships of shade-tolerance variation with morphology and growth are therefore likely to be understood better by a consideration of ontogenetic patterns, rather than focusing on a single narrow age or size class. Because of the well established relationship of leaf life span with shade-tolerance variation (Williams et al. 1989; King 1994; Walters & Reich 1999), similar ontogenetic trends in LAR and associated traits can be expected in other humid evergreen forests. Likewise, there are grounds for predicting similar biomass-allocation patterns in other humid forests where, after the vulnerable first year of life, survival in the understorey should be favoured in plants expressing traits maximizing capture of light, rather than water or nutrients (Tilman 1985; Smith & Huston 1989).


Thanks to FONDECYT for support through grant no. 1980084, to the Millennium Center for Advanced Studies in Ecology and Research on Biodiversity, and to Teresa Parada for help with field work. Comments by David King, Javier Figueroa, Jim Dalling and Rebecca Montgomery helped improve the manuscript.