Foliage area and crown nitrogen turnover in temperate rain forest juvenile trees of differing shade tolerance

Authors


*Present address and correspondence: Christopher H. Lusk, Departamento de Botánica, Universidad de Concepción, Casilla 160-C, Concepción, Chile (fax + 56 41 246005; e-mail clusk@udec.cl).

Summary

1 Shade-tolerant angiosperm trees are reported to require relatively fertile soils in temperate forests. We explored the possibility that high foliage allocation in shade-tolerant species could result in higher whole-plant nitrogen demands than in light-demanding trees of comparable diameters. We measured foliage mass and area, leaf life span, and nitrogen (N) content of fresh foliage and leaf litter for juveniles of 11 evergreen angiosperms in a Chilean temperate rain forest. This permitted estimation of annual nitrogen losses in leaf litter fall by individuals of a given diameter.

2 Leaf life spans were longest in shade-tolerant species. The highest leaf N levels were found in small short-lived early successional trees, whereas among longer-lived species there was no general relationship of leaf N with shade tolerance level. Shade-tolerant species had lower N resorption efficiencies, and therefore higher leaf litter N concentrations on an area basis, than light-demanding associates.

3 Foliage mass, foliage area and total crown N pool were strongly positively related to shade tolerance level.

4 Interspecific variation in annual N losses in leaf litter fall was more closely related to foliage area (R2 = 0.52) than to N concentration of leaf litter (R2 = 0.31) or leaf lifetimes (R2 = 0.01). Although the short-lived early successional species Embothrium coccineum had the highest annual crown N losses of the 11 species, shade-tolerant species had higher annual losses than light-demanding overstorey dominants of comparable longevities.

5 The results are consistent with the proposal that the costs of obtaining enough N for crown maintenance and expansion may constrain the fitness of shade-tolerant angiosperm trees in late successional stages on infertile sites, when soil nutrient availability is reduced by increased uptake and sequestration in biomass and litter. On the other hand, high N loss rates may be more sustainable for short-lived early colonists that complete their life cycles in the initial stages of secondary succession, when nutrient availabilities are often relatively high.

Introduction

Shade-tolerant late-successional tree species typically have low maximal growth rates (Bazzaz 1979; Pacala et al. 1994) and might therefore be expected to have low nutrient requirements (Grime 1979; Reich et al. 1995). However, a number of studies have shown that highly shade-tolerant angiosperm trees are actually scarce on infertile soils, a pattern reported both from northern temperate forests (Spurr & Barnes 1980; Keddy & MacLennan 1990; Franklin et al. 1993) and also from temperate rain forests in the coast ranges of southern Chile (Lusk 1996a,b). As shade-tolerant trees are also often drought-sensitive (Smith & Huston 1989), and as availabilities of nutrients and water are highly correlated in many landscapes, it is difficult to rule out the possibility that this pattern is primarily a response to site water balance. However, rainfall is high in the Chilean coast range forests (Almeyda & Saez 1958) and most of the trees that are common on infertile sites in this region are drought-sensitive species confined to humid maritime climates (Weinberger 1973; Weinberger et al. 1973), making water limitation unlikely in this case.

Studies of leaf physiology provide few clues as to why shade-tolerant angiosperm trees might be averse to low soil fertility. Instantaneous photosynthetic nutrient-use efficiency at leaf level is generally low in shade-tolerant species (Seemann et al. 1987; Pons et al. 1994; Reich et al. 1995a) yet their leaf nitrogen and phosphorous levels are usually similar to, or lower than, those of shade-intolerant trees (Popma et al. 1992; Reich et al. 1995a).

A consideration of nutrient demand at crown level, rather than leaf level, may help understand this pattern. Comparative studies of plant nutrient loss rates have usually focused on the main components of mean residence time of nutrients in the foliage biomass, i.e. leaf life span and nutrient resorption efficiency (Berendse & Aerts 1987; Escudero et al. 1992; Aerts 1996). Few if any studies have addressed the possible nutritional implications of species differences in biomass allocation to foliage. Shade-tolerant trees tend to have deeper crowns and greater foliage areas than light-demanders of comparable diameter (Ellenberg 1978; Chapman & Gower 1991; Canham et al. 1994). As synthesis of photosynthetic enzymes and pigments requires large nutrient inputs, leaves have higher concentrations of most mineral nutrients (especially nitrogen) than other vegetative organs. Furthermore, although foliage comprises only a small fraction of total standing biomass in trees, it is subject to more rapid turnover than woody tissues, and therefore accounts for an important fraction of annual biomass allocation (King 1991). As foliage turnover is often the principal mechanism of nutrient loss for woody perennials (Miller et al. 1976; Berendse et al. 1987), development and maintenance of a high leaf area by shade-tolerant trees may imply significantly higher whole-plant nitrogen demands than in less-tolerant species. However, a number of studies in evergreen forests have reported long leaf life spans in shade-tolerant trees (Williams et al. 1989; King 1994; Reich et al. 1995a), so although their crown nitrogen pools are likely to be larger than those of light-demanding associates, they may turn over more slowly. Nevertheless, few angiosperms appear to be capable of producing leaves that live more than 4–5 years. (Reich et al. 1995b). This constraint on angiosperm leaf life spans, coupled to high leaf areas in shade-tolerant trees, may therefore result in inevitably high nutrient requirements in these taxa.

We measured foliage mass and area, leaf longevity, and nitrogen concentrations of foliage and leaf litter for 11 evergreen tree species of varying shade tolerance in the Chilean coast range forests, and examined how these traits interact to determine annual nitrogen losses in leaf litter fall. We hypothesized that maintenance of high leaf areas in the most shade-tolerant angiosperm species would be associated with greater nitrogen losses in leaf litter fall than for shade-intolerant species of comparable diameters.

Materials and methods

Study site and species

Fieldwork was carried out mainly on the summit plateau (800–940 m a.s.l.) of the coast range near Valdivia in south-central Chile (40°12′S, 73°26′ W). A temperate maritime climate prevails on the western slopes and summits, with annual rainfall of approximately 4000 mm (Almeyda & Saez 1958). However, soil nitrogen status, drainage and depth vary widely over short distances, with strong effects on stand composition (Lusk 1996b).

Our study species (Table 1) were chosen to include a wide range of shade tolerance, as determined by Donoso (1981) mainly on the basis of regenerative behaviour. The shade-intolerant species included both short-lived early successional taxa (Embothrium coccineum and Ovidia pillo-pillo) and long-lived overstorey dominants (Weinmannia trichosperma and Nothofagus betuloides). These two latter species, although establishing on the summit plateau mainly as post-fire even-aged stands (Lusk 1996b), persist into late successional stages by virtue of their longevity (Lusk 1999). Two mid-tolerant species (Nothofagus nitida and Drimys winteri) also often establish as even-aged stands after fire but in addition regenerate in gaps in older stands. The two most shade-tolerant trees of the summit plateau forests, Laureliopsis philippiana (syn. Laurelia philippiana) and Amomyrtus luma, generally occur as all-sized populations in these forests, even in closed-canopy stands. As the summit plateau forests have low alpha-diversity of tree species, we added data from two shade-tolerant species (Myrceugenia planipes and Aextoxicon punctatum) and one mid-tolerant species (Eucryphia cordifoliaceae) found at lower altitudes in the same range (Table 1). Nomenclature follows Marticorena & Quezada (1985).

Table 1.  Ecological characteristics of study species, and relationships of (log) foliage mass and area (y) with (log) stem basal diameter (x). Species arranged from top to bottom in approximate order of increasing shade tolerance
SpeciesAbbreviationFamilyShade tolerance
level
Life spanProvenanceRegression equations
Foliage mass
Foliage area
Embothrium coccineumEmProteaceaeIntolerantShortPlateauy = 2.32x + 0.81y = 2.41x – 1.25
Weinmannia trichospermaWtCunoniaceaeIntolerantLongPlateauy = 1.76x + 1.11y = 2.37x – 0.94
Nothofagus betuloidesNbFagaceaeIntolerantLongPlateauy = 1.90x + 0.99y = 1.92x – 1.25
Ovidia pillo-pilloOpThymelaeaceaeIntolerantShortPlateauy = 2.03x + 0.77y = 1.64x – 1.02
Eucryphia cordifoliaEuEucryphiaceaeMid-tolerantLongMid-altitudey = 2.27x + 1.04y = 2.17x – 0.99
Nothofagus nitidaNnFagaceaeMid-tolerantLongPlateauy = 1.88x + 1.24y = 2.12x – 1.06
Drimys winteriDwWinteraceaeMid-tolerantModeratePlateauy = 1.85x + 1.35y = 1.54x – 0.70
Laureliopsis philippianaLpMonimiaceaeTolerantLongPlateauy = 2.00x + 1.27y = 2.00x – 0.84
Aextoxicon punctatumApAextoxiaceaeTolerantLongMid-altitudey = 2.18x + 1.25y = 2.03x – 0.71
Amomyrtus lumaAlMyrtaceaeTolerantLongPlateauy = 2.16x + 1.12y = 2.12x – 1.01
Myrceugenia planipesMpMyrtaceaeTolerantModerateMid-altitudey = 2.44x + 1.09y = 2.37x – 0.94

Sampling rationale

The amount of nitrogen lost annually in leaf litter fall (Nloss) from individuals of a given diameter was estimated as:

image

where FA = foliage area (m2), T = foliage turnover per year (inverse of leaf life span) and [N]litter = N content of leaf litter (g m–2). We measured these parameters for juvenile trees (2–7 cm basal diameter), as we expected that differences in crown nitrogen turnover would have to be manifested at a relatively early developmental stage if they were to cause the observed field distribution patterns.

Foliage area and mass measurements

We subjectively chose 12–17 saplings (2.0–5.5 m tall) of each species for sampling of leaf and crown parameters. We constrained the influence of site and stand conditions on leaf and crown traits by selecting only saplings growing beneath small canopy openings (5–10 m diameter) and by avoiding low fertility sites.

The basal diameter of each sapling was measured above the root flanges. Total height, diameter at breast height and crown depth (to the lowest live branch) were also measured. Foliage area and mass were measured for all species except Drimys winteri by removing all foliage-bearing branchlets and separating a subsample (20–30%) into leaves and branchlets. After drying for 48–72 h at 65 °C, the coefficient of partition between leaves and branchlets in this subsample was used to estimate the dry weight of leaves in the whole crown. A Li-COR 3000 leaf area meter (Li-Cor Inc., Lincoln, Nebraska) was used to measure the area of at least 30 fresh leaves from each sapling, which were then dried and weighed to permit determination of leaf mass per unit area (LMA) and estimation of whole-plant leaf area. Leaves of Drimys winteri were large enough and sufficiently few in number to permit a less destructive sampling procedure. For this species, all leaves were counted on each individual, the crown divided into upper and lower strata, and a random sample of 15 leaves removed from each stratum. The average area of these 30 leaves, and their dry weight to area ratio, were used to estimate total foliage area and mass of the crown.

Nitrogen concentrations of foliage and leaf litter

Total leaf nitrogen (N) was determined for four saplings of each species by the Kjeldahl method. Although leaf life span varied widely among the 11 species, we used leaves of a similar physiological age range for all species: in sampling a branchlet, we excluded the outermost (newest) c. 20% of the leaves and the innermost (oldest) c. 40%. Total N concentration was also determined for recently fallen leaf litter collected beneath the same four individuals of each species. Resorption efficiency was estimated as the percentage difference between N per unit area of foliage and N per unit area of leaf litter. As leaves can lose as much as 30% of their total dry weight during senescence, resorption efficiency estimates based on the quantity of nutrient per unit leaf area are more accurate than those determined on a mass basis (Chapin 1989).

Leaf life span

Leaf life span was estimated by monitoring leaf survivorship on five to eight saplings per species. Damage by herbivory, human interference and a tree fall during the monitored period subsequently reduced this sample size to four to seven saplings per species. All leaves on one mid-crown branch of each individual (40–150 leaves) were marked with nail polish, and survivorship recorded on two subsequent visits, 6 and 12 months later. Mean leaf life span (years) was estimated as the inverse of mortality over 12 months.

Statistical procedures

Species-specific relationships of foliage mass and area with basal diameter were examined by regression (JMP Statistical Software, SAS Institute, Carey, MA). Although untransformed values of both foliage mass and area were strongly correlated with stem basal area for all species, there were significant problems with heterogeneity of variances. These problems were eliminated by relating foliage parameters to basal diameter (instead of area), and by log-transforming (base 10) both foliage and diameter data. Two-way nested anova (JMP Statistical Software, SAS Institute) was used to examine the overall relationships of foliage mass and area with stem diameter, shade tolerance level, and species within shade tolerance levels, again using log-transformed data.

Calculation of the standard error of the product of random variables can be problematic when more than two variables are involved, as in the case of annual crown nitrogen loss, which was estimated as the product of three variables measured with error (foliage area, litter nitrogen content and foliage turnover rate). Therefore, a bootstrap procedure (Efron & Tibshirani 1993) was adopted to calculate the standard errors of annual crown nitrogen losses and crown nitrogen pool. The algorithm was as follows: (i) in each iteration, bootstrap samples were obtained from the original samples of each variable, the mean of each variable was computed, and then the derived parameter was estimated; (ii) this procedure was repeated 200 times, as recommended by Efron & Tibshirani (1993); (iii) the standard error was estimated as the bootstrap standard deviation of the 200 repetitions of the derived parameter.

Results

Foliage mass and area

Both foliage mass and foliage area showed highly significant overall relationships with stem basal diameter (Table 2; Figs 1 and 2). Shade tolerance classes showed evidence of significant intercept differences in the relationships of foliage parameters with basal diameter, but no evidence of systematic slope differences (Table 2). Species within shade tolerance classes did not show significant differences in either intercepts or slopes of these relationships (Table 2), suggesting that shade tolerance differences were much more important than phylogenetic relationships as determinants of foliage mass and area. Relationships of foliage mass and area with basal diameter were highly significant (P < 0.0001) for all species (Figs 1 and 2).

Table 2.  Summary of anova to test for significant overall effects of stem basal diameter on foliage mass and area, as well as differences between shade tolerance levels (intolerant, mid-tolerant and tolerant) and between species within shade tolerance levels
Dependent variableFoliage massFoliage area 
Whole model
d.f.2121
P< 0.0001< 0.0001
R20.900.88
Source of variation
Diameter
d.f.11
P< 0.0001< 0.0001
F-ratio787.42639.20
Shadetol
d.f.22
P0.01360.0189
F-ratio4.444.08
Diameter × Shadetol
d.f.22
P0.54730.6004
F-ratio0.610.51
Species (Shadetol)
d.f.88
P0.50920.6372
F-ratio0.910.76
Diameter × Species (Shadetol)
d.f.88
P0.56870.3362
F-ratio0.841.15
Figure 1.

Relationships of foliage mass with stem basal diameter, for juveniles of 11 temperate rain forest trees. Species are grouped to maximize visibility. Triangles represent shade-tolerant species, circles represent mid-tolerant species, and squares represent shade-intolerant taxa. P < 0.0001 for all relationships, and R2 ranges from 0.79 to 0.91. Species abbreviations as in Table 1.

Figure 2.

Relationships of foliage area with stem basal diameter, for juveniles of 11 temperate rain forest trees. Species are grouped to maximize visibility (. P < 0.0001 for all relationships, and R2 ranges from 0.69 to 0.90. Species abbreviations as in Table 1, and symbols as in Fig. 1.

Foliage mass at a given diameter showed a strong positive relationship with reported shade tolerance, with the shade-tolerant species carrying 60–300% more foliage mass than their intolerant associates, and mid-tolerant species occupying an intermediate position (Fig. 3). Similar patterns were also apparent in the foliage area data: leaf areas of shade-tolerant species were between 15% and 320% higher than those of light-demanding species at the same diameter (Fig. 4). The rank order of species for leaf area comparisons differed from that found for foliage mass, because of wide interspecific variation in LMA (Table 3).

Figure 3.

Estimated foliage mass (mean (1 SE) at 5.0 cm basal diameter, for 11 temperate rain forest trees. Species are grouped according to reported shade tolerance differences: intolerant species to the left, mid-tolerant species at the centre, and tolerant species to the right. Species abbreviations as in Table 1.

Figure 4.

Estimated foliage area (mean (1 SE) at 5.0 cm basal diameter, for 11 temperate rain forest trees. Species groupings as in Fig. 3, and abbreviations as in Table 1.

Table 3.  Foliage and leaf litter parameters of juveniles of 11 temperate evergreen tree species (mean ± SE). Species arranged from top to bottom in approximate order of increasing shade tolerance. n = 12–17 for crown depth, n = 4–7 for leaf life span, and n = 4 for other parameters
Foliage parametersLeaf litter parameters
SpeciesCrown depth (%)Leaf life-span (months)LMA (g m–2)N/mass (%)N/area (g m–2)LMA (g m–2)N/mass (%)N/area (mg cm–2)N resorption (%)
E. coccineum75.9 ± 3.313 ± 1103.2 ± 9.71.74 ± 0.031.80 ± 0.1791.3 ± 8.90.62 ± 0.070.55 ± 0.0669.3 ± 4.1
W. trichosperma68.0 ± 3.624 ± 4149.2 ± 12.70.90 ± 0.071.32 ± 0.05141.9 ± 12.10.37 ± 0.050.50 ± 0.0361.8 ± 2.6
N. betuloides66.3 ± 4.228 ± 2167.6 ± 10.51.19 ± 0.121.96 ± 0.08160.6 ± 7.70.34 ± 0.010.54 ± 0.0372.6 ± 2.6
O. pillo-pillo75.4 ± 2.725 ± 198.8 ± 2.52.14 ± 0.082.12 ± 0.1279.4 ± 3.00.68 ± 0.050.54 ± 0.0574.6 ± 2.6
E. cordifolia77.9 ± 2.943 ± 2122.5 ± 7.60.99 ± 0.051.20 ± 0.05117.9 ± 10.20.46 ± 0.050.55 ± 0.0854.5 ± 6.9
N. nitida82.1 ± 2.738 ± 2150.7 ± 5.11.16 ± 0.091.75 ± 0.14142.7 ± 7.20.35 ± 0.020.50 ± 0.0371.2 ± 1.9
D. winteri80.4 ± 3.042 ± 4178.0 ± 8.10.87 ± 0.041.55 ± 0.08173.9 ± 7.40.34 ± 0.030.58 ± 0.0362.7 ± 3.5
L. philippiana83.9 ± 2.347 ± 4133.9 ± 4.11.56 ± 0.112.09 ± 0.18111.3 ± 1.90.78 ± 0.050.87 ± 0.0558.5 ± 5.1
A. punctatum92.7 ± 2.054 ± 4126.8 ± 6.00.96 ± 0.061.22 ± 0.11122.1 ± 7.30.61 ± 0.030.75 ± 0.0738.9 ± 6.6
A. luma76.9 ± 1.842 ± 2141.0 ± 13.21.01 ± 0.101.42 ± 0.15131.7 ± 14.70.43 ± 0.020.56 ± 0.0360.5 ± 4.4
M. planipes88.0 ± 2.045 ± 7110.5 ± 3.81.14 ± 0.051.26 ± 0.09100.5 ± 0.50.64 ± 0.020.64 ± 0.0249.1 ± 1.7

Crown depth

Mean crown depth did not vary greatly between species, ranging from 66% of total stem height in Nothofagus betuloides to 92% in Aextoxicon punctatum (Table 3). There was a rather weak tendency for crowns to be deeper in shade-tolerant species. A Tukey–Kramer HSD test showed that the mean crown depth for shade-tolerant species (approximately 85%) was significantly higher than that for intolerant species (approximately 71%) (P < 0.05) but that the mean for mid-tolerant species (approximately 80%) was not significantly different from those of either of the other groups.

Leaf lifetimes

Mean leaf lifetimes generally increased with increasing shade tolerance (Table 3). Shade-tolerant species held their leaves for 42–54 months on average, mid-tolerant species for 38–43 months, and light-demanders for only 13–28 months. The mean leaf lifetime for the light-demanding species (23 months) was significantly shorter than the means for both tolerant and mid-tolerant species (47 and 41 months, respectively), whereas the latter two groups did not differ significantly (Tukey–Kramer HSD test).

Leaf mass per unit area

Mean LMA ranged from 99 cm2 g–1 in Ovidia pillo-pillo to 177 cm2 g–1 in Drimys winteri (Table 3). LMA showed no overall trend in relation to shade tolerance, and there were no significant differences in mean LMA between shade tolerance groups (Tukey–Kramer HSD test). However, the two lowest LMA values were recorded for the short-lived early successional species Embothrium coccineum and Ovidia pillo-pillo.

Leaf nitrogen

Mean leaf N concentration on a mass basis ranged from 0.87% in Drimys winteri to 2.14% in Ovidia pillo-pillo (Table 3). Although no overall trend in relation to shade tolerance was apparent, the two highest values were recorded for the short-lived early successional trees Embothrium coccineum and Ovidia pillo-pillo. On an area basis, mean leaf N content ranged from 1.2 g m–2 in Eucryphia cordifolia to 2.1 g m–2 in Ovidia pillo-pillo. There were no significant differences in mean leaf N between shade tolerance groups, on either mass or area bases (Tukey–Kramer HSD test).

Crown nitrogen pools

Total crown nitrogen pool at 5 cm basal diameter was estimated as the product of foliage area at that diameter and mean foliar N content on an area basis. Crown N pools varied nearly fourfold among the 11 species, and tended to be largest in shade-tolerant species (Fig. 5). However, this trend was not as consistent as those seen for foliage mass and leaf areas, because of considerable interspecific variation in leaf N levels (Table 3). Among the three shade tolerance groupings, the only significant difference in means was found between the four light-demanders and the four shade-tolerant species (Tukey–Kramer HSD test).

Figure 5.

Estimated crown N pool (mean (1 SE) at 5.0 cm basal diameter, for 11 temperate rain forest trees. Species groupings as in Fig. 3, and abbreviations as in Table 1.

Resorption and n content of litter fall

Leaf litter N levels on a mass basis ranged from 0.37% in Nothofagus betuloides to 0.78% in Laureliopsis philippiana (Table 3). There was somewhat less variation on an area basis, from 0.50 g m–2 in Nothofagus nitida and Weinmannia trichosperma to 0.87 g m–2 in Laureliopsis philippiana. On a mass basis, there were no significant differences between mean N content of litter of the three shade tolerance groupings, but on an area basis litter of shade-tolerant species had significantly higher average N content (0.70 g m–2) than shade-intolerant species (mean 0.53 g m–2) (Tukey–Kramer HSD test).

N resorption efficiencies calculated on an area basis ranged from 39% in Aextoxicon punctatum to 75% in Ovidia pillo-pillo (Table 2). Resorption efficiencies showed a strong negative relationship with shade tolerance level: the mean percentage resorption of the four shade-tolerant species (52%) was significantly lower than that of the four intolerant species (70%) (Tukey–Kramer HSD test).

Annual n losses in leaf litter fall

Estimated annual N losses through crown turnover at 5 cm basal diameter varied nearly fivefold among the 11 species (Fig. 6). Although there were no statistically significant differences between the means of the three shade tolerance groups (Tukey–Kramer HSD test), relatively high N losses were estimated for most of the shade-tolerant species, with three of the four species in this group among the four highest rankings for N losses. Most of the shade-intolerant species had low N losses, with the marked exception of the short-lived light-demander Embothrium coccineum, which showed the highest N loss rate of any species, reflecting a much faster foliage turnover rate than that of any of its associates.

Figure 6.

Estimated annual N losses in leaf litter fall (mean (1 SE) from saplings of 5.0 cm basal diameter, for 11 temperate rain forest trees. Species groupings as in Fig. 3, and abbreviations as in Table 1.

Discussion

Foliage area and mass in relation to shade tolerance

In agreement with data reported from northern temperate forests (Chapman & Gower 1991; Canham et al. 1994), juveniles of shade-tolerant Chilean rain forest species carried higher foliage mass and area than light-demanding associates (Fig. 1), thus supporting the main assumption underlying our hypothesis about crown nutrient turnover. Foliage area differences in relation to shade tolerance presumably reflect differential responses to self-shading, determined by leaf-level light compensation points (Givnish 1988) and/or differential ability to harvest sunflecks (Chazdon & Pearcy 1991). Net carbon deficit in heavily shaded foliage layers of shade-intolerant trees is likely to result in shut-down of lower and inner crown branches, whereas shaded leaves of tolerant species may make an important contribution to whole-plant carbon gain (Schulze et al. 1977a,b).

Although crown depth tended to be greater in shade-tolerant species (Table 3), and was highly correlated with both foliage mass and area (both R2 = 0.79), the observed interspecific differences were small compared with those found for foliage mass and area. Other factors besides crown depth (differences in branching density, leaf packing and/or leaf life span) must therefore also contribute significantly to interspecific variation in total leaf area and mass. Leaf life span was in fact highly correlated (R2 = 0.73, P = 0.0008) with foliage mass, and to a lesser extent with leaf area (R2 = 0.51, P = 0.014). Large differences in crown depth in relation to shade tolerance (Canham et al. 1994) may be more characteristic of larger individuals with greater self-shading.

Leaf lifetimes and successional status

Leaf lifetimes have been predicted to play an important role in adaptation to sun and shade in evergreen forests (Coley et al. 1985; King 1994). Rapid foliage turnover should be advantageous in well-lit situations suitable for fast growth, whereas long lifetimes should reduce the energetic cost of crown maintenance in shaded understoreys. Several empirical studies in tropical forests have produced results consistent with such a trade-off (Williams et al. 1989; King 1994; Reich et al. 1995a) but comparable data from temperate evergreen forests are very scarce. Our results are consistent with the predictions outlined above, with the leaves of shade-tolerant species living 1.5–4 times as long as those of light-demanding associates (Table 3). However, comparisons of larger groups of species are required to explore more thoroughly the ecology of leaf lifetimes in temperate evergreen forests, in particular to determine if generalized differences exist between small short-lived early successional species and long-lived light-demanding overstorey species.

Leaf n and resorption

Few data are available on nutrition of temperate broad-leaved evergreen forests, but the leaf N levels reported here are of a similar range (0.9–2.1%, mean 1.24%) to those reviewed by Wardle (1991) for rain forest trees in New Zealand. Nitrogen resorption efficiencies found in our study (39–75%, mean 61%) were generally higher than the average reported for evergreens (47%) in a very extensive review by Aerts (1996).

Although foliage N concentration showed no systematic trend in relation to shade tolerance, shade-tolerant species were consistently less effective at resorbing N during leaf senescence, leading to high N losses per unit area of leaf litter (Table 3). Reich et al. (1995a) found similar patterns in a comparative study of nutrition along a successional sequence in a tropical evergreen forest. These authors reported high leaf N concentrations (> 2%) in the initial short-lived colonists of disturbed sites, but no consistent differences between longer-lived light-demanding and shade-tolerant canopy tree species. Reich et al. (1995a) also reported a decline in N resorption efficiencies along the successional sequence, paralleling our results. However, it is unclear whether this trend is related to shade tolerance per se, or to the parallel trend in leaf lifetimes; resorption efficiencies in our study were in fact strongly negatively correlated with leaf lifetimes (R2 = 0.59, P = 0.006). Comparative studies of resorption in deciduous forests, by eliminating the confounding factor of variation in leaf lifetimes, should be useful for testing for possible relationships between shade tolerance level per se and resorption efficiency, but no large data sets of this nature seem to be available. On the other hand, Aerts's (1996) review of 44 data sets indicated significantly lower average N resorption efficiencies in evergreens than in deciduous perennials, confirming the generality of the relationship of resorption efficiency with leaf lifetime differences.

Shade tolerance, leaf area and crown nutrient turnover

Although our comparison of 11 evergreen angiosperms did not show a consistent overall relationship between shade tolerance level and crown nitrogen demand, we found that shade-tolerant trees did have higher annual nitrogen losses in leaf litter fall than associated light-demanding overstorey species of comparable longevities (Fig. 6). In spite of the slow foliage turnover rates of the shade-tolerant species, their large foliage areas (Fig. 4) and low N resorption efficiencies (Table 3) resulted in higher crown N losses at a given diameter than those of associated light-demanding overstorey trees, even at the early developmental stages examined in this study. The importance of foliage area for nitrogen requirements is borne out by the fact that interspecific variation in annual crown N losses was more closely related to foliage area at a given diameter (R2 = 0.52, P = 0.01) than to N content of leaf litter on an area basis (R2 = 0.31, P = 0.08) or leaf lifetimes (R2 = 0.01, P = 0.74). Crown depth differences between species may increase with tree size (Canham et al. 1994), probably leading to more pronounced differences in foliage mass (and hence nitrogen losses) between shade-tolerant and light-demanding trees.

However, results also indicate that, in contrast to long-lived light-demanding canopy species, some short-lived early successional species have very high crown nitrogen loss rates. Embothrium coccineum was found to have the highest nitrogen loss rates among the 11 species studied (Fig. 6), and this pattern may well be common among species of this type. The high growth potential of short-lived early colonists in secondary successions is typically associated with high leaf N levels and rapid foliage turnover (Reich et al. 1995a), a trait combination that inevitably implies a high crown nitrogen demand. However, the short life span and small maximum size of such species may reduce their chances of experiencing severe nutrient deficiency, especially as these species generally complete their life cycles during the early stages of secondary succession when, even on oligotrophic sites, nutrient availability may be relatively high (Vitousek et al. 1989). It is also noteworthy that, although Embothrium coccineum is found over a wide range of soil fertility, its stature is much reduced on poor sites. Specimens up to 50 cm d.b.h. and 15 m tall are occasionally found on deep fertile soils, but on oligotrophic sites it is reduced to a stunted treelet, rarely exceeding 5 m tall, and is usually greatly outnumbered by its less nutrient-demanding ecological analogue Ovidia pillo-pillo.

While high nutrient requirements may not greatly limit the fitness of an early colonist in secondary succession (e.g. Embothrium coccineum) they are likely to be of greater significance for shade-tolerant species that normally become dominant in the later stages of stand development, when soil nutrient availability generally declines as a result of increased uptake and sequestration in biomass and litter (Vitousek et al. 1989). This may be especially true of stands on infertile soils, where low litter decomposability is likely. In this context, our data are consistent with the proposal that the cost of obtaining enough nitrogen for crown maintenance and expansion may be an important constraint on the fitness of shade-tolerant trees on low fertility sites. The ability of shade-tolerant conifers to grow on poorer sites than their angiosperm counterparts (Lusk 1996a,b) presumably reflects lower nutrient acquisition costs as a result of longer leaf life span.

A more consistent relationship between shade tolerance and crown nitrogen losses seems almost inevitable in deciduous forests. In evergreen forests, relatively long leaf lifetimes moderate nutrient loss rates in shade-tolerant species (Table 3). However, in forests dominated by deciduous species there are no major interspecific differences in leaf lifetimes, yet leaf area trends are similar to those in evergreen forests, in that deciduous shade-tolerant trees such as Acer saccharum and Fagus grandifolia carry much larger foliage loads than more light-demanding taxa such as Quercus spp. and Fraxinus spp. (Chapman & Gower 1991; Canham et al. 1994). It is therefore difficult to avoid the prediction that shade-tolerant trees in deciduous forests will be shown to have much higher crown nitrogen loss rates than their light-demanding associates.

Acknowledgements

We thank Pancho Matus, Tim Brodribb and Pepe Loyola for field help, Rubén Roa for statistical advice and for writing the bootstrap resampling program, and Javier Figueroa, Lohengrin Cavieres and two anonymous referees for helpful comments on the manuscript. Research was supported by grants DIUT 454–26 and FONDECYT 1980084.

Received 17 November 1998 revision accepted 13 May 1999

Ancillary