Growth properties of 16 non-pioneer rain forest tree species differing in sapling architecture
Summary
1. Sapling architecture may be an important determinant of performance traits, such as light interception and height growth, but few studies have examined the direct relationship between sapling architecture and growth properties. To study this relationship and the potential for strategic diversification, we analysed the growth properties in saplings of 16 Bornean tree species that differ in architecture.
2. Annual net production significantly differed amongst species and was positively correlated with total above-ground dry mass, total leaf area and crown area. In contrast, the net assimilation rate was weakly but negatively correlated with these architectural traits. The net assimilation rate was virtually independent of leaf size and specific leaf area. Relationships between sapling architecture and relative growth rate in mass were weak.
3. The relative growth rate in height did not significantly differ amongst species, although their total dry mass, a proxy for extension cost, varied fourfold across species for a given sapling height. This is because the proportional increase in net production with total dry mass, which is based on a larger total leaf area and larger crown area, cancelled out the higher extension cost. All architectural traits, including leaf size and specific leaf area, failed to predict height growth rate.
4. Synthesis. Relative growth rates in both mass and height were relatively independent of sapling architecture. Of the architectural traits, leaf size, specific leaf area and stem diameter were poor predictors of growth properties, even though they were considered functionally important. These results clearly reject the classic hypothesis that architectural variation leads to a trade-off between height growth and light interception, at least for the species that are under shaded conditions. However, functional variation ranging from species with high net production and low net assimilation rates (in saplings of equal height) to species with the opposite traits, which was accompanied by architectural variation in total dry mass and related size factors, may be important for the coexistence of these tree species. The possibility that small total dry mass may be advantageous in height growth under well-lit conditions should be examined in future studies.
Introduction
The morphological diversity of tree juveniles has attracted the attention of many ecologists because it may reflect the diversified regeneration niche of coexisting species (Horn 1971; Givnish 1984; Kohyama 1987; Falster & Westoby 2005; Poorter et al. 2006). A well-known hypothesis relates conspicuous variation in crown architecture to interspecific differences in growth strategy (Givnish 1984; Kohyama 1987; King 1990). Less-branched juveniles with a cylindrical crown, often associated with large leaves, a thinner stem, smaller above-ground mass and larger trunk mass fraction (Kohyama 1987; King 1998; Yamada et al. 2000), may have an advantage in terms of height growth because of low extension costs. However, their architecture may be inferior for light capture under shaded conditions because their total leaf area is relatively small, or self-shading may be inevitable due to the small crowns. In contrast, well-branched juveniles with a flat crown, often associated with small leaves, a thicker stem, larger above-ground mass and larger foliage and branch mass fractions, may be superior in light capture and physical stability, whereas they may have a disadvantage in height growth because they require larger biomass to reach a given height owing to their intensive allocation to lateral growth.
Rapid extension is essential for out-competing other saplings under relatively well-lit conditions, whereas efficient light capture and durable structure are required for survival under shaded conditions. Such functional trade-offs between height growth and light interception and/or physical strength would enable niche partitioning along light gradients within a forest (Givnish 1984). Furthermore, Kohyama (1987) argued that architectural differences promote stable coexistence even in shade-tolerant species. Amongst such species, ‘optimistic’ species intensively allocate their limited carbon budgets to trunk extension and have cylindrical crowns. These species maintain a certain level of extension growth even under shaded conditions, with the concomitant risk of relatively low survival, and thereby enjoy an advantage in asymmetric competition after gap formation. In contrast, ‘pessimistic’ species intensively allocate their carbon budgets to trunk thickening and to branches to construct large crowns. They are superior in light capture and survival under shaded conditions, and thus successfully regenerate if shaded conditions persist. In the temporally unpredictable light environment of mature forests, this variation in sapling architecture may provide alternative means of successful regeneration for even shade-tolerant species. These hypotheses make specific predictions concerning the relationship between sapling architecture and growth properties: crown area and total leaf area are positively correlated with mass growth per leaf area (i.e. net assimilation rate, NAR) and/or the relative growth rate of mass (RGRMass), whereas the construction of large, leafy crowns and thick trunks imposes higher extension costs and thus pose a disadvantage in terms of the relative growth rate in height (RGRHeight).
However, in a comparative analysis of the sapling architecture of 56 non-pioneer species in a Bornean rain forest, Aiba & Nakashizuka (2007) found that the biomass fraction amongst plant components (i.e. leaves, branches and trunk) was virtually independent of crown area, which was strongly correlated with total above-ground dry mass, total leaf area and stem thickness. These results indicated that saplings with higher extension costs may have a proportionally larger leaf mass and area. The strong correlation between total leaf area and crown area suggests that the extent of self-shading, and thus NAR, may be similar amongst species. Uniform NAR would lead to a proportional increase in net production per individual in costly structured saplings, and thus the higher extension cost may be cancelled out. Such findings suggest markedly different relationships between sapling architecture and growth properties: NAR is independent of total above-ground mass, total leaf area and crown area; as a result, net production monotonically increases with total above-ground mass; and therefore both RGRMass and RGRHeight are similar amongst species and are independent of above-ground mass and related size factors such as total leaf area, crown area and stem thickness.
Actual measurement of sapling growth properties by continuous monitoring is essential to verify the alternative possibilities. Previous studies on sapling growth have demonstrated that the responses of these traits to light levels differ significantly between pioneer and non-pioneer species (King 1991, 1994; Poorter 2001; Lusk 2002), and these differences can be explained in part by interspecific differences in specific leaf area (SLA). Similarly, for current-year seedlings, many studies have shown that the leaf area ratio and SLA are key determinants of RGRMass (Kitajima 1994; Cornelissen et al. 1996; Reich et al. 1998) and are important as a basis for light-gradient partitioning. However, in saplings whose aerial architecture is complex, growth properties can differ depending on sapling geometry (e.g. the extent of branching, leaf arrangement and stem thickness), even if their leaf physiology is similar. Understanding the role of such traits would be especially important in comparisons of species that share a specific light environment and similar leaf physiology. However, interspecific differences in the growth properties of non-pioneer saplings and relationships between these traits and the widespread variation in aerial architecture are poorly understood (but see Coomes & Grubb (1998) for the relationship between crown architecture and height growth rate).
In this study, we measured for over 2 years the growth properties (i.e. net production, biomass allocation, RGRMass, NAR, RGRHeight and leaf area accumulation) of 16 non-pioneer tree species that are relatively abundant in the shaded understorey of a Bornean tropical rain forest. To test the alternative hypotheses described earlier in this study, we determined the correlations between architecture and growth properties using both interspecific and phylogenetic methods.
Methods and materials
Study site and species
The study was conducted at Lambir Hills National Park, Sarawak, Malaysia (4°2′ N, 113°50′ E; 150 m a.s.l.). Consistently high temperatures with no distinct dry season characterize the climate of this region. The mean annual temperature and annual precipitation are 26 °C and 2700 mm respectively. The 6949-ha park is covered by primary lowland mixed dipterocarp forest on nutrient-poor sandy clay or clay-rich soils. Dipterocarpaceae dominate the canopy and emergent layers, and Euphorbiaceae, Burseraceae and Myristicaceae are abundant in the lower layers. In a 52-ha permanent plot located in the park, more than 1200 species were recorded, suggesting that this forest is one of most diverse in the world.
We focused on 16 tree species that are relatively abundant as saplings in the understorey (Table 1). The species belong to 12 families and largely differ in their maximum heights, but all are shade-tolerant in the sense that juveniles are abundant even under a closed canopy. The majority of these species were also examined in a previous study of the above-ground architecture of 56 tree species (Aiba & Nakashizuka 2007).
| Genus | Species | Family | Authority | n | Maximum height (m) | GSF (%) |
|---|---|---|---|---|---|---|
| Agrostistachys | longifolia | Euphorbiaceae | (Wight) Benth. & Hook.f | 10 | 6 | 7.2 ± 1.6 |
| Allanathospermum | borneense | Simaroubaceae | Forman | 11 | 40 | 9.0 ± 0.9 |
| Ardisia | sp | Myrsinaceae | 10 | 5 | 9.0 ± 1.7 | |
| Artocarpus | anisophyllus | Moraceae | Miq. | 13 | 45 | 9.5 ± 1.5 |
| Dillenia | excelsa | Dilleniaceae | (Jack) Gilg. | 12 | 40 | 8.5 ± 1.2 |
| Dipterocarpus | globosus | Dipterocarpaceae | Vasque | 9 | 65 | 10.0 ± 1.1 |
| Dryobalanops | aromatica | Dipterocarpaceae | Gaertn.f., nom. cons. | 14 | 65 | 8.5 ± 1.5 |
| Goniothalamus | velutinus | Annonaceae | Airy Shaw | 9 | 4 | 9.1 ± 1.0 |
| Heritierra | sumatrana | Malvaceae | (Miq.) Kosterm. | 9 | 50 | 9.8 ± 1.5 |
| Mallotus | eucaustus | Euphorbiaceae | Airy Shaw | 9 | 6 | 7.2 ± 1.3 |
| Pentace | laxiflora | Malvaceae | Merrill | 11 | 30 | 9.6 ± 1.5 |
| Ryparosa | wallichii | Flacourtiaceae | Ridl. | 8 | 3 | 10.1 ± 0.7 |
| Santiria | laevigata | Burseraceae | Blume | 12 | 30 | 8.9 ± 2.0 |
| Swintonia | foxworthyi | Anacardiaceae | Elmer | 15 | 50 | 9.5 ± 1.5 |
| Tapoides | villamirii | Euphorbiaceae | (Merr.) Airy Shaw | 10 | 9 | 9.0 ± 1.4 |
| Teijsmaniodendron | coriaceum | Verbenaceae | (C.B. Clarke) Kosterm. | 10 | 6 | 9.4 ± 0.8 |
Measurements of sapling growth
In March 2005, we selected 8–15 saplings per species (172 individuals in total), depending on availability. All individuals were sampled from a c. 20-ha area on a relatively flat terrace of sandy clay soil. The area appeared to be covered by a relatively homogeneous plant assemblage, characterized by the relative dominance of an emergent dipterocarp species, Dryobalanops aromatica. The mean height of selected saplings was 1.50 ± 0.04 m (mean ± SD). Individuals badly damaged by herbivory or with evidence of past severe stem breakage were rejected. The height to the highest living shoot tip, crown width in two perpendicular directions (including the greatest width) and stem diameter at about 1 cm above the ground in two perpendicular directions were measured. The points at which stem diameter was measured were marked with paint for later remeasurement. The trunk was also marked with waterproof tape about 20 cm below the top, and the distance from the mark to the tip was measured in later censuses for accurate height growth measurement. For branched individuals, branches were numbered, and the length (distance from the base to the tip) and diameter at 2 cm from the base at right angles were measured. Secondary and more distal branches, if present, were also measured. Leaves were counted and marked to distinguish new leaves in later censuses. For sympodial species (Ryparosa and Teijsmaniodendron), the shoot with the tallest tip was treated as the trunk, and other shoots were considered branches. This definition did not affect our conclusions because no changes in leader shoots were observed and allocation to branches was small (see Results). Hemispherical photographs were taken just above each sapling to measure the light environment. Global site factors (GSFs) were calculated from the photographs, using Gap Light Analyzer software (Frazer et al. 1999). The light environment measured again in the final census was similar to that in the first census; thus, averaged values were used for analyses. The GSF had a mean of 9.0% and ranged from 4.1% to 13.5%, indicating that all studied individuals grew under relatively shaded conditions at this site, where the canopy is relatively sparse compared with that of other rain forests (cf. Delissio & Primack 2003). In a later analysis, the mean GSF for each species was used as an index of species light demand. Saplings were revisited in September 2005, March 2006 and September 2006 to record viability, elongation of leader shoots and numbers of new and dead leaves and branches.
In June 2007, the above-ground parts of saplings were harvested after measurements of height, stem diameter, branch length, branch diameter, crown diameter and leaf number. The saplings were dissected into trunk, branches and leaves, and weighed after oven-drying to constant weight. For the trunk and branches, the relationships between the volume (calculated as a cone, using diameter and length) and mass were used to estimate the mass at the start of the census. Several typical, fully expanded leaves (2–17, depending on leaf size) were dried separately after being digitally scanned to obtain individual leaf area and SLA. The scanned images were processed using the software package lia32 (Yamamoto 2003), and SLA was calculated using area and mass. We estimated the total leaf area at the start and end of the census by multiplying individual leaf area by leaf number. Initial leaf mass was estimated in the same way. The crown area was approximated as an ellipse, and stem diameters were geometrically averaged. We verified the reliability of our estimations by comparing the actual dry mass with estimates from the final census for leaf mass, trunk mass and branch mass. The intercepts and slopes of the major axes of each pair of variables did not significantly differ from 0 and 1, respectively, and the r2-values were 0.98, 0.97 and 0.97 for leaf mass, trunk mass and branch mass respectively. These results indicate that the dry mass of the plant components in the initial census was successfully estimated using this method. Based on these values, net production, allocation, RGRMass, NAR, RGRHeight and the relative rate of leaf area accumulation (LAAR) were calculated. Net production was calculated as MLast + MLost − MInitial, where MLast and MInitial are the total dry mass at the end and start of the census, respectively, and MLost is the estimated dry mass of newly produced branches and leaves that were lost before the last census. Leaves and branches that existed at the time of the first census and died during the census period were not included in the calculation, although some previous studies included them (e.g. King 1994). Thus, net production and allocation in this study express biomass (of above-ground parts) acquired by assimilation during the study period and the fraction allocated to each plant component respectively. RGRMass, RGRHeight and LAAR were calculated as (ln MLast − ln MInitial)/t, (ln HLast − ln HInitial)/t and (ln ALast − ln AInitial)/t, respectively, where H and A are the height and total leaf area, respectively, and t is the census period (in days for RGRMass and years for other values). NAR was defined as (MLast − MInitial)/AMean/t, where AMean is the mean total leaf area during the census period. Net production, RGRMass and NAR were underestimated and the allocation to components was overestimated because we did not account for the growth of underground parts in this study. These discrepancies did not substantially affect our results in comparison with previous studies of sapling growth that also ignored underground parts (King 1991, 1994; Poorter 2001). The maximum heights of species were obtained from published records (Whitmore 1972; Soepadmo & Wong 1995). For a few species with no published information, maximum heights were based on our field observations.
Statistical analysis
Interspecific differences in architecture and growth properties were tested using analysis of covariance (ancova), where GSF and initial plant height served as co-variables. Because the effects of GSF and initial height were significant for many traits, adjusted mean values rather than arithmetic mean values were used in later analyses. Adjusted mean values were calculated as estimates for 150.7-cm tall individuals growing under 9.0% GSF (values are grand mean values of analysed saplings). Using both interspecific- and phylogenetic-based methods (Felsenstein 1985), Pearson correlation coefficients were calculated amongst sapling traits, including adult stature and mean GSF for each species, which can be an index of the light demand of species. The phylogenetic analysis was performed by calculating the phylogenetic independent contrasts amongst traits using the ape package (Bolker et al. 2007) available in the statistical environment r (R Development Core Team 2007). A phylogenetic tree of the 16 species was constructed using the on-line application Phylomatic (Webb & Donoghue 2005). As the 16 species are largely different in their adult stature (Table 1), separate analyses were also performed for understorey species (maximum height < 10 m) and taller species (maximum height > 10 m). The results of these separate analyses generally agreed with those for all species, and thus are given as Supporting Information. All statistical analyses were performed in the statistical environment r (R Development Core Team 2007).
Results
Architecture of saplings
Total dry mass, total leaf area, crown area, individual leaf area, stem diameter and SLA differed significantly amongst the 16 study species (ancova, P < 0.0001; Table 2). The adjusted mean values of total dry mass, total leaf area, crown area, individual leaf area, stem diameter and SLA exhibited 4-, 8-, 6-, 26- and 2-fold interspecific variation amongst the 16 species respectively. Total dry mass was significantly positively correlated with total leaf area (P < 0.0001 in both interspecific and phylogenetic analyses; Table 4), crown area (P < 0.05 in both analyses) and stem diameter (P < 0.0001 in both analyses), but not with leaf mass fraction (r = 0.34, P = 0.19; data not shown) or leaf area ratio (r = −0.18, P = 0.50; data not shown), similar to correlations found in a previous study (Aiba & Nakashizuka 2007). Further details of the architectural differences amongst species are described in Aiba & Nakashizuka (2007).
| Total dry mass (g) | Total leaf area (m2) | Crown area (m2) | Stem diameter (mm) | Leaf size (cm2) | SLA (m2 g−1) | |
|---|---|---|---|---|---|---|
| Agrostistachys | 105.4 ± 17.1 | 0.61 ± 0.14 | 0.27 ± 0.03 | 9.3 ± 0.8 | 129 ± 16 | 0.011 ± 0.001 |
| Allanathospermum | 62.6 ± 5.0 | 0.28 ± 0.04 | 0.26 ± 0.03 | 8.9 ± 0.5 | 67 ± 4 | 0.013 ± 0.001 |
| Ardisia | 83.7 ± 6.3 | 0.33 ± 0.04 | 0.29 ± 0.03 | 12.6 ± 0.4 | 135 ± 7 | 0.012 ± 0.000 |
| Artocarpus | 41.1 ± 3.7 | 0.16 ± 0.03 | 0.29 ± 0.03 | 9.6 ± 0.4 | 269 ± 41 | 0.017 ± 0.001 |
| Dillenia | 102.7 ± 7.8 | 0.43 ± 0.06 | 0.42 ± 0.04 | 12.3 ± 0.5 | 228 ± 18 | 0.014 ± 0.001 |
| Dipterocarpus | 46.5 ± 5.9 | 0.15 ± 0.02 | 0.17 ± 0.03 | 9.1 ± 0.7 | 84 ± 7 | 0.012 ± 0.000 |
| Dryobalanops | 35.6 ± 1.7 | 0.21 ± 0.03 | 0.37 ± 0.02 | 7.1 ± 0.2 | 24 ± 2 | 0.019 ± 0.002 |
| Goniothalamus | 67.2 ± 5.6 | 0.25 ± 0.04 | 0.33 ± 0.02 | 11.7 ± 0.5 | 152 ± 20 | 0.013 ± 0.001 |
| Heritierra | 108.5 ± 14.5 | 0.52 ± 0.15 | 0.81 ± 0.14 | 14.0 ± 0.7 | 559 ± 84 | 0.014 ± 0.001 |
| Mallotus | 60.6 ± 8.4 | 0.37 ± 0.06 | 0.34 ± 0.05 | 9.0 ± 0.8 | 64 ± 9 | 0.014 ± 0.000 |
| Pentace | 44.5 ± 3.2 | 0.21 ± 0.03 | 0.45 ± 0.04 | 8.1 ± 0.4 | 37 ± 2 | 0.015 ± 0.000 |
| Ryparosa | 124.9 ± 19.1 | 0.50 ± 0.10 | 0.59 ± 0.16 | 15.0 ± 1.4 | 116 ± 23 | 0.012 ± 0.001 |
| Santiria | 61.1 ± 4.3 | 0.35 ± 0.04 | 0.72 ± 0.07 | 10.8 ± 0.3 | 635 ± 58 | 0.022 ± 0.003 |
| Swintonia | 30.4 ± 1.3 | 0.10 ± 0.01 | 0.15 ± 0.01 | 7.7 ± 0.2 | 48 ± 3 | 0.013 ± 0.000 |
| Tapoides | 138.0 ± 9.1 | 0.84 ± 0.08 | 0.51 ± 0.04 | 15.1 ± 0.6 | 235 ± 7 | 0.014 ± 0.000 |
| Teijsmaniodendron | 59.5 ± 4.4 | 0.31 ± 0.04 | 0.22 ± 0.05 | 9.3 ± 0.4 | 144 ± 22 | 0.012 ± 0.001 |
| Species | ** | ** | ** | ** | ** | ** |
| GSF | *(+) | *(+) | NS | NS | NS | NS |
| Initial height | **(+) | NS | NS | **(+) | *(+) | NS |
| Total dry mass | Total leaf area | Crown area | Leaf size | SLA | Stem diameter | Net production | Leaf allocation | RGRMass | NAR | RGRHeight | LAAR | GSF | Maximum height | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total dry mass | 0.92** | 0.51* | 0.29 | −0.32 | 0.87** | 0.71* | 0.63* | −0.42 | −0.60* | −0.03 | −0.03 | −0.07 | −0.50* | |
| Total leaf area | 0.91** | 0.50* | 0.31 | −0.16 | 0.71* | 0.84** | 0.63* | −0.18 | −0.56* | 0.14 | 0.08 | −0.28 | −0.50* | |
| Crown area | 0.67* | 0.69* | 0.76* | 0.45 | 0.61* | 0.54* | 0.53* | −0.04 | −0.45 | −0.02 | 0.09 | 0.18 | −0.04 | |
| Leaf size | 0.48 | 0.55* | 0.82** | 0.48 | 0.46 | 0.47 | 0.18 | 0.00 | −0.18 | 0.06 | −0.21 | 0.15 | 0.07 | |
| SLA | −0.10 | 0.12 | 0.55* | 0.52 | −0.18 | 0.00 | 0.09 | 0.35 | 0.04 | 0.13 | 0.20 | −0.04 | 0.39 | |
| Stem diameter | 0.89** | 0.76** | 0.73* | 0.60* | 0.03 | 0.47 | 0.44 | −0.37 | −0.37 | −0.15 | −0.04 | 0.27 | −0.42 | |
| Net production | 0.63* | 0.78** | 0.64* | 0.62* | 0.27 | 0.46 | 0.52* | 0.13 | −0.30 | 0.22 | 0.05 | −0.39 | −0.46 | |
| Leaf allocation | 0.46 | 0.42 | 0.62* | 0.32 | 0.33 | 0.34 | 0.38 | −0.25 | −0.70* | −0.01 | 0.39 | −0.24 | −0.32 | |
| RGRMass | −0.55* | −0.26 | −0.05 | −0.03 | 0.49 | −0.46 | 0.11 | −0.19 | 0.68* | 0.42 | 0.48 | −0.26 | −0.29 | |
| NAR | −0.64* | −0.56* | −0.47 | −0.31 | 0.04 | −0.45 | −0.20 | −0.61* | 0.70* | 0.11 | 0.06 | −0.06 | −0.15 | |
| RGRHeight | −0.28 | −0.08 | 0.01 | −0.02 | 0.34 | −0.30 | 0.01 | 0.01 | 0.49 | 0.10 | 0.22 | −0.64* | −0.32 | |
| LAAR | −0.41 | −0.27 | −0.06 | −0.25 | 0.42 | −0.36 | −0.18 | 0.28 | 0.60* | 0.25 | 0.42 | −0.10 | −0.38 | |
| GSF | 0.36 | 0.17 | 0.13 | 0.14 | −0.30 | 0.56* | −0.13 | −0.06 | −0.44 | −0.13 | −0.66* | −0.18 | 0.30 | |
| Maximum height | −0.13 | −0.16 | −0.17 | −0.06 | −0.07 | −0.12 | −0.29 | −0.19 | −0.43 | −0.29 | −0.15 | −0.35 | −0.01 |
Performance of saplings
Of the 172 saplings studied, one individual of Ardisia died from unknown causes during the census period. Six additional saplings were badly damaged by falling debris and were thus excluded from analyses.
Annual net production significantly differed amongst the 16 species (P < 0.0001; Table 3) and increased significantly with both GSF and initial height within species. The adjusted mean calculated for the average-sized individual (150.7-cm high in the initial census) under the average light environment (9.0% GSF) exhibited fourfold variation, from 6.6 g year−1 for Dipterocarpus to 40.5 g year−1 for Agrostistachys. The allocation of net production to leaf production also significantly differed amongst species (P < 0.0001) and significantly increased with GSF within species. The adjusted mean varied from 27% for Goniothalamus to 81% for Ryparosa. Allocation to branches was relatively small (mean 2.3%, maximum 7.3%; Dryobalanops); thus, most of the above-ground net production was partitioned to leaves and trunk in these species.
| Net production (g year−1) | Leaf allocation (%) | RGRMass (mg g−1 day−1) | NAR (g m−2 day−1) | RGRHeight (m m−1 year−1) | LAAR (m2 m−2year−1) | |
|---|---|---|---|---|---|---|
| Agrostistachys | 40.5 ± 9.4 | 66 ± 8 | 0.15 ± 0.04 | 0.028 ± 0.011 | 0.023 ± 0.004 | 0.02 ± 0.11 |
| Allanathospermum | 11.9 ± 2.3 | 36 ± 7 | 0.12 ± 0.03 | 0.030 ± 0.007 | 0.035 ± 0.014 | −0.02 ± 0.04 |
| Ardisia | 22.4 ± 3.2 | 43 ± 7 | 0.19 ± 0.04 | 0.058 ± 0.015 | 0.025 ± 0.007 | 0.03 ± 0.07 |
| Artocarpus | 9.2 ± 1.9 | 43 ± 4 | 0.16 ± 0.03 | 0.043 ± 0.005 | 0.016 ± 0.005 | 0.14 ± 0.09 |
| Dillenia | 17.9 ± 3.4 | 66 ± 6 | 0.08 ± 0.03 | 0.018 ± 0.008 | 0.029 ± 0.004 | 0.05 ± 0.09 |
| Dipterocarpus | 7.5 ± 2.2 | 42 ± 8 | 0.10 ± 0.03 | 0.033 ± 0.011 | 0.008 ± 0.003 | −0.04 ± 0.03 |
| Dryobalanops | 11.9 ± 1.4 | 47 ± 5 | 0.19 ± 0.04 | 0.037 ± 0.009 | 0.015 ± 0.006 | 0.05 ± 0.05 |
| Goniothalamus | 15.0 ± 1.8 | 27 ± 6 | 0.18 ± 0.03 | 0.055 ± 0.010 | 0.023 ± 0.005 | 0.02 ± 0.06 |
| Heritierra | 30.8 ± 9.3 | 51 ± 7 | 0.13 ± 0.05 | 0.022 ± 0.008 | 0.017 ± 0.005 | −0.03 ± 0.10 |
| Mallotus | 17.7 ± 3.5 | 62 ± 5 | 0.21 ± 0.08 | 0.035 ± 0.012 | 0.043 ± 0.015 | 0.18 ± 0.11 |
| Pentace | 19.9 ± 3.5 | 55 ± 3 | 0.22 ± 0.06 | 0.043 ± 0.015 | 0.025 ± 0.011 | 0.25 ± 0.12 |
| Ryparosa | 16.8 ± 5.4 | 81 ± 11 | 0.08 ± 0.04 | 0.019 ± 0.007 | 0.006 ± 0.016 | 0.11 ± 0.07 |
| Santiria | 28.3 ± 5.0 | 59 ± 3 | 0.21 ± 0.04 | 0.041 ± 0.008 | 0.032 ± 0.008 | 0.04 ± 0.04 |
| Swintonia | 6.6 ± 1.0 | 28 ± 5 | 0.15 ± 0.03 | 0.051 ± 0.012 | 0.021 ± 0.004 | −0.01 ± 0.03 |
| Tapoides | 35.2 ± 4.3 | 58 ± 2 | 0.16 ± 0.02 | 0.028 ± 0.004 | 0.027 ± 0.004 | 0.12 ± 0.04 |
| Teijsmaniodendron | 15.8 ± 1.9 | 42 ± 5 | 0.21 ± 0.04 | 0.046 ± 0.007 | 0.023 ± 0.009 | 0.10 ± 0.06 |
| Species | ** | ** | NA | * | NS | * |
| GSF | **(+) | *(+) | * | **(+) | *(+) | NS |
| Initial height | *(+) | NS | NS | *(−) | NS | NS |
The responses of RGRMass to GSF were significantly different amongst species (P < 0.05); therefore, interspecific differences in elevation of the regression lines were not testable. The adjusted mean ranged from 0.08 mg g−1 day−1 for Dillenia and Ryparosa to 0.22 mg g−1 day−1 for Pentace. NAR differed significantly amongst species (P < 0.05) and was positively correlated with GSF (P < 0.0001) and negatively correlated with initial height (P < 0.05). The adjusted mean varied from 0.018 g m−2 day−1 for Ryparosa to 0.058 g m−2 day−1 for Ardisia. RGRHeight increased with GSF (P < 0.05) but did not significantly differ amongst species. The adjusted mean ranged from 0.006 m m−1 year−1 for Ryparosa to 0.043 m m−1 year−1 for Mallotus. LAAR was significantly different amongst species (P < 0.05), and the adjusted mean varied from −0.04 m2 m−2 year−1 for Dipterocarpus to 0.25 m2 m−2 year−1 for Pentace. LAAR was slightly negative for four species, indicating that total leaf area in these species decreased during the census period under the average light environment, although not significantly so.
Relationship between architecture and growth properties
Net production was positively correlated with total dry mass, total leaf area and crown area in both the interspecific and phylogenetic analyses (Table 4). Allocation to leaves was also positively correlated with these three architectural traits in the interspecific analysis (P < 0.05; the correlation with crown area was also significant in the phylogenetic analysis, P < 0.05). NAR was negatively correlated with total dry mass and total leaf area (P < 0.05 in both analyses). NAR was also weakly but negatively correlated with crown area (P < 0.1 in the interspecific analysis). RGRMass was weakly negatively correlated with total dry mass in the phylogenetic analysis (P < 0.05). Correlations between RGRHeight or LAAR and architectural traits were weak. Of the architectural traits, leaf size, SLA and stem diameter were poorly correlated with sapling growth.
Relationships amongst growth properties
Net production was weakly positively correlated with allocation to leaves in the interspecific analysis (P < 0.05; Table 4). In both analyses, allocation to leaves was weakly negatively correlated with NAR (P < 0.05; Table 4). RGRHeight was virtually independent of net production, allocation to leaves, NAR and LAAR. RGRMass was weakly positively correlated with NAR and LAAR (P < 0.05).
The mean GSF for each species, which may reflect the light demand of a species, was relatively independent of growth properties and was weakly negatively correlated with only RGRHeight (P < 0.05). Adult stature of the study species, which was weakly negatively correlated with total dry mass, total leaf area and stem diameter, was weakly negatively correlated with net production (P < 0.1).
All of these relationships, including those between architecture and growth, were mostly similar even when taller species were separately analysed from understorey species (see Tables S1 and S2 in Supporting Information). One notable difference is that, in separate analyses, positive correlations between leaf size and net production were rather strong in both taller species and understorey species, especially in the phylogenetic analyses.
Discussion
The ranges of the measured growth properties in this study were similar to those reported for non-pioneer species under shaded conditions in other tropical rain forests. RGRMass and NAR were quite similar to values for non-pioneer species in a Bolivian forest (Poorter 2001). Allocation to leaves in this study was comparable with that for Costa Rican (King 1991) and Panamanian (King 1994) species. RGRHeight was similar to values for Amazonian species (Coomes & Grubb 1998). The positive responses of net production, RGRMass, NAR and RGRHeight to GSF within species indicated that even small differences in light conditions under a closed canopy have significant effects on the growth of these tree species. In this study, allocation to leaves also increased with GSF, although previous studies have shown that saplings allocate more to leaves at more shaded sites, possibly to maintain their carbon balance (King 1991, 1994; Lusk 2002). The shorter leaf longevity in less shaded locations in this study (data not shown) appears to be a proximate cause of the higher allocation to leaves, although the ultimate cause is not clear. The limited range of light level in this study (4.1–13.5%) compared with that in other studies may be responsible for this difference.
Net assimilation rate was weakly negatively correlated with total dry mass, total leaf area and even crown area (Table 4). These results indicate that carbon gain per leaf area was maximized in species with small total leaf area, whereas carbon gain (including lost biomass) per individual was maximized in species with large total leaf area. Although crown area was correlated with total leaf area, crown area varied less than leaf area; therefore, the leaf area index, which is the ratio of total leaf area to crown area, increased with total leaf area (r = 0.56, P < 0.05; data not shown). These findings suggest that leaves were more crowded on leafy species; thus, the efficiency of light interception decreased with total leaf area, crown area and total dry mass. NAR was also weakly negatively correlated with allocation to leaves. Whilst it appears reasonable that high NAR enables saplings to reduce allocation to leaves and/or that high allocation to leaves results in more self-shaded crowns, higher allocation to leaves does not result in higher LAAR or leaf mass fraction in these species. This result implies that some species do not allocate more to leaf production to construct more leafy crowns but to compensate for low net production, short leaf longevity, low SLA or, most likely, a combination of these factors. Because the negative correlation between NAR and leaf allocation cannot, therefore, be ascribed to differences in the quantity of leaves, the ecological meaning of the correlation remains ambiguous.
Net assimilation rate was virtually independent of leaf size, which is an important factor for leaf display within the crown. Using a three-dimensional simulation model, Pearcy et al. (2004) demonstrated that the light capture efficiencies of seven shade-tolerant Psychotria species coexisting in a Panamanian forest were rather similar, whereas their leaf sizes and crown architecture greatly differed. Different crown structures may potentially realize similar light capture efficiencies under shaded conditions.
Such relationships between NAR and sapling architecture are not completely consistent with our prediction that NAR would be independent of total leaf area, crown area and total dry mass. However, interspecific differences in NAR (threefold variation) were relatively small compared with those in total leaf area (eightfold variation). As a result, net production virtually proportionally increased with total dry mass and total leaf area amongst species, as expected. Due to this relationship, RGRMass was less variable amongst species and was virtually independent of sapling architecture. Our results demonstrated, possibly for the first time, that large crowns do not lead to large NAR or RGRMass, whereas crown size is positively correlated with net production.
As a result of the strong correlation between total dry mass and net production, RGRHeight did not significantly differ amongst species (Table 3), as proposed in Aiba & Nakashizuka (2007). Although total dry mass, a proxy for the extension cost, varied fourfold amongst species (Table 2), species with costly structures were proportionately more productive and thereby cancelled out the larger extension cost. These results clearly reject the classic hypothesis that saplings face a strategic trade-off between height growth and lateral expansion, and further demonstrate that crown size is virtually neutral with respect to height growth rate, at least for these study species under shaded conditions. In 12 Amazonian tree species, Coomes & Grubb (1998) found uniformly slow growth, independent of crown architecture, which is similar to our results. RGRHeight was virtually independent also of individual leaf size and SLA, which represents other aspects of architectural variation. Large, often compound leaves enable saplings to construct large crowns with little to no allocation to branches and have thus been considered an adaptation for rapid height growth (Givnish 1978). However, the independence of RGRHeight from leaf size is reasonable because construction costs are not necessarily low for large or compound leaves compared with that of small leaves on branches, at least under shaded conditions (Aiba & Nakashizuka 2007). In contrast to our results, in a comparative analysis of 53 Bolivian tropical tree species, Poorter & Bongers (2006) found that SLA can be a good predictor of sapling height growth rate. This discrepancy might have been due in part to the fact that Poorter & Bongers (2006) included typical pioneer species in their comparisons and measured their growth under relatively well-lit conditions. Because height growth rates increase with light availability, it is uncertain whether the positive correlation between SLA and height growth is also valid under uniform light conditions.
Nonetheless, architectural differences may still lead to differences in RGRHeight amongst the study species. Although the responses of RGRHeight to light gradients were similar amongst species, such responses may have differed if saplings growing in gap sites had been included. Species that behave similarly under shaded conditions often exhibit different responses to increasing radiation (Coomes & Grubb 1998; Poorter 2001; Lusk 2002). Sapling architecture may be responsible for the extent of such responses to light. Coomes & Grubb (1998) demonstrated that RGRHeight is correlated with crown architecture only under well-lit conditions. Thus, an examination of the responses of growth properties along broad light gradients in forests would be quite valuable, although finding enough individuals of comparable size in gaps in the hyper-diverse Southeast Asian rain forests may be difficult.
Overall, the observed relationships between architecture and growth properties were relatively similar to those predicted by the hypotheses of Aiba & Nakashizuka (2007), in contrast to the predictions of the classic hypotheses of Givnish (1984) and Kohyama (1987). For example, we did not observe interspecific differences in RGRHeight, and both RGRMass and RGRHeight were relatively independent of sapling architecture. Of the architectural traits that are most often considered functionally important, leaf size, SLA and stem diameter were poor predictors of growth properties. Furthermore, we observed no trade-offs between RGRHeight and other growth properties. The fact that our site is the least demographically diverse of the major long-term tropical research sites (Condit et al. 2006) may have been partially responsible for this result. Future studies at demographically more diverse sites, such as Barro Colorado Island, Panama, may reveal different relationships. We, however, observed a new potentially important functional trade-off characterized by wide variation in species from those with large net production and low NAR to those with the opposite traits. Such variation was accompanied by architectural variation in total dry mass, total leaf area, crown area and stem diameter. What is the ecological consequence of such functional and architectural variation? Net production is inextricably linked to the absolute size of plant components. That is, higher net production results in larger components, which in turn enable higher net production. Species with higher net production were often characterized by larger crowns and thicker stems (Table 4). Larger components offer a variety of benefits to juvenile trees. For example, large crowns may provide not only larger carbon budgets but also an ability to shade neighbouring competitors (Sterck et al. 2001). Thicker stems are important in avoiding stem breakage caused by falling debris (King 1990; Poorter & Werger 1999). Larger root systems may be advantageous in below-ground competition. In addition, juveniles of short species, which spend their entire life span under shaded conditions, tend to have larger total dry mass, larger crowns and higher net production compared with similar-height juveniles of taller species (Table 4). Several previous studies have also demonstrated that shorter species have larger crowns and/or thicker stems (e.g. Poorter et al. 2003; Aiba & Nakashizuka 2007), suggesting the importance of high net production and/or large components for long-term survival under shaded conditions.
The merits of high NAR, small total dry mass, small crowns or thinner stems were not clear from the results of this study. The fact that these traits characterize taller species suggests the possibility that these properties, especially low extension costs, are advantageous for height growth under better light conditions, as discussed earlier in this study. Horn (1971) argued that cylindrical crowns are relatively advantageous under well-lit conditions. Thus, small-crowned species may achieve relatively higher net production whilst keeping their extension costs low under better light conditions. In such cases, RGRHeight after gap formation would be maximized in small-crowned species, and architectural and functional variation may contribute to the stable coexistence of species under spatio-temporally heterogeneous light environments, as hypothesized by Kohyama (1987).
Acknowledgements
Three anonymous referees provided helpful comments on a previous version of this manuscript. This study was supported in part by a Research Fellowship for Young Scientists from the Japan Society for the Promotion of Science and a Research Institute for Humanity and Nature project (P2-2). This study was complied with the current laws of Malaysia.
